• Author / Uploaded
• Emilio Vicente

Citation preview

 i

Schottenfeld and Fraumeni Cancer Epidemiology and Prevention

ii

 iii

Schottenfeld and Fraumeni Cancer Epidemiology and Prevention Fourth Edition Lead Editor

MICHAEL J. THUN, MD, MS Epidemiology and Surveillance Research (Retired) American Cancer Society Atlanta, Georgia

Co-​Editors

MARTHA S. LINET, MD, MPH Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

JAMES R. CERHAN, MD, PHD Department of Health Sciences Research Mayo Clinic Rochester, Minnesota

1

CHRISTOPHER HAIMAN, SCD Department of Preventive Medicine Keck School of Medicine, University of Southern California Los Angeles, California

DAVID SCHOTTENFELD, MD, MSC Department of Epidemiology (Retired) University of Michigan School of Public Health Ann Arbor, Michigan

Project Manager

ANNELIE M. LANDGREN, MPH, PMP

iv

1 Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and certain other countries. Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America. © Oxford University Press 2018 Third Edition published 2006 Second edition published 1996 First edition published 1982 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by license, or under terms agreed with the appropriate reproduction rights organization. Inquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above. You must not circulate this work in any other form and you must impose this same condition on any acquirer. Library of Congress Cataloging-in-Publication Data Names: Thun, Michael J., editor. | Linet, Martha S., editor. | Cerhan, James R., editor. | Haiman, Christopher, editor. | Schottenfeld, David, editor. Title: Schottenfeld and Fraumeni Cancer Epidemiology and Prevention / lead editor, Michael J. Thun ; co-editors, Martha S. Linet, James R. Cerhan, Christopher Haiman, David Schottenfeld ; project manager, Annelie M. Landgren. Other titles: Cancer epidemiology and prevention Description: Fourth edition. | New York, NY : Oxford University Press, [2018] | Preceded by Cancer epidemiology and prevention / edited by David Schottenfeld, Joseph F. Fraumeni Jr. 3rd ed. 2006. | Includes bibliographical references and index. Identifiers: LCCN 2017038170 | ISBN 9780190238667 (hardcover : alk. paper) Subjects: | MESH: Neoplasms—epidemiology | Neoplasms—prevention & control Classification: LCC RA645.C3 | NLM QZ 220.1 | DDC 614.5/999—dc23 LC record available at https://lccn.loc.gov/2017038170 9 8 7 6 5 4 3 2 1 Printed by Sheridan Books, Inc., United States of America

 v

Contents

Acknowledgments  Contributors  Preface 

ix xi xix

1. Introduction  Michael J. Thun, Martha S. Linet, James R. Cerhan, Christopher A. Haiman, and David Schottenfeld

1

I  BASIC CONCEPTS 2. Biology of Neoplasia  Michael Dean and Karobi Moitra

9

3. Morphological and Molecular Classification of Human Cancer  Mark E. Sherman, Melissa A. Troester, Katherine A. Hoadley, and William F. Anderson

19

4. Genomic Landscape of Cancer: Insights for Epidemiologists  Christopher J. Maher and Elaine R. Mardis

43

5. Genetic Epidemiology of Cancer  Kathryn L. Penney, Kyriaki Michailidou, Deanna Alexis Carere, Chenan Zhang, Brandon Pierce, Sara Lindström, and Peter Kraft

53

6. Application of Biomarkers in Cancer Epidemiology  Roel Vermeulen, Douglas A. Bell, Dean P. Jones, Montserrat Garcia-​Closas, Avrum Spira, Teresa W. Wang, Martyn T. Smith, Qing Lan, and Nathaniel Rothman

77

7. Causal Inference in Cancer Epidemiology  Steven N. Goodman and Jonathan M. Samet

97

II  THE MAGNITUDE OF CANCER 8. Patterns of Cancer Incidence, Mortality, and Survival  Ahmedin Jemal, D. Maxwell Parkin, and Freddie Bray

107

9. Socioeconomic Disparities in Cancer Incidence and Mortality  Candyce Kroenke and Ichiro Kawachi

141

10. The Economic Burden of Cancer in the United States  K. Robin Yabroff, Gery P. Guy Jr., Matthew P. Banegas, and Donatus U. Ekwueme

169

vi

vi

Contents

III  THE CAUSES OF CANCER 11. Tobacco  Michael J. Thun and Neal D. Freedman

185

12. Alcohol and Cancer Risk  Susan M. Gapstur and Philip John Brooks

213

13. Ionizing Radiation  Amy Berrington de González, André Bouville, Preetha Rajaraman, and Mary Schubauer-​Berigan

227

14. Ultraviolet Radiation  Adèle C. Green and David C. Whiteman

249

15. Electromagnetic Fields  Maria Feychting and Joachim Schüz

259

16. Occupational Cancer  Kyle Steenland, Shelia Hoar Zahm, and A. Blair

275

17. Air Pollution  Jonathan M. Samet and Aaron J. Cohen

291

18. Water Contaminants  Kenneth P. Cantor, Craig M. Steinmaus, Mary H. Ward, and Laura E. Beane Freeman

305

19. Diet and Nutrition  Marjorie L. McCullough and Walter C. Willett

329

20. Obesity and Body Composition  NaNa Keum, Mingyang Song, Edward L. Giovannucci, and A. Heather Eliassen

351

21. Physical Activity, Sedentary Behaviors, and Risk of Cancer  Steven C. Moore, Charles E. Matthews, Sarah Keadle, Alpa V. Patel, and I-​Min Lee

377

22. Hormones and Cancer  Robert N. Hoover, Amanda Black, and Rebecca Troisi

395

23. Pharmaceutical Drugs Other Than Hormones  Marie C. Bradley, Michael A. O’Rorke, Janine A. Cooper, Søren Friis, and Laurel A. Habel

411

24. Infectious Agents  Silvia Franceschi, Hashem B. El-​Serag, David Forman, Robert Newton, and Martyn Plummer

433

25. Immunologic Factors  Eric A. Engels and Allan Hildesheim

461

IV  CANCERS BY TISSUE OF ORIGIN 26. Nasopharyngeal Cancer  Ellen T. Chang and Allan Hildesheim

489

27. Cancer of the Larynx  Andrew F. Olshan and Mia Hashibe

505

28. Lung Cancer  Michael J. Thun, S. Jane Henley, and William D. Travis

519

29. Oral Cavity, Oropharynx, Lip, and Salivary Glands  Mia Hashibe, Erich M. Sturgis, Jacques Ferlay, and Deborah M. Winn

543

30. Esophageal Cancer  William J. Blot and Robert E. Tarone

579

 vi

vii

Contents

31. Stomach Cancer  Catherine de Martel and Julie Parsonnet

593

32. Cancer of the Pancreas  Samuel O. Antwi, Rick J. Jansen, and Gloria M. Petersen

611

33. Liver Cancer  W. Thomas London, Jessica L. Petrick, and Katherine A. McGlynn

635

34. Biliary Tract Cancer  Jill Koshiol, Catterina Ferreccio, Susan S. Devesa, Juan Carlos Roa, and Joseph F. Fraumeni, Jr.

661

35. Small Intestine Cancer  Jennifer L. Beebe-​Dimmer, Fawn D. Vigneau, and David Schottenfeld

671

36. Cancers of the Colon and Rectum  Kana Wu, NaNa Keum, Reiko Nishihara, and Edward L.Giovannucci

681

37. Anal Cancer  Andrew E. Grulich, Fengyi Jin, and I. Mary Poynten

707

38. Leukemias  Martha S. Linet, Lindsay M. Morton, Susan S. Devesa, and Graça M. Dores

715

39. Hodgkin Lymphoma  Henrik Hjalgrim, Ellen T. Chang, and Sally L. Glaser

745

40. The Non-​Hodgkin Lymphomas  James R. Cerhan, Claire M. Vajdic, and John J. Spinelli

767

41. Multiple Myeloma  Mark P. Purdue, Jonathan N. Hofmann, Elizabeth E. Brown, and Celine M. Vachon

797

42. Bone Cancers  Lisa Mirabello, Rochelle E. Curtis, and Sharon A. Savage

815

43. Soft Tissue Sarcoma  Marianne Berwick and Charles Wiggins

829

44. Thyroid Cancer  Cari M. Kitahara, Arthur B. Schneider, and Alina V. Brenner

839

45. Breast Cancer  Louise A. Brinton, Mia M. Gaudet, and Gretchen L. Gierach

861

46. Ovarian Cancer  Shelley S. Tworoger, Amy L. Shafrir, and Susan E. Hankinson

889

47. Endometrial Cancer  Linda S. Cook, Angela L. W. Meisner, and Noel S. Weiss

909

48. Cervical Cancer  Rolando Herrero and Raul Murillo

925

49. Vulvar and Vaginal Cancers  Margaret M. Madeleine and Lisa G. Johnson

947

50. Choriocarcinoma  Julie R. Palmer

953

51. Renal Cancer  Wong-​Ho Chow, Ghislaine Scelo, and Robert E. Tarone

961

52. Bladder Cancer  Debra T. Silverman, Stella Koutros, Jonine D. Figueroa, Ludmila Prokunina-​Olsson, and Nathaniel Rothman

977

53. Prostate Cancer  Catherine M. Tangen, Marian L. Neuhouser, and Janet L. Stanford

997

vi

viii

Contents

54. Testicular Cancer  Katherine A. McGlynn, Ewa Rajpert-​De Meyts, and Andreas Stang

1019

55. Penile Cancer  Morten Frisch

1029

56. Nervous System  E. Susan Amirian, Quinn T. Ostrom, Yanhong Liu, Jill Barnholtz-​Sloan, and Melissa L. Bondy

1039

57. Melanoma  Bruce K. Armstrong, Claire M. Vajdic, and Anne E. Cust

1061

58. Keratinocyte Cancers  Anala Gossai, Dorothea T. Barton, Judy R. Rees, Heather H. Nelson, and Margaret R. Karagas

1089

59. Childhood Cancers  Eve Roman, Tracy Lightfoot, Susan Picton, and Sally Kinsey

1119

60. Multiple Primary Cancers  Lindsay M. Morton, Sharon A. Savage, and Smita Bhatia

1155

V  CANCER PREVENTION AND CONTROL 61. Framework for Understanding Cancer Prevention  Michael J. Thun, Christopher P. Wild, and Graham Colditz

1193

62. Primary Prevention of Cancer 

1205

62.1. Tobacco Control  Jeffrey Drope, Clifford E. Douglas, and Brian D. Carter

1207

62.2. Prevention of Obesity and Physical Inactivity  Ambika Satija and Frank B. Hu

1211

62.3. Prevention of Infection-​Related Cancers  Marc Bulterys, Julia Brotherton, and Ding-​Shinn Chen

1217

62.4. Protection from Ultraviolet Radiation  Robyn M. Lucas, Rachel E. Neale, Peter Gies, and Terry Slevin

1221

62.5. Preventive Therapy  Jack Cuzick

1229

62.6. Regulation  Jonathan M. Samet and Lynn Goldman

1239

63. Cancer Screening  Jennifer M. Croswell, Russell P. Harris, and Barnett S. Kramer

1255

Index 1271

 ix

Acknowledgments

We are indebted to the more than 190 chapter authors who generously contributed their time, labor, and expertise to produce this comprehensively updated fourth edition. The multi-​authored text reflects the increasingly interdisciplinary and collaborative nature of the field; it provides a resource for researchers seeking to harness the unprecedented advances in genetic and molecular research into large-​scale population studies of cancer etiology, and ultimately into effective preventive interventions. We owe special thanks to Ms. Annelie Landgren, whose energy, enthusiasm, and organizational expertise as project manager have been invaluable in bringing this text to completion. We also thank Dr. Stephen Chanock for his early and unfailing encouragement and for supporting the critical infrastructure necessary for such a collaborative enterprise. This book would not have been possible without the generous forbearance of our spouses and families. Finally, Michael Thun thanks Dr. Lynne Moody for her insights as a sounding board throughout this process.

ix

x

 xi

Contributors

E. Susan Amirian, PhD

Jennifer L. Beebe-​Dimmer, PhD, MPH

Dan L Duncan Cancer Center Baylor College of Medicine Houston, Texas

Wayne State University School of Medicine Karmanos Cancer Institute Detroit, Michigan

William F. Anderson, MD, MPH

Douglas A. Bell, PhD

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Environmental Epigenomics, Immunity, Inflammation and Disease Laboratory National Institute of Environmental Health Sciences Research Triangle Park, North Carolina

Samuel O. Antwi, PhD Department of Health Sciences Research Mayo Clinic College of Medicine Rochester, Minnesota

Bruce K. Armstrong, MD, PhD* School of Public Health The University of Sydney Sydney, New South Wales, Australia

Matthew P. Banegas, PhD, MPH Center for Health Research Kaiser Permanente Portland, Oregon

Jill Barnholtz-​Sloan, PhD Case Comprehensive Cancer Center Case Western Reserve University School of Medicine Cleveland, Ohio

Dorothea T. Barton, MD Department of Surgery Dartmouth-​Hitchcock Medical Center Lebanon, New Hampshire

Laura E. Beane Freeman, PhD Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Amy Berrington de González, DPhil Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Marianne Berwick, PhD Department of Internal Medicine University of New Mexico Albuquerque, New Mexico

Smita Bhatia, MD, MPH Institute of Cancer Outcomes and Survivorship University of Alabama at Birmingham, School of Medicine Birmingham, Alabama

Amanda Black, PhD, MPH Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

A. Blair, PhD Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

William J. Blot, PhD Vanderbilt-​Ingram Cancer Center Nashville, Tennessee

* Retired

xi

xi

xii

Contributors

Melissa L. Bondy, PhD

James R. Cerhan, MD, PhD, (Editor)

Department of Medicine, Section of Epidemiology and Population Sciences Baylor College of Medicine Houston, Texas

Department of Health Sciences Research Mayo Clinic Rochester, Minnesota

AndrÉ Bouville, PhD*

Ellen T. Chang, ScD

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Center for Health Sciences Exponent Inc. Menlo Park, California

Marie C. Bradley, PhD, MScPH

Hepatitis Research Center National Taiwan University Hospital Taipei, Taiwan

Division of Cancer Control and Population Sciences National Cancer Institute Bethesda, Maryland

Freddie Bray, PhD

Ding-​Shinn Chen, MD

Wong-​Ho Chow, PhD

Section of Cancer Surveillance International Agency for Research on Cancer Lyon, France

Department of Epidemiology The University of Texas MD Anderson Cancer Center Houston, Texas

Alina V. Brenner, MD, PhD

Aaron J. Cohen, MPH, DSc‡

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Health Effects Institute Boston, Massachusetts

Louise A. Brinton, PhD

Division of Public Health Services Washington University St. Louis, Missouri

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Philip John Brooks, PhD Laboratory of Neurogenetics National Institute on Alcohol Abuse and Alcoholism, NIH Bethesda, Maryland

Julia Brotherton, MD, PhD National HPV Vaccination Program Register Victorian Cytology Service East Melbourne, Victoria, Australia

Elizabeth E. Brown, PhD, MPH Department of Pathology University of Alabama at Birmingham Birmingham, Alabama

Marc Bulterys, MD, PhD HIV/​Hepatitis Department World Health Organization Geneva, Switzerland

Kenneth P. Cantor, PhD, MPH* Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Deanna Alexis Carere, ScD, CGC Department of Pathology and Molecular Medicine McMaster University Hamilton, Ontario, Canada

Brian D. Carter, MPH Epidemiology Research Program American Cancer Society Atlanta, Georgia * Retired ‡ Consultant/​Contractor

Graham Colditz, MD, DrPH

Linda S. Cook, PhD Department of Internal Medicine University of New Mexico Albuquerque, New Mexico

Janine A. Cooper, PhD School of Pharmacy Queen’s University Belfast Belfast, Northern Ireland

Jennifer M. Croswell, MD, MPH Patient-​Centered Outcomes Research Institute Washington, DC

Rochelle E. Curtis, MA Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Anne E. Cust, PhD School of Public Health and Melanoma Institute Australia The University of Sydney Sydney, New South Wales, Australia

Jack Cuzick, PhD Wolfson Institute of Preventive Medicine Queen Mary University of London London, United Kingdom

Catherine de Martel, MD, PhD Infections and Cancer Epidemiology Group International Agency for Research on Cancer Lyon, France

 xi

Contributors

Michael Dean, PhD

David Forman, PhD

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

International Agency for Research on Cancer Lyon, France

Susan S. Devesa, PhD*

International Agency for Research on Cancer Lyon, France

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Graça M. Dores, MD§ Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Clifford E. Douglas, JD Center for Tobacco Control American Cancer Society Atlanta, Georgia

Jeffrey Drope, PhD Economic & Health Policy Research American Cancer Society Atlanta, Georgia

Donatus U. Ekwueme, PhD, MS National Center for Chronic Disease Prevention and Health Promotion Centers for Disease Control and Prevention Atlanta, Georgia

A. Heather Eliassen, ScD Brigham & Women’s Hospital and Harvard Medical School Harvard TH Chan School of Public Health Boston, Massachusetts

Hashem B. El-​Serag, MD, MPH Gastroenterology and Hepatology Baylor College of Medicine Houston, Texas

Eric A. Engels, MD Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Jacques Ferlay, MSc Section of Cancer Surveillance International Agency for Research on Cancer Lyon, France

Catterina Ferreccio, MD, MPH Division of Public Health and Family Medicine School of Medicine, Pontificia Universidad Católica de Chile Santiago, Chile

Maria Feychting, PhD Institute of Environmental Medicine Karolinska Institutet Stockholm, Sweden

Jonine D. Figueroa, PhD Usher Institute of Population Health Sciences and Informatics, CRUK Edinburgh Centre University of Edinburgh Edinburg, United Kingdom * Retired § Adjunct

Silvia Franceschi, MD

Joseph F. Fraumeni, Jr., MD Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Neal D. Freedman, PhD Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Søren Friis, MD Statistics and Pharmacoepidemiology Danish Cancer Society Research Center Copenhagen, Denmark

Morten Frisch, MD, PhD, DrSci(Med) Department of Epidemiology Research Statens Serum Institut Copenhagen, Denmark

Susan M. Gapstur, PhD, MPH Epidemiology Research Program American Cancer Society Atlanta, Georgia

Montserrat Garcia-​Closas, MD, DrPH Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Mia M. Gaudet, PhD Epidemiology Research Program American Cancer Society Atlanta, Georgia

Gretchen L. Gierach, PhD Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Peter Gies, PhD Australian Radiation Protection and Nuclear Safety Agency Melbourne, Victoria, Australia

Edward L. Giovannucci, MD, ScD Departments of Nutrition and Epidemiology Harvard TH Chan School of Public Health Boston, Massachusetts

Sally L. Glaser, PhD Cancer Prevention Institute of California Fremont, California

Lynn Goldman, MD, MS, MPH Milken Institute School of Public Health George Washington University Washington, DC

xiii

vxi

xiv

Contributors

Steven N. Goodman, MD, PhD

Henrik Hjalgrim, MD, PhD, DrSci(med)

Department of Medicine, Clinical and Translational Research Stanford University School of Medicine Stanford, California

Department of Epidemiology Research Statens Serum Institut Copenhagen, Denmark

Anala Gossai, MPH, PhD

Katherine A. Hoadley, PhD

Geisel School of Medicine Dartmouth College Hanover, New Hampshire

Department of Genetics, Lineberger Comprehensive Cancer Center University of North Carolina at Chapel Hill Chapel Hill, North Carolina

Adèle C. Green, MD, PhD

Jonathan N. Hofmann, PhD

Population Health Division QIMR Berghofer Medical Research Institute Brisbane, Queensland, Australia

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Andrew E. Grulich, PhD

Robert N. Hoover, MD, ScD

Kirby Institute The University of New South Wales Sydney, New South Wales, Australia

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Gery P. Guy Jr, PhD, MPH

Frank B. Hu, MD, PhD

National Center for Chronic Disease Prevention and Health Promotion Centers for Disease Control and Prevention Atlanta, Georgia

Departments of Nutrition and Epidemiology Harvard TH Chan School of Public Health Boston, Massachusetts

Laurel A. Habel, PhD, MPH

Department of Public Health North Dakota State University Fargo, North Dakota

Division of Research Kaiser Permanente Northern California Oakland, California

Christopher A. Haiman, ScD, (Editor) Department of Preventive Medicine Keck School of Medicine of University of Southern California Los Angeles, California

Susan E. Hankinson, ScD Department of Biostatistics and Epidemiology University of Massachusetts Amherst, Massachusetts

Russell P. Harris, MD, MPH§ Lineberger Comprehensive Cancer Center University of North Carolina School of Medicine Chapel Hill, North Carolina

Mia Hashibe, PhD Department of Family and Preventive Medicine Huntsman Cancer Institute, University of Utah School of Medicine Salt Lake City, Utah

S. Jane Henley, MSPH Division of Cancer Prevention and Control US Centers for Disease Control and Prevention Atlanta, Georgia

Rolando Herrero, MD, PhD Prevention and Implementation Group International Agency for Research on Cancer Lyon, France

Allan Hildesheim, PhD Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland §

adjunct

Rick J. Jansen, MS, PhD

Ahmedin Jemal, DVM, PhD Surveillance and Health Services Research Program American Cancer Society Atlanta, Georgia

Fengyi Jin, PhD Kirby Institute The University of New South Wales Sydney, New South Wales, Australia

Lisa G. Johnson, PhD, MPH Division of Public Health Sciences Fred Hutchinson Cancer Research Center Seattle, Washington

Dean P. Jones, PhD Department of Medicine Emory University Atlanta, Georgia

Margaret R. Karagas, PhD Department of Epidemiology Geisel School of Medicine at Dartmouth Hanover, New Hampshire

Ichiro Kawachi, MD, PhD Department of Social and Behavioral Sciences Harvard School of Public Health Boston, Massachusetts

Sarah Keadle, PhD, MPH Kinesiology Department California Polytechnic State University San Luis Obispo, California

 xv

Contributors

NaNa Keum, ScD

Yanhong Liu, PhD

Department of Nutrition Harvard TH Chan School of Public Health Boston, Massachusetts

Department of Medicine, Section of Epidemiology and Population Sciences Baylor College of Medicine Houston, Texas

Sally Kinsey, MD, FRCP Department of Pediatric Hematology Leeds Teaching Hospitals NHS Trust Leeds, United Kingdom

W. Thomas London, MD*

Cari M. Kitahara, PhD

Robyn M. Lucas, MBChB, MPH&TM, PhD

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Radiation Health Services Branch The Australian National University Canberra, The Australian Capital Territory, Australia

Jill Koshiol, PhD

Margaret M. Madeleine, PhD

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Division of Public Health Sciences Fred Hutchinson Cancer Research Center Seattle, Washington

Stella Koutros, PhD

Christopher J. Maher, PhD

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

McDonnell Genome Institute Washington University School of Medicine St. Louis, Missouri

Peter Kraft, PhD

Elaine R. Mardis, PhD

Departments of Epidemiology and Biostatistics Harvard TH Chan School of Public Health Boston, Massachusetts

McDonnell Genome Institute Washington University School of Medicine St. Louis, Missouri

Barnett S. Kramer, MD, MPH

Charles E. Matthews, PhD

Division of Cancer Prevention National Cancer Institute Bethesda, Maryland

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Candyce Kroenke, ScD, MPH

Marjorie L. McCullough, ScD, RD

Division of Research Kaiser Permanente Northern California Oakland, California

Epidemiology Research Program American Cancer Society Atlanta, Georgia

Qing Lan, MD, MPH

Katherine A. McGlynn, PhD

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

I-​Min Lee, MBBS, ScD

Angela L. W. Meisner, MPH

Brigham and Women’s Hospital Harvard Medical School Boston, Massachusetts

New Mexico Tumor Registry University of New Mexico Albuquerque, New Mexico

Tracy Lightfoot, PhD

Kyriaki Michailidou, PhD

Department of Health Sciences University of York York, United Kingdom

Department of Electron Microscopy/​Molecular Pathology The Cyprus Institute of Neurology and Genetics Nicosia, Cyprus

Sara Lindström, PhD

Lisa Mirabello, PhD

Department of Epidemiology University of Washington Seattle, Washington

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Martha S. Linet, MD, MPH, (Editor)

Karobi Moitra, PhD

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Trinity Washington University Washington, DC

* Retired

Fox Chase Cancer Center Philadelphia, Pennsylvania

xv

xvi

xvi

Contributors

Steven C. Moore, PhD

Alpa V. Patel, PhD

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Epidemiology Research Program American Cancer Society Atlanta, Georgia

Lindsay M. Morton, PhD

Kathryn L. Penney, ScD

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

ScD Department of Medicine Brigham and Women’s Hospital /​Harvard Medical School Boston, Massachusetts

Raul Murillo, MD

Gloria M. Petersen, PhD

Prevention and Implementation Group International Agency for Research on Cancer Lyon, France

Department of Health Sciences Research Mayo Clinic College of Medicine Rochester, Minnesota

Rachel E. Neale, PhD

Jessica L. Petrick, PhD

Population Health Division QIMR Berghofer Medical Research Institute Brisbane, Queensland, Australia

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Heather H. Nelson, MPH, PhD

Susan Picton, BMBS, FRCPCH

Division of Epidemiology, Masonic Cancer Center University of Minnesota, Twin Cities Minneapolis, Minnesota

Department of Pediatric Oncology Leeds Teaching Hospitals NHS Trust Leeds, United Kingdom

Marian L. Neuhouser, PhD

Brandon Pierce, PhD

Division of Public Health Sciences Fred Hutchinson Cancer Research Center Seattle, Washington

Departments of Public Health Sciences and Human Genetics University of Chicago Chicago, Illinois

Robert Newton, PhD

Martyn Plummer, PhD

MRC/​UVRI Uganda Research Unit Entebbe, Uganda

International Agency for Research on Cancer Lyon, France

Reiko Nishihara, PhD

I. Mary Poynten, PhD Kirby Institute

Department of Pathology Brigham and Women’s Hospital Boston, Massachusetts

The University of New South Wales Sydney, New South Wales, Australia

Michael A. O’Rorke, PhD

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

School of Medicine Dentistry and Biomedical Sciences Queen’s University Belfast Belfast, Northern Ireland

Andrew F. Olshan, PhD Department of Epidemiology Gillings School of Global Public Health University of North Carolina Chapel Hill, North Carolina

Quinn T. Ostrom, MA, MPH Case Comprehensive Cancer Center Case Western Reserve University School of Medicine Cleveland, Ohio

Julie R. Palmer, ScD Slone Epidemiology Center at Boston University Boston, Massachusetts

D. Maxwell Parkin, MD, DSc

Ludmila Prokunina-​Olsson, PhD

Mark P. Purdue, PhD Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Preetha Rajaraman, PhD Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Ewa Rajpert-​De Meyts, MD, PhD Department of Growth & Reproduction Copenhagen University Hospital Rigshospitalet Copenhagen, Denmark

Judy R. Rees, BM, BCh, PhD

Nuffield Department of Public Health University of Oxford Oxford, United Kingdom

Department of Epidemiology Geisel School of Medicine at Dartmouth Hanover, New Hampshire

Julie Parsonnet, MD

Juan Carlos Roa, MD, Msc

Department of Medicine Stanford University School of Medicine Stanford, California

Department of Pathology School of Medicine, Pontificia Universidad Católica de Chile Santiago, Chile

 xvi

Contributors

Eve Roman, PhD

Martyn T. Smith, PhD

Department of Health Sciences University of York York, United Kingdom

School of Public Health University of California at Berkeley Berkeley, California

Nathaniel Rothman, MD, MPH

Mingyang Song, MD, ScD

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Clinical and Translational Epidemiology Unit and Division of Gastroenterology Massachusetts General Hospital and Harvard Medical School Boston, Massachusetts

Jonathan M. Samet, MD Department of Preventive Medicine Keck School of Medicine of University of Southern California Los Angeles, California

John J. Spinelli, PhD Cancer Control Research

Ambika Satija, ScD

Avrum Spira, MD, MSc

Department of Nutrition Harvard TH Chan School of Public Health Boston, Massachusetts

Division of Computational Biomedicine Boston University School of Medicine Boston, Massachusetts

Sharon A. Savage, MD

Janet L. Stanford, PhD

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Division of Public Health Sciences Fred Hutchinson Cancer Research Center Seattle, Washington

Ghislaine Scelo, PhD

Andreas Stang, MD, MPH

Genetic Epidemiology Group International Agency for Research on Cancer Lyon, France

Institute of Medical Informatics, Biometry and Epidemiology University Hospital Essen Essen, Germany

Arthur B. Schneider, MD, PhD*

Kyle Steenland, PhD

Section of Endocrinology, Diabetes and Metabolism University of Illinois at Chicago College of Medicine Chicago, Illinois

Rollins School of Public Health Emory University Atlanta, Georgia

David Schottenfeld, MD, MSc (Editor)*

School of Public Health University of California Berkeley Berkeley, California

Department of Epidemiology University of Michigan School of Public Health Ann Arbor, Michigan

Mary Schubauer-​Berigan, PhD Division of Surveillance Hazard Evaluation and Field Studies Centers for Disease Control and Prevention Atlanta, Georgia

Joachim Schüz, PhD Section of Environment and Radiation International Agency for Research on Cancer Lyon, France

Amy L. Shafrir, ScD Division of Adolescent and Young Adult Medicine Boston Children’s Hospital Boston, Massachusetts

Mark E. Sherman, MD Health Sciences Research Mayo Clinic College of Medicine Jacksonville, Florida

Debra T. Silverman, ScD, ScM Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Terry Slevin, MPH Cancer Council Western Australia Perth, Western Australia, Australia * Retired

British Columbia Cancer Agency Vancouver, British Columbia, Canada

Craig M. Steinmaus, MD, MPH

Erich M. Sturgis, MD, MPH Department of Head and Neck Surgery The University of Texas MD Anderson Cancer Center Houston, Texas

Catherine M. Tangen, DrPH Division of Public Health Sciences Fred Hutchinson Cancer Research Center Seattle, Washington

Robert E. Tarone, PhD* International Epidemiology Institute Rockville, Maryland

Michael J. Thun, MD, MS (Editor)* Epidemiology and Surveillance Research American Cancer Society Atlanta, Georgia

William D. Travis, MD Department of Pathology Memorial Sloan Kettering Cancer Center New York, New York

Melissa A. Troester, PhD Department of Epidemiology Lineberger Comprehensive Cancer Center University of North Carolina at Chapel Hill Chapel Hill, North Carolina

xvii

xvii

xviii

Contributors

Rebecca Troisi, ScD, MA

David C. Whiteman, MD, PhD

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Population Health Division QIMR Berghofer Medical Research Institute Brisbane, Queensland, Australia

Shelley S. Tworoger, PhD

Charles Wiggins, PhD, MPH

Harvard Medical School and the Brigham and Women’s Hospital Harvard TH Chan School of Public Health Boston, Massachusetts

Department of Internal Medicine University of New Mexico Albuquerque, New Mexico

Celine M. Vachon, PhD

Christopher P. Wild, PhD

Department of Health Sciences Research Mayo Clinic Rochester, Minnesota

Director’s Office International Agency for Research on Cancer Lyon, France

Claire M. Vajdic, PhD

Walter C. Willett, MD, DrPH

Centre for Big Data Research in Health University of New South Wales Sydney, New South Wales, Australia

Department of Nutrition Harvard TH Chan School of Public Health Boston, Massachusetts

Roel Vermeulen, PhD

Deborah M. Winn, PhD

Institute for Risk Assessment Sciences Utrecht University Utrecht, The Netherlands

Division of Cancer Control and Population Sciences National Cancer Institute Bethesda, Maryland

Fawn D. Vigneau, JD, MPH

Kana Wu, MD, PhD

Wayne State University School of Medicine Karmanos Cancer Institute Detroit, Michigan

Department of Nutrition Harvard TH Chan School of Public Health Boston, Massachusetts

Teresa W. Wang, PhD

K. Robin Yabroff, PhD

Division of Computational Biomedicine Boston University School of Medicine Boston, Massachusetts

Division of Cancer Control and Population Sciences National Cancer Institute Bethesda, Maryland

Mary H. Ward, PhD

Shelia Hoar Zahm, ScD‡

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Division of Cancer Epidemiology and Genetics National Cancer Institute Bethesda, Maryland

Noel S. Weiss, MD, DrPH

Chenan Zhang, PhD

Department of Epidemiology University of Washington Seattle, Washington

Department of Epidemiology and Biostatistics University of California, San Francisco San Francisco, California

‡

Consultant/​Contractor

x i

Preface

The Schottenfeld and Fraumeni text on Cancer Epidemiology and Prevention has served as the premier reference text for population research on the causes and prevention of cancers since the publication of the first edition in 1982 (Schottenfeld and Fraumeni, 1982). It is written for colleagues pursuing careers in research in cancer epidemiology and, more broadly, in preventive oncology. The founding editors, Dr. David Schottenfeld, now emeritus professor of epidemiology at the University of Michigan, and Dr. Joseph Fraumeni, recently retired as the director of the Division of Cancer Epidemiology and Genetics at the National Cancer Institute (NCI), updated their landmark text in 1996 and 2006 (Schottenfeld and Fraumeni, 1996, 2006). The current edition again provides a comprehensive update of research advances in cancer epidemiology, prevention, and related fields in the past 10–​15 years, and honors the founding editors in the title. The new editorial team is led by Dr. Michael Thun (editor-​in-​chief), formerly with the American Cancer Society, and includes four senior co-​editors: Drs. Martha Linet from NCI, James Cerhan from the Mayo Clinic, Christopher Haiman from the University of Southern California, and David Schottenfeld. We are also deeply indebted to the internationally recognized experts who authored the 63 chapters. Without their generous effort and commitment, this updated synthesis would not be possible.

xix

x

 1

1 Introduction MICHAEL J. THUN, MARTHA S. LINET, JAMES R. CERHAN, CHRISTOPHER A. HAIMAN, AND DAVID SCHOTTENFELD

I

n this introduction, we provide an overview of the text and highlight cross-​cutting developments and new opportunities that are transforming our understanding of the causes and prevention of cancer. As in previous editions, the text is grouped into five major parts: “Basic Concepts,” “The Magnitude of Cancer,” “The Causes of Cancer,” “Cancers by Tissue of Origin,” and “Cancer Prevention and Control.” Part I first describes research advances in understanding “the biology of neoplasia,” including the progressive disruption of genetic and epigenetic controls that regulate cell growth, division, and survival (Chapter 2). Advances in high-​throughput technologies have greatly expanded the ability to identify germline and somatic mutations and to relate these to etiology, prognosis, and treatment. Tumor classification is also changing for certain cancers, as data on the molecular features and lineage of the neoplastic cells is combined with information on the primary anatomic location and the morphologic, histopathologic and clinical characteristics of the tumor (Chapter 3). The “landscape” of genomic and epigenomic alterations in tumor tissue has been cataloged for multiple human cancers (Chapter 4), revealing both the singularity of individual cancer genomes and the commonality of genetic alterations that drive cancer in different tissues. Chapter 5 describes advances in research on inherited genomic variants that affect cancer risk. Genome-​wide association studies (GWAS) have identified more than 700 germline loci associated with increased or decreased risk for various types of cancer, although the risk estimates for almost all are small to modest. Innovations in genomics and other “OMIC” technologies are identifying biomarkers that reflect internal exposures, biological processes, and intermediate outcomes in large population studies (Chapter 6). While research in many of these areas is still in its infancy, mechanistic and molecular insights are extending the traditional criteria for inferring causation in epidemiologic studies of cancer (Chapter 7). Part 2 of the book discusses the global public health impact of cancer and its relationship to demographic trends, changing risk factors, socioeconomic disparities, and economic development. It considers the direct and indirect costs of cancer in the United States to illustrate the economic burden in a high income country. Parts 3–​5 of the book discuss the growing list of exposures known to affect cancer risk, the epidemiology of over 30 types of cancer by tissue of origin, and the encouraging progress in cancer prevention and control. Major developments in these areas are discussed below, beginning with those that affect the public health impact of cancer.

MAJOR NEW DEVELOPMENTS Global Trends in Cancer Risk and Burden Part II, “The Magnitude of Cancer,” provides a global public health perspective on cancer. The human and economic costs of cancer are increasing worldwide (http://​globocan.iarc.fr). The World Health Organization (WHO) estimates that 14 million new cases and 8.2 million deaths from cancer occurred in 2012. This burden is projected to increase to 24 million cases and 13 million deaths annually by 2035 (Ferlay et al., 2013). Chapter 8 decribes the disproportionate increase in the cancer burden in low-​and middle-​income countries (LMICs), which can least afford additional health-​related, social and financial

costs. In 2012, these countries accounted for over half (57%) of all incident cancers; this is projected to increase to nearly two-​thirds (65%) by 2035. Much of the increase will result from the growth and aging of populations, since LMICs currently comprise about 80% of the world’s population, and large numbers of young adults are now surviving to older ages, when cancer becomes more common. In addition to the effect of demographic changes, cancer incidence and mortality rates are increasing in LMICs because of the widespread adoption of Western patterns of diet, physical inactivity, excess body fat, delayed reproduction, and tobacco smoking, especially of manufactured cigarettes. As countries advance economically, the incidence rates of cancers traditionally associated with Westernization (e.g., breast, colorectum, lung, and prostate) increase more rapidly than the decrease in cancers caused partly or wholly by infectious agents (e.g., stomach, liver, uterine cervix). Survival after a diagnosis of cancer is also lower in LMICs than in high-​resource countries, because of later stage at diagnosis, a higher proportion of tumors diagnosed clinically rather than incidentally, and limited access to standard and state-​of-​ the-​art treatment protocols. In economically developed countries, the incidence rates of most cancers are either stabilizing at a high level or decreasing, depending on the temporal trends of underlying risk factors and utilization of cancer screening. Despite the decreasing rates, the disease burden, or number of cancer cases and deaths, continues to increase. The increasing burden results from the aging and growth of populations, and the decline in competing causes of death from circulatory and infectious diseases. Mortality rates are decreasing more rapidly than incidence rates for many cancer sites due to a combination of prevention, early detection, and improvements in treatment. Part III, “The Causes of Cancer,” discusses 15 broad categories of exposure that affect cancer risk. These include exposures that are typically considered “environmental” by the public (chemical carcinogens, ionizing radiation, occupational exposures, pollutants in air and drinking water), as well as exposures that are less widely recognized as carcinogenic (infectious agents, metabolic factors, body composition, reproductive and other hormones, pharmaceutical drugs, and immunological conditions). All of these exposures are “environmental” in the sense that they are acquired after conception rather than inherited. Some are genotoxic and damage the structure of DNA or alter DNA repair; others modify gene expression, induce oxidative stress and/​ or chronic inflammation, suppress host immunity, immortalize cells, modulate receptors, and/​or alter cell proliferation, cell death, or nutrient supply (Smith et al., 2016). Although some exposures are conventionally perceived as “lifestyle choices”, they are by no means entirely voluntary. For example, behavioral risk factors such as tobacco smoking, energy imbalance, and physical inactivity are strongly influenced by factors in the social, economic, and cultural environment, beginning in early childhood. Physiologic addiction is a major driver of tobacco use at all ages. Part IV of the book describes “Cancers by Tissue of Origin” for 33 anatomic sites, multiple primary tumors, and cancers in children. Rapid advances in discovering the molecular events that drive certain forms of cancer are transforming clinical diagnoses and treatment, and affecting tumor classification. This will influence future endpoints in etiologic studies and population-​based cancer surveillance.

1

2

2 Introduction Part V, “Cancer Prevention and Control,” discusses the impact of interventions that effectively reduce carcinogenic exposures or disrupt the multistage progression of tumors. It focuses on interventions that demonstrably reduce cancer risk in the general population, rather than in special circumstances or high-​risk subgroups. Examples of these are discussed in Chapters 61–​63. In all cases, the design and implementation of preventive measures require translational research to ensure safety, optimize feasibility and impact, and critically evaluate all stages of the process.

Cancer Prevention and Control A growing number of population-​level preventive interventions are proving to be highly effective, as confirmed by the decreases in incidence as well as mortality rates from certain cancers (Chapters 61–​63). Tobacco control has reduced the age-​standardized incidence rate of lung cancer by up to 40% among men in high-​and middle-​income countries. Increased screening for colorectal cancer and removal of precursor lesions is credited for the 30% decrease in the incidence rates at this site in the United States. Universal neonatal vaccination against hepatitis B virus (HBV) has markedly decreased the prevalence of chronic HBV infection and liver cancer at younger ages in high-​risk areas of East Asia and will yield maximal benefits against cancer in the future. The development of safe and effective vaccines against human papillomavirus (HPV) and less expensive and less onerous screening tests for cervical cancer have greatly expanded opportunities to prevent HPV-​related cancers among women in many LMICs. Increased funding is becoming available for application research and cancer preventive services in LMICs. Cancer prevention presents both opportunities and challenges, as discussed in Part V of the text. The best practices developed for tobacco control provide an encouraging model of how health-​related policies can address the behavioral causes of cancer. However, these must be tailored to fit the particular social, economic, and other considerations that affect the exposure (Chapter 61).

Advances in Genomics and other OMICs Technological advances in high-​ throughput genotyping/​ sequencing and gene expression arrays have transformed research on both inherited (germline) susceptibility variants and the largely acquired (somatic) mutations in tumor tissue. Epidemiologic studies of cancer genetics have focused mainly on germline variants associated with cancer risk and etiology, whereas clinical and basic researchers have characterized the landscape of somatic alterations in tumor cells that drive the development and progression of cancer.

Germline Susceptibility Variants

The tools to identify inherited genetic susceptibility variants have advanced enormously since publication of the previous edition of this text in 2006. At that time, studies involved either high-​risk families or the evaluation of a small number of pre-​specified “candidate genes” in case-control studies of sporadic cancers in the general population. The candidate gene approach was largely unsuccessful in identifying robust associations for several reasons, including small sample size, limited statistical power, failure to account for multiple testing (generating negative and false positive results, respectively), and limited biologic knowledge to inform the selection of candidate genes. Following the completion of the Human Genome Project in 2003, genome-​wide maps of single nucleotide polymorphisms (SNPs) became available. Advances in high-​ throughput genotyping technology, combined with knowledge about the structure of genetic linkage disequilibrium, created opportunities to conduct exploratory (hypotheis-​free or “agnostic”) surveys across the entire genome. Over the past decade, GWAS have robustly identified more than 700 common (i.e., minor allele frequency >5%) susceptibility loci associated with cancer risk, as discussed for specific sites in Part IV, “Cancers by Tissue of Origin.” Because GWAS test millions of alleles across the genome, they require stringent criteria

(“genome-​wide significance”), large sample size, and replication in more than one study to exclude chance associations. Most of the associations identified through GWAS are modest (per allele ORs: 1.5–​2.0) or weak (ORs 800 vs. ≤ 110 μg/​L) predicted 5-​fold higher rates of lung and 8-​fold higher rates of bladder cancer decades later (Steinmaus et al., 2014).

Postnatal and Later Exposures Major mechanisms for postnatal exposures have included childhood infection, lifestyle behaviors, and secondhand exposure to tobacco smoke. Population-​ level dietary trends, combined with reductions in physical activity, have led to an unprecedented rise in childhood obesity, which has disproportionately affected children from low SES households. The childhood obesity epidemic in the United States (and in many other parts of the globe) has dire implications for future cancer risk and disparities in cancer incidence. The latency period for cancer incidence is typically decades, but these factors may lead to an accelerated risk of the development of cancer early in life, particularly cancers (e.g., colon cancer) that share many of the same risk factors as CVD and diabetes.

Childhood and Adolescent Infection.  There is substantial

evidence that infections are likely causal factors in several types of cancer, including lymphoma, cancers of the liver, nasopharynx, cervix, and stomach. Earlier studies showed that these types of cancers accounted for up to 20% of cancer worldwide (Eckhart, 1998), though recent studies indicate that cancers caused by infections are more than three times more prevalent in developing (26%) than in developed (8%) countries (Thun, DeLancey, Center, Jemal, & Ward, 2010). Viruses linked to human tumors include Epstein-​Barr virus (EBV) (B-​ cell lymphomas, Burkitt’s lymphoma, nasopharyngeal cancer, Hodgkin lymphoma, T-​cell lymphomas, and gastric cancer); hepatitis B virus (hepatocellular carcinoma); papillomavirus types including 16, 18, 31, 33, 35, 39, 45, 52, 56, 58 (cervical and anogenital cancer); and HTLV-​1 (adult T-​cell leukemia) (Chelimo, Wouldes, Cameron, & Elwood, 2013; F.  Liu et  al., 2016; G.  S. Taylor, Long, Brooks, Rickinson, & Hislop, 2015; Zane & Jeang, 2014). Transmission and timing of infection have been linked to SES.

EBV is involved in about 35%–​50% of cases of Hodgkin disease. Though most adults have had an EBV infection, and are thus carriers of these viral genes (Evans & Mueller, 1997), higher SES groups tend to be infected at later ages than those of lower SES. EBV infection at a later age is associated with more severe clinical sequelae, namely infectious mononucleosis and Hodgkin lymphoma, in higher SES populations (Gutensohn, 1982). Findings strongly indicate that chronic Helicobacter pylori infection is involved in the development of non-​cardia gastric adenocarcinoma and mucosal-​ associated B-​ cell lymphoma (International Agency for Research on Cancer Ad Hoc Working Group, 1994; Nightingale & Gruber, 1994). About two-​thirds of the world’s population and half of all adults over age 50 in the United States are infected with H. pylori (Correa & Piazuelo, 2011). Since the 1930s, modernization, leading to clean water, fewer children sharing a bed, smaller families, and possibly the increasing use of antibiotics in children (Blaser, 1999), has contributed to reductions in H.  pylori transmission and prevalence (Parsonnet, 1995). The infection is typically acquired in childhood and there are fewer children in the United States today with H. pylori infection than in previous periods. H. pylori is found more often in lower socioeconomic groups in the United States, many of whom report one or more risk factors for infection, including a history of crowding in childhood, a mother who carries H. pylori, a large number of siblings, presence of older siblings (< 4-​year age difference), and unclean water sources. In turn, several of these factors are associated with lower SES (Goodman & Correa, 1995).

Lifestyle Behaviors and Obesity.  The US Centers for Disease

Control and Prevention reported that the number of overweight children more than doubled in the last three decades of the twentieth century (Centers for Disease Control and Prevention, 1997). More recent evidence suggests that rates of obesity have stabilized in the past couple of decades (Ogden, Carroll, Fryar, & Flegal, 2015; Ogden, Carroll, Kit, & Flegal, 2012; Ogden et  al., 2016; Ogden, Lamb, Carroll, & Flegal, 2010), but overweight and obesity are more prevalent among children and adolescents from lower SES backgrounds, and 12.7 million children and teens ages 2–​18 (17% of the entire population) are currently overweight or obese (Ogden, Carroll, Kit, & Flegal, 2014). Maternal obesity, also predominant in adults from lower SES backgrounds, is strongly linked with childhood obesity (Strauss & Knight, 1999). Low SES children often live in areas with large numbers of fast food restaurants and stores that sell primarily low-​quality, nutritionally poor, and calorie-​dense foods. Studies also have shown that children in lower SES families are more likely to be exposed to advertisements that promote eating large portions of obesity-​promoting foods, including sugar-​sweetened beverages, fast food, and processed foods (Harris, Pomeranz, Lobstein, & Brownell, 2009; Piernas & Popkin, 2011; Powell, Szczypka, & Chaloupka, 2010; Powell, Wada, & Kumanyika, 2014). Childhood overweight and obesity in turn may track into adulthood overweight and obesity, which are major causes of cancer in later life. Dietary habits, which begin in childhood, are also predictive of future dietary patterns. Lack of local availability of fresh produce, cultural patterns of food consumption (R. E.  Lee & Cubbin, 2002), norms accepting of overweight (Becker, Yanek, Koffman, & Bronner, 1999; Crawford, Story, Wang, Ritchie, & Sabry, 2001), and sedentary behaviors are each related to lower SES and overweight. In addition to habits that are developed in the context of a lack of resources, current overweight predicts future overweight. Girls who are overweight have earlier maturational timing, which is associated with later overweight (Adair & Gordon-​Larsen, 2001), and is also an independent risk factor for breast cancer (Cancer, 2012). Adolescents who are obese are far more likely to be obese as adults, with long-​term, multiple consequences for morbidity (Dietz, 1998a, 1998b; Micic, 2001). Patterns established in childhood and adolescence therefore predispose people from lower SES backgrounds to worse health outcomes, including cancer, in later adulthood.

 15



Socioeconomic Disparities in Cancer Incidence and Mortality

Smoking and Secondhand Exposure to  Cigarette Smoke.  Children from lower SES backgrounds are more likely to

be exposed to environmental tobacco smoke (ETS) in their homes (K. M. Emmons et al., 2001; Schuster, Franke, & Pham, 2002). Though a meta-​analysis found no increased association of parental smoking and lung cancer of the grown child (Boffetta, Tredaniel, & Greco, 2000), ETS has been associated with an increased risk of lung cancer in spouses (adult non-​smoking women) (Kreuzer et al., 2002; R. Taylor, Cumming, Woodward, & Black, 2001). Low-​SES youth are subsequently more prone to take up smoking compared to youth from more affluent backgrounds. Factors that may be related to the higher initiation of smoking among low-​SES youth include parental smoking, though especially for white adolescents (Griesler & Kandel, 1998; Tyas & Pederson, 1998), greater exposure to peer norms of smoking (Alexander, Piazza, Mekos, & Valente, 2001; Tyas & Pederson, 1998), as well as greater targeting by the tobacco industry, a strong predictor of adolescent onset of smoking (Altman, Levine, Coeytaux, Slade, & Jaffe, 1996; Pierce, Choi, Gilpin, Farkas, & Berry, 1998). Low-​SES youth are also more likely to start smoking at an earlier age, which has been linked to greater difficulty in quitting later on (Pampel, Mollborn, & Lawrence, 2014). Furthermore, it has been hypothesized that early smoking occurring during adolescence, a “critical period” in lung development, may be particularly hazardous, in that tobacco carcinogens may induce genetic alterations that make the early smoker more susceptible to the damaging effects of continued smoking (Wiencke & Kelsey, 2002).

Environmental Exposures. A meta-​analysis and a registry-​ based case control study showed that postnatal exposure to traffic pollution and air toxics, particularly benzene, was related to a higher risk of childhood leukemia (Amigou et  al., 2011; Filippini, Heck, Malagoli, Del Giovane, & Vinceti, 2015). Recently, there has also been a growing research emphasis on the potential cancer-​causing effects of plasticizers, chemicals, and flame retardants. Though exposure in daily life is nearly ubiquitous, there is evidence that children from low-​ SES households have higher exposures to flame retardants because of exposure to household furniture in poor condition (Schreder, Uding, & La Guardia, 2016). These chemicals have been shown to cause cancer in animal studies, though researchers are only beginning to understand the potential impact of these unregulated chemicals on the development of cancer and on the implications for the SES-​cancer gradient (Costa, Pellacani, Dao, Kavanagh, & Roque, 2015). Research in humans is not well established. In addition to direct effects of chemicals on risk of cancer, plasticizers have been hypothesized to be related to higher estrogens in children, leading to earlier menarche in girls, and potentially an increasing risk of breast cancer in girls of low SES (M. Lee, 2012).

Adult Exposures Lifestyle and Related Factors Major lifestyle factors in adult life linked to cancer incidence include smoking, poor diet, alcohol consumption, physical inactivity, and overweight or obesity. Many of these factors are associated with each other and tend to cluster together. Several health-​related behaviors (e.g., smoking, heavy alcohol consumption, overeating) are used by individuals to cope with stress, a fact that has not escaped the notice of manufacturers and advertisers who target vulnerable audiences. In combination with physical inactivity, an imprudent diet can increase the risk of being overweight or obese. The changing temporal socioeconomic patterns in US lung and colorectal cancer mortality may be attributable in part to changing socioeconomic patterns in these factors.

Smoking.  Smoking continues to be among the most important fac-

tors responsible for the SES gradient in cancer. Cigarette smoking is the leading determinant of lung cancer; 90% of those with lung cancer

155

were smokers (Pesch et  al., 2012). Reducing or eliminating smoking would prevent the bulk of lung cancer cases, as well as drastically reduce socioeconomic disparities in lung cancer incidence. In response to declining cigarette consumption in the United States, particularly among higher SES groups, the tobacco industry substantially increased expenditures on marketing and advertising promotions, such as “Marlboro Miles” and Camel Cash,” and tobacco promotional items (Federal Trade Commission, 1999). After the Master Settlement Agreement of 1998 banned the distribution of branded merchandise, tobacco companies invested in discounts and point-​of-​sale advertising, which are disproportionately targeted to people—​and especially youth—​ from lower SES backgrounds (Robert, Cheney, & Azad, 2009). Within the past decade, electronic cigarettes (e-​cigarettes) have also come into use. Several studies have reported on disparities in awareness and use of e-​cigarettes by income levels (Carroll Chapman & Wu, 2014) but the links to SES and to health are as yet unclear.

Diet.  Higher SES is positively correlated with higher quality diet,

including greater consumption of fruit, vegetables, and legumes (Centers for Disease Control and Prevention, 2000; Pechey & Monsivais, 2016), and lower consumption of foods with high salt/​ fat/​sugar content (Pechey & Monsivais, 2016). Factors such as fat intake, red meat consumption, inadequate vegetable consumption, high caloric intake, physical inactivity, and heavy alcohol consumption have been suggested as important risk factors for colorectal cancer (Vargas & Thompson, 2012). Each of these factors has been related to lower SES, both at the individual level and the area (neighborhood) level (Centers for Disease Control and Prevention, 2000; Diez-​Roux et  al., 1999; Kamimoto, Easton, Maurice, Husten, & Macera, 1999; Li et al., 2000). However, the apparent correlation between low SES and unhealthy diet has not always been direct. A study of dietary trends by Popkin and colleagues (Popkin, Siega-​Riz, & Haines, 1996) found large differences in dietary quality in 1965, with whites of high SES eating the least, and blacks of low SES eating the most healthful diet as measured by a diet quality index. By the 1989–​1991 survey, the diets of all groups studied were relatively similar. Of particular note, grain and legume consumption declined among low SES blacks since the 1960s but increased among higher SES whites. The study also indicated that people from both lower and higher SES backgrounds showed improvements in diet, based on increases in fruit and vegetable intake and then-​ current recommendations for reductions in fat intake. Higher SES groups initially had more improvements to make, but dietary quality was similar across SES groups by the early 1990s after the advent of recommendations to reduce fat (≤ 30%), saturated fat (< 10%), and cholesterol (< 300 mg/​day) intake; to increase consumption of fruits and vegetables (5+ per day) and complex starches (6+ servings/​day); to limit sodium intake (≤ 2400 mg/​day); and to maintain proper calcium (RDA) and protein (< 2 times RDA) intake (Popkin et al., 1996). In 1998, Shi noted that changing patterns in diet may have reflected a greater receptivity of higher SES people to public health messages about consuming foods of high nutritional quality (L. Shi, 1998). Although the authors did not discuss the consumption of junk food, studies throughout the 2000s have shown that the potential benefits of relatively healthier diets in blacks compared with whites in the 1960s were reversed in the following decades, related to increased intakes of low-​quality foods (Bahr, 2007; Darmon & Drewnowski, 2008, 2015; Hiza, Casavale, Guenther, & Davis, 2013; Siega-​Riz & Popkin, 2001; Thompson et al., 2009; Vargas & Thompson, 2012; C. Y. Wang et al., 2014). Thus the rise in obesity prevalence among lower SES groups has been influenced by increased consumption of low-​quality foods and overall caloric consumption, even while efforts were made toward meeting dietary recommendations (Harnack, Jeffery, & Boutelle, 2000; Heini & Weinsier, 1997; Siega-​Riz & Popkin, 2001). Trends toward increased consumption may be due to increases in availability and relative affordability of food items, increased marketing of low-​quality food items (Chandon & Wansink, 2012; C. Y. Wang et al., 2014), and increases in portion sizes (Nielsen & Popkin, 2003). Associated with these, changes may be attributed to changes in agricultural practices,

156

156

Part II: The Magnitude of Cancer

subsidizing of corn production, and food technologies leading to cheap production of low-​quality foods (Franck, Grandi, & Eisenberg, 2013; Rickard, Okrent, & Alston, 2013).

Alcohol.  Though wealthier people are more likely to engage in

moderate alcohol consumption, which has been linked to lower levels of insulin, body weight, and risk of diabetes (Gepner et al., 2015; Wakabayashi, 2014), heavy alcohol consumption is more predominant among lower SES groups (Esser et al., 2014). Considerable evidence suggests a connection between heavy alcohol consumption and increased risk for cancer, with an estimated 5.8% of total cancer deaths attributable directly or indirectly to alcohol (Rothman, 1980). Alcohol consumption is an established cause of cancers of the mouth, pharynx, larynx, esophagus, liver, and, in women, breast (Boffetta, Hashibe, La Vecchia, Zatonski, & Rehm, 2006). For each of these cancers, risk increases substantially with intake of more than 2 drinks per day, though regular consumption of even a few drinks per week has been associated with an increased risk of breast cancer in women (Shield, Soerjomataram, & Rehm, 2016). Other research suggests that other lifestyle factors associated with low SES, in combination with heavy alcohol consumption, such as malnutrition (Giskes, Turrell, Patterson, & Newman, 2002; Su & Arab, 2001) or smoking, may interact to further increase risk of certain types of cancer (Jones, Bates, McCoy, & Bellis, 2015). Alcohol consumption combined with low folate consumption appears to predict higher levels of colon cancer (Vargas & Thompson, 2012), and combined with tobacco use, increases risk of cancers of the mouth, larynx, and esophagus more than the independent effect of either drinking or smoking (Hashibe et al., 2009).

Obesity and Physical Activity. Overweight and obesity are

associated with several cancers, including postmenopausal breast cancer, colon cancer, endometrial cancer, prostate cancer, renal cell carcinoma, esophageal adenocarcinoma, ovarian, cervical, and thyroid cancers. In addition to changing diets, the disproportionate prevalence of obesity among people from lower SES backgrounds may help to account for the changing SES gradient for colorectal cancer. As indicated, obesity has increased dramatically over the past several decades, though the prevalence has stabilized over the past decade or so (Flegal, Carroll, Kit, & Ogden, 2012; Flegal, Carroll, Ogden, & Curtin, 2010; Yanovski & Yanovski, 2011), but lower SES groups, especially women, are much more likely to be overweight than people from higher SES backgrounds (Robbins, Vaccarino, Zhang, & Kasl, 2000; Sharpe, Whitaker, Alia, Wilcox, & Hutto, 2016). As socioeconomic disparities in overweight have widened, the SES gradient in colorectal cancer incidence and mortality is likely to increase. Similarly, the gradient will likely change as well for other cancers related to obesity. Higher SES is generally positively correlated with physical activity, which in turn has been associated with lower levels of obesity-​related cancers. As an independent risk factor, the evidence for decreased risk with higher physical activity is classified as convincing for breast, colon, and endometrial cancers, probable for prostate cancer, and possible for lung cancers (American Cancer Society, 2016; Clague & Bernstein, 2012; Friedenreich & Orenstein, 2002). In recent years, research has been extended to examine the potential effects of physical activity on other types of cancer. The hypothesized biological mechanisms for the association between physical activity and cancer include changes in endogenous sex hormone levels and growth factors, decreased obesity and central adiposity, and possibly changes in immune function. Central adiposity has been implicated in metabolic conditions that promote carcinogenesis. Evidence is also increasing that exercise influences other aspects of the cancer process, including cancer detection, coping, rehabilitation, and survival (Friedenreich & Orenstein, 2002). Based on existing evidence, the American Cancer Society has issued physical activity guidelines for cancer prevention, generally recommending at least 30 minutes of moderate-​to-​vigorous intensity physical activity on 5 or more days per week, or vigorous activity for 75 minutes or more per week (American Cancer Society, 2016).

Sexual Behavior.  Aspects of sexual behavior and sexually trans-

mitted infections (STIs) have been linked to reproductive cancers such as cervical and prostate cancers (Castellsague et al., 2002). There are limited data available for assessing rates of STIs by SES, and when data exist, associations are generally weaker than those seen for race/​ ethnicity and STIs, though some studies do show elevated rates of STIs among disadvantaged groups (Harling, Subramanian, Barnighausen, & Kawachi, 2013). Cervical cancer is more common among women from lower SES backgrounds, who are less likely to receive regular screening and early treatment (Churilla et al., 2016). Other important factors related to the development of cervical cancer include a woman’s history of sexual exposure, her partners’ history of sexual exposure, use of barrier methods of contraception (protective), and the age at which a woman becomes sexually active (Chelimo et al., 2013). Risk of cervical cancer is also increased when a woman is exposed to the virus before full maturation of the cervix. In turn, several of these factors have been shown to be related to SES, especially level of education. In a study by Hogan, Sun, and Cornwell (2000), adolescent females whose parents were better educated (greater than high school education) were 28% less likely to initiate sexual intercourse and 52% more likely to use a contraceptive at first intercourse. Similar findings in other studies also suggest that females with lower income or parental educational attainment are more likely to have an early age at first intercourse (Blum et al., 2000; Lammers, Ireland, Resnick, & Blum, 2000; Santelli, Lowry, Brener, & Robin, 2000; S. Singh, Darroch, & Frost, 2001), and subsequently they have more and/​or concurrent sexual partners (Coker et al., 1994; O’Donnell, O’Donnell, & Stueve, 2001) and are less likely to use barrier contraception (Bankole, Darroch, & Singh, 1999). Roura and colleagues (2014) noted that compared to women who never smoked, women who currently smoked were 1.9 times more likely to develop cervical cancer, due to the damage cigarette smoking causes to cervical cells.

Reproductive Factors

Reproductive factors associated with SES and with cancer incidence include age at menarche, parity, age at first birth, age at menopause, and use of exogenous hormones. Higher SES women have been shown consistently to have higher rates of breast cancer. They are more likely to have first children at a later age and to have fewer children. They have been more likely to take exogenous hormones after menopause. These factors have in turn been associated with an increased breast cancer risk. Early age at menarche is also associated with higher risk of breast cancer. Higher SES has been historically associated with earlier age at menarche, which has been observed both within countries, as well as in comparisons of developed versus less developed countries. On the other hand, with the obesity epidemic growing disproportionately among low-​SES children, early menarche may increasingly weigh against lower SES women in the United States, thereby leading to a further narrowing of the SES gap in breast cancer incidence in the future. As previously discussed, declines in the use of exogenous hormones after 2003 led to declines in breast cancer incidence, but particularly in women of higher SES who were taking hormone therapy at a higher rate than women of low SES.

Occupational Exposures

Blue-​collar and manual workers experience greater exposure than white-​collar workers to chemicals, diesel fumes, dyes, and other agents in the workplace, including inorganic gases, organic compounds, solvents, mineral and wood dusts, silica, metals, and bioaerosols, which are established carcinogens (Boffetta et al., 2000; Kauppinen, Teschke, Savela, Kogevinas, & Boffetta, 1997; Kogevinas & Porta, 1997; US Department of Health and Human Services, 2002; Weston, Aronson, Siemiatycki, Howe, & Nadon, 2000). Blue-​collar workers are also more likely to be exposed to environmental tobacco smoke (ETS) (Curtin, Morabia, & Bernstein, 1998). Percent of white-​collar workers was a predictor of more restrictive smoking policies in a national sample of worksites (K.M. Emmons,

 157



Socioeconomic Disparities in Cancer Incidence and Mortality

2000). Furthermore, the interactive effects of smoking and chemical exposures may put workers of low SES at additional increased risk of cancer. The combination of asbestos exposure and cigarette smoking, both correlated with low SES, has been associated with a synergistically increased risk of lung cancer (Gustavsson et al., 2002). Though not all shift workers occupy lower SES positions (e.g., healthcare professionals), shift work is more common among persons from lower SES backgrounds (e.g., jobs in the service industry, transport, security work, manufacturing, etc.) (Boggild, Suadicani, Hein, & Gyntelberg, 1999). Longer durations (> 20 years) of rotating night-​ shift work have been linked to immune dysfunction (Nakata, 2012) and rotating night-​shift work has been linked to increased risks of breast and colon cancer (Schernhammer et  al., 2001, 2003a). In particular, working 30 or more years on the night shift was associated with a moderate elevation in risk of breast cancer (RR = 1.36; 95% CI: 1.04, 1.78) and working 15 or more years on rotating night shifts was associated with an increase in colon cancer (RR = 1.35; 95% CI: 1.03, 1.77). The mechanism is believed to be through the suppression of melatonin production, caused by prolonged exposure to light during night-​time (melatonin, in turn, is believed to have oncostatic action).

The Social Environment Neighborhood Environments.  Much empirical work on SES

and cancer to date has focused on individual-​level exposures and relationships. However, increasing attention has been paid more recently to the independent contribution of area-​level socioeconomic factors on health outcomes (Kawachi & Berkman, 2003)  including cancer (Gomez et  al., 2015), and research to identify contextual effects on health outcomes has expanded markedly over the past decade. A study by Merkin and colleagues (Merkin, Stevenson, & Powe, 2002) found that living in areas with lower levels of education and income increased the odds of presenting with advanced-​stage breast cancer by 50% for black women and by 75% for white women. One study found that those in severe poverty or near poverty were, respectively, 3.0 and 1.6 times more likely to live in areas of higher-​than-​expected incidence of late-​stage breast cancer when compared with women living in non-​poverty (MacKinnon et al., 2007). However, a recent study in Michigan in 1992–​2009 found that women with breast cancer living in low-​SES areas showed greater declines in mortality rates compared to those living in high-​SES areas, resulting in a narrowing of the SES-​ mortality gap (Akinyemiju et al., 2013). Community and neighborhood level factors, including the availability of green space (for physical activity), access to stores and grocers, as well as social norms and social cohesion, may have an influence on health above and beyond individual factors, including individual SES (Kawachi & Berkman, 2003). Special study designs and analytical techniques (multilevel analysis) are required to detect the presence of contextual effects. Obstacles abound in drawing causal inferences from such data. Nevertheless, redirecting the focus of interventions from individuals toward improving the quality of the places where they reside may result in new potential solutions to improve outcomes. The potential existence of contextual influences on cancer risk offers a promising avenue for cancer prevention.

Social Networks.  One major way through which SES may have

differential effects on cancer mortality is through effects on social networks. Those with larger personal social networks, defined as the web of social relationships that surround an individual (Berkman & Glass, 2000), have lower cancer mortality (Beasley et  al., 2010; Ellwardt, van Tilburg, Aartsen, Wittek, & Steverink, 2015; C. H. Kroenke et al., 2006, 2013; Rottenberg et  al., 2014)  and specifically breast cancer mortality, since there is almost no work on social ties and cancer outcomes for cancers other than breast cancer. Persons with larger social networks are said to be more socially integrated, and those with smaller networks are more socially isolated. Social isolation may be plausibly hypothesized to influence outcomes through several mechanisms. First, social relationships and interactions influence behaviors and health outcomes by

157

“contagion” (K. Smith & Christakis, 2008; Valente, 2009), through norms (Berkman & Glass, 2000; K. Smith & Christakis, 2008), peer modeling (Bandura, 1986), and via access to “social capital”—​defined as the informational and material resources exchanged through social ties (Moore, Daniel, Paquet, Dube, & Gauvin, 2009). Depending on network structure (Granovetter, 1973), cancer patients may have different levels of access to resources, referrals, advice, opportunities to participate in clinical trials, and knowledge about the side effects and outcomes of treatment. Identification with social network members may increase the likelihood of adopting behaviors (Christakis & Fowler, 2007; Pachucki, Jacques, & Christakis, 2011) through influences on shared behaviors and norms. Social networks may also bring costs or burdens related to social roles that can adversely influence health (C. H. Kroenke et al., 2012). Larger social networks can increase caregiving obligations, disproportionately true among women of lower SES. While potentially rewarding, caregiving can be physically and emotionally demanding, leading to poorer self-​care and worse health outcomes (Cannuscio et al., 2002; Kiecolt-​Glaser et al., 1987; S. Lee, Colditz, Berkman, & Kawachi, 2003; S. Lee, Kawachi, & Grodstein, 2004; Schulz & Beach, 1999). The quality of social relationships also appears to differ by SES; high-​strain relationships, more common in those of low SES, can lead to higher levels of stress. People of different SES also appear to depend differently on social networks (August & Sorkin, 2011; Bureau, 2011). Women with greater education have larger social networks but a greater proportion of distal members. In contrast, women of lower SES are more likely to have smaller but denser networks of people who live in close proximity and have more intense, frequent contact with greater interdependency. In men, their spouses are often the most critical member of their social network influencing their health; marital status has been reliably associated with cancer outcomes (Hanske et al., 2016; Jin et al., 2016; R. L. Shi et al., 2016; van Jaarsveld, Miles, Edwards, & Wardle, 2006; L.  Wang, Wilson, Stewart, & Hollenbeak, 2011). However, men of lower SES are less likely to be married (Ajrouch, Blandon, & Antonucci, 2005; Choi & Marks, 2011; Gomez et  al., 2016; Villingshoj, Ross, Thomsen, & Johansen, 2006). Larger social networks predict more optimal cancer mortality risk factors (C. H. Kroenke et al., 2016) and lower cancer mortality generally, but associations may depend substantially on the quality of, and obligations entailed by, those relationships, as well as sociodemographic factors.

Psychosocial Factors Several mechanisms have been hypothesized through which psychosocial factors (such as stress, depression, and social support) may influence cancer outcomes. First, stress and depression are related to health behaviors that are linked to cancer, including cigarette smoking, excessive alcohol intake, and low levels of physical activity. Cigarettes are a relatively affordable and accessible means of stress reduction, particularly in the context of poverty (K. M. Emmons, 2000). A recent meta-​analysis also found that levels of job strain and social support predicted physical activity (Griep et al., 2015). However, psychosocial factors may also have direct effects on cancer mortality through a variety of biological mechanisms (Lutgendorf & Sood, 2011).

Stress and Depression. Psychosocial factors have been also

hypothesized to directly influence immune function (Spiegel, Sephton, Terr, & Stites, 1998). Acting via the hypothalamic-​pituitary-​adrenal axis in a complex feedback loop between the central nervous (CNS) and immune systems, stress increases cortisol levels, which adversely affects immune function (Segerstrom & Miller, 2004). Frequent cortisol release that occurs with chronic stress may lead to persistently high cortisol levels (Kirschbaum et al., 1995), leading to immune suppression. Because the immune system is involved in the body’s immune surveillance mechanism (e.g., eliminating mutated cells), immune dysfunction may lead to more rapid development of cancer. Stress may also promote cancer through DNA damage, faulty DNA repair, inhibition of apoptosis, effects on endocrine parameters, or somatic mutation

158

Part II: The Magnitude of Cancer

SES is inversely related to the likelihood of obtaining cancer screening (Katz & Hofer, 1994; Katz, Zemencuk, & Hofer, 2000). In their study, Challenges in Meeting Healthy People 2020 Objectives for Cancer-​ Related Preventive Services, National Health Interview Survey, 2008 and 2010, Brown and colleagues noted that only women in the highest income group (with incomes ≥ 400% of the federal poverty level or FPL) surpassed the 2020 Healthy People’s goal of having 81.1% of all women engage in breast cancer screening. Among women whose incomes less than 200% FPL, less than 62% had screened for breast cancer (US Department of Health and Human Services, 2000). As a consequence, the poor often present with cancer at later, less treatable stages. Bradley and colleagues found that persons < 65 years who were insured by Medicaid had a greater risk of late-​stage diagnosis and death for breast, cervical, colon, lung, and prostate cancer than those not insured by Medicaid (Brewster et  al., 2001). People from lower SES backgrounds generally present and are diagnosed at later stages, which explains, in part, their higher mortality rates after cancer (Boscoe et al., 2016; F. Wang, Luo, & McLafferty, 2010). Even holding cancer stage constant, the poor have been shown to have worse survival (Bradley, Given, & Roberts, 2002). A higher prevalence of comorbid conditions may complicate recovery. They have historically had more limited access to healthcare (Bindman et al., 1995; Hargraves, 2002) and have often received less aggressive treatment (Bristow et al., 2013; Mahal et al., 2014; N. Wang et al., 2015). The Affordable Care Act (ACA), signed into law in 2010, expanded health care to 20 million previously uninsured people (Obama, 2016) and held promise in reducing SES disparities in preventive services and treatment (O’Keefe, Meltzer, & Bethea, 2015) particularly through its expansion of affordable Medicaid coverage. However, these advances may be reversed with the repeal of the ACA and phase out of Medicaid financing intended by the current administration and majority Republican congress (Kaplan, 2017). The socioeconomic pattern of colorectal cancer mortality began to cross over (from higher rates in persons of higher SES up to the 1980s, 1.05

1

0.95

0.9

0.85

0.8

0.75

0.7

0.65

8 06

–0

5 20

01

–0

0 20

19

96

–0

5 91

–9

0 19

86

–9

5 19

81

–8

0 19

19

76

–8

5

0.6

–7

informational, and appraisal support (Berkman & Glass, 2000)—​is one means through which social relationships may influence cancer survival. In addition, Sherbourne and Stewart (1991) identified affectionate support and “positive social interaction” in patients with chronic illness. Tangible support could include rides to the hospital, trips to the pharmacy, or provision of healthy meals (Hirschman & Bourjolly, 2005; Woloshin et al., 1997). Network members may provide informational support through referrals to physicians and clinics or alternative types of treatment, or they may buffer stress (Cohen & Wills, 1985) through provision of emotional support. Observational studies of women with breast cancer have typically, though not always, found increased mortality rates among those with low levels of social support, adjusted for stage of disease and treatment factors (Chou, Stewart, Wild, & Bloom, 2012; Goodwin et al., 1996; Maunsell, Brisson, & Deschenes, 1995; Reynolds et al., 1994). However, promising results from early trials (Spiegel, Bloom, Kraemer, & Gottheil, 1989) suggesting that peer support may improve cancer survival in metastatic breast cancer patients have not been borne out by more recent studies (Goodwin et al., 2001; Spiegel et al.,

Access to Healthcare and Screening

71

Social Support.  Social support—​for example, tangible, emotional,

2007). There is little work evaluating other types of social interventions on cancer mortality.

19

(Forlenza & Baum, 2000; Jenkins, Van Houten, & Bovbjerg, 2014). These may be precursors to certain types of cancer such as hormonal (Riley, 1981; Rowse, Weinberg, Bellward, & Emerman, 1992)  and lymphatic cancers (Fox, Goldblatt, & Jones, 1985; Levav et al., 2000). Despite plausible mechanisms linking psychosocial factors to cancer outcomes, the empirical evidence has been decidedly mixed. For example, certain types of job stress have been hypothesized to be linked to increased disease risk, including cancer. According to Karasek (Karasek & Theorell, 1990), jobs that are both high in psychological demands and low in decision-​making authority (or control) are hypothesized to result in job strain. High-​strain jobs are typically over-​represented in lower SES occupations, such as assembly-​line jobs and certain kinds of jobs in the clerical and service sector. In turn, job strain has been shown to raise the likelihood of deleterious behavioral coping responses (such as cigarette smoking, alcohol abuse), thereby increasing the risk of disease outcomes. While the majority of epidemiological studies of job strain and CVD have found support for such an association, the link to cancer still remains elusive. For example, in the Harvard Nurses’ Health Study, Achat and colleagues (Achat, Kawachi, Byrne, Hankinson, & Colditz, 2000)  found no evidence of an association between job strain and breast cancer incidence in a cohort of nurses. In 2007, in a study with data from The Women’s Lifestyle and Health Cohort Study (n = 36,332), Kuper and colleagues (Kuper, Yang, Theorell, & Weiderpass, 2007) found that women with low job control and high job demands (“job strain”) had only a slightly higher risk of breast cancer than women with high job control and low demands (“low strain”) (HR = 1.2; CI: 0.9, 1.6). Though the number of breast cancer cases was relatively small in both studies (n  =  219 in Achat et  al., and n  =  767 in Kuper et  al.), a follow-​up study by Schernhammer and colleagues (Schernhammer et  al., 2003b), with 1,030 breast cancer cases, found an inverse, rather than the expected positive, association between job strain and breast cancer. Consistent with the finding in the Schernhammer study, Kroenke and colleagues also found that high stress from caregiving was related to a possible lower risk of breast cancer (Kroenke et al., 2004). In a meta-​analysis of 116,056 European men and women, job stress was not significantly associated with colorectal, breast, or prostate cancer. It was positively associated with lung cancer in age-​and sex-​adjusted analyses, but was no longer significantly associated after adjustment for cigarette smoking. There was a suggestion of an inverse association of stress with breast and prostate cancer, though these associations were not significant (Heikkila et  al., 2013). Gradus and colleagues found no association between post-​traumatic stress disorder and cancer (Gradus et al., 2015). Equivocal effects of stress on cancer may be due to the result of the combination of behavioral, hormonal, and other biological effects and the mixed result on cancer incidence, or the possible methodological issues inherent with the lack of evaluation of coping in combination with stress, and their influences on outcomes.

Mortality Rate Ratios (MRR)

158

5-year groups Average Diffusion

Diffusion 1SD above average

Diffusion 1SD below average

Figure  9–​4. The modified impact of SES (MMRs) with three different state-​level diffusion speeds, 1971–​2008.

 159



Socioeconomic Disparities in Cancer Incidence and Mortality

then lower rates thereafter) prior to the dissemination of colorectal screening guidelines (G. K. Singh, Miller, & Hankey, 2002), suggesting that other factors besides screening and health insurance status are responsible for the reversal in the SES gradient for this disease (Niu, Roche, Pawlish, & Henry, 2013). Screening for colorectal cancer has increased since the previous version of this chapter (Meissner, Breen, Klabunde, & Vernon, 2006), moreso among those of higher SES, though the SES differential is smaller in states with rapid diffusion of information. Figure  9-​4 shows that states with a higher propensity for diffusion show smaller socioeconomic disparities for colorectal cancer mortality over the 40-​year study period, while those states with a propensity toward slow and moderate rates of diffusion show wider disparities, suggesting that this mechanism may moderate socioeconomic disparities (A. Wang, Clouston, Rubin, Colen, & Link, 2012) (Figure 9–​4).

Biological Influences Related to Aging While there are many biological mechanisms through which socioeconomic disadvantage may influence cancer incidence and mortality, we briefly focus on two areas of research that have mushroomed since the last version of this chapter—​epigenetics and telomeres. Epigenetics is the study of heritable changes in gene expression, without changes to the underlying DNA, that lead to phenotypic changes in living organisms; DNA methylation is believed to provide a stable and comprehensive record of epigenetic influence (Hochberg et al., 2011). Telomeres are nucleoprotein structures and the protective caps at the ends of linear eukaryotic chromosomes that are required for genomic stability (Chiodi & Mondello, 2016; Meena, Rudolph, & Gunes, 2015). We address these areas in particular because of their potential relevance to cumulative socioeconomic disadvantage or “weathering” (Geronimus, 1992) and their potential in future research.

Epigenetics and DNA Methylation. Socioeconomic dis-

advantage may be expressed through epigenetic modifications—​ variations in genetic expression, triggered by changes in DNA methylation, resulting from external and/​ or environmental influences (Bhattacharjee, Shenoy, & Bairy, 2016; Sapienza & Issa, 2016; Trimarchi, Mouangsavanh, & Huang, 2011). Researchers in Canada and the United Kingdom found that socioeconomic position in childhood is associated with DNA methylation differences many years later. Socioeconomic disadvantage in childhood appears to have a greater influence on DNA methylation patterns than socioeconomic disadvantage in adulthood (Borghol et al., 2012). Thus, having a disadvantaged (or advantaged) background may have profound ramifications for health over the life course, even if socioeconomic circumstances improve in adulthood. Socioeconomic disadvantage has been related to DNA methylation of genes involved in inflammation (Needham et al., 2015; Stringhini et al., 2015), a major predictor of tumor progression. Abnormal DNA methylation patterns have been associated with development of various cancers, including head and neck cancers, breast cancer, and prostate cancer (Montenegro et  al., 2016; Nelson et  al., 2007; Shaw et al., 2008; Veeck & Esteller, 2010).

Telomeres.  Low SES in adulthood has been weakly associated

with telomere length (Robertson et al., 2013), though social disadvantage has been linked to shorter telomeres in children (Mitchell et al., 2014). Chronic stress, common among those of low SES, has also been related to telomere attrition (Epel et  al., 2004). Aberrations in telomere maintenance contribute to cancer development (Basu et al., 2013) and have predicted cancers of the breast, gastrointestinal tract, larynx, bladder, and ovary, as well as chronic lymphocytic leukemia and cutaneous melanoma and non-​ melanoma skin cancers (Caini et al., 2015; Kotsopoulos et al., 2014; Shen et al., 2012; Valls-​Bautista, Pinol-​Felis, Rene-​Espinet, Buenestado-​Garcia, & Vinas-​Salas, 2015; R. Wei, DeVilbiss, & Liu, 2015; Willeit, Willeit, Kloss-​Brandstatter, Kronenberg, & Kiechl, 2011; Willeit et  al., 2010). Telomere shortening has been related to cancer progression and earlier mortality (Duggan et  al., 2014; Kammori et  al., 2015; Svenson et  al., 2008;

159

Svenson, Oberg, Stenling, Palmqvist, & Roos, 2016; Weischer et al., 2013; Willeit et al., 2010, 2011; Zhang et al., 2015).

A Note on SES versus Race SES is correlated with race. African-​Americans and Hispanics are disproportionately represented in low SES groups, whether measured by education, income, or occupation. A great deal of work has explored differential cancer outcomes by race, predominantly focusing on differences between blacks and whites. In the absence of SES data in routine sources of data, there has been a tendency to use race as a proxy for SES (Williams, 1997). Though the practice of conflating SES with race/​ethnicity has historically posed a barrier to understanding the etiological relationships of each to cancer outcomes, researchers have more carefully distinguished SES from race in more recent research. For some cancers, SES may entirely “explain” the association of race and cancer. In a study of Detroit women, Bradley and colleagues found, after accounting for SES, that race was not related to stage at presentation or survival after breast cancer diagnosis (Bradley et al., 2002). Also, after controlling for income, black women were as likely, or more likely, than white women to report having had a recent mammogram. Black women from higher SES backgrounds experienced similar rates of breast cancer compared to affluent white women. In other cases, SES has explained part but not all of the racial difference in risk (Bristow et al., 2013; Robbins et al., 2000). Tellingly, race and SES do not always vary in the same direction with respect to cancer incidence and mortality. When incidence rates were stratified by race and census block SES, Krieger and colleagues (1999) found marked heterogeneity in the relationships of SES and race to individual cancer sites. This was also true in a recent analysis in the California Cancer Registry (Yin et al., 2010). The strength of association between SES and cancer may also vary by race. In a study by Yost et al. (Yost, Perkins, Cohen, Morris, & Wright, 2001), SES was positively related to breast cancer incidence, but the effect was stronger for Hispanic and Asian women than it was for whites and blacks. This chapter has focused specifically on the influence of SES on cancer outcomes, and fortunately, since the previous edition of this chapter, there has been a greater tendency, and a greater recognition of the need, to evaluate the independent and joint effects of SES and race/​ethnicity on cancer outcomes. Research that evaluates the cross-​ classification of SES and race, as well as other factors including urban/​ rural status, nativity, and residence in ethnic enclaves (Schupp, Press, & Gomez, 2014), may be the most helpful in trying to understand health disparities.

NEEDED RESEARCH Vice President Joseph Biden called for and President Barack Obama launched a “moonshot” initiative to cure cancer (Conrads & Petricoin, 2016) that would focus on expediting collaborative clinical and laboratory research and encourage information sharing. However, given that lifestyle and environmental factors have a substantial impact on global cancer incidence and mortality, and that most of these risk factors are related to socioeconomic factors, national efforts in cancer prevention could be facilitated by focusing on socioeconomic disparities (Horwitz, 2016) and by evaluating the effects of policies on cancer-​ related health behaviors. Understanding the effects of SES on cancer over the life course will necessitate research on the cumulative impact of SES on outcomes using more complex analytic methodological tools (S. Liu, Jones, & Glymour, 2010). There is a need for research that considers other SES measures such as familial wealth. Furthermore, there is a need to explore more fully the combined effects of multiple sociodemographic factors including SES, urban/​rural status (G. K.  Singh, Williams, Siahpush, & Mulhollen, 2011), race/​ethnicity, and immigration history on outcomes. To understand the ways in which area-​and neighborhood-​level environments affect cancer, more specific research

160

160

Part II: The Magnitude of Cancer

will be needed on social and built-​in environments (Hernandez & Blazer, 2006). Study of environmental factors and cancer outcomes pertinent to SES disparities may also need to begin to consider the implications of the progression of climate change on cancer (Portier et al., 2010).

CONCLUSION The relationship of SES to cancer is complex and dynamic, and continues to evolve. It can be difficult to make generalizations about the relationship, because of the heterogeneity of cancer types and outcome measures, as well as the differences in approaches to measuring SES itself (different indicators, different levels of analysis). Whether the association is causal remains controversial in some cases, even though several plausible pathways have been identified by which SES could causally affect the onset and prognostic course of specific cancers. However, socioeconomic disadvantage is often associated with higher cancer incidence and worse prognosis once diagnosed. SES may be a “fundamental” determinant of health outcomes, and this appears true throughout the cancer spectrum—​from cancer incidence, to detection, treatment, and survival. Importantly, there is evidence that the diffusion of knowledge can mitigate socioeconomic disparities. Investigations of the link between SES and cancer will continue to benefit from more sophisticated approaches to the measurement of SES, including multidimensional measures, measures of familial resources, and more sophisticated evaluation of socioeconomic trajectories over the life course, as well as better specification and tests of etiological mechanisms. Such an exercise is of more than passing academic interest, since any program or policy to reduce the burden of cancer must also pay attention to mitigating the existing socioeconomic disparities in cancer outcomes.

ACKNOWLEDGMENTS We would like to thank Circe Le Compte and Jon Pfefferle for their assistance in updating references. References Achat, H., Kawachi, I., Byrne, C., Hankinson, S., & Colditz, G. (2000). A prospective study of job strain and risk of breast cancer. Int J Epidemiol, 29(4), 622–​628. Adair, L. S., & Gordon-​Larsen, P. (2001). Maturational timing and overweight prevalence in US adolescent girls. Am J Public Health, 91(4), 642–​644. Adler, N. E., Boyce, T., Chesney, M. A., Cohen, S., Folkman, S., Kahn, R. L., & Syme, S. L. (1993). Socioeconomic status and health: the challenge of the gradient. Am Psychol, 49, 15–​24. Adler, N. E., & Rehkopf, D. H. (2008). U.S.  disparities in health:  descriptions, causes, and mechanisms. Annu Rev Public Health, 29, 235–​252. doi:10.1146/​annurev.publhealth.29.020907.090852 Ajrouch, K. J., Blandon, A. Y., & Antonucci, T. C. (2005). Social networks among men and women: the effects of age and socioeconomic status. J Gerontol B Psychol Sci Soc Sci, 60(6), S311–​S317. Akinyemiju, T. F., Soliman, A. S., Copeland, G., Banerjee, M., Schwartz, K., & Merajver, S. D. (2013). Trends in breast cancer stage and mortality in Michigan (1992–​2009) by race, socioeconomic status, and area healthcare resources. PLoS One, 8(4), e61879. doi:10.1371/​journal.pone.0061879 Albano, J. D., Ward, E., Jemal, A., Anderson, R., Cokkinides, V. E., Murray, T., . . . Thun, M. J. (2007). Cancer mortality in the United States by education level and race. J Natl Cancer Inst, 99(18), 1384–​1394. doi:10.1093/​ jnci/​djm127 Alexander, C., Piazza, M., Mekos, D., & Valente, T. (2001). Peers, schools, and adolescent cigarette smoking. J Adolesc Health, 29(1), 22–​30. Altman, D. G., Levine, D. W., Coeytaux, R., Slade, J., & Jaffe, R. (1996). Tobacco promotion and susceptibility to tobacco use among adolescents aged 12 through 17  years in a nationally representative sample. Am J Public Health, 86(11), 1590–​1593. American Cancer Society. (2016). Physical activity and cancer. Retrieved from:  www.cancer.org/​NUPA:  http://​www.cancer.org/​acs/​groups/​content/​@nho/​documents/​document/​physicalactivityandcancerpdf.pdf. Amigou, A., Sermage-​Faure, C., Orsi, L., Leverger, G., Baruchel, A., Bertrand, Y.,  .  .  .  Clavel, J. (2011). Road traffic and childhood leukemia:  the

ESCALE study (SFCE). Environ Health Perspect, 119(4), 566–​572. doi:10.1289/​ehp.1002429 Ater, J. (2015). Adult survivorship of pediatric cancers. In L. Foxhall & M. Rodriguez (Eds.), Advances in cancer survivorship management, MD Anderson cancer care series (pp. 41–​56). New York: Springer. August, K. J., & Sorkin, D. H. (2011). Support and influence in the context of diabetes management:  Do racial/​ethnic differences exist? J Health Psychol, 16(5), 711–​ 721. doi:1359105310388320 [pii] 10.1177/​ 1359105310388320 Bahr, P. R. (2007). Race and nutrition: an investigation of Black-​White differences in health-​related nutritional behaviours. Sociol Health Illn, 29(6), 831–​856. doi:10.1111/​j.1467-​9566.2007.01049.x Bailey, H. D., Infante-​Rivard, C., Metayer, C., Clavel, J., Lightfoot, T., Kaatsch, P., . . . Schuz, J. (2015). Home pesticide exposures and risk of childhood leukemia: findings from the childhood leukemia international consortium. Int J Cancer, 137(11), 2644–​2663. doi:10.1002/​ijc.29631 Bandura, A. (1986). Social foundations of thought and action: a social cognitive theory. Englewood Cliffs, NJ: Prentice-​Hall. Bankole, A., Darroch, J. E., & Singh, S. (1999). Determinants of trends in condom use in the United States, 1988–​1995. Fam Plann Perspect, 31(6), 264–​271. Barrera, M., Shaw, A. K., Speechley, K. N., Maunsell, E., & Pogany, L. (2005). Educational and social late effects of childhood cancer and related clinical, personal, and familial characteristics. Cancer, 104(8), 1751–​1760. doi:10.1002/​cncr.21390 Basu, N., Skinner, H. G., Litzelman, K., Vanderboom, R., Baichoo, E., & Boardman, L. A. (2013). Telomeres and telomere dynamics: relevance to cancers of the GI tract. Expert Rev Gastroenterol Hepatol, 7(8), 733–​748. doi:10.1586/​17474124.2013.848790 Beaglehole, R., & Bonita, R. (1997). Public health at the crossroads: achievements and prospects. Cambridge; New York: Cambridge University Press. Beasley, J. M., Newcomb, P. A., Trentham-​Dietz, A., Hampton, J. M., Ceballos, R. M., Titus-​Ernstoff, L., . . . Holmes, M. D. (2010). Social networks and survival after breast cancer diagnosis. J Cancer Surviv, 4(4), 372–​380. doi:10.1007/​s11764-​010-​0139-​5 Becker, D. M., Yanek, L. R., Koffman, D. M., & Bronner, Y. C. (1999). Body image preferences among urban African Americans and whites from low income communities. Ethn Dis, 9(3), 377–​386. Berkman, L., & Glass, T. A. (2000). Social integration, social networks, social support, and health. In L. Berkman & I. Kawachi (Eds.), Social epidemiology (pp. 137–​173). New York: Oxford University Press. Bhattacharjee, D., Shenoy, S., & Bairy, K. L. (2016). DNA methylation and chromatin remodeling:  the blueprint of cancer epigenetics. Scientifica (Cairo), 2016, 6072357. doi:10.1155/​2016/​6072357 Bindman, A. B., Grumbach, K., Osmond, D., Komaromy, M., Vranizan, K., Lurie, N., . . . Stewart, A. (1995). Preventable hospitalizations and access to health care. JAMA, 274(4), 305–​311. Blaser, M. J. (1999). Where does Helicobacter pylori come from and why is it going away? JAMA, 282(23), 2260–​2262. Blum, R. W., Beuhring, T., Shew, M. L., Bearinger, L. H., Sieving, R. E., & Resnick, M. D. (2000). The effects of race/​ethnicity, income, and family structure on adolescent risk behaviors. Am J Public Health, 90(12), 1879–​1884. Boffetta, P., Hashibe, M., La Vecchia, C., Zatonski, W., & Rehm, J. (2006). The burden of cancer attributable to alcohol drinking. Int J Cancer, 119(4), 884–​887. doi:10.1002/​ijc.21903 Boffetta, P., Tredaniel, J., & Greco, A. (2000). Risk of childhood cancer and adult lung cancer after childhood exposure to passive smoke:  a meta-​ analysis. Environ Health Perspect, 108(1), 73–​82. Boggild, H., Suadicani, P., Hein, H. O., & Gyntelberg, F. (1999). Shift work, social class, and ischaemic heart disease in middle aged and elderly men; a 22 year follow up in the Copenhagen Male Study. Occup Environ Med, 56(9), 640–​645. Borghol, N., Suderman, M., McArdle, W., Racine, A., Hallett, M., Pembrey, M.,  .  .  .  Szyf, M. (2012). Associations with early-​life socio-​economic position in adult DNA methylation. Int J Epidemiol, 41(1), 62–​74. doi:10.1093/​ije/​dyr147 Boscoe, F. P., Henry, K. A., Sherman, R. L., & Johnson, C. J. (2016). The relationship between cancer incidence, stage, and poverty in the United States. Int J Cancer, 139(3), 607–​612. doi:10.1002/​ijc.30087 Bradley, C. J., Given, C. W., & Roberts, C. (2002). Race, socioeconomic status, and breast cancer treatment and survival. J Natl Cancer Inst, 94(7), 490–​496. Braveman, P. (2011). Accumulating knowledge on the social determinants of health and infectious disease. Public Health Rep, 126(Suppl 3), 28–​30. Braveman, P. A., Cubbin, C., Egerter, S., Chideya, S., Marchi, K. S., Metzler, M., & Posner, S. (2005). Socioeconomic status in health research:  one size does not fit all. JAMA, 294(22), 2879–​2888.

 16



Socioeconomic Disparities in Cancer Incidence and Mortality

Brewster, D. H., Thomson, C. S., Hole, D. J., Black, R. J., Stroner, P. L., & Gillis, C. R. (2001). Relation between socioeconomic status and tumour stage in patients with breast, colorectal, ovarian, and lung cancer: results from four national, population based studies. BMJ, 322(7290), 830–​831. Bristow, R. E., Powell, M. A., Al-​Hammadi, N., Chen, L., Miller, J. P., Roland, P. Y., . . . Cliby, W. A. (2013). Disparities in ovarian cancer care quality and survival according to race and socioeconomic status. J Natl Cancer Inst, 105(11), 823–​832. doi:10.1093/​jnci/​djt065 Bureau, U. S. C. (2011). US Census Bureau, 2005–​2009 American Community Survey. Retrieved from: http://​factfinder.census.gov/​servlet/​STTable?_​ bm=y&-​geo_​id=01000US&-​qr_​name=ACS_​2009_​5YR_​G00_​S1201&-​ ds_​name=ACS_​2009_​5YR_​G00_​. Byers, T. E., Wolf, H. J., Bauer, K. R., Bolick-​Aldrich, S., Chen, V. W., Finch, J. L., . . . Patterns of Care Study, G. (2008). The impact of socioeconomic status on survival after cancer in the United States:  findings from the National Program of Cancer Registries Patterns of Care Study. Cancer, 113(3), 582–​591. doi:10.1002/​cncr.23567 Caini, S., Raimondi, S., Johansson, H., De Giorgi, V., Zanna, I., Palli, D., & Gandini, S. (2015). Telomere length and the risk of cutaneous melanoma and non-​melanoma skin cancer:  a review of the literature and meta-​analysis. J Dermatol Sci, 80(3), 168–​174. doi:10.1016/​ j.jdermsci.2015.08.003 Cannuscio, C. C., Jones, C., Kawachi, I., Colditz, G. A., Berkman, L., & Rimm, E. (2002). Reverberations of family illness: a longitudinal assessment of informal caregiving and mental health status in the Nurses’ Health Study. Am J Public Health, 92(8), 1305–​1311. Carroll Chapman, S. L., & Wu, L. T. (2014). E-​cigarette prevalence and correlates of use among adolescents versus adults: a review and comparison. J Psychiatr Res, 54, 43–​54. doi:10.1016/​j.jpsychires.2014.03.005 Castellsague, X., Bosch, F. X., Munoz, N., Meijer, C. J., Shah, K. V., de Sanjose, S., . . . Franceschi, S. (2002). Male circumcision, penile human papillomavirus infection, and cervical cancer in female partners. N Engl J Med, 346(15), 1105–​1112. Centers for Disease Control and Prevention. (1997). Update:  prevalence of overweight among children, adolescents, and adults—​ United States, 1988–​1994. MMWR Morb Mortal Wkly Rep, 46, 199–​202. Centers for Disease Control and Prevention. (2000). Behavioral Risk Factor Surveillance System:  Prevalence data:  Nationwide—​ 2000 nutrition. Retrieved from: http://​apps.nccd.cdc.gov/​brfss/​income.asp?cat=NU&yr= 2000&qkey=4340&state=US. Chandon, P., & Wansink, B. (2012). Does food marketing need to make us fat? A  review and solutions. Nutr Rev, 70(10), 571–​593. doi:10.1111/​ j.1753-​4887.2012.00518.x Chang, C. M., Su, Y. C., Lai, N. S., Huang, K. Y., Chien, S. H., Chang, Y. H., . . . Lee, C. C. (2012). The combined effect of individual and neighborhood socioeconomic status on cancer survival rates. PLoS One, 7(8), e44325. doi:10.1371/​journal.pone.0044325 PONE-​D-​12-​15961 [pii] Chang, E. T., Yang, J., Alfaro-​Velcamp, T., So, S. K., Glaser, S. L., & Gomez, S. L. (2010). Disparities in liver cancer incidence by nativity, acculturation, and socioeconomic status in California Hispanics and Asians. Cancer Epidemiol Biomarkers Prev, 19(12), 3106–​3118. doi:10.1158/​1055-​9965. EPI-​10-​0863 Chelimo, C., Wouldes, T. A., Cameron, L. D., & Elwood, J. M. (2013). Risk factors for and prevention of human papillomaviruses (HPV), genital warts and cervical cancer. J Infect, 66(3), 207–​217. doi:10.1016/​ j.jinf.2012.10.024 Chiodi, I., & Mondello, C. (2016). Telomere and telomerase stability in human diseases and cancer. Front Biosci (Landmark Ed), 21, 203–​224. Choi, H., & Marks, N. F. (2011). Socioeconomic status, marital status continuity and change, marital conflict, and mortality. J Aging Health, 23(4), 714–​742. doi:10.1177/​0898264310393339 Chou, A. F., Stewart, S. L., Wild, R. C., & Bloom, J. R. (2012). Social support and survival in young women with breast carcinoma. Psychooncology, 21(2), 125–​133. doi:10.1002/​pon.1863 Christakis, N. A., & Fowler, J. H. (2007). The spread of obesity in a large social network over 32  years. N Engl J Med, 357(4), 370–​379. doi:NEJMsa066082 [pii] 10.1056/​NEJMsa066082 Chuang, S. C., Gallo, V., Michaud, D., Overvad, K., Tjonneland, A., Clavel-​ Chapelon, F., . . . Vineis, P. (2011). Exposure to environmental tobacco smoke in childhood and incidence of cancer in adulthood in never smokers in the European Prospective Investigation into Cancer and Nutrition. Cancer Causes Control, 22(3), 487–​494. doi:10.1007/​s10552-​010-​9723-​2 Churilla, T., Egleston, B., Dong, Y., Shaikh, T., Murphy, C., Mantia-​Smaldone, G., . . . Anderson, P. (2016). Disparities in the management and outcome of cervical cancer in the United States according to health insurance status. Gynecol Oncol, 141(3), 516–​523. doi:10.1016/​j.ygyno.2016.03.025

161

Clague, J., & Bernstein, L. (2012). Physical activity and cancer. Curr Oncol Rep, 14(6), 550–​558. doi:10.1007/​s11912-​012-​0265-​5 Clegg, L. X., Reichman, M. E., Miller, B. A., Hankey, B. F., Singh, G. K., Lin, Y. D.,  .  .  .  Edwards, B. K. (2009). Impact of socioeconomic status on cancer incidence and stage at diagnosis:  selected findings from the surveillance, epidemiology, and end results:  National Longitudinal Mortality Study. Cancer Causes Control, 20(4), 417–​435. doi:10.1007/​ s10552-​008-​9256-​0 Cohen, S., & Wills, T. A. (1985). Stress, social support, and the buffering hypothesis. Psychol Bull, 98(2), 310–​357. Coker, A. L., Richter, D. L., Valois, R. F., McKeown, R. E., Garrison, C. Z., & Vincent, M. L. (1994). Correlates and consequences of early initiation of sexual intercourse. J Sch Health, 64(9), 372–​377. Collaborative Group on Hormonal Factors in Breast Cancer (2012). Menarche, menopause, and breast cancer risk:  individual participant meta-​ analysis, including 118 964 women with breast cancer from 117 epidemiological studies. Lancet Oncol, 13(11), 1141–​1151. doi:10.1016/​ S1470-​2045(12)70425-​4 Conrads, T. P., & Petricoin, E. F. (2016). The Obama administration’s cancer moonshot:  a call for proteomics. Clin Cancer Res. doi:10.1158/​1078-​ 0432.CCR-​16-​0688 Cook, M. B., Gamborg, M., Aarestrup, J., Sorensen, T. I., & Baker, J. L. (2013). Childhood height and birth weight in relation to future prostate cancer risk: a cohort study based on the copenhagen school health records register. Cancer Epidemiol Biomarkers Prev, 22(12), 2232–​2240. doi:10.1158/​ 1055-​9965.EPI-​13-​0712 Correa, P., & Piazuelo, M. B. (2011). Helicobacter pylori infection and gastric adenocarcinoma. US Gastroenterol Hepatol Rev, 7(1), 59–​64. Costa, L. G., Pellacani, C., Dao, K., Kavanagh, T. J., & Roque, P. J. (2015). The brominated flame retardant BDE-​47 causes oxidative stress and apoptotic cell death in vitro and in vivo in mice. Neurotoxicology, 48, 68–​76. doi:10.1016/​j.neuro.2015.03.008 Crawford, P. B., Story, M., Wang, M. C., Ritchie, L. D., & Sabry, Z. I. (2001). Ethnic issues in the epidemiology of childhood obesity. Pediatr Clin North Am, 48(4), 855–​878. Currie, J., & Hyson, R. (1999). Is the impact of health shocks cushioned by socioeconomic status? The case of low birthweight. Am Econ Rev, 89, 245–​250. Curtin, F., Morabia, A., & Bernstein, M. (1998). Lifetime exposure to environmental tobacco smoke among urban women: differences by socioeconomic class. Am J Epidemiol, 148(11), 1040–​1047. Darmon, N., & Drewnowski, A. (2008). Does social class predict diet quality? Am J Clin Nutr, 87(5), 1107–​1117. Darmon, N., & Drewnowski, A. (2015). Contribution of food prices and diet cost to socioeconomic disparities in diet quality and health:  a systematic review and analysis. Nutr Rev, 73(10), 643–​660. doi:10.1093/​nutrit/​ nuv027 Davey-​ Smith, G., Gunnell, D., & Ben-​ Shlomo, Y. (2001). Life-​ course approaches to socio-​economic differentials in cause-​specific adult mortality. In D. Leon & G. Walt (Eds.), Poverty, inequality and health:  an international perspective (pp. 88–​124). Oxford: Oxford University Press. De Flora, S., & Bonanni, P. (2011). The prevention of infection-​associated cancers. Carcinogenesis, 32(6), 787–​795. doi:10.1093/​carcin/​bgr054 De Flora, S., Crocetti, E., Bonanni, P., Ferro, A., Vitale, F., Vaccines, . . . Public, H. (2015). Incidence of infection-​associated cancers in Italy and prevention strategies. Epidemiol Prev, 39(4 Suppl 1), 14–​20. Denniston, M. M., Jiles, R. B., Drobeniuc, J., Klevens, R. M., Ward, J. W., McQuillan, G. M., & Holmberg, S. D. (2014). Chronic hepatitis C virus infection in the United States, National Health and Nutrition Examination Survey 2003 to 2010. Ann Intern Med, 160(5), 293–​300. doi:10.7326/​ M13-​1133 Dietz, W. H. (1998a). Childhood weight affects adult morbidity and mortality. J Nutr, 128(2 Suppl), 411S–​414S. Dietz, W. H. (1998b). Health consequences of obesity in youth: childhood predictors of adult disease. Pediatrics, 101(3 Pt 2), 518–​525. Diez-​Roux, A. V., Nieto, F. J., Caulfield, L., Tyroler, H. A., Watson, R. L., & Szklo, M. (1999). Neighbourhood differences in diet: the Atherosclerosis Risk in Communities (ARIC) Study. J Epidemiol Community Health, 53(1), 55–​63. DiMartino, L. D., Birken, S. A., & Mayer, D. K. (2016). The relationship between cancer survivors’ socioeconomic status and reports of follow-​ up care discussions with providers. J Cancer Educ. doi:10.1007/​ s13187-​016-​1024-​3 Doubeni, C. A., Laiyemo, A. O., Major, J. M., Schootman, M., Lian, M., Park, Y., . . . Sinha, R. (2012). Socioeconomic status and the risk of colorectal cancer: an analysis of more than a half million adults in the National

162

162

Part II: The Magnitude of Cancer

Institutes of Health-​ AARP Diet and Health Study. Cancer, 118(14), 3636–​3644. doi:10.1002/​cncr.26677 Doubeni, C. A., Schootman, M., Major, J. M., Stone, R. A., Laiyemo, A. O., Park, Y.,  .  .  .  Schatzkin, A. (2012). Health status, neighborhood socioeconomic context, and premature mortality in the United States:  The National Institutes of Health-​AARP Diet and Health Study. Am J Public Health, 102(4), 680–​688. doi:10.2105/​AJPH.2011.300158 Du, X. L., Fang, S., & Meyer, T. E. (2008). Impact of treatment and socioeconomic status on racial disparities in survival among older women with breast cancer. Am J Clin Oncol, 31(2), 125–​ 132. doi:10.1097/​ COC.0b013e3181587890 Du, X. L., Lin, C. C., Johnson, N. J., & Altekruse, S. (2011). Effects of individual-​level socioeconomic factors on racial disparities in cancer treatment and survival: findings from the National Longitudinal Mortality Study, 1979–​2003. Cancer, 117(14), 3242–​3251. doi:10.1002/​cncr.25854 Duggan, C., Risques, R., Alfano, C., Prunkard, D., Imayama, I., Holte, S., . . . McTiernan, A. (2014). Change in peripheral blood leukocyte telomere length and mortality in breast cancer survivors. J Natl Cancer Inst, 106(4), dju035. doi:10.1093/​jnci/​dju035 Eckersley, R., Dixon, J., & Douglas, B. (2001). The social origins of health and well-​being. Cambridge: Cambridge University Press. Eckhart, W. (1998). Viruses and human cancer. Sci Prog, 81(Pt 4), 315–​328. Elder, G. H., Clipp, E. C., Brown, J. S., Martin, L. R., & Friedman, H. W. (2009). The life-​long mortality risks of World War II experiences. Res Aging, 31(4), 391–​412. doi:10.1177/​0164027509333447 Elder, G. H., Jr., Shanahan, M. J., & Clipp, E. C. (1994). When war comes to men’s lives: life-​course patterns in family, work, and health. Psychol Aging, 9(1), 5–​16. Elks, C. E., Loos, R. J., Hardy, R., Wills, A. K., Wong, A., Wareham, N. J., . . . Ong, K. K. (2012). Adult obesity susceptibility variants are associated with greater childhood weight gain and a faster tempo of growth: the 1946 British Birth Cohort Study. Am J Clin Nutr, 95(5), 1150–​1156. doi:10.3945/​ajcn.111.027870 Ellwardt, L., van Tilburg, T., Aartsen, M., Wittek, R., & Steverink, N. (2015). Personal networks and mortality risk in older adults: a twenty-​year longitudinal study. PLoS One, 10(3), e0116731. doi:10.1371/​journal.pone.0116731 Emmons, K. M. (2000). Health behaviors in a social context. In L. F. Berkman & I. Kawachi (Eds.), Social epidemiology (pp. 242–​266). New York: Oxford University Press. Emmons, K. M., Wong, M., Hammond, S. K., Velicer, W. F., Fava, J. L., Monroe, A. D., & Evans, J. L. (2001). Intervention and policy issues related to children’s exposure to environmental tobacco smoke. Prev Med, 32(4), 321–​331. doi:10.1006/​pmed.2000.0822 Enewold, L., Horner, M. J., Shriver, C. D., & Zhu, K. (2014). Socioeconomic disparities in colorectal cancer mortality in the United States, 1990–​2007. J Community Health, 39(4), 760–​766. doi:10.1007/​s10900-​014-​9824-​z Epel, E. S., Blackburn, E. H., Lin, J., Dhabhar, F. S., Adler, N. E., Morrow, J. D., & Cawthon, R. M. (2004). Accelerated telomere shortening in response to life stress. Proc Natl Acad Sci U S A, 101(49), 17312–​17315. doi:10.1073/​pnas.0407162101 Esser, M. B., Hedden, S. L., Kanny, D., Brewer, R. D., Gfroerer, J. C., & Naimi, T. S. (2014). Prevalence of alcohol dependence among US adult drinkers, 2009–​2011. Prev Chronic Dis, 11, E206. doi:10.5888/​pcd11.140329 Evans, A. S., & Mueller, N. E. (1997). Malignant lymphomas. In A. S. Evans & R. Kaslow (Eds.), Viral infections of humans: epidemiology and control (4th ed.). New York: Plenum Publishers. Faggiano, F., Partanen, T., Kogevinas, M., & Boffetta, P. (1997). Socioeconomic differences in cancer incidence and mortality. IARC Sci Publ, 138, 65–​176. Farrell, P., & Fuchs, V. R. (1982). Schooling and health: the cigarette connection. J Health Econ, 1(3), 217–​230. Federal Trade Commission. (1999). Report to Congress for 1997 Pursuant to the Federal Cigarette Labeling and Advertising Act. Retrieved from: https://​ www.ftc.gov/​sites/​default/​files/​documents/​reports/​1999-​report-​cigarette-​ sales-​advertising-​and-​promotion-​covering-​1997-​section-​iii.recommendation-​regarding-​ftc-​cigarette-​test-​method/​1997cigarettereport.pdf. Filippini, T., Heck, J. E., Malagoli, C., Del Giovane, C., & Vinceti, M. (2015). A review and meta-​analysis of outdoor air pollution and risk of childhood leukemia. J Environ Sci Health C Environ Carcinog Ecotoxicol Rev, 33(1), 36–​66. doi:10.1080/​10590501.2015.1002999 Flegal, K. M., Carroll, M. D., Kit, B. K., & Ogden, C. L. (2012). Prevalence of obesity and trends in the distribution of body mass index among US adults, 1999–​2010. JAMA, 307(5), 491–​497. doi:10.1001/​jama.2012.39 Flegal, K. M., Carroll, M. D., Ogden, C. L., & Curtin, L. R. (2010). Prevalence and trends in obesity among US adults, 1999–​2008. JAMA, 303(3), 235–​ 241. doi:10.1001/​jama.2009.2014

Forlenza, M. J., & Baum, A. (2000). Psychosocial influences on cancer progression:  alternative cellular and molecular mechanisms. Curr Opin Psychiatry, 13(6), 639–​645. Fox, A. J., Goldblatt, P. O., & Jones, D. R. (1985). Social class mortality differentials: artefact, selection or life circumstances? J Epidemiol Community Health, 39(1), 1–​8. Franck, C., Grandi, S. M., & Eisenberg, M. J. (2013). Agricultural subsidies and the American obesity epidemic. Am J Prev Med, 45(3), 327–​333. doi:10.1016/​j.amepre.2013.04.010 Freese, J., & Lutfey, K. (2011). Fundamental causality: challenges of an animating concept for medical sociology. In B. A. Pescosolido, J. Martin, J. McLeod, & A. Rogers (Eds.), Handbook of the sociology of health, illness, and healing:  a blueprint for the 21st cenutry (pp. 67–​ 81). New York: Springer. Friedenreich, C. M., & Orenstein, M. R. (2002). Physical activity and cancer prevention: etiologic evidence and biological mechanisms. J Nutr, 132(11 Suppl), 3456S–​3464S. Fry, R. (2015). More millennials living with parents despite improved job market. Retrieved from http://​www.pewsocialtrends.org/​2015/​07/​29/​ more-​millennials-​living-​with-​family-​despite-​improved-​job-​market/​ Fuchs, V. R. (1983). How we live: an economic perspective on Americans from birth to death. Cambridge, MA: Harvard University Press. Gepner, Y., Golan, R., Harman-​ Boehm, I., Henkin, Y., Schwarzfuchs, D., Shelef, I., . . . Shai, I. (2015). Effects of initiating moderate alcohol intake on cardiometabolic risk in adults with type 2 diabetes: a 2-​year randomized, controlled trial. Ann Intern Med, 163(8), 569–​579. doi:10.7326/​ M14-​1650 Geronimus, A. T. (1992). The weathering hypothesis and the health of African-​ American women and infants: evidence and speculations. Ethn Dis, 2(3), 207–​221. Giskes, K., Turrell, G., Patterson, C., & Newman, B. (2002). Socioeconomic differences among Australian adults in consumption of fruit and vegetables and intakes of vitamins A, C and folate. J Hum Nutr Diet, 15(5), 375–​385; discussion 387–​390. Glymour, M., Avendano, M., & Kawachi, I. (2014). Socioeconomic status and health. In L. Berkman, I. Kawachi, & M. Glymour (Eds.), Social epidemiology, 2nd ed. Oxford: Oxford University Press. Gomez, S. L., Hurley, S., Canchola, A. J., Keegan, T. H., Cheng, I., Murphy, J. D., . . . Martinez, M. E. (2016). Effects of marital status and economic resources on survival after cancer:  a population-​based study. Cancer, 122(10), 1618–​1625. doi:10.1002/​cncr.29885 Gomez, S. L., Shariff-​Marco, S., DeRouen, M., Keegan, T. H., Yen, I. H., Mujahid, M.,  .  .  .  Glaser, S. L. (2015). The impact of neighborhood social and built environment factors across the cancer continuum: current research, methodological considerations, and future directions. Cancer, 121(14), 2314–​2330. doi:10.1002/​cncr.29345 Goodman, K. J., & Correa, P. (1995). The transmission of Helicobacter pylori:  a critical review of the evidence. Int J Epidemiol, 24(5), 875–​887. Goodwin, P. J., Leszcz, M., Ennis, M., Koopmans, J., Vincent, L., Guther, H.,  .  .  .  Hunter, J. (2001). The effect of group psychosocial support on survival in metastatic breast cancer. N Engl J Med, 345(24), 1719–​1726. doi:10.1056/​NEJMoa011871 Goodwin, P. J., Leszcz, M., Koopmans, J., Arnold, A., Doll, R., Chochinov, H.,  .  .  .  Pritchard, K. I. (1996). Randomized trial of group psychosocial support in metastatic breast cancer:  the BEST study. Breast-​ Expressive Supportive Therapy study. Cancer Treat Rev, 22(Suppl A), 91–​96. Gottlieb, L., Sandel, M., & Adler, N. E. (2013). Collecting and applying data on social determinants of health in health care settings. JAMA Intern Med, 173(11), 1017–​1020. doi:10.1001/​jamainternmed.2013.560 Gradus, J. L., Farkas, D. K., Svensson, E., Ehrenstein, V., Lash, T. L., Milstein, A.,  .  .  .  Sorensen, H. T. (2015). Posttraumatic stress disorder and cancer risk:  a nationwide cohort study. Eur J Epidemiol, 30(7), 563–​568. doi:10.1007/​s10654-​015-​0032-​7 Granovetter, M. (1973). The strength of weak ties. Am J Sociol, 78(6), 1360–​1380. Greenland, S. (2000). An introduction to instrumental variables for epidemiologists. Int J Epidemiol, 29(4), 722–​729. Griep, R. H., Nobre, A. A., Alves, M. G., da Fonseca Mde, J., Cardoso Lde, O., Giatti, L., . . . Chor, D. (2015). Job strain and unhealthy lifestyle: results from the baseline cohort study, Brazilian Longitudinal Study of Adult Health (ELSA-​Brasil). BMC Public Health, 15, 309. doi:10.1186/​ s12889-​015-​1626-​4 Griesler, P. C., & Kandel, D. B. (1998). Ethnic differences in correlates of adolescent cigarette smoking. J Adolesc Health, 23(3), 167–​180.

 163



Socioeconomic Disparities in Cancer Incidence and Mortality

Grossman, M. (1975). The correlation between health and schooling. In N. E. Terleckyj (Ed.), Household production and consumption (pp. 147–​223). New York: Columbia University Press. Grusky, D. B. (1994). Social stratification in sociological perspective. Boulder, CO: Westview Press. Gupta, S., Wilejto, M., Pole, J. D., Guttmann, A., & Sung, L. (2014). Low socioeconomic status is associated with worse survival in children with cancer: a systematic review. PLoS One, 9(2), e89482. doi:10.1371/​journal.pone.0089482 PONE-​D-​13-​42635 [pii] Gustavsson, P., Nyberg, F., Pershagen, G., Scheele, P., Jakobsson, R., & Plato, N. (2002). Low-​dose exposure to asbestos and lung cancer: dose-​ response relations and interaction with smoking in a population-​based case-​referent study in Stockholm, Sweden. Am J Epidemiol, 155(11), 1016–​1022. Gutensohn, N. M. (1982). Social class and age at diagnosis of Hodgkin’s disease:  new epidemiologic evidence for the “two-​disease hypothesis.” Cancer Treat Rep, 66(4), 689–​695. Hanske, J., Meyer, C. P., Sammon, J. D., Choueiri, T. K., Menon, M., Lipsitz, S. R., . . . Trinh, Q. D. (2016). The influence of marital status on the use of breast, cervical, and colorectal cancer screening. Prev Med, 89, 140–​145. doi:10.1016/​j.ypmed.2016.05.017 Hargraves, J. L. (2002). The Insurance Gap and Minority Health Care, 1997–​2001: Center for Studying Health Systems Change Data Bulletin (Tracking report #2). Retrieved from:  http://​www.hschange.org/​ CONTENT/​443/​443.pdf. Harling, G., Subramanian, S., Barnighausen, T., & Kawachi, I. (2013). Socioeconomic disparities in sexually transmitted infections among young adults in the United States:  examining the interaction between income and race/​ethnicity. Sex Transm Dis, 40(7), 575–​581. doi:10.1097/​ OLQ.0b013e31829529cf Harnack, L. J., Jeffery, R. W., & Boutelle, K. N. (2000). Temporal trends in energy intake in the United States:  an ecologic perspective. Am J Clin Nutr, 71(6), 1478–​1484. Harper, S., Lynch, J., Meersman, S. C., Breen, N., Davis, W. W., & Reichman, M. C. (2009). Trends in area-​socioeconomic and race-​ethnic disparities in breast cancer incidence, stage at diagnosis, screening, mortality, and survival among women ages 50  years and over (1987–​2005). Cancer Epidemiol Biomarkers Prev, 18(1), 121–​131. doi:10.1158/​1055-​9965. epi-​08-​0679 Harris, J. L., Pomeranz, J. L., Lobstein, T., & Brownell, K. D. (2009). A crisis in the marketplace:  how food marketing contributes to childhood obesity and what can be done. Annu Rev Public Health, 30, 211–​225. doi:10.1146/​annurev.publhealth.031308.100304 Hashibe, M., Brennan, P., Chuang, S. C., Boccia, S., Castellsague, X., Chen, C.,  .  .  .  Boffetta, P. (2009). Interaction between tobacco and alcohol use and the risk of head and neck cancer:  pooled analysis in the International Head and Neck Cancer Epidemiology Consortium. Cancer Epidemiol Biomarkers Prev, 18(2), 541–​ 550. doi:10.1158/​ 1055-​9965.epi-​08-​0347 Hastert, T. A., Beresford, S. A., Sheppard, L., & White, E. (2015). Disparities in cancer incidence and mortality by area-​level socioeconomic status: a multilevel analysis. J Epidemiol Community Health, 69(2), 168–​176. doi:jech-​2014-​204417 [pii] 10.1136/​jech-​2014-​204417 Heikkila, K., Nyberg, S. T., Theorell, T., Fransson, E. I., Alfredsson, L., Bjorner, J. B., . . . Consortium, I. P.-​W. (2013). Work stress and risk of cancer: meta-​analysis of 5700 incident cancer events in 116,000 European men and women. BMJ, 346, f165. doi:10.1136/​bmj.f165 Heini, A. F., & Weinsier, R. L. (1997). Divergent trends in obesity and fat intake patterns: the American paradox. Am J Med, 102(3), 259–​264. Hernandez, L. M., & Blazer, D. G. (2006). The impact of social and cultural environment on health. In Genes, behavior, and the social environment:  moving beyond the nature/​nurture debate (p. xv). Washington, DC: National Academies Press. Hersh, A. L., Stefanick, M. L., & Stafford, R. S. (2004). National use of postmenopausal hormone therapy: annual trends and response to recent evidence. JAMA, 291(1), 47–​53. doi:10.1001/​jama.291.1.47 Hirschman, K. B., & Bourjolly, J. N. (2005). How do tangible supports impact the breast cancer experience? Soc Work Health Care, 41(1), 17–​32. doi:10.1300/​J010v41n01_​02 Hiza, H. A., Casavale, K. O., Guenther, P. M., & Davis, C. A. (2013). Diet quality of Americans differs by age, sex, race/​ethnicity, income, and education level. J Acad Nutr Diet, 113(2), 297–​306. doi:10.1016/​j.jand.2012.08.011 Hochberg, Z., Feil, R., Constancia, M., Fraga, M., Junien, C., Carel, J. C.,  .  .  .  Albertsson-​Wikland, K. (2011). Child health, developmental plasticity, and epigenetic programming. Endocr Rev, 32(2), 159–​224. doi:10.1210/​er.2009-​0039

163

Hogan, D. P., Sun, R., & Cornwell, G. T. (2000). Sexual and fertility behaviors of American females aged 15–​19 years: 1985, 1990, and 1995. Am J Public Health, 90(9), 1421–​1425. Hogue, C. J. (2011). Environmental health disparities in children. Rev Environ Health, 26(3), 139–​140. Horwitz, R. I. (2016). Equity in cancer care and outcomes of treatment: a different type of cancer moonshot. JAMA, 315(12), 1231–​1232. doi:10.1001/​ jama.2016.2242 Institute of Medicine. (2011). Institute of Medicine Committee on Public Health Strategies to Improve Health (9780309161275 9780309161282). Retrieved from: http://​www.ncbi.nlm.nih.gov/​pubmed/​24983050. Kindig, D. A., Panzer, A. M., & Nielsen-​Bohlman, L. (2004). Health literacy: a prescription to end confusion. National Academies Press. Washington, D.C. Retrieved from: https://​www.nap.edu/​read/​10883/​chapter/​1 International Agency for Research on Cancer Ad Hoc Working Group. (1994). Schistosomes, liver flukes and Helicobacter pylori. IARC Monogr Eval Carcinog Risks Humans, 61. Jenkins, F. J., Van Houten, B., & Bovbjerg, D. H. (2014). Effects on DNA damage and/​or repair processes as biological mechanisms linking psychological stress to cancer risk. J Appl Biobehav Res, 19(1), 3–​23. doi:10.1111/​ jabr.12019 Jin, J. J., Wang, W., Dai, F. X., Long, Z. W., Cai, H., Liu, X. W., . . . Wang, Y. N. (2016). Marital status and survival in patients with gastric cancer. Cancer Med, 5(8), 1821–​1829. doi:10.1002/​cam4.758 Jones, L., Bates, G., McCoy, E., & Bellis, M. A. (2015). Relationship between alcohol-​attributable disease and socioeconomic status, and the role of alcohol consumption in this relationship: a systematic review and meta-​ analysis. BMC Public Health, 15, 400. doi:10.1186/​s12889-​015-​1720-​7 Kamimoto, L. A., Easton, A. N., Maurice, E., Husten, C. G., & Macera, C. A. (1999). Surveillance for five health risks among older adults: United States, 1993–​1997. MMWR CDC Surveill Summ, 48(8), 89–​130. Kammori, M., Sugishita, Y., Okamoto, T., Kobayashi, M., Yamazaki, K., Yamada, E., & Yamada, T. (2015). Telomere shortening in breast cancer correlates with the pathological features of tumor progression. Oncol Rep, 34(2), 627–​632. doi:10.3892/​or.2015.4063 Kanjilal, S., Gregg, E. W., Cheng, Y. J., Zhang, P., Nelson, D. E., Mensah, G., & Beckles, G. L. (2006). Socioeconomic status and trends in disparities in 4 major risk factors for cardiovascular disease among US adults, 1971–​2002. Arch Intern Med, 166(21), 2348–​2355. doi:10.1001/​ archinte.166.21.2348 Kaplan, T. (2017). Abby Goodnough and Jennifer Steinhauer. March 9, 2017. G.O.P. Health Care Bill Moves Closer After Vote at 4 A.M. New  York Times https://​www.nytimes.com/​2017/​03/​09/​us/​politics/​health-​bill-​ clears-​ h ouse- ​p anel-​ i n-​ p re- ​ d awn-​ h ours.html?hp&action=click&pgty pe=Homepage&clickSource=story- ​ h eading&module=first-​ c olumn-​ region®ion=top-​news&WT.nav=top-​news Karasek, R., & Theorell, T. (1990). Healthy work. New York: Basic Books. Katz, S. J., & Hofer, T. P. (1994). Socioeconomic disparities in preventive care persist despite universal coverage: breast and cervical cancer screening in Ontario and the United States. JAMA, 272(7), 530–​534. Katz, S. J., Zemencuk, J. K., & Hofer, T. P. (2000). Breast cancer screening in the United States and Canada, 1994: socioeconomic gradients persist. Am J Public Health, 90(5), 799–​803. Kauppinen, T., Teschke, K., Savela, A., Kogevinas, M., & Boffetta, P. (1997). International data base of exposure measurements in the pulp, paper and paper product industries. Int Arch Occup Environ Health, 70(2), 119–​127. Kawachi, I., & Berkman, L. F. (2003). Neighborhoods and health. New York: Oxford University Press. Kent, E. E., Breen, N., Lewis, D. R., de Moor, J. S., Smith, A. W., & Seibel, N. L. (2015). US trends in survival disparities among adolescents and young adults with non-​ Hodgkin lymphoma. Cancer Causes Control, 26(8), 1153–​1162. doi:10.1007/​s10552-​015-​0609-​1 Kiecolt-​Glaser, J. K., Glaser, R., Shuttleworth, E. C., Dyer, C. S., Ogrocki, P., & Speicher, C. E. (1987). Chronic stress and immunity in family caregivers of Alzheimer’s disease victims. Psychosom Med, 49(5), 523–​535. Kinsey, T., Jemal, A., Liff, J., Ward, E., & Thun, M. (2008). Secular trends in mortality from common cancers in the United States by educational attainment, 1993–​2001. J Natl Cancer Inst, 100(14), 1003–​1012. doi:10.1093/​jnci/​djn207 Kirschbaum, C., Prussner, J. C., Stone, A. A., Federenko, I., Gaab, J., Lintz, D.,  .  .  .  Hellhammer, D. H. (1995). Persistent high cortisol responses to repeated psychological stress in a subpopulation of healthy men. Psychosom Med, 57(5), 468–​474. Kish, J. K., Yu, M., Percy-​Laurry, A., & Altekruse, S. F. (2014). Racial and ethnic disparities in cancer survival by neighborhood socioeconomic status in Surveillance, Epidemiology, and End Results (SEER) Registries. J Natl

164

164

Part II: The Magnitude of Cancer

Cancer Inst Monogr, 2014(49), 236–​243. doi:10.1093/​jncimonographs/​ lgu020 Kogevinas, M., Pearce, N., Susser, M., & Boffetta, P. (1997). Social inequalities and cancer. IARC Scientific Publications(138). Kogevinas, M., & Porta, M. (1997). Socioeconomic differences in cancer survival: a review of the evidence. IARC Sci Publ(138), 177–​206. Koh, H. K. (2011). The ultimate measures of health. Public Health Rep, 126(Suppl 3), 14–​15. Kostova, D., Andes, L., Erguder, T., Yurekli, A., Keskinkılıç, B., Polat, S., . . . Asma S; Office on Smoking and Health, National Center for Chronic Disease Prevention and Health Promotion, CDC. Cigarette prices and smoking prevalence after a tobacco tax increase—Turkey, 2008 and 2012. MMWR Morb Mortal Wkly Rep. 2014 May 30, 63(21), 457–461. PubMed PMID: 24871250. Kotsopoulos, J., Prescott, J., De Vivo, I., Fan, I., McLaughlin, J., Rosen, B., . . . Narod, S. A. (2014). Telomere length and mortality following a diagnosis of ovarian cancer. Cancer Epidemiol Biomarkers Prev, 23(11), 2603–​2606. doi:10.1158/​1055-​9965.EPI-​14-​0885 Kreuzer, M., Heinrich, J., Kreienbrock, L., Rosario, A. S., Gerken, M., & Wichmann, H. E. (2002). Risk factors for lung cancer among nonsmoking women. Int J Cancer, 100(6), 706–​713. Krieger, N. (2005). Defining and investigating social disparities in cancer:  critical issues. Cancer Causes Control, 16(1), 5–​14. doi:10.1007/​ s10552-​004-​1251-​5 Krieger, N., Chen, J. T., & Ebel, G. (1997). Can we monitor socioeconomic inequalities in health? A survey of U.S. health departments’ data collection and reporting practices. Public Health Reports, 112, 481–​491. Krieger, N., Chen, J. T., Kosheleva, A., & Waterman, P. D. (2012). Shrinking, widening, reversing, and stagnating trends in US socioeconomic inequities in cancer mortality for the total, black, and white populations: 1960–​2006. Cancer Causes Control, 23(2), 297–​319. doi:10.1007/​ s10552-​011-​9879-​4 Krieger, N., Chen, J. T., & Waterman, P. D. (2010). Decline in US breast cancer rates after the Women’s Health Initiative: socioeconomic and racial/​ethnic differentials. Am J Public Health, 100(Suppl 1), S132–​139. doi:10.2105/​ AJPH.2009.181628 Krieger, N., Chen, J. T., Waterman, P. D., Rehkopf, D. H., Yin, R., & Coull, B. A. (2006). Race/​ethnicity and changing US socioeconomic gradients in breast cancer incidence:  California and Massachusetts, 1978–​2002 (United States). Cancer Causes Control, 17(2), 217–​226. doi:10.1007/​ s10552-​005-​0408-​1 Krieger, N., Quesenberry, C., Jr., Peng, T., Horn-​Ross, P., Stewart, S., Brown, S.,  .  .  .  Ward, F. (1999). Social class, race/​ethnicity, and incidence of breast, cervix, colon, lung, and prostate cancer among Asian, Black, Hispanic, and White residents of the San Francisco Bay Area, 1988–​92 (United States). Cancer Causes Control, 10(6), 525–​537. Kroenke, C. (2008). Socioeconomic status and health: youth development and neomaterialist and psychosocial mechanisms. Soc Sci Med, 66(1), 31–​42. doi:10.1016/​j.socscimed.2007.07.018 Kroenke, C. H., Hankinson, S. E., Schernhammer, E. S., Colditz, G. A., Kawachi, I., & Holmes, M. D. (2004). Caregiving stress, endogenous sex steroid hormone levels, and breast cancer incidence. Am J Epidemiol, 159(11), 1019–​1027. doi:10.1093/​aje/​kwh148 Kroenke, C. H., Kubzansky, L. D., Schernhammer, E. S., Holmes, M. D., & Kawachi, I. (2006). Social networks, social support, and survival after breast cancer diagnosis. J Clin Oncol, 24(7), 1105–​1111. Kroenke, C. H., Michael, Y., Tindle, H., Gage, E., Chlebowski, R., Garcia, L., . . . Caan, B. J. (2012). Social networks, social support and burden in relationships, and mortality after breast cancer diagnosis. Breast Cancer Res Treat, 133(1), 375–​385. doi:10.1007/​s10549-​012-​1962-​3 Kroenke, C. H., Michael, Y. L., Shu, X. O., Poole, E. M., Kwan, M. L., Nechuta, S., . . . Chen W. Y. (2017). Post-diagnosis social networks, and lifestyle and treatment factors in the After Breast Cancer Pooling Project. Psychooncology, 26(4), 544-552. doi: 10.1002/pon.4059. Epub 2016 Jan 8. PubMed PMID: 26749519; PubMed Central PMCID: PMC4938778. Kroenke, C. H., Quesenberry, C., Kwan, M. L., Sweeney, C., Castillo, A., & Caan, B. J. (2013). Social networks, social support, and burden in relationships, and mortality after breast cancer diagnosis in the Life After Breast Cancer Epidemiology (LACE) study. Breast Cancer Res Treat, 137(1), 261–​271. doi:10.1007/​s10549-​012-​2253-​8 Kubzansky, L. D., Seeman, T. E., & Glymour, M. M. (2014). Biological pathways linking social conditions and health:  plausible mechanisms and emerging puzzles. In L. F. Berkman, I. Kawachi, & M. M. Glymour (Eds.), Social epidemiology (2nd ed., pp. 512–​561). New York: Oxford University Press. Kuehni, C. E., Strippoli, M. P., Rueegg, C. S., Rebholz, C. E., Bergstraesser, E., Grotzer, M.,  .  .  .  Swiss Pediatric Oncology Group (SPOG). (2012).

Educational achievement in Swiss childhood cancer survivors compared with the general population. Cancer, 118(5), 1439–​1449. doi:10.1002/​cncr.26418 Kunkle, B., Bae, S., Singh, K. P., & Roy, D. (2014). Increased risk of childhood brain tumors among children whose parents had farm-​related pesticide exposures during pregnancy. JP J Biostat, 11(2), 89–​101. Kuper, H., Yang, L., Theorell, T., & Weiderpass, E. (2007). Job strain and risk of breast cancer. Epidemiology, 18(6), 764–​ 768. doi:10.1097/​ EDE.0b013e318142c534 Lammers, C., Ireland, M., Resnick, M., & Blum, R. (2000). Influences on adolescents’ decision to postpone onset of sexual intercourse: a survival analysis of virginity among youths aged 13 to 18 years. J Adolesc Health, 26(1), 42–​48. Le Cornet, C., Fervers, B., Dalton, S. O., Feychting, M., Pukkala, E., Tynes, T., . . . Schuz, J. (2015). Testicular germ cell tumours and parental occupational exposure to pesticides:  a register-​based case-​control study in the Nordic countries (NORD-​TEST study). Occup Environ Med, 72(11), 805–​811. doi:10.1136/​oemed-​2015-​102860 Lee, D., & Jackson, M. (2015). The relationship between lifetime health trajectories and socioeconomic attainment in middle age. Soc Sci Res, 54, 96–​112. doi:10.1016/​j.ssresearch.2015.06.023 Lee, M. (2012). Endocrine-​ disrupting compounds and pubertal development: the effects of environmental compounds on the timing of puberty in humans. In V. R. Preedy (Ed.), Handbook of growth and growth monitoring in health and disease. New York: Springer. Lee, R. E., & Cubbin, C. (2002). Neighborhood context and youth cardiovascular health behaviors. Am J Public Health, 92(3), 428–​436. Lee, S., Colditz, G. A., Berkman, L. F., & Kawachi, I. (2003). Caregiving and risk of coronary heart disease in U.S. women: a prospective study. Am J Prev Med, 24(2), 113–​119. doi:S0749379702005822 [pii] Lee, S., Kawachi, I., & Grodstein, F. (2004). Does caregiving stress affect cognitive function in older women? J Nerv Ment Dis, 192(1), 51–​57. doi:10.1097/​01.nmd.0000106000.02232.30 Leong, N. M., Mignone, L. I., Newcomb, P. A., Titus-​Ernstoff, L., Baron, J. A., Trentham-​Dietz, A., . . . Egan, K. M. (2003). Early life risk factors in cancer: the relation of birth weight to adult obesity. Int J Cancer, 103(6), 789–​791. doi:10.1002/​ijc.10886 Levav, I., Kohn, R., Iscovich, J., Abramson, J. H., Tsai, W. Y., & Vigdorovich, D. (2000). Cancer incidence and survival following bereavement. Am J Public Health, 90(10), 1601–​1607. Levy, D. T., Mumford, E. A., & Compton, C. (2006). Tobacco control policies and smoking in a population of low education women, 1992–​2002. J Epidemiol Community Health, 60(Suppl 2), 20–​26. Li, R., Serdula, M., Bland, S., Mokdad, A., Bowman, B., & Nelson, D. (2000). Trends in fruit and vegetable consumption among adults in 16 US states:  Behavioral Risk Factor Surveillance System, 1990–​1996. Am J Public Health, 90(5), 777–​781. Liberatos, P., Link, B. G., & Kelsey, J. L. (1988). The measurement of social class in epidemiology. Epidemiol Rev, 10, 87–​121. Litt, J. S., Gerry Taylor, H., Margevicius, S., Schluchter, M., Andreias, L., & Hack, M. (2012). Academic achievement of adolescents born with extremely low birth weight. Acta Paediatr, 101(12), 1240–​1245. doi:10.1111/​j.1651-​2227.2012.02790.x Liu, F., Wang, X. W., Chen, L., Hu, P., Ren, H., & Hu, H. D. (2016). Systematic review with meta-​analysis: development of hepatocellular carcinoma in chronic hepatitis B patients with hepatitis B surface antigen seroclearance. Aliment Pharmacol Ther, 43(12), 1253–​1261. doi:10.1111/​apt.13634 Liu, L., Cozen, W., Bernstein, L., Ross, R. K., & Deapen, D. (2001). Changing relationship between socioeconomic status and prostate cancer incidence. J Natl Cancer Inst, 93(9), 705–​709. Liu, S., Jones, R. N., & Glymour, M. M. (2010). Implications of lifecourse epidemiology for research on determinants of adult disease. Public Health Rev, 32(2), 489–​511. Lochner, K., & Kawachi, I. (2000). Socioeconomic status. In D. Hunter & G. A. Colditz (Eds.), Cancer prevention: the causes and prevention of cancer (Vol. 1). Dordrecht, The Netherlands: Kluwer Academic Publishers. Lope, V., Garcia-​ Esquinas, E., Perez-​ Gomez, B., Altzibar, J. M., Gracia-​ Lavedan, E., Ederra, M., . . . Pollan, M. (2016). Perinatal and childhood factors and risk of breast cancer subtypes in adulthood. Cancer Epidemiol, 40, 22–​30. doi:10.1016/​j.canep.2015.11.004 Lutgendorf, S. K., & Sood, A. K. (2011). Biobehavioral factors and cancer progression: physiological pathways and mechanisms. Psychosom Med, 73(9), 724–​730. doi:10.1097/​PSY.0b013e318235be76 Lynch, J., & Kaplan, G. A. (2000). Socioeconomic position. In L. F. Berkman & I. Kawachi (Eds.), Social epidemiology (pp. 13–​35). New York: Oxford University Press. Ma, J., Xu, J., Anderson, R. N., & Jemal, A. (2012). Widening educational disparities in premature death rates in twenty six states in the United

 165



Socioeconomic Disparities in Cancer Incidence and Mortality

States, 1993–​2007. PLoS One, 7(7), e41560. doi:10.1371/​ journal. pone.0041560 MacKinnon, J. A., Duncan, R. C., Huang, Y., Lee, D. J., Fleming, L. E., Voti, L., . . . Wilkinson, J. D. (2007). Detecting an association between socioeconomic status and late stage breast cancer using spatial analysis and area-​based measures. Cancer Epidemiol Biomarkers Prev, 16(4), 756–​ 762. doi:10.1158/​1055-​9965.EPI-​06-​0392 Mahal, B. A., Aizer, A. A., Ziehr, D. R., Hyatt, A. S., Lago-​Hernandez, C., Chen, Y. W., . . . Nguyen, P. L. (2014). The association between insurance status and prostate cancer outcomes: implications for the Affordable Care Act. Prostate Cancer Prostatic Dis, 17(3), 273–​279. doi:10.1038/​pcan.2014.23 Malecki, K. C., Bekkedal, M., Hanrahan, L., Stephenson, L., Werner, M., & Anderson, H. A. (2006). Linking childhood cancer with potential environmental exposure determinants. WMJ, 105(2), 32–​35. Manson, J. E., Hsia, J., Johnson, K. C., Rossouw, J. E., Assaf, A. R., Lasser, N. L., . . . Women’s Health Initiative Investigators. (2003). Estrogen plus progestin and the risk of coronary heart disease. N Engl J Med., 349(6), 523–​534. Maule, M., Zugna, D., Migliore, E., Alessi, D., Merletti, F., Onorati, R.,  .  .  .  Spadea, T. (2016). Surviving a childhood cancer:  impact on education and employment. Eur J Cancer Prev. doi:10.1097/​ CEJ.0000000000000258. [Epub ahead of print] Maunsell, E., Brisson, J., & Deschenes, L. (1995). Social support and survival among women with breast cancer. Cancer, 76(4), 631–​637. McCormack, V. A., dos Santos Silva, I., De Stavola, B. L., Mohsen, R., Leon, D. A., & Lithell, H. O. (2003). Fetal growth and subsequent risk of breast cancer: results from long term follow up of Swedish cohort. BMJ, 326(7383), 248. McDonough, P., Duncan, G. J., Williams, D., & House, J. (1997). Income dynamics and adult mortality in the United States, 1972 through 1989. Am J Public Health, 87(9), 1476–​1483. Meena, J., Rudolph, K. L., & Gunes, C. (2015). Telomere dysfunction, chromosomal instability and cancer. Recent Results Cancer Res, 200, 61–​79. doi:10.1007/​978-​3-​319-​20291-​4_​3 Meissner, H. I., Breen, N., Klabunde, C. N., & Vernon, S. W. (2006). Patterns of colorectal cancer screening uptake among men and women in the United States. Cancer Epidemiol Biomarkers Prev, 15(2), 389–​394. doi:10.1158/​ 1055-​9965.EPI-​05-​0678 Merkin, S. S., Stevenson, L., & Powe, N. (2002). Geographic socioeconomic status, race, and advanced-​stage breast cancer in New  York City. Am J Public Health, 92(1), 64–​70. Merletti, F., Galassi, C., & Spadea, T. (2011). The socioeconomic determinants of cancer. Environ Health, 10(Suppl 1), S7. doi:10.1186/​ 1476-​069X-​10-​S1-​S7 Micic, D. (2001). Obesity in children and adolescents:  a new epidemic? Consequences in adult life. J Pediatr Endocrinol Metab, 14(Suppl 5), 1345–​1352; discussion 1365. Mitchell, C., Hobcraft, J., McLanahan, S. S., Siegel, S. R., Berg, A., Brooks-​ Gunn, J., . . . Notterman, D. (2014). Social disadvantage, genetic sensitivity, and children’s telomere length. Proc Natl Acad Sci U S A, 111(16), 5944–​5949. doi:10.1073/​pnas.1404293111 Montenegro, M. F., Gonzalez-​Guerrero, R., Sanchez-​Del-​Campo, L., Pinero-​ Madrona, A., Cabezas-​Herrera, J., & Rodriguez-​Lopez, J. N. (2016). Targeting the epigenetics of the DNA damage response in breast cancer. Cell Death Dis, 7, e2180. doi:10.1038/​cddis.2016.85 Moore, S., Daniel, M., Paquet, C., Dube, L., & Gauvin, L. (2009). Association of individual network social capital with abdominal adiposity, overweight and obesity. J Public Health (Oxf), 31(1), 175–​183. doi:fdn104 [pii] 10.1093/​pubmed/​fdn104 Nakata, A. (2012). Psychosocial job stress and immunity: a systematic review. Methods Mol Biol, 934, 39–​75. doi:10.1007/​978-​1-​62703-​071-​7_​3 Needham, B. L., Smith, J. A., Zhao, W., Wang, X., Mukherjee, B., Kardia, S. L.,  .  .  .  Diez Roux, A. V. (2015). Life course socioeconomic status and DNA methylation in genes related to stress reactivity and inflammation: the multi-​ethnic study of atherosclerosis. Epigenetics, 10(10), 958–​ 969. doi:10.1080/​15592294.2015.1085139 Nelson, W. G., Yegnasubramanian, S., Agoston, A. T., Bastian, P. J., Lee, B. H., Nakayama, M., & De Marzo, A. M. (2007). Abnormal DNA methylation, epigenetics, and prostate cancer. Front Biosci, 12, 4254–​4266. Nielsen, S. J., & Popkin, B. M. (2003). Patterns and trends in food portion sizes, 1977–​1998. JAMA, 289(4), 450–​453. Nightingale, T. E., & Gruber, J. (1994). Helicobacter and human cancer. J Natl Cancer Inst, 86(20), 1505–​1509. Niu, X., Roche, L. M., Pawlish, K. S., & Henry, K. A. (2013). Cancer survival disparities by health insurance status. Cancer Med, 2(3), 403–​411. doi:10.1002/​cam4.84 O’Donnell, B. L., O’Donnell, C. R., & Stueve, A. (2001). Early sexual initiation and subsequent sex-​related risks among urban minority youth:  the reach for health study. Fam Plann Perspect, 33(6), 268–​275.

165

O’Keefe, E. B., Meltzer, J. P., & Bethea, T. N. (2015). Health disparities and cancer: racial disparities in cancer mortality in the United States, 2000–​ 2010. Front Public Health, 3, 51. doi:10.3389/​fpubh.2015.00051 Obama, B. (2016). United States health care reform: progress to date and next steps. JAMA, 316(5), 525–​532. doi:10.1001/​jama.2016.9797 Ogden, C. L., Carroll, M. D., Fryar, C. D., & Flegal, K. M. (2015). Prevalence of obesity among adults and youth:  United States, 2011–​2014. NCHS Data Brief, 219, 1–​8. Ogden, C. L., Carroll, M. D., Kit, B. K., & Flegal, K. M. (2012). Prevalence of obesity and trends in body mass index among US children and adolescents, 1999–​2010. JAMA, 307(5), 483–​ 490. doi:10.1001/​ jama.2012.40 Ogden, C. L., Carroll, M. D., Kit, B. K., & Flegal, K. M. (2014). Prevalence of childhood and adult obesity in the United States, 2011–​2012. JAMA, 311(8), 806–​814. doi:10.1001/​jama.2014.732 Ogden, C. L., Carroll, M. D., Lawman, H. G., Fryar, C. D., Kruszon-​Moran, D., Kit, B. K., & Flegal, K. M. (2016). Trends in obesity prevalence among children and adolescents in the United States, 1988–​1994 through 2013–​ 2014. JAMA, 315(21), 2292–​2299. doi:10.1001/​jama.2016.6361 Ogden, C. L., Lamb, M. M., Carroll, M. D., & Flegal, K. M. (2010). Obesity and socioeconomic status in children and adolescents:  United States, 2005–​2008. NCHS Data Brief, 51, 1–​8. Pachucki, M. A., Jacques, P. F., & Christakis, N. A. (2011). Social network concordance in food choice among spouses, friends, and siblings. Am J Public Health, 101(11), 2170–​2177. doi:10.2105/​AJPH.2011.300282 Pais, J. (2014). Cumulative structural disadvantage and racial health disparities: the pathways of childhood socioeconomic influence. Demography, 51(5), 1729–​1753. doi:10.1007/​s13524-​014-​0330-​9 Pampel, F. C., Krueger, P. M., & Denney, J. T. (2010). Socioeconomic disparities in health behaviors. Annu Rev Sociol, 36, 349–​370. doi:10.1146/​ annurev.soc.012809.102529 Pampel, F. C., Mollborn, S., & Lawrence, E. M. (2014). Life course transitions in early adulthood and SES disparities in tobacco use. Soc Sci Res, 43, 45–​59. doi:10.1016/​j.ssresearch.2013.08.005 Pamuk, E., Makuc, D., Heck, K., Reuben, C., & Lochner, K. (1998). Socioeconomic status and health chartbook: health, United States, 1998. Hyattsville, MD: National Center for Health Statistics. Parise, C. A., & Caggiano, V. (2013). Disparities in race/​ethnicity and socioeconomic status: risk of mortality of breast cancer patients in the California Cancer Registry, 2000–​ 2010. BMC Cancer, 13, 449. doi:10.1186/​ 1471-​2407-​13-​449 Parker, J. D., Schoendorf, K. C., & Kiely, J. L. (1994). Associations between measures of socioeconomic status and low birth weight, small for gestational age, and premature delivery in the United States. Ann Epidemiol, 4(4), 271–​278. Parsonnet, J. (1995). The incidence of Helicobacter pylori infection. Aliment Pharmacol Ther, 9(Suppl 2), 45–​51. Pearce, N., & Matos, E. (1994). Occupational cancer in developing countries: introduction. IARC Sci Publ, 129, 1–​3. Pechey, R., & Monsivais, P. (2016). Socioeconomic inequalities in the healthiness of food choices: Exploring the contributions of food expenditures. Prev Med, 88, 203–​209. doi:10.1016/​j.ypmed.2016.04.012 Pesch, B., Kendzia, B., Gustavsson, P., Jockel, K. H., Johnen, G., Pohlabeln, H., . . . Bruning, T. (2012). Cigarette smoking and lung cancer: relative risk estimates for the major histological types from a pooled analysis of case-​control studies. Int J Cancer, 131(5), 1210–​1219. doi:10.1002/​ ijc.27339 Peters, S., Glass, D. C., Greenop, K. R., Armstrong, B. K., Kirby, M., Milne, E., & Fritschi, L. (2014). Childhood brain tumours: associations with parental occupational exposure to solvents. Br J Cancer, 111(5), 998–​1003. doi:10.1038/​bjc.2014.358 Petridou, E. T., Sergentanis, T. N., Perlepe, C., Papathoma, P., Tsilimidos, G., Kontogeorgi, E.,  .  .  .  Adami, H. O. (2015). Socioeconomic disparities in survival from childhood leukemia in the United States and globally: a meta-​analysis. Ann Oncol, 26(3), 589-​597. doi:10.1093/​annonc/​mdu572 Phelan, J. C., Link, B. G., & Tehranifar, P. (2010). Social conditions as fundamental causes of health inequalities:  theory, evidence, and policy implications. J Health Soc Behav, 51(Suppl), S28–​S40. doi:10.1177/​ 0022146510383498 Pierce, J. P., Choi, W. S., Gilpin, E. A., Farkas, A. J., & Berry, C. C. (1998). Tobacco industry promotion of cigarettes and adolescent smoking. JAMA, 279(7), 511-​515. Piernas, C., & Popkin, B. M. (2011). Increased portion sizes from energy-​dense foods affect total energy intake at eating occasions in US children and adolescents: patterns and trends by age group and sociodemographic characteristics, 1977–​2006. Am J Clin Nutr, 94(5), 1324–​1332. doi:10.3945/​ ajcn.110.008466

16

166

Part II: The Magnitude of Cancer

Platz, E. A., Giovannucci, E., Rimm, E. B., Curhan, G. C., Spiegelman, D., Colditz, G. A., & Willett, W. C. (1998). Retrospective analysis of birth weight and prostate cancer in the Health Professionals Follow-​up Study. Am J Epidemiol, 147(12), 1140–​1144. Popkin, B. M., Siega-​Riz, A. M., & Haines, P. S. (1996). A comparison of dietary trends among racial and socioeconomic groups in the United States. N Engl J Med, 335(10), 716–​720. Portier, C. J., K., T. T., Carter, S. R., Dilworth, C. H., Grambsch, A. E., Gohlke, J.,  .  .  .  Whung, P.-​Y. (2010). A human health perspective on climate change: a report outlining the research needs on the human health effects of climate change. Retrieved from: www.niehs.nih.gov/​climatereport. Powell, L. M., Szczypka, G., & Chaloupka, F. J. (2010). Trends in exposure to television food advertisements among children and adolescents in the United States. Arch Pediatr Adolesc Med, 164(9), 794–​802. doi:10.1001/​ archpediatrics.2010.139 Powell, L. M., Wada, R., & Kumanyika, S. K. (2014). Racial/​ethnic and income disparities in child and adolescent exposure to food and beverage television ads across the U.S.  media markets. Health Place, 29, 124–​131. doi:10.1016/​j.healthplace.2014.06.006 Praestegaard, C., Kjaer, S. K., Nielsen, T. S., Jensen, S. M., Webb, P. M., Nagle, C. M., . . . Ovarian Cancer Association, C. (2016). The association between socioeconomic status and tumour stage at diagnosis of ovarian cancer: a pooled analysis of 18 case-​control studies. Cancer Epidemiol, 41, 71–​79. doi:10.1016/​j.canep.2016.01.012 Pudrovska, T., & Anikputa, B. (2012). The role of early-​life socioeconomic status in breast cancer incidence and mortality: unraveling life course mechanisms. J Aging Health, 24(2), 323–​344. doi:10.1177/​0898264311422744 Rehkopf, D., Glymour, M., & Osypuk, T. (2016). The consistency assumption for causal inference in social epidemiology:  when a rose is not a rose. Curr Epidemiol Rep, 3(1), 63–​71. Reynolds, P., Boyd, P. T., Blacklow, R. S., Jackson, J. S., Greenberg, R. S., Austin, D. F., . . . Edwards, B. K. (1994). The relationship between social ties and survival among black and white breast cancer patients. National Cancer Institute Black/​ White Cancer Survival Study Group. Cancer Epidemiol Biomarkers Prev, 3(3), 253–​259. Rickard, B. J., Okrent, A. M., & Alston, J. M. (2013). How have agricultural policies influenced caloric consumption in the United States? Health Econ, 22(3), 316–​339. doi:10.1002/​hec.2799 Riley, V. (1981). Psychoneuroendocrine influences on immunocompetence and neoplasia. Science, 212(4499), 1100–​1109. Robbins, J. M., Vaccarino, V., Zhang, H., & Kasl, S. V. (2000). Excess type 2 diabetes in African-​American women and men aged 40–​74 and socioeconomic status:  evidence from the Third National Health and Nutrition Examination Survey. J Epidemiol Community Health, 54(11), 839–​845. Robert, J., Cheney, M. K., & Azad, M. R. (2009). Point-​of-​sale marketing of tobacco products:  taking advantage of the socially disadvantaged? J Health Care Poor Underserved, 20(2), 489–​506. doi:10.1353/​hpu.0.0147 Robertson, T., Batty, G. D., Der, G., Fenton, C., Shiels, P. G., & Benzeval, M. (2013). Is socioeconomic status associated with biological aging as measured by telomere length? Epidemiol Rev, 35, 98–​111. doi:10.1093/​ epirev/​mxs001 Ross, C. E., & Wu, C. (1995). The links between education and health. Am Sociol Rev, 60, 719–​745. Rothman, K. J. (1980). The proportion of cancer attributable to alcohol consumption. Prev Med, 9(2), 174–​179. Rottenberg, Y., Litwin, H., Manor, O., Paltiel, A., Barchana, M., & Paltiel, O. (2014). Prediagnostic self-​assessed health and extent of social networks predict survival in older individuals with cancer: a population based cohort study. J Geriatr Oncol, 5(4), 400–​407. doi:10.1016/​j.jgo.2014.08.001 Roura, E., Castellsague, X., Pawlita, M., Travier, N., Waterboer, T., Margall, N.,  .  .  .  Riboli, E. (2014). Smoking as a major risk factor for cervical cancer and pre-​cancer: results from the EPIC cohort. Int J Cancer, 135(2), 453–​466. doi:10.1002/​ijc.28666 Rowse, G. J., Weinberg, J., Bellward, G. D., & Emerman, J. T. (1992). Endocrine mediation of psychosocial stressor effects on mouse mammary tumor growth. Cancer Lett, 65(1), 85–​93. Sadana, R., & Harper, S. (2011). Data systems linking social determinants of health with health outcomes: advancing public goods to support research and evidence-​based policy and programs. Public Health Rep, 126(Suppl 3), 6–​13. Sandhu, M. S., Luben, R., Day, N. E., & Khaw, K. T. (2002). Self-​reported birth weight and subsequent risk of colorectal cancer. Cancer Epidemiol Biomarkers Prev, 11(9), 935–​938. Santelli, J. S., Lowry, R., Brener, N. D., & Robin, L. (2000). The association of sexual behaviors with socioeconomic status, family structure, and race/​ ethnicity among US adolescents. Am J Public Health, 90(10), 1582–​1588.

Sapienza, C., & Issa, J. P. (2016). Diet, nutrition, and cancer epigenetics. Annu Rev Nutr, 36, 665–​681. doi:10.1146/​annurev-​nutr-​121415-​112634 Schernhammer, E. S., Laden, F., Speizer, F. E., Willett, W. C., Hunter, D. J., Kawachi, I., & Colditz, G. A. (2001). Rotating night shifts and risk of breast cancer in women participating in the nurses’ health study. J Natl Cancer Inst, 93(20), 1563–​1568. Schernhammer, E. S., Laden, F., Speizer, F. E., Willett, W. C., Hunter, D. J., Kawachi, I.,  .  .  .  Colditz, G. A. (2003a). Night-​shift work and risk of colorectal cancer in the nurses’ health study. J Natl Cancer Inst, 95(11), 825–​828. Schernhammer, E. S., Hankinson, S., Rosner, B., Kroenke, C. H., Speizer, F. E., Willett, W. C., . . . Kawachi, I. (2003b). Job stress and breast cancer risk: the Nurses’ Health Study. Unpublished data. Schottenfeld, D., & Beebe-​Dimmer, J. (2015). The cancer burden attributable to biologic agents. Ann Epidemiol, 25(3), 183–​187. doi:10.1016/​ j.annepidem.2014.11.016 Schreder, E. D., Uding, N., & La Guardia, M. J. (2016). Inhalation a significant exposure route for chlorinated organophosphate flame retardants. Chemosphere, 150, 499–​504. doi:10.1016/​j.chemosphere.2015.11.084 Schulz, R., & Beach, S. R. (1999). Caregiving as a risk factor for mortality:  the Caregiver Health Effects Study. JAMA, 282(23), 2215–​2219. doi:joc91040 [pii] Schupp, C. W., Press, D. J., & Gomez, S. L. (2014). Immigration factors and prostate cancer survival among Hispanic men in California: does neighborhood matter? Cancer, 120(9), 1401–​1408. doi:10.1002/​cncr.28587 Schuster, M. A., Franke, T., & Pham, C. B. (2002). Smoking patterns of household members and visitors in homes with children in the United States. Arch Pediatr Adolesc Med, 156(11), 1094–​1100. Scragg, R., Laugesen, M., & Robinson, E. (2002). Cigarette smoking, pocket money and socioeconomic status:  results from a national survey of 4th form students in 2000. N Z Med J, 115(1158), U108. Segerstrom, S. C., & Miller, G. E. (2004). Psychological stress and the human immune system:  a meta-​analytic study of 30  years of inquiry. Psychol Bull, 130(4), 601–​630. doi:10.1037/​0033–​2909.130.4.601 Shariff-​Marco, S., Yang, J., John, E. M., Sangaramoorthy, M., Hertz, A., Koo, J.,  .  .  .  Gomez, S. L. (2014). Impact of neighborhood and individual socioeconomic status on survival after breast cancer varies by race/​ethnicity:  the Neighborhood and Breast Cancer Study. Cancer Epidemiol Biomarkers Prev, 23(5), 793–​811. doi:10.1158/​1055-​9965.EPI-​13-​0924 Sharpe, P. A., Whitaker, K., Alia, K. A., Wilcox, S., & Hutto, B. (2016). Dietary intake, behaviors and psychosocial factors among women from food-​ secure and food-​insecure households in the United States. Ethn Dis, 26(2), 139–​146. doi:10.18865/​ed.26.2.139 Shaw, R. J., Hall, G. L., Lowe, D., Liloglou, T., Field, J. K., Sloan, P., & Risk, J. M. (2008). The role of pyrosequencing in head and neck cancer epigenetics: correlation of quantitative methylation data with gene expression. Arch Otolaryngol Head Neck Surg, 134(3), 251–​256. doi:10.1001/​ archoto.2007.50 Shebl, F. M., Capo-​Ramos, D. E., Graubard, B. I., McGlynn, K. A., & Altekruse, S. F. (2012). Socioeconomic status and hepatocellular carcinoma in the United States. Cancer Epidemiol Biomarkers Prev, 21(8), 1330–​1335. doi:10.1158/​1055-​9965.EPI-​12-​0124 Shen, J., Gammon, M. D., Terry, M. B., Bradshaw, P. T., Wang, Q., Teitelbaum, S. L.,  .  .  .  Santella, R. M. (2012). Genetic polymorphisms in telomere pathway genes, telomere length, and breast cancer survival. Breast Cancer Res Treat, 134(1), 393–​400. doi:10.1007/​s10549-​012-​2058-​9 Sherbourne, C. D., & Stewart, A. L. (1991). The MOS social support survey. Soc Sci Med, 32(6), 705–​714. Shi, L. (1998). Sociodemographic characteristics and individual health behaviors. South Med J, 91(10), 933–​941. Shi, R. L., Qu, N., Lu, Z. W., Liao, T., Gao, Y., & Ji, Q. H. (2016). The impact of marital status at diagnosis on cancer survival in patients with differentiated thyroid cancer. Cancer Med, 5(8), 2145–​ 2154. doi:10.1002/​ cam4.778 Shield, K. D., Soerjomataram, I., & Rehm, J. (2016). Alcohol use and breast cancer:  a critical review. Alcohol Clin Exp Res, 40(6), 1166–​1181. doi:10.1111/​acer.13071 Shrestha, A., Ritz, B., Wilhelm, M., Qiu, J., Cockburn, M., & Heck, J. E. (2014). Prenatal exposure to air toxics and risk of Wilms’ tumor in 0-​to 5-​year-​old children. J Occup Environ Med, 56(6), 573–​578. doi:10.1097/​ JOM.0000000000000167 Siega-​Riz, A. M., & Popkin, B. M. (2001). Dietary trends among low socioeconomic status women of childbearing age in the United States from 1977 to 1996: a comparison among ethnic groups. J Am Med Womens Assoc, 56(2), 44–​48, 72. Sieh, W., Rothstein, J. H., McGuire, V., & Whittemore, A. S. (2014). The role of genome sequencing in personalized breast cancer prevention. Cancer

 167



Socioeconomic Disparities in Cancer Incidence and Mortality

Epidemiol Biomarkers Prev, 23(11), 2322–​2327. doi:10.1158/​1055-​9965. EPI-​14-​0559 Singh, G. K., Miller, B. A., & Hankey, B. F. (2002). Changing area socioeconomic patterns in U.S.  cancer mortality, 1950–​1998. Part II:  Lung and colorectal cancers. J Natl Cancer Inst, 94(12), 916–​925. Singh, G. K., Miller, B. A., Hankey, B. F., Feuer, E. J., & Pickle, L. W. (2002). Changing area socioeconomic patterns in U.S. cancer mortality, 1950–​1998. Part I: All cancers among men. J Natl Cancer Inst, 94(12), 904–​915. Singh, G. K., Williams, S. D., Siahpush, M., & Mulhollen, A. (2011). Socioeconomic, rural-​urban, and racial inequalities in US cancer mortality: Part I: All cancers and lung cancer and Part II: Colorectal, prostate, breast, and cervical cancers. J Cancer Epidemiol, 2011, 107497. doi:10.1155/​2011/​107497 Singh, S., Darroch, J. E., & Frost, J. J. (2001). Socioeconomic disadvantage and adolescent women’s sexual and reproductive behavior:  the case of five developed countries. Fam Plann Perspect, 33(6), 251–​258, 289. Sistino, J. J., & Ellis, C., Jr. (2011). Effects of health disparities on survival after neonatal heart surgery: why should racial, ethnic, gender, and socioeconomic status be included in the risk analysis? J Extra Corpor Technol, 43(4), 232–​235. Smith, K., & Christakis, N. (2008). Social networks and health. Ann Rev Sociol, 34, 405–​429. Smith, N. R., Jensen, B. W., Zimmermann, E., Gamborg, M., Sorensen, T. I., & Baker, J. L. (2016). Associations between birth weight and colon and rectal cancer risk in adulthood. Cancer Epidemiol, 42, 181–​185. doi:10.1016/​j.canep.2016.05.003 Spiegel, D., Bloom, J. R., Kraemer, H. C., & Gottheil, E. (1989). Effect of psychosocial treatment on survival of patients with metastatic breast cancer. Lancet, 2(8668), 888–​891. Spiegel, D., Butler, L. D., Giese-​Davis, J., Koopman, C., Miller, E., DiMiceli, S., . . . Kraemer, H. C. (2007). Effects of supportive-​expressive group therapy on survival of patients with metastatic breast cancer: a randomized prospective trial. Cancer, 110(5), 1130–​1138. doi:10.1002/​cncr.22890 Spiegel, D., Sephton, S. E., Terr, A. I., & Stites, D. P. (1998). Effects of psychosocial treatment in prolonging cancer survival may be mediated by neuroimmune pathways. Ann N Y Acad Sci, 840, 674–​683. Spycher, B. D., Feller, M., Roosli, M., Ammann, R. A., Diezi, M., Egger, M., & Kuehni, C. E. (2015). Childhood cancer and residential exposure to highways: a nationwide cohort study. Eur J Epidemiol, 30(12), 1263–​1275. doi:10.1007/​s10654-​015-​0091-​9 Steenland, K., Henley, J., Calle, E., & Thun, M. (2004). Individual-​and area-​ level socioeconomic status variables as predictors of mortality in a cohort of 179,383 persons. Am J Epidemiol, 159(11), 1047–​1056. doi:10.1093/​ aje/​kwh129 Steenland, K., Henley, J., & Thun, M. (2002). All-​cause and cause-​specific death rates by educational status for two million people in two American Cancer Society cohorts, 1959–​1996. Am J Epidemiol, 156(1), 11–​21. Steenland, K., Hu, S., & Walker, J. (2004). All-​cause and cause-​specific mortality by socioeconomic status among employed persons in 27 US states, 1984–​1997. Am J Public Health, 94(6), 1037–​1042. Steinkellner, A. R., Denison, S. E., Eldridge, S. L., Lenzi, L. L., Chen, W., & Bowlin, S. J. (2012). A decade of postmenopausal hormone therapy prescribing in the United States:  long-​ term effects of the Women’s Health Initiative. Menopause, 19(6), 616–​621. doi:10.1097/​ gme.0b013e31824bb039 Steinmaus, C., Ferreccio, C., Acevedo, J., Yuan, Y., Liaw, J., Duran, V., . . . Smith, A. H. (2014). Increased lung and bladder cancer incidence in adults after in utero and early-​life arsenic exposure. Cancer Epidemiol Biomarkers Prev, 23(8), 1529–​1538. doi:10.1158/​1055-​9965.EPI-​14-​0059 Strauss, R. S., & Knight, J. (1999). Influence of the home environment on the development of obesity in children. Pediatrics, 103(6), e85. Stringhini, S., Polidoro, S., Sacerdote, C., Kelly, R. S., van Veldhoven, K., Agnoli, C., . . . Vineis, P. (2015). Life-​course socioeconomic status and DNA methylation of genes regulating inflammation. Int J Epidemiol, 44(4), 1320–​1330. doi:10.1093/​ije/​dyv060 Su, L. J., & Arab, L. (2001). Nutritional status of folate and colon cancer risk:  evidence from NHANES I  epidemiologic follow-​up study. Ann Epidemiol, 11(1), 65–​72. Svenson, U., Nordfjall, K., Stegmayr, B., Manjer, J., Nilsson, P., Tavelin, B., . . . Roos, G. (2008). Breast cancer survival is associated with telomere length in peripheral blood cells. Cancer Res, 68(10), 3618–​3623. doi:10.1158/​0008-​5472.CAN-​07-​6497 Svenson, U., Oberg, A., Stenling, R., Palmqvist, R., & Roos, G. (2016). Telomere length in peripheral leukocytes is associated with immune cell tumor infiltration and prognosis in colorectal cancer patients. Tumour Biol, 37(8), 10877–​10882. doi:10.1007/​s13277-​016-​4987-​0

167

Tao, L., Ladabaum, U., Gomez, S. L., & Cheng, I. (2014). Colorectal cancer mortality among Hispanics in California:  differences by neighborhood socioeconomic status and nativity. Cancer, 120(22), 3510–​3518. doi:10.1002/​cncr.28837 Taylor, G. S., Long, H. M., Brooks, J. M., Rickinson, A. B., & Hislop, A. D. (2015). The immunology of Epstein-​ Barr virus-​ induced disease. Annu Rev Immunol, 33, 787–​ 821. doi:10.1146/​ annurev-​immunol-​032414-​112326 Taylor, R., Cumming, R., Woodward, A., & Black, M. (2001). Passive smoking and lung cancer: a cumulative meta-​analysis. Aust N Z J Public Health, 25(3), 203–​211. Thompson, F. E., McNeel, T. S., Dowling, E. C., Midthune, D., Morrissette, M., & Zeruto, C. A. (2009). Interrelationships of added sugars intake, socioeconomic status, and race/​ethnicity in adults in the United States: National Health Interview Survey, 2005. J Am Diet Assoc, 109(8), 1376–​1383. doi:10.1016/​j.jada.2009.05.002 Thun, M. J., DeLancey, J. O., Center, M. M., Jemal, A., & Ward, E. M. (2010). The global burden of cancer:  priorities for prevention. Carcinogenesis, 31(1), 100–​110. doi:10.1093/​carcin/​bgp263 Trichopoulos, D. (1990). Hypothesis:  does breast cancer originate in utero? Lancet, 335(8695), 939–​940. Trimarchi, M. P., Mouangsavanh, M., & Huang, T. H. (2011). Cancer epigenetics:  a perspective on the role of DNA methylation in acquired endocrine resistance. Chin J Cancer, 30(11), 749–​ 756. doi:10.5732/​ cjc.011.10128 Tyas, S. L., & Pederson, L. L. (1998). Psychosocial factors related to adolescent smoking: a critical review of the literature. Tob Control, 7(4), 409–​420. US Department of Health and Human Services. (2000). Healthy people 2010: understanding and improving health. 2nd ed. Retrieved from: http://​ www.health.gov/​healthypeople/​document/​html/​volume1/​03cancer.htm. NTP (National Toxicology Program). 2002. Report on Carcinogens, 10th Ed. U.S. Department of Health and Human Services, Public Health Service, National Toxicology Program, Research Triangle Park, NC. Valente, T. W. (2009). Social networks and health: models, methods, and applications. New York: Oxford University Press. Valls-​Bautista, C., Pinol-​Felis, C., Rene-​Espinet, J. M., Buenestado-​Garcia, J., & Vinas-​Salas, J. (2015). In colon cancer, normal colon tissue and blood cells have altered telomere lengths. J Surg Oncol, 111(7), 899–​904. doi:10.1002/​jso.23894 van Hasselt, M., Kruger, J., Han, B., Caraballo, R. S., Penne, M. A., Loomis, B., & Gfroerer, J. C. (2015). The relation between tobacco taxes and youth and young adult smoking: what happened following the 2009 U.S. federal tax increase on cigarettes? Addict Behav, 45, 104–​109. doi:10.1016/​ j.addbeh.2015.01.023 van Jaarsveld, C. H., Miles, A., Edwards, R., & Wardle, J. (2006). Marriage and cancer prevention: does marital status and inviting both spouses together influence colorectal cancer screening participation? J Med Screen, 13(4), 172–​176. Van Maele-​Fabry, G., Hoet, P., & Lison, D. (2013). Parental occupational exposure to pesticides as risk factor for brain tumors in children and young adults:  a systematic review and meta-​analysis. Environ Int, 56, 19–​31. doi:10.1016/​j.envint.2013.02.011 Van Maele-​Fabry, G., Lantin, A. C., Hoet, P., & Lison, D. (2010). Childhood leukaemia and parental occupational exposure to pesticides:  a systematic review and meta-​analysis. Cancer Causes Control, 21(6), 787–​809. doi:10.1007/​s10552-​010-​9516-​7 Vargas, A. J., & Thompson, P. A. (2012). Diet and nutrient factors in colorectal cancer risk. Nutr Clin Pract, 27(5), 613–​623. doi:10.1177/​ 0884533612454885 Vatten, L. J., Maehle, B. O., Lund Nilsen, T. I., Tretli, S., Hsieh, C. C., Trichopoulos, D., & Stuver, S. O. (2002). Birth weight as a predictor of breast cancer:  a case-​control study in Norway. Br J Cancer, 86(1), 89–​91. Veeck, J., & Esteller, M. (2010). Breast cancer epigenetics: from DNA methylation to microRNAs. J Mammary Gland Biol Neoplasia, 15(1), 5–​17. doi:10.1007/​s10911-​010-​9165-​1 Villingshoj, M., Ross, L., Thomsen, B. L., & Johansen, C. (2006). Does marital status and altered contact with the social network predict colorectal cancer survival? Eur J Cancer, 42(17), 3022–​ 3027. doi:10.1016/​ j.ejca.2006.04.027 Vohra, J., Marmot, M. G., Bauld, L., & Hiatt, R. A. (2016). Socioeconomic position in childhood and cancer in adulthood: a rapid-​review. J Epidemiol Community Health, 70(6), 629–​634. doi:10.1136/​jech-​2015-​206274 von Ehrenstein, O. S., Heck, J. E., Park, A., Cockburn, M., Escobedo, L., & Ritz, B. (2015). In utero and early-​life exposure to ambient air toxics and childhood brain tumors:  a population-​based case-​control study in California, USA. Environ Health Perspect, 124(7), 1093–​1099.

168

168

Part II: The Magnitude of Cancer

Wakabayashi, I. (2014). Light-​to-​moderate alcohol drinking reduces the impact of obesity on the risk of diabetes mellitus. J Stud Alcohol Drugs, 75(6), 1032–​1038. Wang, A., Clouston, S. A., Rubin, M. S., Colen, C. G., & Link, B. G. (2012). Fundamental causes of colorectal cancer mortality:  the implications of informational diffusion. Milbank Q, 90(3), 592–​ 618. doi:10.1111/​ j.1468-​0009.2012.00675.x Wang, C. Y., de Dieu Tapsoba, J., Anderson, M. L., Vernon, S. W., Chubak, J., Fuller, S., & Green, B. B. (2014). Time to screening in the systems of support to increase colorectal cancer screening trial. Cancer Epidemiol Biomarkers Prev, 23(8), 1683–​1688. doi:10.1158/​1055-​9965.EPI-​14-​0270 Wang, F., Luo, L., & McLafferty, S. (2010). Healthcare access, socioeconomic factors and late-​stage cancer diagnosis: an exploratory spatial analysis and public policy implication. Int J Public Pol, 5(2–​3), 237–​258. doi:10.1504/​ IJPP.2010.030606 Wang, L., Wilson, S. E., Stewart, D. B., & Hollenbeak, C. S. (2011). Marital status and colon cancer outcomes in US Surveillance, Epidemiology and End Results registries:  does marriage affect cancer survival by gender and stage? Cancer Epidemiol, 35(5), 417–​ 422. doi:10.1016/​ j.canep.2011.02.004 Wang, N., Cao, F., Liu, F., Jia, Y., Wang, J., Bao, C., . . . Cheng, Y. (2015). The effect of socioeconomic status on health-​care delay and treatment of esophageal cancer. J Transl Med, 13, 241. doi:10.1186/​s12967-​015-​0579-​9 Wei, F., Miglioretti, D. L., Connelly, M. T., Andrade, S. E., Newton, K. M., Hartsfield, C. L.,  .  .  .  Buist, D. S. (2005). Changes in women’s use of hormones after the Women’s Health Initiative estrogen and progestin trial by race, education, and income. J Natl Cancer Inst Monogr, 35, 106–​112. doi:10.1093/​jncimonographs/​lgi047 Wei, R., DeVilbiss, F. T., & Liu, W. (2015). Genetic polymorphism, telomere biology and non-​small lung cancer risk. J Genet Genomics, 42(10), 549–​ 561. doi:10.1016/​j.jgg.2015.08.005 Weischer, M., Nordestgaard, B. G., Cawthon, R. M., Freiberg, J. J., Tybjaerg-​ Hansen, A., & Bojesen, S. E. (2013). Short telomere length, cancer survival, and cancer risk in 47102 individuals. J Natl Cancer Inst, 105(7), 459–​468. doi:10.1093/​jnci/​djt016 Weston, T. L., Aronson, K. J., Siemiatycki, J., Howe, G. R., & Nadon, L. (2000). Cancer mortality among males in relation to exposures assessed through a job-​exposure matrix. Int J Occup Environ Health, 6(3), 194–​ 202. doi:10.1179/​oeh.2000.6.3.194 Wiencke, J. K., & Kelsey, K. T. (2002). Teen smoking, field cancerization, and a “critical period” hypothesis for lung cancer susceptibility. Environ Health Perspect, 110(6), 555–​558. Wigle, D. T., Turner, M. C., & Krewski, D. (2009). A systematic review and meta-​ analysis of childhood leukemia and parental occupational pesticide exposure. Environ Health Perspect, 117(10), 1505–​1513. doi:10.1289/​ehp.0900582 Wilkinson, R. G. (1996). Unhealthy societies:  the afflictions of inequality. London: Routledge.

Willeit, P., Willeit, J., Kloss-​ Brandstatter, A., Kronenberg, F., & Kiechl, S. (2011). Fifteen-​ year follow-​ up of association between telomere length and incident cancer and cancer mortality. JAMA, 306(1), 42–​44. doi:10.1001/​jama.2011.901 Willeit, P., Willeit, J., Mayr, A., Weger, S., Oberhollenzer, F., Brandstatter, A.,  .  .  .  Kiechl, S. (2010). Telomere length and risk of incident cancer and cancer mortality. JAMA, 304(1), 69–​75. doi:10.1001/​jama.2010.897 Williams, D. R. (1997). Commentary on monitoring socioeconomic status. Public Health Reports, 112, 492–​494. Wilson, C. L., Stratton, K., Leisenring, W. L., Oeffinger, K. C., Nathan, P. C., Wasilewski-​Masker, K.,  .  .  .  Ness, K. K. (2014). Decline in physical activity level in the Childhood Cancer Survivor Study cohort. Cancer Epidemiol Biomarkers Prev, 23(8), 1619–​1627. doi:10.1158/​1055-​9965. EPI-​14-​0213 Woloshin, S., Schwartz, L. M., Tosteson, A. N., Chang, C. H., Wright, B., Plohman, J., & Fisher, E. S. (1997). Perceived adequacy of tangible social support and health outcomes in patients with coronary artery disease. J Gen Intern Med, 12(10), 613–​618. Yanovski, S. Z., & Yanovski, J. A. (2011). Obesity prevalence in the United States:  up, down, or sideways? N Engl J Med, 364(11), 987–​989. doi:10.1056/​NEJMp1009229 Yen, I. H., & Moss, N. (1999). Unbundling education: a critical discussion of what education confers and how it lowers risk for disease and death. Ann N Y Acad Sci, 896, 350–​351. Yin, D., Morris, C., Allen, M., Cress, R., Bates, J., & Liu, L. (2010). Does socioeconomic disparity in cancer incidence vary across racial/​ ethnic groups? Cancer Causes Control, 21(10), 1721–​1730. doi:10.1007/​ s10552-​010-​9601-​y Yost, K., Perkins, C., Cohen, R., Morris, C., & Wright, W. (2001). Socioeconomic status and breast cancer incidence in California for different race/​ethnic groups. Cancer Causes Control, 12(8), 703–​711. Zane, L., & Jeang, K. T. (2014). HTLV-​1 and leukemogenesis:  virus-​cell interactions in the development of adult T-​cell leukemia. Recent Results Cancer Res, 193, 191–​210. doi:10.1007/​978-​3-​642-​38965-​8_​11 Zhang, C., Chen, X., Li, L., Zhou, Y., Wang, C., & Hou, S. (2015). The association between telomere length and cancer prognosis:  evidence from a meta-​analysis. PLoS One, 10(7), e0133174. doi:10.1371/​ journal. pone.0133174 Zibbell, J. E., Iqbal, K., Patel, R. C., Suryaprasad, A., Sanders, K. J., Moore-​ Moravian, L., . . . ; Centers for Disease Control and Prevention (CDC). (2015). Increases in hepatitis C virus infection related to injection drug use among persons aged ≤30  years—​ Kentucky, Tennessee, Virginia, and West Virginia, 2006–​2012. MMWR Morb Mortal Wkly Rep, 64(17), 453–​458. Ziol-​Guest, K. M., Duncan, G. J., & Kalil, A. (2009). Early childhood poverty and adult body mass index. Am J Public Health, 99(3), 527–​532. doi:10.2105/​ajph.2007.130575

 169

10

The Economic Burden of Cancer in the United States K. ROBIN YABROFF, GERY P. GUY JR., MATTHEW P. BANEGAS, AND DONATUS U. EKWUEME

OVERVIEW With an aging and growing population and improved early detection and survival following cancer diagnosis in the United States, the prevalence of cancer survivorship is expected to increase in the future. Based only on population trends, national expenditures for cancer care are projected to increase as well, from $124.6 billion in 2010 to $157.8 billion in 2020 (both in 2010 US dollars). When trends toward greater intensity of healthcare service use and more expensive treatments are considered, expenditures are projected to be even greater. Thus, estimating and projecting the economic burden of cancer, including healthcare expenditures and productivity losses for patients and their families, are increasingly important issues for healthcare policymakers, healthcare systems, physicians, employers, and society overall. In this chapter, we describe measures of the economic burden of cancer, including (1) direct costs, resulting from the use of resources for medical care for cancer; (2) indirect costs, resulting from the loss of economic resources and opportunities associated with morbidity and mortality due to cancer and its treatment; and (3) psychosocial or intangible costs, such as pain and suffering. Consistent with the intensity of treatment for initial care, recurrence, and end-​of-​life care, costs are highest in the initial period following diagnosis and, among patients who die from their disease, at the end of life, following a U-​shaped curve. We describe how this temporal cost pattern influences differences in per-​person and aggregate cost estimates for cancers with excellent prognosis and longer survival (e.g., breast and prostate) and cancers with poorer prognosis (e.g., pancreas and lung) and we present recent per-​person and national estimates of direct and indirect costs overall and by cancer site. Finally, we describe publicly available data sources and methodologic issues with measuring economic burden and identify key areas for future research.

INTRODUCTION In 2014, an estimated 14.5 million Americans were alive with a history of cancer (American Cancer Society, 2016; Howlander et  al., 2016). The absolute number of cancer survivors is expected to be substantially larger in the future because cancer incidence increases with age, and the US population is both aging and growing (de Moor et al., 2013; Mariotto et al., 2011). In addition, advances in screening, detection, and treatment are associated with improved survival following cancer diagnosis. These developments will also result in increased cancer prevalence. It is projected that the number of cancer survivors will increase to 19 million by 2024 (American Cancer Society, 2016). This increasing number of cancer survivors will receive medical care throughout the trajectory of their cancer experiences, including short-​and long-​term effects of disease and its treatment. As shown in Figure 10–1, based only on population aging and growth, national expenditures for cancer care are projected to increase from $124.6 billion in 2010 to $157.8 billion by 2020 (both in 2010 US dollars), a 27% increase (Mariotto et al., 2011). When recent trends toward greater intensity of healthcare service use (Warren et al., 2008; Dinan et  al., 2010; Conti et  al., 2014) and increasing costs of cancer care (Bach 2009; Conti et  al., 2014; Dinan et  al., 2010; Elkin and Bach, 2010; Howard et  al., 2010; Schrag, 2004; Tangka et  al., 2010; Warren et al., 2008; Woodward et al., 2007; Wong et al., 2009)

are considered, expenditures in 2020 are projected to be even greater (Mariotto et al., 2011). Thus, estimating and projecting the economic burden of cancer, including healthcare expenditures and productivity losses for patients and their families, are increasingly important issues for healthcare policymakers, healthcare systems, physicians, employers, and society overall. In this chapter, we describe measures of the economic burden of cancer and approaches for estimating economic burden, and we present recent national estimates. We also describe publicly available data sources and discuss methodologic issues with measuring economic burden, and identify key areas for future research.

WHY MEASURE THE ECONOMIC BURDEN OF CANCER? Measurement of the economic burden of cancer is important at many levels for medical and non-​medical resource allocation, reimbursement decisions, and the evaluation of specific programs throughout the course of cancer care, from early detection and diagnosis to treatment, survivorship, and end-​of-​life care. Descriptive studies of the various components of the economic burden of cancer also serve as an initial step in elucidating relevant domains in cost-​effectiveness analysis of cancer prevention and control interventions and identifying priorities and resource needs in the planning of cancer prevention and control strategies and programs.

HOW IS THE ECONOMIC BURDEN OF CANCER MEASURED? Illness and disease create an economic burden for patients, family and friends, employers, the healthcare system, and society. Cancer and its treatment may result in pain and suffering, morbidity and reduced quality of life, financial losses for the patient and family, and in some cases, premature mortality. The economic burden of cancer also includes the loss of economic resources and opportunities from the perspective of employers and society more broadly. The economic burden of disease is measured by cost, the monetary valuation of resources used to treat disease and the loss of economic opportunities related to disease occurrence and treatment. Three categories of cost domains are typically identified: (1) direct costs, resulting from the use of resources for medical care for cancer as well as care for lasting and late effects of cancer; (2) indirect costs, resulting from the loss of economic resources and opportunities associated with morbidity and mortality due to cancer and its treatment; and (3) psychosocial (intangible) costs, such as pain and suffering. Examples of these categories of costs are listed in Table 10–1 and are described in greater detail in the following subsections.

Direct Costs Direct costs include the monetary value of resources used for medical care in the early detection, diagnosis, and treatment of disease, as well as for rehabilitation, surveillance, and end-​of-​life care. The distinguishing characteristic of a direct cost is that it represents resource

169

170

170

PART II:  THE MAGNITUDE OF CANCER Breast Colorectal Lymphoma Lung Prostate Leukemia Ovary Brain Bladder Kidney Head_Neck Uterus Melanoma Pancreas Stomach Cervix Esophagus 0

5

10

15

20

25

Billion US$ Expenditures in 2010

Additional Expenditures in 2020

Figure 10–1.  Projected increase in national expenditures in 2020 by cancer site (in billion US$): Based on population changes only. Source: Mariotto AB, Yabroff KR, Shao Y, Feuer EJ, Brown ML. 2011. Projections of the costs of cancer care in the United States: 2010–​2020. J Natl Cancer Inst, 103, 117–​128.

utilization associated with a direct monetary payment. These payments are usually recorded in the financial records of a healthcare insurance system, healthcare providers, or an individual household. Such costs include medical resources provided under public or private insurance systems, as well as individual out-​of-​pocket expenditures. Direct costs for medical care include hospitalizations, emergency room care, outpatient care, nursing home and home healthcare, and the services of primary care physicians and specialists as well as other health practitioners (Table 10–1). Cancer treatments, such as chemotherapy, immunotherapy, and radiation therapy, supportive agents, rehabilitation counseling, and other rehabilitation costs, are also included in direct Table 10–1.  Cost Domains DIRECT COSTS

INDIRECT COSTS

Medical

Hospitalizations Physician visits Home healthcare Hospice care Chemotherapy Radiation Rehabilitation Supportive agents

Non-​Medical

Transportation to and from medical care Housekeeping services Costs of relocating Alterations to property

Morbidity

Time lost from work/​lost productivity Time spent seeking medical care Caregiver time or changes in caregiver productivity

Mortality

Economic productivity lost due to premature death

INTANGIBLE/​PSYCHOSOCIAL COSTS

Pain Suffering Grief

costs. The majority of studies of the economic burden of cancer report direct medical costs (Yabroff et al., 2012), in part, because these data are more readily available. Studies of direct medical costs are typically based on billing system data from insurers, hospital discharge data, and large nationally representative surveys. In addition to costs for the provision of medical care, the direct non-​medical costs are those costs borne by patients, caregivers, and families. These costs include those associated with transportation to and from health providers, child and dependent care costs, and costs of relocating temporarily or permanently to improve access to specific treatments or facilities. Illness can force a family to incur expenses as part of caring for a cancer patient, including those for routine household tasks, special diets or clothing, items for rehabilitation, alterations of property, and vocational, social, and family counseling services. Expenditures for retraining or re-​education and interest lost on withdrawal of savings or interest charges on funds borrowed to pay illness-​ related expenses are also considered direct non-​medical costs. These data are not systematically collected as part of billing systems, and as a result, few studies report direct non-​medical costs.

Indirect Costs Indirect costs are the time and economic output lost or forgone due to disease and its treatment and any late and lasting effects of treatment on the usual activities of patients, family, and friends, including employment, housekeeping, volunteer activities, and leisure. These costs are not reflected by direct monetary transactions but do reflect the use of economic resources in response to disease and its treatment that could be used for other purposes in the absence of disease. Indirect costs are measured as the value of losses of economic output due to morbidity or premature mortality from disease. Cancer survivors may not be able to fully participate in work or other usual activities because of morbidity associated with the illness. Family members and others may spend time caring for the patient rather than pursuing other activities, may make unwanted job changes, or may miss opportunities for promotion and education.

 17



171

The Economic Burden of Cancer in the United States

Estimated morbidity costs of cancer can include losses measured by the value of forgone earnings and the imputed value of lost housekeeping services (measured by wages that would have to be paid to replace those services). A related type of indirect cost is the time the patient and/​or family spend receiving medical care, known as patient or caregiver “time costs.” Longitudinal morbidity data and patient time data are not collected systematically as part of billing systems, and as a result, few studies have reported these costs. However, several studies have reported productivity losses associated with morbidity costs in a given year using nationally representative cross-​sectional data (Ekwueme et al., 2014; Guy et al., 2013, 2014). Patient time costs have been estimated from medical service frequencies (e.g., hospitalizations, physician visits, chemotherapy) combined with time estimates of medical service duration and values for that time (e.g., median wages) (Yabroff et al., 2007, 2014). Mortality costs are the present value of future output lost because of premature death due to cancer. One of the main components of mortality costs is the remaining life expectancy in the absence of cancer death—​the person-​years of life lost (PYLL). The other main component of mortality costs is the time value of the PYLL. All measures of indirect costs require a method for estimating the value of time associated with morbidity, patient time, or the PYLL due to early death from cancer. Two general approaches are commonly used to value time: the human-​capital approach and the willingness-​to-​pay approach. In the human-​capital approach, sex-​and age-​specific earnings are combined with expected productivity trends and time lost to estimate unrealized lifetime earnings. This approach explicitly values the time of individuals with higher earning potential as greater than those with less earning potential, which is one of its limitations. The willingness-​ to-​pay approach, in contrast, incorporates both lost productivity and the intrinsic value of life by estimating the amount an average individual would be willing to pay for an additional year of life. Because incidence and mortality rates for most cancer sites are highest in the elderly, a population that is less likely to be in the workforce than their younger counterparts, comparing results from these two approaches for valuing time is particularly relevant for estimating the burden of cancer (Ramsey, 2008). Few studies have systematically measured indirect costs associated with cancer care. However, studies that have attempted to measure components of these costs have reported that these frequently unmeasured costs may be substantial (Bradley et al., 2008; Ekwueme et al., 2008; Li et al., 2010; Yabroff et al., 2007, 2008a).

Psychosocial or Intangible Costs Disease may also impact the quality of life of the patient, family, and friends in ways that are not reflected in the categories of direct or indirect costs. These are referred to as psychosocial or intangible costs. As the result of a cancer diagnosis, its treatment, and late or lasting effects of treatment, patients may suffer from reduced sexual functioning, disfigurement, disability, pain, or fears of life-​threatening disease. Patients and their families may make changes in life plans, which can induce anxiety, reduce self-​esteem and feelings of well-​being, or create resentment and family conflict. The combination of financial hardship and psychosocial problems can be especially devastating. For the purposes of cost-​effectiveness analysis, psychosocial cost is conceptualized as a quality of life outcome and is measured in terms of decrements of quality-​adjusted life-​years or in terms of utility. Developing appropriate concepts and measures of the quality of life ramifications of cancer and cancer treatment is currently an active area of cancer research.

WHAT ARE THE TEMPORAL PATTERNS OF THE ECONOMIC BURDEN OF CANCER? When measured longitudinally starting from cancer diagnosis, the monthly patterns of medical care use and the associated costs of cancer change over time. As noted earlier, and consistent with the

$18,000 $16,000 $14,000 $12,000

36 Month Survival

$10,000

48 Month Survival

$8,000

60 Month Survival

$6,000

72 Month Survival

$4,000

84 Month Survival

$2,000 $0 0

12

24

36

48

60

72

84

96

96 Month Survival

Month After Diagnosis

Figure 10–2.  Monthly Medicare payments for colorectal cancer patients by length of survival. Source: Yabroff KR, Warren JL, Schrag D, Mariotto M, Meekins A, Topor M, Brown ML. Comparison of approaches for estimating incidence costs of care for colorectal cancer patients. Med Care. 2009 Jul;47(7 Suppl 1):S56-​63. doi: 10.1097/​MLR.0b013e3181a4f482. PubMed PMID: 19536010.

intensity of treatment for initial care, recurrence, and end-​of-​life care, cancer costs are highest in the initial period following diagnosis and, among patients who die from their disease, at the end of life (Brown et al., 2002; Fireman et al., 1997; Riley et al., 1995; Taplin et al., 1996; Yabroff et al., 2007, 2008b, 2009). Costs are lowest in the period between the initial and end-​of-​life periods, following a U-​ shaped curve. As shown in Figure 10–2, this U-​shaped medical care cost pattern is consistent across cohorts of cancer patients with very different survival times following diagnosis (Yabroff et  al., 2009). The width and height of this U-​shaped cost curve vary by cancer site, stage at diagnosis, and patient age (Brown et al., 2002; Riley et al., 1995; Taplin et al., 1996; Yabroff et al., 2007, 2008b, 2009). All of these factors affect summary measures of costs and inform the methods used to estimate these costs.

INCIDENCE AND PREVALENCE CANCER COSTS Direct, indirect, and intangible cancer costs are typically reported starting at diagnosis for a group of cancer patients defined by clinical characteristics (incidence costs), or for all cancer survivors alive in a specific year (prevalence costs). An incidence cost estimate in 2015 will include only those patients diagnosed within this year, and costs will begin accumulating at the high-​cost diagnosis period for all patients. Incidence cost estimates can range from periods of less than 1 year to a lifetime (for patients diagnosed in 2015 in this example). A  prevalence cost estimate in 2015 will include all cancer patients alive at any point during the year, but will only include the portion of the cost trajectory that occurs during 2015. Prevalence cost estimates are typically reported as annual estimates or for a particular year. Importantly, the high-​and low-​cost portions of the U-​shaped curve included in the prevalence cost estimate will vary by cancer site. For example, because long-​term survival following breast and prostate cancer diagnosis is high, in any given year, the majority of breast and prostate cancer survivors will be in the low-​cost bottom of the U-​shaped curve. In contrast, because survival following pancreatic cancer diagnosis is generally poor, in any given year the majority of pancreatic cancer survivors will be in the high-​cost period following diagnosis or at the end-​of-​life, the arms of the “U.” As with other measures of disease burden, per capita incidence and prevalence cost estimates in a given year will be relatively similar when survival is short, because the vast majority of individuals will have high costs in either arm of the “U,” with few individuals with low costs at the bottom of the “U” (e.g., pancreatic cancer). For cancers with long survival (e.g., breast and prostate cancers), per-​capita prevalence costs

172

172

PART II:  THE MAGNITUDE OF CANCER

in a given year will be significantly lower than per-​capita incidence costs in that year, because the prevalence cost estimate will be predominantly composed of survivors at the bottom of the “U,” where costs are lowest. Both incidence and prevalence cost measures can be useful for resource allocation and policy and program planning. Incidence costs are commonly used in cost-​effectiveness models, for decisions about specific therapies or understanding patient and treatment trajectories, whereas prevalence costs are most commonly used in understanding the overall impact of disease on local, federal, or health plan budgets.

APPROACHES FOR ESTIMATING CANCER COSTS Several approaches are commonly used for estimating incidence and prevalence cancer costs. As with other measures of disease burden, cohort studies, both retrospective and prospective, can be used for estimating incidence costs. Cross-​sectional data can be used for estimating prevalence costs, although in some situations, such as the use of billing records (without linkage to cancer registry data), the prevalence estimate will reflect “treated prevalence,” rather than prevalence cost measured from all cancer survivors alive in a specific year. Only patients receiving treatment in a specific year can be identified from claims for cancer treatment. Other approaches for estimating incidence and prevalence costs include modeling or microsimulation. This approach is typically based on well-​defined hypothetical cohorts of cancer patients in a natural history of disease model where the probability of medical events (e.g., hospitalizations for a side effect of treatment) is combined with cost estimates for each of the medical events (e.g., average hospitalization cost for patients with that side effect of treatment). Another approach, the phase of care approach, uses the U-​shaped cost pattern described in Figure 10–2 to divide healthcare services, costs, and observation time for cancer patients into clinically relevant periods or phases in relation to the date of diagnosis and the date of death. Three phases, corresponding to the arms of the “U” and the bottom of the “U” (i.e., initial, last year of life, and continuing phases, respectively), have been commonly used. Phase-​specific estimates can then be used as an input to estimate either incidence or prevalence costs. Phase-​specific cost estimates can be combined with survival probabilities following diagnosis to yield modeled incidence costs (Yabroff et al., 2009) or with phase-​specific cancer prevalence estimates in a given year to produce prevalence costs for that year (Mariotto et  al., 2011). The phase-​of-​care approach is a more efficient use of data than a cohort approach because patients can contribute to multiple phases and this can lead to more robust estimates, particularly for less common cancer sites.

WHAT ARE SOME RECENT ESTIMATES OF THE COSTS OF CANCER CARE? In this section, we present recent estimates of the direct medical costs of cancer and report both incidence and prevalence cost estimates. We also present estimates of indirect costs of cancer, including annual productivity losses due to employment disability, missed workdays, and lost household productivity, and mortality costs associated with early death from cancer.

Estimates of Direct Medical Costs A number of studies have estimated the direct medical costs associated with cancer (Bernard et  al., 2011; Ekwueme et  al., 2011; Guy et al., 2013, 2014; Howard et al., 2004; Mariotto et al., 2011; Short et al., 2011). These studies include estimates of incidence costs among the newly diagnosed and annual prevalence costs among all cancer patients alive in a given year.

Incidence Cost Estimates

Costs associated with cancer for newly diagnosed elderly patients in the first year following diagnosis range from a little more than $10,000

per person for breast and prostate cancer in women and men, respectively, to more than $25,000 per person for colorectal, lung, and pancreas cancers (in 2004 dollars; Yabroff et  al., 2008b; Figures 10–3a and 10–3b). Differences in 1-​year incidence costs by cancer site reflect differences in the distribution of stage of disease at diagnosis, survival following diagnosis, and treatment intensity, especially for hospitalizations, by cancer site. The majority of breast and prostate cancer patients are diagnosed with early stage disease, have long survival, and do not require extensive hospitalization, whereas the majority of lung and pancreatic cancer patients are diagnosed with advanced disease and have relatively short survival, often requiring extensive hospitalization. These per-​person incidence estimates can be applied to national estimates of the number of newly diagnosed elderly cancer patients to generate a national incidence cost estimate for a specific year. Even though the per-​ person 1-​ year estimates of the incidence costs for pancreas cancer are among the highest ($29,814 in men and $30,357 in women), approximately 21,000 elderly patients were diagnosed with pancreatic cancer in 2004, yielding a national 1-​year incidence cost estimate of $642 million. On the other hand, the per-​person costs associated with female breast cancer and prostate cancer are relatively low ($11,748 and $10,626), but more than 77,000 elderly female breast cancer patients and 118,000 elderly prostate cancer patients were diagnosed in 2004, yielding much larger national 1-​year incidence cost estimates of $905 million and $1,254 million, respectively.

Prevalence Cost Estimates

Several studies have reported prevalence costs for cancer survivors (Bernard et  al., 2011; Guy et  al., 2013, 2014; Howard et  al., 2004; Short et al., 2011). Mean annual direct medical costs have been consistently reported to be greater for cancer survivors than for individuals without a history of cancer (Bernard et al., 2011; Guy et al., 2013, 2014; Howard et al., 2004; Short et al., 2011). In addition, as shown in Figure 10–4, mean annual prevalence costs for the recently diagnosed are greater than for the previously diagnosed and individuals without a cancer history in those aged 18–​64 years and 65+ years (Guy et al., 2013). Mean annual prevalence costs are higher for the older age group than in the younger age group, reflecting greater comorbidity burden and healthcare utilization in those ages 65+, for individuals with and without cancer. The distribution of costs by type of medical care service also varies between cancer survivors and individuals without a cancer history (Guy et al., 2013). Among those aged 18–​64 years, ambulatory care and inpatient care are the most common services for recently diagnosed cancer survivors (46.2% and 40.5%, respectively), with prescription medications contributing only a small portion of costs (8.1%). For previously diagnosed cancer survivors aged 18–​64, prescription medications contribute a larger portion of costs (18.9%); and for individuals without a cancer history, the contribution of ambulatory care (33.9%), inpatient care (27.4%), and prescription medications (22.6%) are more similar. The distribution of costs by service type is similar for the older age group, with ambulatory and inpatient care the most common service category in recently diagnosed cancer survivors (41.2% and 41.9%, respectively); prescriptions are a larger portion for the previously diagnosed (18.9%), and ambulatory care (25.9%), inpatient care (34.3%), and prescription medications (24.3%) have similar distributions for those without a cancer history. The distribution of direct medical costs by payer type is quite different between individuals aged 18–​ 64  years and 65+ years (Figures 10–5a and 10–5b). In the younger group with and without cancer, private health insurance is the predominant payer for medical care, whereas in the older group, the Medicare program is the predominant payer. A study using the phase-​of-​care approach to estimate national prevalence costs in the United States reported costs associated with cancer of $124.6 billion dollars (in 2010 dollars) (Mariotto et al., 2011). The highest costs were for female breast ($16.5 billion), colorectal ($14.1 billion), lymphoma ($12.1 billion), lung ($12.1 billion) and prostate ($11.9 billion) cancers, reflecting the absolute number of cancer

 173



The Economic Burden of Cancer in the United States

173

(a) Bladder

$14,313 $11,748

Breast Colorectal

$25,903

Corpus uteri

$16,485 $22,633

Head and Neck Kidney

$23,836

Liver

$25,398

Lung

$26,764

Pancreas

$30,557 $0

$5,000

$10,000

$15,000

$20,000

$30,000

$25,000

$35,000

(b) Bladder

$13,466

Colorectal

$25,711

Head and Neck

$21,052

Kidney

$23,920

Liver

$24,086

Lung

$26,449

Pancreas

Figure 10–3.  (a) Direct medical costs in year following diagnosis, by cancer site in elderly female patients. (b) Direct medical costs in year following diagnosis, by cancer site in elderly male patients. Note: Calculations

made using the linked SEER-​Medicare data, 1999–​2003. Estimates are the net estimates based on controls. Source:  Yabroff KR, Lamont EB, Mariotto A, Warren JL, Meekins A, Topor M, Brown ML. 2008. Cost of care for elderly cancer patients in the United $35,000 States. J Natl Cancer Inst, 100, 630–​641.

$29,814

Prostate

$10,626 $0

$5,000

$10,000

$15,000

$20,000

survivors by phases of care and phase-​specific cost estimates by cancer site. A larger proportion of prevalence costs occurred in the continuing phase of care for cancers with longer survival, such as breast and prostate cancers, than for those with short survival, such as lung and pancreas cancers.

Indirect Costs To date, the majority of studies estimating indirect costs associated with cancer have reported prevalence estimates. These studies have mainly reported productivity losses from employment disability (being unable to work due to health), missed worked days among the employed, and lost household productivity (Guy et  al., 2013, 2014) or mortality costs associated with premature death due to cancer (Bradley et al., 2008; Ekwueme et al., 2008; Li et al., 2010; Yabroff et  al., 2008a). As shown in Figures 10–6a and 10–6b, total per-​person productivity losses are highest among recently diagnosed cancer survivors aged 18–​64 ($4,694 in 2010 US dollars) and 65+ ($6,133 in 2010 US dollars), compared with the previously diagnosed, and individuals without a cancer history. As shown in Figure 10–7, per-​person productivity losses are higher for male and female cancer survivors

$25,000

$30,000

($3,719 and $4,033 in 2011 US dollars, respectively) than for males and females without a cancer history ($2,260 and $2,703 in 2011 US dollars, respectively) (Ekwueme et al., 2014). For all groups, employment disability is the largest component of productivity loss, followed by missed workdays and lost household productivity. A number of studies have estimated mortality costs and the productivity losses associated with premature death due to cancer (Bradley et al., 2008; Ekwueme et al., 2008; Li et al., 2010; Yabroff et al., 2008a). Table 10–2 lists the number of cancer deaths in 2005, person-​years of life lost (PYLL), and present value of lifetime earnings by cancer site, using the human capital approach for valuing the PYLLs (Bradley et  al., 2008). Because this approach uses gender-​ and age-​specific earnings to value the years with unrealized earnings, cancers that affect individuals with lower earnings potential will have less productivity loss than cancers that affect individuals with higher earnings potential. For example, the majority of men who die from prostate cancer are ages 65 and older, an age when most are retired and would not otherwise be earning wages from work in the future. Although the number of prostate cancer deaths is higher than the number of pancreas cancer deaths, the PYLL is lower, and the present value of lifetime earnings is lower ($3.3 billion vs. $6.6 billion).

Ages 18–64

174

Recently diagnosed

$17,170

Previously diagnosed No cancer history

$6,485 $3,611

Ages 65+

Recently diagnosed

$23,441

Previously diagnosed

$12,357

No cancer history $0

$8,724 $5,000

$10,000

$15,000

$20,000

$25,000

Figure 10–4.  Mean annual direct medical costs in cancer survivors and individuals without a cancer history. Notes: Calculations using the Medical Expenditure Panel Survey–​Household Component, 2008–​2010. Estimates are predicted margins from a generalized linear regression model with a gamma distribution and a log link controlling for age, sex, race/​ethnicity, and number of comorbid conditions. All monetary amounts are in 2010 dollars. Source: Guy GP Jr, Ekwueme DU, Yabroff KR, et al. 2013. The economic burden of cancer survivorship among adults in the United States. J Clin Oncol, 31(30), 3749–3757. PMCID: PMC3795887.

(a) Recently diagnosed cancer survivors

No cancer history

7.3% 6.1%

9.8% 16.9%

17.1%

Out-of-pocket

10.6%

Private health insurance Medicare

2.8%

8.2%

Medicaid 66.7%

Other

$17,170 (95% Cl: $13,433, $20,907)

$3,611 (95% Cl: $3,486, $3,736)

(b) Recently diagnosed cancer survivors 1.1%

10%

54.6%

No cancer history

6.9%

8.7% 4.4%

11.3%

13.9%

Out-of-pocket 13.1%

Private health insurance Medicare Medicaid 70.7%

$23,441 (95% Cl: $18,367, $28,514)

Other 59.9%

$8,724 (95% Cl: $8,302, $9,147)

Figure 10–5.  (a) Annual medical expenditures by payer type, for individuals aged 18–​64 years. (b) Annual medical expenditures by payer type, for individuals aged 65+ years. Notes: Calculations using the Medical Expenditure Panel Survey–​Household Component, 2008–​2010. Estimates are predicted margins from a generalized linear regression model with a gamma distribution and a log link controlling for age, sex, race/​ethnicity, and number of comorbid conditions. All monetary amounts are in 2010 dollars. Source: Guy GP Jr, Ekwueme DU, Yabroff KR, Dowling EC, Li C, Rodriguez J, DeMoor J, Virgo K. 2013. The economic burden of cancer survivorship in the United States. J Clin Oncol, 31(30), 3749–​3757.

 175



175

The Economic Burden of Cancer in the United States (a)

$7,000

$5,000

Employment disability

Missed work days

Missed work days

$6,000

Lost household productivity

$2,664 $2,833

$2,000 $1,000 $0

US dollars

US dollars

$4,000

$3,000 $2,961

$2,831

$2,000 $2,109

$1,585

$1,644

Lost household productivity

$4,000

$5,000

$3,000

Employment disability

$386

$447 $313

$321 $134

Recently diagnosed cancer survivors

Previously diagnosed cancer survivors

No cancer history

$1,862 $1,000

$0

$686

$597

$386

$933 $201

Cancer history

No cancer history

$291 Cancer history

Women Employment disability

(b) $7,000

Missed work days Lost household productivity

$6,000

US dollars

$4,350

$3,000

$4,485 $3,777

$2,000 $1,000 $0

$1,109 $676

Men

Figure 10–7. Annual lost productivity among cancer survivors and individuals without a cancer history. Notes: Calculations using the Medical Expenditure Panel Survey–​ Household Component, 2008–​ 2011. All monetary amounts are in 2010 dollars. Source: Ekwueme DU, Yabroff KR, Guy GP Jr, et  al. 2014. Medical costs and produc­tivity losses of cancer survivors: United States, 2008–2011. MMWR Morb Mortal Wkly Rep, 63 (23), 505–510. PMID: 24918485.

$5,000 $4,000

$267 $131 No cancer history

$428 $382

Recently Previously diagnosed diagnosed cancer survivors cancer survivors

$331 $301 No cancer history

Figure 10–6. (a) Annual lost productivity among cancer survivors and individuals without a cancer history, aged 18–​64 years. (b) Annual lost productivity among cancer survivors and individuals without a cancer history, aged 65+ years. Notes: Calculations using the Medical Expenditure Panel Survey–​House­hold Component, 2008–​2010. All monetary amounts are in 2010 dollars. Source: Guy GP Jr, Ekwueme DU, Yabroff KR, Dowling EC, Li C, Rodriguez J, DeMoor J, Virgo K. 2013. The economic burden of cancer survivorship in the United States. J Clin Oncol, 31(30), 3749–​3757.

Other approaches to value the PYLL, such as the willingness-​to-​pay approach, incorporate both lost productivity and the intrinsic value of life and place an equivalent value on all years of life. Estimates of the value of life lost for prostate cancer are substantially higher when the willingness-​to-​pay approach is used to value the PYLL (Yabroff et al., 2008a) than a human capital approach because older men dying of prostate cancer have lower earnings potential. The direct medical cost and indirect cost estimates presented in this section were calculated from a variety of existing data sources. In the next section, we describe common data sources for estimating economic burden of cancer in the United States in more detail.

WHAT DATA SOURCES ARE USED FOR ESTIMATING THE ECONOMIC BURDEN OF CANCER? In this section, we discuss different types of data commonly used for estimating the economic burden of cancer in the United States, including hospital discharge data, health insurance claims, and household surveys (Lund et al., 2009; Yabroff et al., 2011). Selected publicly available data sources that are commonly used to estimate aspects of the burden

of cancer are listed in Table 10–3, including the Healthcare Costs and Utilization Program (H​CUP) discharge data, the linked Surveillance Epidemiology and End Results cancer registry–​ Medicare insurance claims data (SEER-​Medicare), the Medical Expenditure Panel Survey (MEPS), and the Medicare Current Beneficiary Survey (MCBS). Features of these data sources are described in the following, including population coverage, patient characteristics, information about cancer diagnosis, and the availability of information about the burden of cancer, including direct, indirect, and intangible costs. We also discuss whether these data sources can be used to estimate incidence and prevalence costs associated with cancer. Table 10–2.  Lost Productivity Due to Cancer Deaths in the US Among Adults Aged 20 Years and Older by Cancer Site (2005)

Lung and bronchus Female breast Colon and rectum Pancreas Brain and other nervous system Leukemia Non-​Hodgkin’s lymphoma Liver and intrahepatic bile duct Kidney and renal pelvis Head and neck Prostate Stomach Melanoma of the skin Ovary Cervix uteri Urinary bladder Hodgkin lymphoma Source: Bradley et al., 2008.

Number of Deaths in 2005

Person-​Years of Life Lost (in thousands)

Present Value of Lifetime Earnings (in billions of dollars)

166,755 44,546 61,240 32,054 13,607

2569.8 874.0 861.4 472.5 314.2

$36.131 $12.096 $10.652 $6.610 $5.743

22,945 23,850

390.1 362.3

$5.722 $5.511

14,502

242.8

$4.420

12,819 10,830 33,642 13,351 8062 15,253 4338 13,275 1414

206.2 181.4 300.5 201.3 155.1 274.5 114.9 150.5 36.4

$3.424 $3.413 $3.301 $3.225 $3.194 $2.824 $1.827 $1.826 $0.833

176

176

PART II:  THE MAGNITUDE OF CANCER

Table 10–3.  Characteristics of Selected Publicly Available Data Sources/​Research Resources in the US for Estimating Economic Burden of Cancer

Description DATA CHARACTERISTICS National or nationally representative Individual-​level longitudinal data Approximate number of cancer survivors in 2015 Duration of information

SEER-​Medicare

MEPS

H​CUP

MCBS

SEER Tumor Registries Linked to Medicare Claims

Nationally Representative In-​Person Survey with Provider Data Collection

Inpatient Discharge Data from Sampled Hospitals

National Survey Linked to Medicare Claims

Geographically defined √ > 1,000,000

√ 5 panels over 2 years < 2000

√ > 1,000,000

√ √ < 2000

2 years

Health insurance type

Medicare eligibility through death Medicare fee-​for-​service only

All payers

Hospital admission through discharge All payers

All payers

PATIENT INFORMATION Age distribution

Aged 65+ or disabled (any age)

Aged 18+

All ages

√

√

Self-​report, procedure or diagnosis codes

Procedure or diagnosis codes Self-​report, procedure or diagnosis codes

√

Inpatient hospital only

√

√

√

√

√ √ √ √

√

√ √ √* √

Information about patients without cancer In cancer registry regions CANCER INFORMATION Cancer diagnosis Stage at diagnosis Treatment

Registry, procedure or diagnosis codes √ √

TYPE OF COST ESTIMATE Incidence Prevalence in a specific year

√ √

COST DOMAINS Direct Medical Cost Components Hospital Physician and other outpatient services Outpatient pharmacy Out of pocket Indirect Cost Components Productivity loss (e.g., days lost from work) Patient time Caregiver time Intangible costs

√ √ √*

Aged 65+ or disabled (any age) √

√ √ √

√

*Data on Medicare Part D prescription drug services are available starting in 2006. Before 2006, drugs administered parenterally, and their administration was covered by Medicare Part B, as were Prodrugs, the oral drug equivalent of drugs administered parenterally. SEER-​Medicare = Surveillance Epidemiology and End Results–​Medicare; MEPS = Medical Expenditure Panel Survey; H-​CUP = Healthcare Costs and Utilization Program; MCBS = Medicare Current Beneficiary Survey.

Healthcare Costs and Utilization Project (HCUP) The HCUP is a family of national and state healthcare databases and tools sponsored by the Agency for Healthcare Quality and Research, and includes the National (Nationwide) Inpatient Sample (NIS) and State Inpatient Databases (SID), Nationwide Emergency Department Sample and State Emergency Department Databases, and State Ambulatory Care Databases. The NIS and SID include medical care use and associated costs for individuals of all ages with all types of health insurance, but for only one aspect of medical care, inpatient hospitalizations (Agency for Healthcare Research and Quality, 2016). These NIS data have been available every year since 1988. The unit of observation is the hospitalization rather than the individual patient, so multiple hospitalizations for the same person are unique observations in the NIS and cannot be linked at the individual level to provide longitudinal information about care. In some states, linkage for multiple hospitalizations and across HCUP databases is available, although these data have not been commonly used to estimate cancer burden. Although millions of cancer survivors can be identified in the NIS by diagnosis and procedure codes, information about date of diagnosis, stage, additional cancers, pre-​hospitalization comorbidity, and other factors that influence patient eligibility for specific treatments is not available. Further, information about cancer survivors treated outside the hospital inpatient setting is unavailable in

the NIS. The NIS has been mainly used to estimate the use and costs of surgery among hospitalized cancer survivors (Newton and Ewer, 2010; Seifeldin and Hantsch, 1999). Because information about the date of cancer diagnosis is not available, the estimates from the NIS are prevalence estimates. Although the number of hospital days can be used to identify patient time in the hospital, these data have rarely been used for estimating any types of costs other than direct medical costs.

Linked SEER-​Medicare Data The linked SEER-​Medicare data are made available by the National Cancer Institute and contain longitudinal information about medical care and associated payments received before, during, and after cancer diagnosis for Medicare beneficiaries (Warren et al., 2002). Information about cancer diagnoses is available for the years that each geographically defined registry has been part of the SEER program, starting in 1973. Medicare claims since 1991 are available for two cohorts of beneficiaries included in the SEER-​Medicare data: cancer survivors and a sample of Medicare beneficiaries residing in the SEER areas who do not have cancer. Because more than 95% of individuals over the age of 65 in the United States are enrolled in the Medicare program, and SEER registries currently cover approximately 26% of the US population, these data contain information about cancer care for a large

 17



The Economic Burden of Cancer in the United States

portion of the elderly in the United States. More than 3 million cancer survivors can be identified from SEER-​Medicare with detailed information about cancer stage, grade, and histology for each diagnosis. Identification of incident cancer patients and patients with multiple cancers are additional strengths of the linkage of health insurance program data with cancer registry data. As with other health insurance program data (also known as administrative data), Medicare claims do not contain any information about individuals without health insurance or the utilization and costs of specific services covered by other insurance programs or those services paid entirely out of pocket. Medicare payments represent approximately 51%–​ 65% of all direct medical care costs in the elderly (Crystal et al., 2000; Guy et  al., 2013), with the remaining costs covered by other insurers or by out-​of-​pocket payments. These linked SEER-​Medicare data have been used extensively to estimate the incidence and prevalence costs of medical care associated with cancer in the elderly (Mariotto et al., 2011; Warren et al., 2008). Additionally, the costs associated with patient time have been estimated from patients’ medical service frequencies (e.g., hospital length of stay, physician visits, chemotherapy), combined with average service-​specific time estimates (Yabroff et al., 2007). Notably, the linked SEER-​Medicare data cannot be used to estimate costs associated with cancer care in the working-​age population under the age of 65 (who are only eligible for Medicare in limited situations), and information about employment and health status is unavailable.

Medical Expenditure Panel Survey The Medical Expenditure Panel Survey (MEPS) is an annual nationally representative household survey conducted by the Agency for Healthcare Research and Quality (AHRQ) (Cohen et al., 2009). The MEPS was first fielded in 1996 and consists of three surveys:  the Household Component (HC), the Medical Provider Component, and the Insurance Component. The MEPS-​ HC collects demographic, health status, access to medical care, employment, and healthcare utilization and expenditure data for adults of all ages, with all insurance types (Table 10–3). The MEPS also collects these data for individuals without health insurance. The MEPS-​HC uses an overlapping panel design, in which each panel consists of individuals who are interviewed in person five times (or rounds) over approximately 2.5  years. Data from two panels are combined to produce estimates for each calendar year, and each panel separately produces longitudinal estimates. Through panel 11 in 2007, cancer survivors were only identified in the MEPS if they received care for cancer, or missed work or school or spent a day in bed because of cancer, during the survey period (“treated prevalence”). Only very small numbers of cancer survivors were identified in a given year during this period, although multiple years of the MEPS have been combined to assess economic burden (Howard et al., 2004; Short et al., 2011). Starting in panel 12 and in 2008, a new question was added for adults aged 18 years and older about whether a doctor or other health professional ever told the person that he or she had cancer or a malignancy of any kind. Rather than only identifying the subset of cancer survivors undergoing cancer treatment at the time of the survey, this change now results in the identification of all individuals with a history of cancer, consistent with definitions of cancer survivorship used to estimate cancer prevalence in the United States (de Moor et al., 2013; Mariotto et al., 2011). Notably, the question about a cancer diagnosis is asked for all adults in the family, making the MEPS one of the only data sources to identify cancer history for the family, and these data can be used to assess the effects of cancer on a spouse (Litzelman and Yabroff, 2015). Details about cancer diagnosis, including stage, tumor characteristics, and other prognostic information, are unavailable. Several studies have combined multiple years of the MEPS to describe healthcare utilization and expenditures in cancer survivors receiving cancer care compared with individuals without cancer (Bernard et al., 2011; Finkelstein et al., 2009; Guy et al., 2013, 2014; Howard et al., 2004; Short et al., 2011). These MEPS data have also been used to estimate the costs associated with patient time from medical service frequencies (e.g., hospital length of stay, physician visits, chemotherapy), combined with average service-​specific time estimates (Yabroff et al.,

177

2014). Because date of diagnosis is not available, cost estimates from the MEPS are prevalence cost estimates. In 2011, the MEPS Experiences with Cancer survey was a self-​ administered questionnaire of adult cancer survivors, containing questions on financial hardship related to cancer, its treatment, and lasting and late effects of treatment, such as having to borrow money or go into debt, being unable to cover medical care costs, worry about paying medical bills, and other financial sacrifices (Yabroff et al., 2012). It also included a number of questions related to productivity for cancer survivors and their caregivers. An updated MEPS Experiences with Cancer survey of adult cancer survivors is being fielded in 2016.

Medicare Current Beneficiary Survey The Medicare Current Beneficiary Survey (MCBS) is a longitudinal, nationally representative survey of Medicare beneficiaries in the United States conducted by the Centers for Medicare & Medicaid Services. The MCBS collects information about expenditures and sources of payment for all services used by Medicare beneficiaries, including copayments, deductibles, and non-​covered services. In addition, demographics, health status, access to medical care, and physical functioning are collected about individuals with all types of health insurance. The survey is conducted over a 4-​year period, with multiple interviews a year. The MCBS Cost and Use files link Medicare claims from the population with fee-​for-​service coverage to the survey. Cancer survivors can be identified based on their response to the question “Have you ever been told by a doctor or other health professional that you had cancer or a malignancy of any kind?” or with claims algorithms using diagnosis codes. The survey can be used to supplement the claims data to explore out-​of-​pocket spending or use of strategies to reduce the costs of medications (Davidoff et al., 2013; Nekhlyudov et al., 2011). Cost estimates from the MCBS are prevalence cost estimates. Nationally representative surveys such as the MEPS or the MCBS are limited by small numbers of patients who are newly diagnosed or diagnosed with cancers associated with short survival (e.g., lung) or low incidence (e.g., testicular). Details about cancer diagnosis, including stage, clinical characteristics and other prognostic information, and treatment are generally not available. Because questions are not specific to burden of illness related to cancer (except in the MEPS Experiences with Cancer survey), these MEPS and MCBS data have been previously used to assess employment, insurance coverage, access to care, and health status in cancer survivors compared with individuals without a history of cancer (Davidoff et al., 2013; Dowling et al., 2010; Guy et al., 2013; Nekhlyudov et al., 2011; Yabroff et al., 2004). As described earlier and as illustrated in Table 10.3, commonly used publicly available data sources have different strengths and limitations for estimating the burden of cancer in the United States. No single data source contains comprehensive longitudinal information for large numbers of cancer survivors about multiple healthcare services across insurance type (including the uninsured) in the elderly or in younger populations. Further, few data sources contain information about indirect costs, including patient and caregiver time costs and lost productivity due to morbidity (Lund et al., 2009).

WHAT ARE KEY FACTORS RELATED TO THE MEASUREMENT OF HEALTHCARE COSTS? There are several key factors that are critical to the measurement of the economic burden of cancer, including the perspective of the economic analyses, consideration of current and future costs, and measurement of cancer-​attributable costs. These issues are defined and discussed in following sections.

The Frame of Reference or Perspective The frame of reference of an economic study is the viewpoint from which the analyses are conducted. As shown in Table 10–4, examples of perspectives are those of the patient and family, employers, healthcare insurers or payer, including federal and state programs such as

178

178

PART II:  THE MAGNITUDE OF CANCER

Table 10–4.  Perspectives and Relevant Cost Categories Perspective Cost Category DIRECT COSTS

Patient and Family Medical

Health Insurer/​Payer

Society

√ Payments for services

√

√ Premiums, copayments and other out-​of-​pocket √ Transportation, housekeeping

√ Premiums

Morbidity

√ Reduced productivity, time seeking medical care

√ Reduced productivity, presentism, absenteeism, disability payments, replacement costs

Mortality

√ Lost productivity

√

√

√

Non-​Medical INDIRECT COSTS

Employer

INTANGIBLE/​PSYCHOSOCIAL COSTS

Medicare and Medicaid, and society. Depending on the perspective of the analysis, the types of costs included, sources of data, and study design will vary. While the societal perspective is preferred for purposes of economic analysis for general policy analysis, for specific purposes, such as the impact of new healthcare legislation, it is often useful to define a frame of reference that is narrower than all of society (Gold et al., 1996). For example, while the societal perspective may be the relevant perspective for policy analysis involving approval of an innovative cancer control intervention, the perspective of the Medicare program or a health maintenance organization may be relevant for assessing the financial impact of such a decision on resource requirements required to fund these operations. Not all social costs will be recognized as costs at various organizational levels. Time lost from paid work for unpaid caregiving by the relatives of the cancer patient is a cost to society in terms of lost economic productivity and a cost to the family in terms of lost household income, but it might represent a savings to the Medicare program in terms of a shorter length of hospital stay or less provision of formal home-​care services. As noted previously, the time spent by cancer patients and family members related to cancer and its treatment is rarely fully accounted for among existing studies because these data are not collected systematically. Excluding patient time costs from analyses comparing alternative interventions (e.g., cost-​effectiveness analyses) may result in favoring interventions with large patient-​time cost burdens (Russell, 2009).

Discounting When costs are distributed over time and the purpose of analysis is to assess these costs relative to an initial investment, such as a cancer screening program to promote early detection, economists use an adjustment known as discounting. The purpose of discounting is to adjust the value of costs and benefits (savings) incurred in the future to their present market value because the same resources, if available and invested today, would yield a return if placed in a productive activity. In the United States, the most commonly used value in current health economics literature is 3% (in real terms—​that is, adjusted for inflation). Other industrialized countries have used different discount rates, and there is some discussion about what the appropriate discount rate should be from a social viewpoint. The choice of a discount rate reflects a value judgment: a low discount rate indicates a long time horizon, and a high discount rate indicates a short time horizon. All else being equal, the use of a high discount rate would tend to favor investment in a treatment program where benefits would be realized more rapidly over investment in a prevention program where ongoing costs are incurred well in advance of the realization of any benefits.

Adjusting for Inflation It is often useful to express the cost of cancer in current-​year dollars. Because the rate of inflation for the price of healthcare services has consistently been higher than the rate for the economy as a whole, it

√ √

is necessary to use a price index specific to this sector of the economy. Direct medical cost estimates using multiple years of data (e.g., 2008–​ 2011) are typically inflated using a price index to the latest year of data and reported for that year (e.g., in 2011 US dollars). The healthcare component of the Consumer Price Index (CPI) is often used for this purpose. This index, however, refers only to the components of healthcare expenditures paid for directly by consumers. Other prices indices are used in other settings, such as the Personal Health Care Expenditure Price Index or indices that are designed to reflect healthcare price inflation as experienced by the Medicare program (Agency for Healthcare Research and Quality, 2015a). Note that adjustments for inflation and the discounting of future costs and benefits are distinct and unrelated issues.

Charges, Reimbursements, Expenditures, and Costs The economic concept of cost relates to the use of specific resources in an efficient manner. Direct observation of resource use is rarely available, however. Reflections of cost are available, such as the amount billed, charged, or reimbursed for the provision of healthcare services. The amount billed or charged may not reflect the actual cost associated with providing healthcare services, however, and may have little relationship to underlying costs (Finkler, 1982). For example, some hospitals or other providers may negotiate with health plans for reduced payments for specific patients. Here, the actual “cost” of the service may be unrelated to the amount charged to any of the purchasers. When using claims data, researchers frequently use payments to reflect costs, rather than charges (Yabroff et al., 2013). Charges may also be converted into costs using “cost-​to-​charge” ratios (Agency for Healthcare Research and Quality, 2015b).

Measurement of Costs Attributable to Cancer There are two general approaches to estimating the costs attributable to cancer:  estimating costs for specific events assumed to be cancer-​ related, and comparing costs of all care for cancer patients to other individuals without a cancer history. In the first approach, researchers classify specific costs as being due to cancer or not. This approach can be used in observational studies and is more frequently used in microsimulation modeling, where treatment scenarios are developed as a series of probabilistic events for a well-​defined, hypothetical population of cancer patients. The cost of medical services and procedures defined by the scenario can then be estimated using a cost estimate from another data source. This approach may underestimate the frequency of services that cancer patients receive for seemingly unrelated care. In the second approach, researchers measure attributable costs of cancer care by comparing the costs of all care for cancer patients or survivors to those of non-​cancer patients or patients without a cancer history. The difference is a net cost. The comparison can match cancer patients and individuals without a cancer history by relevant characteristics, such as age, gender, geographic region, and comorbidity status, or can control for these characteristics with multivariable analyses.

 179



The Economic Burden of Cancer in the United States

This approach is particularly advantageous when using administrative data because it minimizes reliance on diagnostic and procedure codes for identifying services as cancer-​related or not. In some situations, such as the estimation of costs associated with lung cancer, this approach may overstate services as attributable to cancer that are more common among lung cancer patients. For example, emphysema is more common among smokers who are also more likely to develop lung cancer. Costs associated with emphysema care would also be included in any net cost of lung cancer.

WHAT ARE EMERGING ISSUES IN THE ECONOMIC BURDEN OF CANCER IN THE UNITED STATES? In this section, we discuss several emerging issues related to the economic burden of cancer in the United States, including precision medicine and financial hardship associated with cancer care.

Precision Medicine The average cost of newly developed cancer treatments, primarily targeted therapies used in precision medicine, has risen dramatically over the past decades (Bach, 2009; Schrag, 2004), and many have price tags of more than $10,000 per month. Advances in the identification of genetic mutations associated with treatment response allow the tailoring of treatment to individual patients. Limiting the use of expensive therapies to patients with genetic profiles associated with treatment response has the potential to reduce the program costs of these treatments. For example, colorectal cancer patients carrying KRAS mutations do not respond to treatment with cetuximab or panitumumab (Amado et al., 2008), and by restricting the use of cetuximab to patients without the KRAS mutations, the incremental cost-​effectiveness ratio is more favorable (Mittmann et al., 2009). However, because the goal of precision medicine is to identify patients most likely to respond to targeted therapies, all potentially eligible patients must be tested, including those for whom treatment will not ultimately be indicated. Finally, the increase in median survival attributed to treatment with many of these targeted therapies, even among those most likely to respond, is typically only a few additional months (Fojo et  al., 2014). Currently, little is known about the utilization and effectiveness of precision medicine therapies in community practice. Important areas for future research include estimating and projecting their uptake, patterns of care, and cost; the appropriate use of genetic tests to personalize therapy; and the impact on patient morbidity and survival.

Financial Hardship Associated with Cancer The use of precision medicine highlights an increasingly important issue—​out-​of-​pocket costs and financial hardship to patients and their families. Because most health insurance plans require some form of cost-​sharing for oral drug therapy (20% co-​insurance is typical), patients and their families with health insurance may face bills of tens of thousands of dollars for a full course of treatment. The costs of cancer care will present additional challenges to cancer survivors without health insurance because they may responsible for the entire treatment cost. Patients may delay treatment or may fail to seek care because of high patient cost-​sharing or financial hardship (Kent et al., 2013; Weaver et al., 2010). Treatment adherence is also poorer among cancer survivors with financial hardship or higher out-​of-​pocket costs (Dusetzina et al., 2014; Neugut et al., 2011). A national study estimated that more than 2 million cancer survivors in the United States did not receive needed medical services because of cost (Weaver et al., 2010). Because health insurance in the United States is predominantly employment-​ based in the working-​ age population, its relationship with the economic burden of cancer survivorship is complex. A cancer diagnosis may limit employment opportunities, which in turn may lead to a loss of health insurance and fewer resources for paying for medical care, further magnifying financial hardships associated with cancer. Cancer survivors are at risk of medical debt and bankruptcy (Banegas et  al., 2016; Ramsey et  al., 2013; Yabroff et  al., 2016), as

179

well as psychological aspects of financial hardship, including stress, anxiety, and worry about their financial situation (Meneses et  al., 2012; Yabroff et al., 2016). Among working-​age cancer survivors, one in four report ever having any material financial hardship (i.e., medical debt, bankruptcy, trouble paying medical bills) and one in three report psychological financial hardship (Yabroff et al., 2016). Informal caregiving may also influence employment opportunities for caregivers, potentially limiting a caregiver’s ability to hold a full-​time position or resulting in higher rates of absenteeism, particularly when the patient travels long distances for specialized care. Alternatively, maintaining health insurance coverage may lead cancer survivors or their employed family caregivers to work longer hours (Bradley et al., 2002; Hollenbeak et al., 2012) or to continue working (Bradley et al., 2007)  and delay retirement. Evaluation of financial hardship and employment and health insurance trajectories in cancer survivors and their families will be an important area for additional research, particularly in relation to changes in health insurance status that may occur with ongoing implementation of the Affordable Care Act. As the prevalence of cancer survivorship increases in the United States, the economic burden of cancer will increase as well. Thus, estimating and projecting the economic burden of cancer, including healthcare expenditures and productivity losses for patients and their families, are increasingly important issues for healthcare policymakers, healthcare systems, physicians, employers, and society overall. Efforts to ensure the comprehensiveness and quality of data resources for estimating the economic burden of cancer, particularly indirect costs, are ongoing. Emerging areas for future research include evaluating the use, effectiveness, and consequences of targeted therapies, and financial hardship for patients and their families. This work will inform efforts by healthcare policymakers, healthcare systems, and employers to improve the cancer survivorship experience in the United States. References Agency for Healthcare Research and Quality. 2015a. Using the appropriate price indices for analyses of health care expenditures or income across multiple years. Available from: http://​meps.ahrq.gov/​about_​meps/​Price_​ Index.shtml. Accessed October 18, 2015. Agency for Healthcare Research and Quality. 2015b. Cost-​to-​charge ratio files. Available from:  https://​www.hcup-​us.ahrq.gov/​db/​state/​costtocharge.jsp. Accessed October 9, 2015. Agency for Healthcare Research and Quality. 2016. Healthcare Cost and Utili­ zation Project (HCUP). Available from: http://​www.ahrq.gov/​research/​data/​ hcup/​index.html. Accessed April 15, 2016. Amado RG, Wolf M, Peeters M, et al. 2008. Wild-​type KRAS is required for panitumumab efficacy in patients with metastatic colorectal cancer. J Clin Oncol, 26(10), 1626–​1634. PMID: 18040272. American Cancer Society. 2016. Cancer treatment and survivorship facts & ­figures  2014–​2015. Available from:  http://​www.cancer.org/​acs/​groups/​ content/​@research/​documents/​document/​acspc-​042801.pdf Accessed April 19, 2016. Bach PB. 2009. Limits on Medicare’s ability to control rising spending on cancer drugs. N Engl J Med, 360(6), 626–​633. PMID: 9176475. Banegas MP, Guy GP Jr, de Moor JS, et al. 2016. For working-​age cancer survivors, medical debt and bankruptcy create financial hardships. Health Aff, 35(1), 54–​61. PMID: 26733701. Bernard DS, Farr SL, Fang Z. 2011. National estimates of out-​of-​pocket health care expenditure burdens among nonelderly adults with cancer: 2001 to 2008. J Clin Oncol, 29(20), 2821–​2826. PMCID: PMC3139395. Bradley CJ, Bednarek HL, Neumark D. 2002. Breast cancer and women’s labor supply. Health Serv Res, 37(5), 1309–​1328. PMCID: PMC1464031. Bradley CJ, Neumark D, Luo Z, Bednarek HL. 2007. Employment-​contingent health insurance, illness, and labor supply of women: evidence from married women with breast cancer. Health Econ, 16(7), 19–​37. PMID: 17177273. Bradley CJ, Yabroff KR, Dahman B, Feuer EJ, Mariotto A, Brown ML. 2008. Productivity costs of cancer mortality in the United States: 2000–​2020. J Natl Cancer Inst, 100(24), 1763–​1770. PMCID: PMC2720777. Brown ML, Riley GF, Schussler N, Etzioni RD. 2002. Estimating health care costs related to cancer treatment from SEER-​Medicare data. Med Care, 40(8 Suppl), IV-​104–​117. PMID: 12187175. Cohen JW, Cohen SB, Banthin JS. 2009. The Medical Expenditure Panel Survey:  a national information resource to support healthcare cost research and inform policy and practice. Med Care, 47(7 Suppl 1), S44–​50. PMID: 19536015.

180

180

PART II:  THE MAGNITUDE OF CANCER

Conti RM, Fein AJ, Bhatta SS. 2014. National trends in spending on and use of oral oncologics, first quarter 2006 through third quarter. Health Aff, 33(10), 1721–​1727. PMCID: PMC4594844. Crystal S, Johnson RW, Harman J, Sambamoorthi U, Kumar R. 2000. Out-​of-​ pocket health care costs among older Americans. J Gerontol B Psychol Sci Soc Sci, 55(1), S51–​S62. PMID: 10728130. Davidoff AJ, Erten M, Shaffer T, et al. 2013. Out-​of-​pocket health care expenditure burden for Medicare beneficiaries with cancer. Cancer, 119(6), 1257–​1265. PMID: 23225522. de Moor JS, Mariotto AB, Parry C, et al. 2013. Cancer survivors in the United States:  prevalence across the survivorship trajectory and implications for care. Cancer Epidemiol Biomarkers Prev, 22(4), 561–​ 570. PMCID: PMC3654837. Dinan MA, Curtis LH, Hammill BG, et al. 2010. Changes in the use and costs of diagnostic imaging among Medicare beneficiaries with cancer, 1999–​ 2006. JAMA, 303(16), 1625–​1631. PMID: 20424253. Dowling E, Yabroff KR, Mariotto AB, McNeel T, Zeruto C, Buckman D. 2010. Burden of illness in adult survivors of childhood cancers: findings from a population-​based national sample. Cancer, 116(15), 3712–​3721. PMCID: PMC2913863. Dusetzina SB, Winn AN, Abel GA, Huskamp HA, Keating NL. 2014. Cost sharing and adherence to tyrosine kinase inhibitors for patients with chronic myeloid leukemia. J Clin Oncol, 32(4), 306–​311. PMID: 24366936. Ekwueme DU, Chesson HW, Zhang KB, Balamurugan A. 2008. Years of potential life lost and productivity costs because of cancer mortality and for specific cancer sites where human papillomavirus may be a risk factor for carcinogenisis: United States, 2003. Cancer, 113(10 Suppl), 2936–​2945. PMID: 18980277. Ekwueme DU, Guy GP Jr, Li C, Rim SH, Parelkar P, Chen SC. 2011. The health burden and economic costs of cutaneous melanoma mortality by race/​ethnicity-​United States, 2000 to 2006. J Am Acad Dermatol, 65(5 Suppl 1), S133–​S143. PMID: 22018062. Ekwueme DU, Yabroff KR, Guy GP Jr, et al. 2014. Medical costs and productivity losses of cancer survivors: United States, 2008–​2011. MMWR Morb Mortal Wkly Rep, 63 (23), 505–​510. PMID: 24918485. Elkin EB, Bach PB. 2010. Cancer’s next frontier: addressing high and increasing costs. JAMA, 303(11), 1086–​1087. PMCID: PMC3647336. Finkelstein EA, Tangka FK, Trogdon JG, Sabatino SA, Richardson LC. 2009. The personal financial burden of cancer for the working age population. Am J Manag Care, 15(11), 801–​806. PMID: 19895184. Finkler SA. 1982. The distinction between cost and charges. Ann Intern Med, 96(1), 102–​109. PMID: 7053682. Fireman BH, Quesenberry CP, Somkin CP, et al. 1997. Cost of care for cancer in a health maintenance organization. Health Care Financ Rev, 18(4), 51–​ 76. PMCID: PMC4194474. Fojo T, Mailankody S, Lo A. 2014. Unintended consequences of expensive cancer therapeutics: the pursuit of marginal indications and a me-​too mentality that stifles innovation and creativity: the John Conley Lecture. JAMA Otolaryngol Head Neck Surg, 140(12), 1225–​1236. PMID: 25068501. Gold MR, Siegel JE, Russell LB, Weinstein MC (Eds.). 1996. Cost-​Effectiveness in Health and Medicine. New York: Oxford University Press. Guy GP Jr, Ekwueme DU, Yabroff KR, et al. 2013. The economic burden of cancer survivorship among adults in the United States. J Clin Oncol, 31(30), 3749–​3757. PMCID: PMC3795887. Guy GP Jr, Yabroff KR, Ekwueme DU, et al. 2014. Estimating the health and economic burden of cancer among those diagnosed as adolescents and young adults. Health Aff, 33(6), 1024–​1031. PMCID: PMC4582764. Hollenbeak CS, Short PF, Moran J. 2011. The implications of cancer survivorship for spousal employment. J Cancer Surviv, 5(3), 226–​ 234. PMCID: PMC4176691. Howard DH, Molinari N-​A, Thorpe KE. 2004. National estimates of medical costs incurred by nonelderly cancer patients. Cancer, 100(5), 883–​891. PMID: 14983481. Howard DH, Kauh J, Lipscomb J. 2010. The value of new chemotherapeutic agents for metastatic colorectal cancer. Arch Intern Med, 170(6), 537–​542. PMID: 20233802. Howlander N, Noone AM, Krapcho M, et  al. 2016. SEER Cancer Statistics Review, 1975–​2013. Bethesda, MD: National Cancer Institute. http://​seer. cancer.gov/​csr/​1975_​2013/​, based on November 2015 SEER data submission, posted to the SEER website, April 2016. Kent EE, Forsythe LP, Yabroff KR, et al. 2013. Are survivors who report cancer-​related financial problems more likely to forgo or delay medical care? Cancer, 119(20), 3710–​3717. PMCID: PMC4552354.

Li C, Ekwueme DU, Rim SH, Tangka FK. 2010. Years of potential life lost and productivity losses from male urogential cancer deaths:  United States, 2004. Urology, 76(3), 528–​535. PMID: 20573389. Litzelman K, Yabroff KR. 2015. How are spousal depressed mood, distress, and quality of life associated with risk of depressed mood in cancer survivors? Longitudinal findings from a national sample. Cancer Epidemiol Biomarkers Prev, 24(6), 969–​977. PMCID: PMC4453017. Lund JL, Yabroff KR, Ibuka Y, et al. 2009. Inventory of data sources for estimating health care costs in the United States. Med Care, 47 (7 Suppl 1), S127–​S142. PMCID: PMC309738. Mariotto AB, Yabroff KR, Shao Y, Feuer EJ, Brown ML. 2011. Projections of the cost of cancer care in the United States: 2010–​2020. J Natl Cancer Inst, 103(2), 117–​128. PMCID: PMC3107566. Meneses K, Azuero A, Hassey L, McNees P, Pisu M. 2012. Does economic burden influence quality of life in breast cancer survivors? Gynecol Oncol, 124(3), 437–​443. PMCID: PMC3278545. Mittmann N, Au H-​J, Tu D, et al. 2009. Prospective cost-​effectiveness analysis of cetuximab in metastatic colorectal cancer: economic evaluation of National Cancer Institute of Canada Clinical Trials Group CO.17 trial. J Natl Cancer Inst, 101(17), 1182–​1192. PMID: 19666851. Nekhlyudov L, Madden J, Graves AJ, Zhang F, Soumerai SB, Ross-​Degnan D. 2011. Cost-​related medication nonadherence and cost-​ saving strategies used by elderly Medicare cancer survivors. J Cancer Surviv, 5(4), 395–​404. PMCID: PMC3767465. Neugut AI, Subar M, Wilde ET, et al. 2011. Association between prescription co-​payment amount and compliance with adjuvant hormonal therapy in women with early-​stage breast cancer. J Clin Oncol, 29(18), 2534–​2542. PMCID: PMC3138633. Newton AN, Ewer SR. 2010. Inpatient cancer treatment: an analysis of financial and nonfinancial performance measures by hospital-​ownership type. J Health Care Finance, 37(2), 56–​80. PMID: 21294439. Ramsey SD. 2008. How should we value lives lost to cancer? J Natl Cancer, 100(24), 1742–​1743. PMID: 19066279. Ramsey S, Blough D, Kirchhoff A, et al. 2013. Washington state cancer patients found to be at greater risk for bankruptcy than people without a cancer diagnosis. Health Aff, 32(6), 1143–​1152. PMCID: PMC4240626. Riley GF, Potosky AL, Lubitz JD, Kessler LG. 1995. Medicare payments from diagnosis to death for elderly cancer patients by stage and diagnosis. Med Care, 33(8), 828–​841. PMID: 7637404. Russell L. 2009. Completing costs: patients’ time. Med Care, 47 (7 Suppl 1), S89–​S93. PMID: 19536025. Sabatino SA, Coates RJ, Uhler RJ, Alley LG, Pollack LA. 2006. Health insurance coverage and cost barriers to needed medical care among US adult cancer survivors age 2 products.

NATURE OF THE EXPOSURE Smoked Tobacco Products Manufactured cigarettes comprise an estimated 84% of global tobacco consumption. Commonly used smoked and non-​combustible tobacco products are listed in Table 11–​1 and are discussed later in this chapter.

Composition of Tobacco Smoke Tobacco smoke is a complex, heterogeneous mixture that contains at least 5300 (and by some estimates as many as 7000) identified chemicals (US Department of Health and Human Services, 2010, 2014).

These include at least 70 chemicals designated by the IARC as having sufficient evidence of carcinogenicity, based on studies of laboratory animals and/​or humans (IARC, 2012a; US Department of Health and Human Services, 2014). Comprehensive reviews of the chemical composition of tobacco smoke are available elsewhere (IARC, 2004). At least seven classes of carcinogens are formed by the combustion of tobacco (Table 11–​2). These include polycyclic aromatic hydrocarbons (PAHs), heterocyclic compounds, N-​nitrosamines, aromatic amines, formaldehyde, phenolic compounds, and a variety of free radicals (IARC, 2004). Others, such as arsenic, cadmium, chromium, nickel, and polonium 210, are incorporated into the tobacco plant from soil or phosphate fertilizers. The concentrations of carcinogens such as tobacco-​specific nitrosamines are increased by fermentation and some forms of curing. The concentration of nicotine and the rapidity with which it can be absorbed depends on the selection of tobacco strains and the pH of the smoke (IARC, 2004; US Department of Health and Human Services, 2010). The chemical characteristics of the smoke have not generally been considered in epidemiologic studies of cancer, largely because the presence and concentration of various constituents have changed over time and are difficult to reconstruct.

Mainstream Versus Sidestream Smoke Active smoking generates both mainstream smoke (MS) and sidestream smoke (SS):  MS is drawn directly from the burning tobacco into the mouth; SS is released from the smoldering tobacco into the ambient air, where it mixes with exhaled MS to make up environmental tobacco smoke (ETS). Involuntary exposure to ETS, also called passive or secondhand smoking, involves similar chemical constituents, although the concentrations may differ by several orders of magnitude (IARC, 2004; US Department of Health and Human Services, 2014).

Nicotine Nicotine is the principal alkaloid present in tobacco and accounts for 0.05%–​4.00% (by weight) of the tobacco leaf (US Department of Health and Human Services, 1988). While nicotine itself is not carcinogenic, physical dependence on nicotine is the main factor that

 19

191

Tobacco 100

2011 2012 2013 2014 2015

Percentage of students

25

20

15

10

5

0

Any

≥ 2 types

E-cigarettes

Cigarettes

Cigars

Hookahs

Smokeless tobacco

Pipe tobacco

Bidis

Tobacco product

Figure 11–​7.  Tobacco use in middle and high school students, United States, 2011–​2015. Source: Singh T et al. (2016).

sustains tobacco use (US Department of Health and Human Services, 2014). Nicotine from tobacco binds with the nicotinic receptors for acetylcholine in the central and peripheral nervous systems. In the central nervous system (CNS), the receptors regulate the release of Table 11–​2. Tobacco Smoke Carcinogens Evaluated in the IARC Monographs Chemical Class

Number of Carcinogens

Polycyclic aromatic hydrocarbons (PAHs) and their heterocyclic analogues N-​Nitrosamines

15

Aromatic amines

12

8

Aldehydes

2

Phenols

2

Volatile hydrocarbons

3

Other organics

12

Inorganic compounds

8

Representative Carcinogens Benzo(a)pyrene (BaP) Dibenz(a,h)anthracene

4-​(Methylnitrosamino)-​1-​(3-​ pyridyl)-​1-​butanone (NNK) N'-​Nitrosonornicotine (NNN) 4-​Aminobiphenyl 2-​Naphthylamine Formaldehyde Acetaldehyde Catechol Caffeic acid Benzene 1,3-​Butadiene Isoprene Ethylene oxide Acrylonitrile Cadmium Polonium-​210

There are many other carcinogens in cigarette smoke that have not been evaluated in an IARC Monograph. Source: IARC (2004a).

neurotransmitters such as dopamine, serotonin, and γ-​aminobutyric acid. Exposure to exogenous nicotine stimulates the production of additional nicotine receptors (Benowitz, 1996). For most users, nicotine addiction from tobacco use represents true drug dependence (US Department of Health and Human Services, 2010). Withdrawal symptoms among cigarette smokers who attempt to quit include anxiety, irritability, weariness, constipation/​diarrhea, insomnia, intense craving, and difficulty concentrating (Balfour and Fagerstrom, 1996). The severity of these symptoms may equal withdrawal from opiates, amphetamines, and cocaine. The strength of the addiction is illustrated by the high failure rate among smokers who attempt to quit. Approximately 70% of current smokers express a desire to quit, yet fewer than 50% try to stop each year (Centers for Disease Control and Prevention, 2000). Unassisted, about 2.5% of smokers succeed in quitting permanently on a single quit attempt. The success rate approximately doubles with appropriate pharmacological and/​or behavioral treatment.

Factors That Influence Exposure Cumulative exposure to the carcinogens in tobacco depends on factors that affect the design and manufacture of the product, and social, economic, and biological determinants of usage.

Bioavailability of Nicotine

Tobacco products vary in their delivery of nicotine in a form that can be rapidly absorbed. Differences in the bioavailability of nicotine influence both the addictiveness of various products and the surface area of epithelium or number of cells exposed to harmful constituents in tobacco smoke. Cigarettes and moist snuff increase plasma nicotine concentration almost immediately (Figure 11–​8) and are most addictive, whereas nicotine replacement products generally provide much slower nicotine uptake (Benowitz, 1996; US Department of Health and Human Services, 2010). Inhalation of cigarette smoke increases

192

192

PART III:  THE CAUSES OF CANCER Tobacco 20

Plasma nicotine concentration (ng/mL)

15

Cigarette (nicotine delivery 1–2 mg)

Oral snuff

Nasal spray (1 mg)

Nicotine gum (Polacrilex 4 mg)

Nico Derm CQ patch (21 mg)

Nicotrol patch (21 mg)

10 5 20 15 10 5 0 0

30

60

90

120

0

30

60

90

120

0

30

60

90

120

Time post administration (minutes)

Figure 11–​8.  Plasma nicotine uptake from various tobacco products.

plasma nicotine and produces discernible CNS effects in as little as 7 seconds owing to the large surface area of the lungs. The uptake of nicotine from various tobacco products depends on the pH of the product. Commercial brands of moist snuff produce a pH range in saliva from 5.39 (at which less than 3% of the nicotine is free or un-ionized) to 8.19 (where at least half of the nicotine is un-ionized and can be rapidly absorbed) (Djordjevic et al., 1994). Nicotine in an alkaline environment, as produced by pipes, cigars, smokeless tobacco products, and cigarettes in the first half of the twentieth century is absorbed through the oral mucosa. Alkaline smoke is too irritating for most people to inhale. In contrast, protonated nicotine, as delivered by contemporary cigarettes, is absorbed more rapidly through respiratory epithelium. Deeper inhalation was made palatable by the development, selection, and mixing of special tobacco blends and other modifications. Inhalation not only accelerates the absorption of nicotine, but also exposes a much larger surface area to the toxic and carcinogenic compounds in smoke.

Other Aspects of Product Design and Composition

As mentioned, the concentration of toxic and carcinogenic chemicals in tobacco smoke is influenced by methods of curing and manufacturing, as well as by agricultural practices. For example, the introduction of reconstituted tobacco in the 1950s allowed manufacturers to use the ribs and stems from tobacco as well as the leaves (US Department of Health and Human Services, 2010). This reduced waste and production costs and somewhat lowered the concentration of polycyclic aromatic hydrocarbons (PAHs), but greatly increased the concentration of tobacco-​specific nitrosamines (TSNAs) (Hoffmann and Hoffmann, 1997; US Department of Health and Human Services, 2014). Tobacco that is flue-​cured and has direct contact with the combustion byproducts from propane gas heaters also has higher concentrations of TSNAs than air-​cured tobacco (Collishaw, 2016). Fermentation of tobacco for use in cigars and moist snuff greatly increases the concentration of TSNAs (US Department of Health and Human Services, 2010). Brands of moist snuff commonly used in the United States have TSNA levels far above the limit allowed in other consumer products (Hoffmann and Hoffmann, 1997). Cigarettes from different countries differ in their concentration of nitrosamines, nitrates, nicotine, and other compounds (Ashley et  al., 2003). Comparative studies of the levels of TSNAs in cigarette tobacco documented that Marlboro cigarettes purchased in 11 of 13 foreign countries had significantly higher TSNA levels than the locally popular non-​U.S. brands from the same country (Ashley

et  al., 2003). Even when measured by machine smoking, the “tar” and nicotine yields of cigarettes vary by a factor of 3 across WHO regions (Calafat et  al., 2004). Although epidemiologic studies have not generally been able to consider measurements of the chemical composition of smoke in different countries when comparing risk, the observed variations in TSNA levels provide a promising target for regulation. As described later in the chapter, urinary markers of TSNA exposure have been associated with esophageal and lung cancer (Yuan et al., 2014).

Ventilated Cigarettes

Cigarettes with ventilated filters were introduced in the United States in the late 1960s. Unlike plain cellulose acetate filters, these were designed with ventilation holes that allowed air to enter and dilute the smoke when the cigarettes were smoked by machines (US Department of Health and Human Services, 2014). Machine-​measured “tar” and nicotine ratings from cigarettes with ventilated filters were substantially lower than machine ratings from a similar cigarette with an unventilated filter or from an unfiltered cigarette. Smokers who used these products could block the ventilation holes to obtain their accustomed level of intake. Nevertheless, they perceived the smoke to be less harsh than that from unfiltered or unventilated cigarettes. Cigarette companies marketed these products to health-​conscious smokers as an alternative to quitting, using the terms “light,” “ultralight,” “mild,” and “low tar” to imply lower health risks. The method of machine smoking was developed by the tobacco industry during the 1930s and was adopted officially by the Federal Trade Commission (FTC) in 1969. It became known as the “FTC protocol” (Institute of Medicine, 2001). Settings were fixed so that the machine took one 2-​second (35 ml) puff per minute until the cigarette was consumed. Tar and nicotine were extracted from a special filter through which the machine “smoked” the cigarette. Tar represents the total particulate matter after removing the nicotine and water (Institute of Medicine, 2001). As shown in Figure 11–​9, the average sales-​weighted tar and nicotine yields of US cigarettes decreased dramatically according to machine measurements. The average “tar” yield per cigarette decreased from 38 mg in 1954 to 12 mg in 1998, while the average nicotine rating decreased from 2.3 mg to 0.9 mg (US Department of Health and Human Services, 1981, 1989; Kozlowski et al., 2001). Initially, the reduction was achieved by adding cellulose acetate filters and using more porous wrapping paper (Hoffmann and Hoffmann, 1997). Further reductions were achieved after 1970 by the use of ventilated filters, “puffed tobacco” that reduced the amount

 193

193

Tobacco 3.0

4.0 Tar Nicotine

3.5

2.5

3.0

Tar (mg)

1.5

2.0 1.5

1.0

Nicotine (mg)

2.0

2.5

1.0 0.5

0.5 0.0 1950 1955

0.0 1960

1965

1970

1975

1980

1985

1990

1995

Year

Figure 11–​9.  Sales-​weighted tar and nicotine values for US cigarettes as measured by a smoking machine using the Federal Trade Commission (FTC) method, 1954–​1998. Note: Values before 1968 are estimated from available data. Courtesy of D. Hoffmann, personal communication. Source: National Cancer Institute Smoking and Tobacco Control Monograph 13 (2002), p. 2.

per cigarette, and modified wrapping paper that burned more rapidly so that the testing machine could take fewer puffs (Kozlowski et al., 2001). Epidemiologists who studied the cancer risks from “reduced-​yield” cigarettes were largely unaware of the changing technologies that affected the machine measurements. Studies conducted before the year 2000 equated “high” and “low” tar cigarettes with unfiltered and filter-​tip cigarettes respectively. It was assumed that a reduction in risk associated with the addition of a filter would correspond to further reductions in risk across much lower levels of machine-​measured “tar” (Harris et  al., 2004). The critical distinction between ventilated and unventilated filters had not yet been disclosed to the public. The limitations of the FTC protocol in measuring the risk of ventilated cigarettes became apparent by the late twentieth century, when age-​standardized lung cancer rates continued to increase in smokers, despite large reductions in machine-​measured yield (Thun et al., 1997; US Department of Health and Human Services, 1989). Studies of salivary cotinine found little relationship between machine-​measured ratings below 1.0 mg nicotine and actual nicotine uptake (Figure 11–​10) (Benowitz, 2001; Jarvis et al., 2001). Moreover, an analysis by Harris and colleagues at the American Cancer Society found no reduction in lung cancer risk from “light” or “ultralight” cigarettes compared to “regular” cigarettes with unventilated cellulose acetate filters, but higher risk from smoking unfiltered cigarettes (Figure 11–​11) (Harris et al., 2004). Based on these findings, the FTC rescinded its approval of machine smoking in 2008 and the Food and Drug Administration (FDA) banned the use of descriptors such as “light,” “mild,” and “low tar” in the United States. The tobacco industry has subsequently circumvented this restriction by introducing color codes that smokers recognize as equivalent to the banned terms (Connolly and Alpert, 2014).

Tobacco Products Other Than Cigarettes Combustible

As mentioned, about 15%–​ 35% of global tobacco consumption involves tobacco products other than manufactured cigarettes (Eriksen et  al., 2012). These are listed in Table 11–​1. Other smoked tobacco products include hand-​ rolled cigarettes, cigars, pipes, and a variety of products smoked widely in Southeast Asia. Hand-​rolled cigarettes made from loose tobacco are used increasingly in the United

States, since they are exempted from the excise taxes on manufactured cigarettes. For tax purposes, cigars are defined as shredded tobacco wrapped in tobacco leaf or paper (Department of the Treasury, 2006). They vary in size from cigarette-​sized cigarillos to cheroots and double coronas (Eriksen et  al., 2012). Small filter-​tip cigars are usually smoked like cigarettes but are less heavily taxed and can be purchased individually. Their use is increasing in the United States, whereas the previously widespread use of pipes is decreasing worldwide (Shafey et al., 2009). Waterpipes (shisha or sheecha) have been used traditionally in Middle-​Eastern countries and in Russia and Vietnam (Giovino et  al., 2012; IARC, 2004), but are increasingly marketed to young people in Western countries using flavors of molasses, apple, banana, vanilla, and other fruits or candies (Tobacco Atlas, 2016). The most common form of smoked tobacco in India involves bidis, traditionally hand-​rolled in dried temburni leaf and tied with a string (Shafey et  al., 2009). Kreteks are clove-​and cocoa-​flavored small cigars that are produced and sold in Indonesia but also marketed worldwide (Gandini et al., 2008).

Non-​Combusted Products

An historical overview of the use of smokeless or non-​combusted tobacco products is provided in IARC Monograph 89 (IARC, 2007)  and more recently in a National Cancer Institute Monograph (National Cancer Institute and Centers for Disease Control and Prevention, 2014). Traditional forms of smokeless or “spit” tobacco used in Western countries include snuff (moist and dry) and chewing tobacco. In the United States alone, several million people use smokeless products (Agaku et  al., 2014; Lee et  al., 2014; Mazurek et  al., 2014). Moist snuff is the most commonly used product in the United States (Agaku and Alpert, 2015); snus, a lower nitrosamine form of moist snuff, is widely used in Nordic countries. Moist snuff consists of finely ground tobacco with 20%–​55% moisture content, often flavored with mint, wintergreen, or raspberry (Borgerding et al., 2012). A pinch (called a “dip” or “rub”) is placed between the gum and the cheek or under the tongue (Shafey et al., 2009). Chewing tobacco is popular among baseball players, male adolescents, and in both sexes in some rural populations in the United States. Products may be flavored with sugar, molasses, or licorice, and in the United States are sold with a range of concentrations of nicotine and other toxicants that increase as users become habituated. Dual use of cigarettes plus smokeless tobacco products is most common among young unmarried men (Lee et al., 2014). This form of poly-​tobacco use increased

194

194

PART III:  THE CAUSES OF CANCER 900 800

Saliva cotinine (ng/ml)

700 600 500 400 300 200 100 0 0.0

0.1

0.2

0.3

0.6 0.7 0.4 0.5 Nicotine yield (mg)

0.8

0.9

1.0

1.1

Figure 11–​10.  Range of salivary cotinine among smokers smoking cigarettes within a given FTC rating for nicotine. Source: Jarvis et al. (2001).

among users under age 26 in the United States between 2002 and 2011 (Fix et al., 2014). The predominant form of tobacco used in Sweden is moist snuff or snus with much lower concentration of nitrosamines than most products in the United States (Nilsson, 1998). Betel quid or pan is widely used in Asia and the Western Pacific, especially by women (Shafey et al., 2009). Betel leaves (Piper betle) are mixed with tobacco, Areca nut (Areca catechu), lime, wood ash, or other substances to form a “quid,” “pan,” or “nass” (IARC, 2004). The mixture is then chewed and/​or retained in the mouth. The use of smokeless tobacco is estimated to cause more than 100,000 deaths annually from cancer, with most deaths occurring in Southeast Asia (Ezzati et al., 2002; Sinha et al., 2016).

Novel Tobacco Products Electronic Nicotine Delivery Devices (ENDS). Novel

2.5

2.5

2.0

2.0 Hazard ratio (95% CI)

Hazard ratio (95% CI)

devices that heat tobacco or nicotine but do not produce smoke have been introduced in recent decades. E-​cigarettes are the most widely used of the electronic nicotine delivery systems (ENDS). The use of e-​cigarettes is greatest in the United States and Europe (Arrazola et al., 2015), but uptake is also occurring in middle-​and low-​income

countries (Palipudi KM et  al., 2015). They were invented by the Chinese pharmacist Hon Lik in the first decade of the 2000s and are principally manufactured in China. According to tobacco industry documents, cigarette manufacturers were working to develop similar products since the early 1990s to serve as alternatives to nicotine-​ replacement therapy and for use in smoke-​free environments, but only recently decided to bring these products to market (Dutra et al., 2016). Users obtain nicotine through vaporization of a liquid solution that typically includes a carrier compound such as propylene glycol, as well as various flavorings (Grana et al., 2014). Many of these flavors, such as “gummy bear” and “jolly rancher” are attractive to youth (US Department of Health and Human Services, 2016). The products come in many shapes and sizes, including both disposable and reusable models, as illustrated in Figure 11–​12. Cross-​sectional studies indicate that ENDS users are exposed to a number of carcinogens, including tobacco specific–​nitrosamines and acrolein (Goniewicz et al., 2014), although at concentrations that are lower and involve fewer carcinogens than tobacco smoke. The ultimate impact of e-​cigarettes and other putative harm reduction products on health is unknown. Although use of these products produces lower exposures to toxic and carcinogenic compounds than

1.5

1.0

0.5

Never smoked regularly

1.5

1.0

0.5

≤ 35 35–54 ≥ 55

≤ 7 8–14 15–21 ≥ 21

Quit smoking at age (years)

Tar level (mg) current smoker

Never smoked regularly

≤ 35 35–54 ≥ 55

≤ 7 8–14 15–21 ≥ 21

Quit smoking at age (years)

Tar level (mg) current smoker

Figure 11–​11.  Hazard ratios for lung cancer in men and women from the CPS-​II cohort by level of machine-​measured “tar” in cigarettes smoked for current smokers or by age at cessation for former smokers. Source: Harris et al. (2004).

 195

195

Tobacco ENDS type

Product characteristics

Disposable e-cigarette

• For short time use • Cheap; widely available from gas stations or super markets • Convenient, as no refill solution or cartridge is needed • Might be a starting point for many users

Rechargeable e-cigarette

• Various sizes from cigarette size to compact flashlight size • Various colors and style • Requires battery with a charger • Requires replacing or refilling nicotine cartridges

Tank system e-cigarette

• High capacity • Battery operated • Lasts longer than typical rechargeable models • Requires refilling the tank with nicotine solution • Often recommended for heavy smokers by retailers

Personal Vaporizer

• Large size and heavy weight • Battery power output control allows the user to adjust the output of the voltage going to the heating element, which has a direct effect on the temperature of the heating element • Usually not recommended by retailers for beginning users

E-cigar

• Looks like an authentic large ciger • Usually disposable

Figure 11–​12.  E-​cigarettes and other ENDs. Source: ASCO, AACR policy statement: Brandon et al., JCO (2015).

active smoking, marketing practices may detrimentally affect cessation and initiation patterns in the population. Smokers who continue to smoke in addition to using e-​cigarettes often use them to delay cessation by circumventing smoke-​free laws. Ultimately, it is unknown whether e-​cigarettes will be used predominantly by smokers or by adolescents and young adults who do not otherwise use tobacco, and whether e-​cigarettes are contributing to the decline in other forms of tobacco smoking among youth in the United States (Figure 11–​7). Accurate prevalence data on the use of e-​cigarettes and other ENDs is only now emerging from the National Health Interview Survey (Figure 11–​13) (Arrazola et al., 2015; QuickStats, 2016). The majority of e-​ cigarette smokers reported continuing to smoke cigarettes, especially at ages > 25 years. A substantial proportion of e-​cigarette users in the youngest age group (18–​24  years) had never smoked cigarettes, but may have used other tobacco products. The percentage who had never smoked cigarettes became much smaller with age. There is considerable variation among countries in awareness and use of e-​cigarettes (Gravely et al., 2014). Only a few studies have examined the effects of e-​cigarette use on cessation of cigarette smoking (Orellana-​Barrios et  al., 2016), with data from randomized trials being particularly scarce. As described in the 2016 Surgeon General’s report (US Department of Health and Human Services, 2016), current knowledge suggests health benefits for current adult cigarette smokers who use e-​cigarettes as a way to quit tobacco use completely or perhaps even as a long-​term replacement for cigarette smoking. Much larger randomized clinical trials are urgently needed, however, to determine whether e-​cigarettes are

effective smoking-​cessation devices. Observational studies are needed to determine the effects of switching from cigarettes to e-​cigarettes on disease risk. As discussed earlier, a substantial proportion of younger e-​cigarette users have never tried cigarettes or other tobacco products. Over the last few years, longitudinal studies of tobacco-​use patterns have been initiated. Published studies so far suggest that youths who use e-​cigarettes may be more likely to use cigarettes subsequently than never e-​cigarette users (Barrington-​Trimis et al., 2016; US Department of Health and Human Services, 2016). Whether these associations persist in future studies and ultimately affect the prevalence of cigarette smoking in the United States and other countries will be a critical gauge of the impact of e-​cigarettes on health.

MEASUREMENT OF EXPOSURE Questionnaire Measures The best and most thoroughly validated external measures of active cigarette smoking derive from self-​ reports (Shields, 2002). Most adults can report whether they have smoked 100 or more cigarettes in their lifetime and whether they now smoke every day or on some days, the definition of current smoking (Centers for Disease Control and Prevention, 1994). A meta-​analysis of 26 studies that evaluated the validity of self-​reported data on tobacco use found that self-​reported smoking status predicted biochemical evidence of active smoking

196

196

PART III:  THE CAUSES OF CANCER 100 90 80

Current cigarette smokers Former cigarette smokers Never cigarette smokers

Percentage

70 60 50 40 30 20 10 0 Total

18–24 years

25–44 years

≥45 years

Age group

Figure 11–​13. Cigarette smoking status among current adult e-​cigarette users, by age group. Sources: National Health Interview Survey (2016) in QuickStats (2016).

with 87% sensitivity and 89% specificity (Patrick et  al., 1994; US Department of Health and Human Services, 2001). The sensitivity and specificity of self-​reported smoking were higher in studies of adults than in those of children. Less than 1% of respondents in the National Health and Nutrition Examination Survey (NHANES) who denied the use of a nicotine-​containing product had an elevated serum cotinine level, after adjustment for passive smoking (Yeager and Krosnick, 2010). Information on tobacco use of comparable quality can be collected online, by paper questionnaire, or by structured interview (Steffen et al., 2014). Smoking history is considered a more sensitive measure of intermittent smoking than are biochemical indices, as the half-​life of cotinine is only about 17 hours (Benowitz, 1996). However, self-​reported data on the number of cigarettes consumed per day are thought to underestimate actual consumption by at least 20%. Estimates of per capita consumption based on questionnaire surveys consistently underestimate consumption based on cigarette sales data by 20%–​30% (Todd, 1978). Despite their quantitative limitations, self-​reported data on number of cigarettes smoked per day, years of smoking, and ages at initiation and cessation have been sufficient to document etiologic relationships between smoking and numerous disease endpoints (IARC, 2004). They may not be sufficient to document relatively small differences in the pathogenicity of cigarettes, however, given potentially large unmeasured variations in smoking behavior. Other measures that might refine estimates of cumulative exposure over a lifetime could consider the intensity of smoking at different ages, as well as daily and non-​daily use, birth cohorts, indices of addiction, the number and duration of failed cessation attempts, and the design characteristics of the product being smoked. Behavioral characteristics such as puff volume, the average number of puffs taken per cigarette, depth of inhalation, and retention time in the lung cannot be reliably measured by questionnaire. Self-​reported data on depth of inhalation have not proven to be reliable, and personal monitoring has not yet been widely used. Exposure assessment is less well developed for non-​ cigarette tobacco products. For example, few epidemiological studies have assessed risk in relation to age at initiation and cessation of use of products other than cigarettes. It is particularly difficult to assess exposures from e-​cigarettes and waterpipes. E-​cigarette users may puff on

a single tank of e-​juice over the course of a day or days. Waterpipe users often share puffs with others over the course of several hours. Future studies of non-​cigarette tobacco products should extensively query usage patterns in order to further delineate the health risks of these products. The FDA Center for Tobacco Products (FDA-​ CTP) recently released a list of 93 harmful and potentially harmful constituents (HPHC) in tobacco products and tobacco smoke that must be reported by tobacco manufacturers and importers for their products (Food and Drug Administration, 2012). This information has not yet been used in epidemiologic studies, but could conceivably be coupled with biomarkers of exposure to estimate exposure to specific chemicals. Exposure to secondhand smoke can be estimated from qualitative information on questionnaires, supplemented by quantitative assessments based on biomarkers or air measurements (Apelberg et  al., 2013; Avila-​Tang et al., 2013a, 2013b) Study participants can describe the smoking status of their spouse, the number of smokers at home, and the extent of exposure at work or in other settings in the recent past (see Chapter  17). Cotinine and other biomarkers of tobacco smoke provide objective evidence of exposure within the past 3–​7 days. Self-​ reported information on exposures in childhood and at various periods throughout life cannot be validated directly, but can provide qualitative information about past exposures.

Measures of Addiction A parameter that has not often been included in epidemiologic studies but that appears informative is the level of nicotine addiction of individual smokers. The Fagerstrom index is a widely used index of addiction in behavioral studies (Fagerstrom, 1978; Fagerstrom et al., 1996) that measures six correlates of physical dependence. It includes questions such as: “How soon after you wake up do you smoke your first cigarette? Do you find it difficult to refrain from smoking in places where it is forbidden? Do you smoke if you are so ill that you are in bed most of the day?” Questions in the Fagerstrom index may correlate with parameters of tobacco exposure that are difficult to measure by questionnaire, such as greater puff volume, depth of inhalation, and retention time in the lung. Including one of more of these questions in epidemiologic studies might prove to be a useful adjunct to the current

 197

Tobacco assessment of lifetime exposure. For example, a recent study found that time to first cigarette was associated with lung cancer in and above standard smoking measures such as smoking duration and cigarettes smoked per day (Gu et al., 2014). The common practice of combining information on the intensity and duration of smoking into a single variable of cumulative exposure (pack-​years or cigarette-​years) is contraindicated. Analyses of both the British Doctors’ Study (Doll and Peto, 1978) and the American Cancer Society Cohorts (Flanders et al., 2003) have shown that the duration of smoking is a much stronger predictor of lung cancer risk than is the number of cigarettes smoked per day. Lung cancer risk increases with the fourth or fifth power of years of smoking but only the second power of cigarettes per day. Researchers increasingly recommend that pack-​years no longer be used as an exposure variable (Leffondre et al., 2002), just as “ever” smoking is no longer considered an optimal category of exposure because of changes in risk after cessation.

Biomarkers Various biomarkers have been used in epidemiologic studies of tobacco to assess absorption, metabolism, excretion, and biologic activity of constituents in smoke (IARC, 2004; US Department of Health and Human Services, 2010).

Biomarkers of Exposure

The most commonly used biomarkers of exposure are those that reflect nicotine absorption. Cotinine is the main proximate metabolite of nicotine and is considered the biomarker of choice for indicating exposure to tobacco during the last 2–​7 days (Avila-​Tang et al., 2013a; Benowitz, 1996). The concentration of cotinine in plasma, saliva, or urine can reliably differentiate active smoking from ETS exposure (IARC, 2004). Other biomarkers, such as thiocyanate in plasma or saliva, carbon monoxide in exhaled alveolar air, and blood carboxyhemoglobin concentrations, are less sensitive and/​or less specific as markers of tobacco exposure than cotinine (Institute of Medicine, 2001). Biomarkers of nicotine metabolism are highly informative in studies of secondhand smoke, but less useful for quantifying the lifetime history of active smoking because of their short half-​life. Other biomarkers reflect the uptake, activation, and excretion of carcinogens in tobacco smoke. These are discussed extensively elsewhere (IARC, 2012a; US Department of Health and Human Services, 2010). Studies that measure metabolites of tobacco-​specific nitrosamines in urine and other bodily fluids (Hecht et al., 2016) document the absorption and metabolic activation of two powerful carcinogens, 4-​(methylnitrosamino)-​1-​(3-​pyridyl)-​1-​butanone (NNK) and N'-​nitrosonornicotine (NNN) following active or passive exposure to tobacco smoke or smokeless products (Hecht et al., 2008). All of these biomarkers of exposure reflect recent time periods rather than lifetime exposure. They also cannot typically distinguish between forms of tobacco, challenging interpretation in populations where dual or poly use is common. Biomarkers are useful, however, for validating questionnaire information on recent smoking, for documenting the absorption of carcinogens from active and passive smoking, and for excluding active smokers from studies of secondhand smoke. However, they do not improve measurably on questionnaire data in epidemiologic studies of smoking and cancer. Serum cotinine has been shown to predict lung cancer risk among smokers (Boffetta et al., 2006), but there is no evidence that it is more predictive than questionnaire-​based measures of smoking. Cotinine and carbon monoxide levels reflect smoke exposure within the last few days, and thiocyanate (from hydrogen cyanide) within the past few weeks (Jarvis et al., 1987). Several studies suggest that measurement of urinary levels of the tobacco-​specific carcinogens NNN and NNK and their metabolites may improve on estimates of carcinogen exposure from active or passive smoking when combined with cotinine and self-​reported smoking (Goniewitz et al., 2011; Vardavas et al., 2013; Yuan et al., 2014). However, these studies have been conducted in only a few populations

197

so far. Further studies in other populations with a range of tobacco use patterns are merited.

Biomarkers of Genotoxicity

DNA adducts provide direct evidence that carcinogenic constituents of tobacco smoke interact with circulating lymphocytes and with tissues affected by smoking-​related cancers (lung, oral cavity, urinary bladder, and uterine cervix). Cigarette smoking increases the mutagenicity of urine, the activation of certain enzymes in body tissue, and the presence of DNA adducts from tobacco-​specific nitrosamines or 4-​aminobiphenyl in lymphocytes, hemoglobin, and tumor tissue (US Department of Health and Human Services, 2010). Adducts bound to cellular macromolecules persist longer than nicotine metabolites after abstinence from tobacco use. The number of chromosomal aberrations in cultured lymphocytes and the extent of lipid peroxidation have been shown to correlate with the number of cigarettes smoked per day (Shields, 2002). Smoking produces multiple distinctive mutational signatures in tumors that are correlated with aspects of cigarette smoking behavior (Alexandrov et al., 2016). The most widely recognized are the G→T transversions in TP53 found in adenocarcinomas and squamous cell carcinomas of the lung in smokers (Alexandrov et al., 2016; US Department of Health and Human Services, 2010). Thirty percent of the TP53 mutations in lung tumors of smokers involve G→T transversions, whereas only 10% of the mutations in lung tumors obtained from non-​smokers are of this type. TP53 mutations in smokers occur preferentially at hotspots where DNA adducts from exposure to polycyclic aromatic hydrocarbons (PAH) are formed and incompletely repaired (US Department of Health and Human Services, 2010). They are found in pre-​neoplastic lesions as well as lung cancer in smokers, indicating that they represent early somatic events (US Department of Health and Human Services, 2010). Other studies have shown clear associations between cigarette smoking and methylation status at numerous genomic regions (Joehanes et al., 2016). These and future molecular studies may provide further insights into the mechanisms by which tobacco causes various types of cancer, may refine associations with particular subtypes of cancer, and may suggest new treatment modalities.

CANCERS CAUSED BY ACTIVE CIGARETTE SMOKING Cancer Sites with “Sufficient Evidence” Active cigarette smoking causes more than 20 different cancer sites or subsites, according to designations by the IARC and the US Surgeon General (IARC, 2012a; US Department of Health and Human Services, 2014). These include cancers of the lung (all histologic subtypes), oral cavity, nasal cavity and accessory sinuses, naso-​, oro-​, and hypopharynx, larynx, esophagus (including both squamous cell carcinoma and adenocarcinoma), stomach, pancreas, colorectum, liver, kidney (including adeno-​and urothelial carcinomas), ureter, urinary bladder, uterine cervix, ovary (mucinous), and acute myeloid leukemia. Even this list may be incomplete, as it does not include sites such as breast cancer or advanced prostate cancer for which the Surgeon General has designated the evidence as “suggestive but not conclusive” (US Department of Health and Human Services, 2014). Table 11–​3 shows the relative risk (RR) estimates for the association between current cigarette smoking and cancer sites that are formally designated as smoking-​related by both the IARC and the Surgeon General (IARC, 2012a; US Department of Health and Human Services, 2014). It also shows the year when each site was classified as causally related to smoking, and the estimated attributable fraction (AF) of deaths from active cigarette smoking in the United States. The number of cancer sites and/​or subsites classified as causally related to smoking has increased since the earliest studies of smoking and cancer (Doll, 1998; US Department of Health and Human Services, 2014). Larger studies with longer follow-​up have identified relationships with cancers that are less common or less strongly associated with smoking than cancers of the lung, larynx, oral cavity,

198

198

PART III:  THE CAUSES OF CANCER

and esophagus. The identification of oncogenic infectious agents has allowed studies to control for confounding by these pathogens with regard to cancers of the liver, stomach, and uterine cervix (IARC, 2004; US Department of Health and Human Services, 2014). Changes in cigarette design have strengthened the association between smoking and adenocarcinoma of the lung (US Department of Health and Human Services, 2014). Studies of tumor subsites have revealed causal relationships with mucinous ovarian cancer and with both adenocarcinoma and squamous cell carcinoma of the esophagus. The recognition of additional smoking-​related cancers may also reflect the aging of birth cohorts with the heaviest lifetime exposure, especially in women (Carter et al., 2015; Pirie et al., 2013; Thun et al., 2013). Undiscovered causal relationships may still exist with rare cancers, uncommon subtypes, or molecularly defined subgroups (Carter et al., 2015).

Cigarette Smoking and Respiratory Tract Cancers The respiratory epithelium is heavily exposed to cigarette smoke. Cancers of the lung and larynx were the first to be designated as causally related to active smoking in men (US Department of Health Education and Welfare, 1964). The associations between active cigarette smoking and cancers of the lung and larynx are the strongest of all sites—​an estimated 77% of laryngeal cancer and 80% of lung cancers in the United States are attributed smoking (Siegel et al., 2015).

Nasal Cavity and Paranasal Sinuses.  The evidence linking

active cigarette smoking to cancers of the nasal cavity and paranasal sinuses was designated as causal by the IARC in 2004 (IARC, 2004). Sinonasal cancers are rare; thus the evidence comes mainly from case-​ control studies. The nasal cavity and paranasal sinuses are exposed to mainstream tobacco smoke only during exhalation and only for a subset of smokers. Evidence for causality includes the presence of dose–​ response relationships in most studies, the decrease in relative risk observed with time since quitting, the absence of likely confounders, and the stronger relationship observed for squamous cell cancers than for adenocarcinomas (IARC, 2004). Active cigarette smoking approximately doubles the incidence of nasal cavity squamous cell carcinomas in case-​control studies conducted in North America, Europe, Japan, and China (IARC, 2004; US Department of Health and Human Services, 2004). Prospective studies of these sites are few, although consistent associations were observed in the large NIH-​AARP cohort (Freedman et al., 2016).

Larynx.  Cancer of the larynx is more strongly associated with active cigarette smoking than any other cancer except lung (Table 11–​3)

(Chapter 27). Smoking was designated “a significant factor in causation” of laryngeal cancer among men in 1964 (US Department of Health Education and Welfare, 1964) and among women in 1980 (US Department of Health and Human Services, 1980). The association between active cigarette smoking and death from laryngeal cancer appears to have strengthened in women but not men in the United States since the 1960s. The relative risk estimates for current versus never cigarette smoking were 14.6 in men and 13.0 in women in Cancer Prevention Study I, a large American Cancer Society cohort followed from 1959 to 1966 (Hammond, 1966). The corresponding relative risk estimates from a pooled analysis of five contemporary US cohorts was 13.9 (CI: 8.3, 23.3) in men and 103.8 (CI: 24.2, 445.5) in women (Carter et al., 2015). The wide confidence interval for women in the contemporary cohorts reflects the low risk of death from laryngeal cancer among female never smokers (only two deaths from this cancer observed in female never smokers). It is important to note that, although the relative risks of cigarette smoking for laryngeal cancer tend to be higher in women than in men, the absolute number of smoking-​attributable deaths from laryngeal cancer annually in the United States is far higher in men (2125) than in women (730), reflecting the particularly low incidence of laryngeal cancer in non-​smoking women (Siegel et al., 2015). Variations in the strength of the association by tumor subsite, gender, age, type of tobacco, and time since cessation are discussed further in Chapter 27. Siegel et al. estimate that 77% of deaths from laryngeal cancer in the United States are attributable to cigarette smoking (Siegel et al., 2015). The combination of tobacco smoking and heavy alcohol consumption creates much higher risk of laryngeal as well as other aerodigestive tract cancers than does smoking alone (Hashibe et al., 2009), as discussed below.

Trachea, Bronchus, and Lung. Cigarette smoking is more

strongly associated with lung cancer than with any other cancer site (Table 11–​3). All histologic subtypes of lung cancer are now strongly associated with cigarette smoking, whereas during the 1950s, adenocarcinoma was reported to be minimally associated (Chapter 28) (Doll et al., 1957; Kreyberg, 1962; Wynder and Hoffmann, 1994). A recent US Surgeon General Report attributed this increase to changes in the design and composition of cigarettes (US Department of Health and Human Services, 2014). The Report identified the two most plausible explanations for the increased association with adenocarcinoma of the lung as the increase in TSNA concentrations in smoke from reconstituted tobacco and the shift in inhalation patterns due to ventilated cigarettes. The evidence for any specific explanation was characterized as suggestive rather than definite. The association between active cigarette smoking and death from lung cancer has strengthened over the last 50 years in the United States

Table 11–​3. Smoking-​Attributable Mortality from Cancer, 2010–​2014 Attributable Deaths

Cancer Type

Year Formally Classified

Lip, oral cavity, pharynx Esophagus Stomach Colorectum Liver Pancreas Larynx Trachea, lung, bronchus Cervix uteri Urinary bladder Kidney, other urinary tract Acute myeloid leukemia

1964/​1971* 1982 2004 2014 2014 1982 1964 1964/​1968* 2004 1979 1982 2004

Relative Risk for Current vs. Never Smoking

%

No.

Men

Women

Men

Women

Men

5.7 3.9 1.9 1.4 2.3 1.6 13.9 25.3 N.A.-​ 3.9 1.8 1.9

5.6 5.1 1.7 1.6 1.8 1.9 103.8 22.9 3.5 3.9 1.2 1.1

48.7 52.3 25.6 11.2 28.1 9.9 72.1 83.2 N.A.-​ 46.5 22.3 23.2

43.0 44.4 10.8 8.0 14.1 14.2 93.4 76.4 22.2 40.9 7.2 3.4

2955 6011 1656 2976 4085 1870 2125 72,164 -​-​-​ 4920 1904 1181

Source: Modified from US Surgeon General report 2014; Carter et al., NEJM 2015; Siegel et al. (JAMA Internal Medicine, 2015). *Lip cancer was classified as causal in 1964, other oropharyngeal cancers in 1971. Lung cancer was classified as causal in men in 1964 and in women in 1968.

Women 1077 1296 476 1992 975 2626 730 53,635 862 1804 350 136

 19

Tobacco and the United Kingdom, following birth cohort patterns in smoking (Pirie et al., 2013; Thun et al., 2013). In men, the relative risk for death from lung cancer in current versus never smokers increased from approximately 12 in the 1960s to about 24 in the 1980s and then plateaued (Thun et al., 2013). The corresponding relative risk for female smokers increased from less than 3 in the 1960s to almost 26 in the contemporary cohorts (2000–​2010). In both sexes, the relative risk estimates for long-​term smokers of 40+ cigarettes per day approached 50. Projections based on age and birth cohort patterns suggest that the average risk will continue to increase among older female smokers in the United States for several decades (Thun et al., 2013; Pirie et al., 2013). Active cigarette smoking is estimated to account for approximately 80% (CI:  79.2, 81.1) percent of lung cancer deaths in the US general population (Siegel et al., 2015), and 96% among current smokers, based on updated hazard ratio estimates (Thun et al., 2013). Incidence and mortality rates and the relative risk estimates associated with smoking have been shown to vary by ethnicity, with the highest rates in the United States found among blacks (Torre et al., 2016). Several factors are thought to contribute to these disparities, as reviewed in Thun and Henley (2006). Menthol cigarettes are smoked more commonly by black smokers than other racial or ethnic groups, but epidemiological studies have not found that these confer higher risk for lung cancer than non-​menthol cigarettes (Blot et al., 2011; Brooks et al., 2003; Rostron, 2012). The association between current cigarette smoking and lung cancer is considerably weaker in Japan and China; it is not yet clear whether this reflects the more recent uptake of manufactured cigarettes in these countries than in the West, less intensive smoking, or, in the case of Japan, the use of charcoal filters.

Cigarette Smoking and Gastrointestinal Cancers Cigarette smoking is causally associated with cancer at all sites in the gastrointestinal tract except the salivary glands and possibly the small intestine (IARC, 2012a). It is also causally associated with cancer of the pancreas and liver. Among the gastrointestinal tract cancers, the association with smoking becomes progressively weaker from the oral cavity to the colorectum (Table 11–​3). The interaction between smoking and alcohol is also strongest for squamous cell cancers of the upper aerodigestive tract (Hashibe, 2010).

Oral Cavity, Oropharynx, Lip, and Salivary Glands.  Active

cigarette smoking was first designated as causally related to cancers of the oral cavity and pharynx by the US Surgeon General in 1982 (US Department of Health and Human Services, 1982) and by the IARC in 1986 (IARC, 1986). Historically, many studies have combined cancers of the oral cavity with those of the oropharynx, even though these are now recognized to have distinctive etiologic and clinical features. More recently, a pooled analysis of over 13,000 cases and 18,000 controls reported similar relative risk estimates for ever smoking with oral cavity cancer (RR 2.87; 95% CI: 2.60, 3.18) and oropharyngeal cancer (RR 3.01; 95% CI: 2.71, 3.35) (Wyss et al., 2013). Unlike other sites, cancers of the salivary gland have not been shown to be associated with cigarette smoking. The etiology of cancers of the oral cavity and oropharynx differ substantially with regard to human papillomavirus (HPV) infection. Whereas HPV is a strong risk factor for cancers of the oropharynx and base of the tongue, it is minimally associated with cancers of the oral cavity. Studies have not yet fully elucidated whether cigarette smoking interacts with HPV infection for cancers of the oropharynx. Active cigarette smoking is estimated to account for almost half (47.0%; CI: 38.6, 55.5) of deaths from cancers of the oral cavity and pharynx in the United States (Siegel et al., 2015). This attributable fraction estimate has not decreased, despite decreases in smoking prevalence and a large increase in HPV-​related oropharyngeal cancers (see Chapter 29) (Chaturvedi et al., 2008, 2013). The high fraction of cases attributed to smoking may increasingly represent an interaction between smoking and HPV, rather than smoking alone. Future studies of oral cavity and

199

oropharyngeal cancers should separate them into etiologically relevant subsites.

Nasopharynx.  Active smoking was first classified as causally

related to nasopharyngeal cancer by the IARC in 2004 (IARC, 2004). A meta-​analysis of 28 case-​control studies and four prospective cohort studies published between 1979 and 2011 reported a meta-​relative risk estimate of 1.60 (95% CI: 1.38, 1.87) for ever versus never smoking, with higher risks associated with greater intensity and longer duration of smoking (Xue et al., 2013). Other studies and heterogeneity by cell type and geographic population are discussed further in Chapter 26.

Esophagus.  Active cigarette smoking was designated as a major

cause of esophageal cancer by the US Surgeon General in 1982 and by the IARC in 1986 (IARC, 1986; US Department of Health and Human Services, 1982). Historically, only cancers of the lung and larynx were more strongly associated with active smoking than esophageal cancer (Vineis et al., 2004). Active cigarette smoking is causally related to both esophageal squamous cell carcinoma and adenocarcinoma, the two main histologic subtypes (IARC, 2004). The association with squamous cell carcinoma is stronger (RR 5.63; 95% CI: 2.7, 11.7) than that with adenocarcinoma (RR 2.77; 95% CI: 1.4, 5.6) in a pooled analysis of smokers with > 60 pack-​years in the Beacon Consortium (Lubin et al., 2012). Alcohol consumption strongly potentiates the relationship of smoking to esophageal squamous cell carcinoma but not adenocarcinoma (see Chapter 30). The overall association between cigarette smoking and esophageal cancer appears to have decreased in high-​income countries, coincident with the shift in cell type toward a larger proportion of adenocarcinoma. Whereas the relative risk estimates in Cancer Prevention Study I  during follow-​up from 1959 to 1966 were 6.76 for men and 7.75 for women (Hammond, 1966), they had decreased to 3.9 (CI: 3.0, 5.0) and 5.1 (CI: 3.5, 7.4) in men and women, respectively, in a pooled analysis of five large contemporary cohorts followed from 2000 to 2010 (Carter et al., 2015). Active cigarette smoking is estimated to account for about half (50.7%, range 44.8% to 56.5%) of deaths from esophageal cancer in the United States (Siegel et al., 2015). Greater detail about the association between cigarette smoking and esophageal cancer is provided in Chapter 30.

Stomach.  The IARC and the US Surgeon General first desig-

nated the evidence linking active smoking to stomach cancer as causal in 2004 (IARC, 2004; US Department of Health and Human Services, 2014). The relative risk estimates for current versus never smokers average approximately 1.6 for men and women combined in more than 20 cohort studies and nearly 40 case-​control studies (IARC, 2004). As described in Chapter  31, the relationship is not confounded by infection with Helicobacter pylori. However, case-​ control studies stratified on H.  pylori seropositivity have reported stronger associations in persons who are seropositive for H. pylori than in uninfected individuals (Brenner et al., 2002; Jedrychowski et al., 1993, 1999; Siman et al., 2001; Wu et al., 2003; Zaridze et al., 2000). Levels of DNA methylation also increase with pack-​years of smoking only among H.  pylori–​infected individuals (Shimazu et  al., 2015). Cigarette smoking increases the risk of cancers throughout the stomach, including cardia and non-​cardia (IARC, 2004; Ladeiras-​Lopes et  al., 2008). No studies have yet examined associations between smoking and four recently identified molecularly defined subtypes of gastric cancer (Cancer Genome Atlas Research, 2014a). Siegel et al. estimate that 19.6% (13.8%–​26.8%) of deaths from stomach cancer among men and women combined in the United States are attributable to cigarette smoking (Siegel et al., 2015).

Colorectum.  Smoking was not associated with colorectal cancer

in large cohort mortality studies conducted in the mid-​twentieth century (Doll and Hill, 1964; Hammond, 1966; Kahn, 1966; Weir and Dunn, 1970). Beginning in the 1980s, studies consistently reported associations between smoking and colorectal adenomatous polyps

20

200

PART III:  THE CAUSES OF CANCER

(IARC, 2004). In the mid-​1990s, Giovannucci and colleagues reported increased frequency of adenomatous polyps and colorectal cancer among smokers in the Harvard Nurses and Health Professionals cohorts (Giovannucci et  al., 1994a, 1994b). They hypothesized that smoking affected an early stage of colorectal neoplasia, and that a long induction period might be needed to observe an association with incidence or mortality rates. The relationship between active smoking and large bowel cancer was designated causal by the IARC in 2012 (IARC, 2012a) and by the US Surgeon General in 2014 (US Department of Health and Human Services, 2014). The association between cigarette smoking and colorectal cancer is weaker than that for most other smoking-​related cancers (Table 11–​3); it is also weaker for cancer of the colon (RR = 1.2; 95% CI: 1.1, 1.3) than rectum (RR = 1.6; 95% CI:1.3, 1.8) (Botteri et  al., 2008). The IARC based its conclusion on the consistent relationship between smoking and adenomatous colorectal polyps and the results of four meta-​analyses that controlled for multiple covariates (Botteri et al., 2008; Huxley et al., 2009; Liang et al., 2009; Tsoi et al., 2009). The 2014 US Surgeon General report reviewed 19 high-​quality prospective studies of smoking in relation to incident colorectal cancer, 9 prospective cohort studies of cancer mortality, and 16 case-​control studies, published during 2002–​2009, plus the meta-​analyses mentioned earlier (US Department of Health and Human Services, 2014). Nearly all studies reported the strongest associations with adenomatous polyps and colorectal cancer in current smokers and intermediate associations in former smokers. The associations are stronger among smokers with early age at initiation and longer duration of smoking. Intriguingly, several studies have suggested stronger associations between cigarette smoking and the subset of colorectal tumors with microsatellite instability (MSI) and CIMP-​or BRAF-​positive tumors (Campbell et al., 2009; Chia et al., 2006; Limsui et al., 2010; Poynter et al., 2009). These findings may point to the carcinogenic mechanism underlying the association. Such relationships, as well as the importance of early life exposure, are discussed in Chapter 36.

Pancreas.  As described in Chapter  32, the association between

active cigarette smoking and pancreatic cancer was designated as causal by the US Surgeon General in 1982 (US Department of Health and Human Services, 1982) and by the IARC in 1986 (IARC, 1986). A recent pooled analysis of five cohorts in the United States reported relative risk estimates of 1.6 in men and 1.9 in women for current versus never cigarette smokers (Carter et  al., 2015). Relative risk is lower for former than for current smokers and decreases with earlier age at cessation (Iodice et al., 2008; Lynch et al., 2009). As noted in Chapter 32, the decrease in pancreatic incidence and mortality rates that occurred among US whites in the late twentieth century, following reductions in smoking, reversed after the year 2000, following large increases in the prevalence of obesity and type 2 diabetes. Recent estimates indicate that active smoking accounts for 10%–​14% of pancreatic cancer in the United States (Siegel et al., 2015; US Department of Health and Human Services, 2014). Further information about the association of active smoking with pancreatic cancer, including potential molecular mechanisms, can be found in Chapter 32.

Liver.  The IARC designated the relationship between active smok-

ing and liver cancer as causal in 2004 (IARC, 2004), as did the US Surgeon General in 2014 (US Department of Health and Human Services, 2014). The latter reviewed 113 studies published through 2012, with particular attention to potential confounding by viral hepatitis and alcohol consumption (US Department of Health and Human Services, 2014). Meta-​analyses found comparable associations in 31 studies that allowed comparisons between current and never smokers (mRR = 1.7; CI: 1.5, 1.9) and in a subset of nine case-​control and four cohort studies in which subjects had no evidence of viral hepatitis (mRR = 1.8; CI: 1.2, 2.7). Other supporting evidence for causality is discussed in Chapter 33. Active smoking has been suggested to interact with other liver cancer risk factors, including viral hepatitis and consumption of alcohol (Chuang et al., 2010; Hassan et al., 2008; Kuper H et al., 2000;

Kuper HE et al., 2000; Marrero et al., 2005), raising concerns about the future burden of liver cancer in low-​and middle-​income countries with continued high prevalence of viral hepatitis and increasing smoking prevalence.

Cigarette Smoking and Urinary Tract Cancers Active cigarette smoking is a well-​established risk factor for cancers throughout the urinary tract, including the kidney (parenchyma and pelvis), ureter, and bladder. The association of these sites with cigarette smoking has long been classified as causal (IARC, 1986; US Department of Health and Human Services, 2004). Current smoking is more strongly associated with urothelial (formerly known as transitional) carcinomas of the renal pelvis (RR > 3) than with tumors of the renal parenchyma (RR about 1.3), as discussed in the following and in Chapter 51.

Renal Adenocarcinoma. Active cigarette smoking has been

designated a cause of renal cell carcinoma by both the IARC and the US Surgeon General (IARC, 2012a; US Department of Health and Human Services, 2014). Relative risk estimates from a recent meta-​ analysis indicate that, compared to never smokers, risk is approximately 36% higher in current smokers and 16% higher among former smokers (Cumberbatch et al., 2016). The classification of renal cell tumors has evolved substantially over the last 30 years to incorporate genetic and molecular tumor characteristics (Chapter 51). The current classification scheme recognizes 15 distinct entities and 4 subtypes (Srigley et al., 2013). Most of the available data on cigarette smoking pertain to clear cell tumors, which comprise over 75% of renal cell carcinomas. Relatively few studies have examined associations by histological subtype. These report associations with clear cell and papillary type, but not with the rarer chromophobe type (Patel et al., 2015; Purdue et al., 2013). Larger studies are needed to characterize variations in the association between smoking and histologic and molecular subtypes of renal cell cancer.

Renal Pelvis and Ureter. Active cigarette smoking is firmly

established as the major cause of urothelial cancers of the renal pelvis and ureter. Relative risk estimates exceed 3.0 in most studies (McLaughlin et al., 1992), including the large prospective NIH-​AARP cohort (Freedman et al., 2016). The estimates of attributable fraction are 50% or higher in many parts of the world.

Urinary Bladder.  The association between active cigarette smok-

ing and cancer of the urinary bladder was first designated as causal by the IARC in 1986 (IARC, 1986) and by the US Surgeon General in 1979 (US Department of Health and Human Services, 1979). Cigarette smoking is the leading cause of bladder cancer in most countries (see Chapter 52). As is the case for lung and laryngeal cancers, the association between smoking and bladder cancer has increased over the last few decades. The relative risk in current compared to never smokers increased from about 3 during the 1990s to approximately 4–​5 in the last decade (Baris et al., 2009; Freedman et al., 2011; Purdue and Silverman, 2016). The relative risk and attributable fraction estimates have historically been higher among men than women (IARC, 2004), but have converged over time in the United States. The attributable fraction was estimated to be approximately 50% in both men and women in the NIH-​AARP cohort (Freedman et al., 2011). Metabolites of heterocyclic aromatic amines, polycyclic aromatic hydrocarbons, and other carcinogens in tobacco are excreted in urine (IARC, 2004; US Department of Health and Human Services, 2010). Studies have investigated interactions between cigarette smoking and inherited variants in metabolizing genes and have documented differences in risk associated with loci that affect N-​acetyltransferase (NAT2) activity (Garcia-​Closas et al., 2013; Kilfoy et al., 2010). These studies, in combination with functional genomic analyses, have provided important clues to bladder carcinogenesis (Chapter 52). Emerging data on the somatic mutations in bladder cancer from The Cancer Genome Atlas (TCGA) and similar projects (Cancer Genome Atlas Research,

 201

Tobacco 2014b) indicate that there are distinct molecularly defined subtypes of bladder cancer. Future studies that examine associations between cigarette smoking and molecularly defined subtypes will likely provide further insights into carcinogenic mechanisms.

Cigarette Smoking and Reproductive Tract Cancers The evidence that cigarette smoking causes cervical cancer and mucinous carcinoma of the ovary is designated as “sufficient” and is discussed here. The evidence that active smoking causes cancers of the breast, vulva and vagina, and aggressive prostate cancer, or reduces the risk of endometrial cancer, are considered “suggestive” and are discussed later.

Uterine Cervix.  Cancer of the uterine cervix is consistently associ-

ated with cigarette smoking. Not until HPV infection was identified as a necessary cause of cervical cancer, however, were concerns about confounding by sexually transmitted diseases addressed directly (IARC, 1995)  (Chapter  48). Later studies investigated cigarette smoking as a cofactor for cervical cancer in HPV-​positive women and observed consistent associations with both invasive squamous cell cancer and carcinoma in situ (Castellsague and Munoz, 2003; International Collaboration of Epidemiological Studies of Cervical Cancer et  al., 2006). The relative risk for squamous cell carcinoma increases in current compared to never smokers with the number of cigarettes smoked per day and with younger age at starting smoking. No association is observed between smoking and adenocarcinoma of the cervix, however. The IARC and the US Surgeon General designated the overall relationship between active cigarette smoking and cervical cancer as causal in 2004 (IARC, 2004; US Department of Health and Human Services, 2004).

Ovary.  Well over 30 epidemiologic studies have investigated the

association of active cigarette smoking with ovarian cancer (IARC, 2012a). Most reported no overall association, although at least four studies have reported a positive association between active cigarette smoking and mucinous ovarian cancer (Collaborative Group on Epidemiological Studies of Ovarian Cancer et al., 2012; Faber et al., 2013; Gram et al., 2012; Wentzensen et al., 2016). Smoking has not been associated with serous or endometrioid subtypes of ovarian cancer; an inverse association has been reported for clear cell tumors (Wentzensen et al, 2016). Etiologic heterogeneity among subtypes of ovarian cancer has also been observed for other established risk factors besides smoking (Wentzensen et al., 2016).

Cigarette Smoking and Hematopoietic Cancers As part of the 2001 WHO classification of hematopoietic and lymphoid neoplasms, “the leukemias” were classified into two major groupings:  myeloid neoplasms (including acute myeloid leukemia, myelodysplastic syndromes, and myeloproliferative neoplasms) (Chapter 38) and lymphoid neoplasms, including lymphoid leukemias (circulating phase) and lymphomas (solid phase) (Chapter 40). Active cigarette smoking has consistently been associated with increased risk of acute myeloid leukemia (AML), as discussed here, but not other hematological or lymphoid malignancies, discussed later in the chapter.

Myeloid Leukemia. The association between active cigarette

smoking and increased risk of AML was designated as causal by both the US Surgeon General and the IARC in 2004. The relative risk for current versus never smoking was 1.40 (95% CI: 1.22,1.60) in a recent meta-​analysis of 23 studies and 7746 cases (Fircanis et al., 2014). Accumulating evidence also suggests that there is also a positive association between smoking and the preleukemic myelodysplastic syndromes. For example, the relative risk for current versus never smoking was 1.81 (95% CI: 1.24, 2.66) in a 2013 meta-​analysis of 14 studies and 2588 cases, with evidence for a dose-​response relationship (Tong et al., 2013).

201

Fewer studies have examined associations with other myeloid neoplasms, and the results have been mixed (Chapter 38). Existing studies have been limited, however, by the uncommon nature of many types and subtypes of myeloid neoplasms and changing classification schemes over time.

Cancer Sites with Suggestive or Limited Evidence Breast

The question of whether a causal relationship exists between breast cancer and exposure to tobacco smoke has been controversial for over 20 years (Chapter 45). The 2014 US Surgeon General Report comprehensively reviewed 15 cohort and 34 case-​control studies published between 2000 and 2012 (US Department of Health and Human Services, 2014), as well as studies previously considered by the IARC (IARC, 2004, 2012a) and the US Surgeon General (US Department of Health and Human Services, 2004). The 2014 Report devoted more than 160 pages of tables and text to this issue. Based on the data then available, it concluded that the evidence regarding active smoking as a cause of breast cancer was “suggestive but not sufficient to infer a causal relationship” (US Department of Health and Human Services, 2014). There was concern about the potential for residual confounding by alcohol consumption and screening and inconsistencies regarding the timing of exposure. An additional eight large cohort studies and meta-​analyses have been published on smoking and breast cancer since the Surgeon General Report (Bjerkaas et al., 2013; Catsburg et al., 2015; Dossus et al., 2014; Gaudet et al., 2013; Gram et al., 2015, 2016; Nyante et al., 2014; Rosenberg et al., 2013). These consistently showed a 10%–​20% higher risk of breast cancer in women who are current compared to never smokers and a somewhat stronger association in women who initiated smoking before first birth. Given the substantial public health importance of the issue, the evidence should be periodically reviewed by expert panels to identify and resolve residual controversies.

Advanced Stage Prostate Cancer

Many studies have investigated associations between active cigarette smoking and prostate cancer (Chapter 53). The relationship is complicated by potential differences in PSA screening and healthcare utilization among smokers and non-​smokers. For incident cancer, a large meta-​ analysis of more than 50,000 incident cases found an inverse association comparing current to never smokers (RR = 0.90; 95% CI: 0.85, 0.96). This association was limited to studies completed after 1995, however, when PSA screening became widely adopted (Islami et al., 2014). In contrast, studies of advanced stage or aggressive prostate cancer have tended to find positive associations with smoking (Zu and Giovannucci, 2009), as have studies of prostate cancer mortality. Current cigarette smoking was associated with increased mortality from prostate cancer in a meta-​analysis of nearly 12,000 prostate cancer deaths (RR:  1.24; 95% CI:  1.18, 1.31), with evidence for a dose-​response relationship (Islami et  al., 2014). There remains uncertainly about whether this association is causal, however. In 2014, the US Surgeon General comprehensively reviewed the evidence and concluded that the available data were suggestive of no causal relationship with incident prostate cancer but higher risk of prostate cancer death (US Department of Health and Human Services, 2014). Future epidemiologic and mechanistic research is certainly warranted.

Vulva and Vagina

Cancers of the vulva and vagina are rare, but active cigarette smoking is associated with increased risk of both sites in the limited number of studies on this issue (Chen et al., 1999; Daling et al., 1992, 2002; Hussain et  al., 2008; Madeleine et  al., 1997). A  strong interaction between cigarette smoking and HPV 16 on the risk of vulvar cancer was reported by Madeleine et al. (1997). More work is needed to characterize the combined effect of smoking and HPV infection on these cancers with respect to the timing of exposure (Chapter 49).

20

202

PART III:  THE CAUSES OF CANCER

Anus

Cancer of the anus, a malignancy with squamous or transitional cell histology, has been repeatedly associated with cigarette smoking, although confounding by HPV has not been excluded (Chapter  37). Future studies with careful control of confounding are needed to resolve the issue.

Biliary Tract

Studies have investigated associations between active smoking and cancers of the biliary tract, including gallbladder, extrahepatic bile duct, and ampulla of Vater (Chapter  34). Neither the IARC nor the US Surgeon General has classified biliary tract cancer as being causally related to tobacco smoking. However, a recent meta-​analysis of 11 studies reported an association between ever smoking and gallbladder cancer (summary RR = 1.45; 95% CI: 1.11, 1.89) (Wenbin et al., 2013). A separate meta-​analysis also found evidence for an association with extrahepatic bile duct cancers (summary RR  =  1.23; 95% CI: 1.01, 1.50) (Ye et al., 2013).

Keratinocyte Carcinomas

Keratinocyte cancers are the most common malignancies in humans. These tumors, which include basal and squamous cell carcinomas, arise from keratinocytes or their precursor cells (Chapter 58). Results from case-​control and cohort studies generally suggest an association between active cigarette smoking and squamous cell carcinomas. A meta-​analysis in 2012 observed a relative risk of 1.54 (95% CI: 1.03, 2.31) comparing current to never smokers (Leonardi-​Bee et al., 2012). In contrast, active smoking does not appear to increase risk of basal cell carcinoma (0.89; 95% CI: 0.77, 1.03).

Cancer Sites Inversely Associated with Cigarette Smoking Endometrium

The incidence of endometrial cancer is inversely associated with active cigarette smoking (IARC, 2004; US Department of Health and Human Services, 2004). The relationship varies by smoking status; risk was 30%–​40% lower in current smokers and 9%–​12% lower among former compared to never smokers in an analysis of the National Institutes of Health-​AARP Diet and Health Study (Felix et al., 2014). The association is also stronger in postmenopausal than premenopausal women (Zhou et al., 2008), but has not been examined in relation to tumor subtypes or molecular markers. A number of mechanisms have been hypothesized, including effects of smoking on circulating hormone levels (discussed in Chapter 47 for endometrial cancer and in Chapter  45 for breast cancer, as well as in the 2014 US Surgeon General’s report) (US Department of Health and Human Services, 2014).

Thyroid

Data from both case-​control and cohort studies suggest an inverse association between active cigarette smoking and thyroid cancer (Chapter 44). For example, the relative risks for current versus never smoking was 0.68 (95% CI: 0.55, 0.85) in a pooled analysis of five prospective US cohort studies (Wiersinga, 2013). The mechanisms that underlie this association are unclear.

Melanoma

Both case-​control and cohort studies have examined the association between active cigarette smoking and melanoma (Chapter 57). Most studies have reported inverse associations. A  2015 meta-​analysis of 23 studies observed pooled relative risks of 0.70 (95% CI: 0.63, 0.78) for current smoking and 0.90 (95% CI: 0.85, 0.95) for former smoking (Li et al., 2015). The mechanisms underlying this association are unclear. Non-​causal explanations, including confounding by sunlight and other risk factors, selection bias, competing risks, and publication bias, have been suggested (Freedman et al., 2003; Li et al., 2015; Thompson et al., 2013).

Cancer Sites with Limited Evidence or No Association Small Intestine

Relatively few studies have examined active smoking in relation to cancers of the small intestine, due to the rarity and heterogeneity of these cancers. However, recent studies do not suggest an association with either adenocarcinoma or carcinoid tumors, which make up the two main histological types of small intestinal cancers (Chapter 35).

Testes

There is limited published evidence on cigarette smoking and cancer of the testes (Brown et  al., 1987; Gallagher et  al., 1995; Henderson et al., 1979), although three relatively recent studies have found associations with marijuana smoking (Daling et  al., 2009; Lacson et  al., 2012; Trabert et al., 2011) (Chapter 54).

Penis

Several case-​control studies have examined the association of active cigarette smoking with penile cancer (Chapter 55); however, the results to date are inconsistent, and confounding by HPV status is a concern.

Non-​Hodgkin Lymphoma

Numerous studies have investigated associations of active smoking with non-​Hodgkin lymphomas (NHL), a heterogeneous group of over 40 lymphoid neoplasms (Chapter 40). Overall, there appears to be no association, although associations with certain subtypes have been noted (Morton et al., 2014).

Hodgkin Lymphoma

Active cigarette smoking is not generally considered to be a risk factor for Hodgkin lymphoma. However, results from a pooled analysis from 12 case-​control studies in the InterLymph consortium (Kamper-​ Jorgensen et  al., 2013), as well as separate meta-​analyses, provide evidence for a positive association (Castillo et al., 2011; Sergentanis et al., 2013), at least for some subtypes (Chapter 39).

Multiple Myeloma

There is no evidence that smoking causes multiple myeloma (Chapter 41).

Nervous System

Most studies have found no association with smoking and glioma or other tumors of the nervous system (Chapter 56).

Soft Tissue Sarcomas

The data regarding cigarette smoking and soft-​tissue sarcoma are limited and mixed (Franceschi and Serraino, 1992; Serraino et al., 1991; Zahm et al., 1992) (Chapter 43).

Fraction of Cancer Deaths Attributed to Cigarette Smoking Estimates of the fraction of all cancer deaths attributable to tobacco use have changed little over the past 30–​40  years. In 1981, Doll and Peto estimated that 30% (acceptable range 25%–​40%) of cancer deaths in the United States could be attributed to tobacco smoking (Doll and Peto, 1981). This estimate has persisted, despite reductions in smoking prevalence and consumption, partly because additional cancer sites have been designated as causally related to smoking (US Department of Health and Human Services, 2014), and partly because of the delayed effect of birth cohort patterns of smoking on cancer risk. Jacobs et  al. recently estimated that the population attributable fraction (PAF) for deaths from all cancers combined was 28.7% when estimated conservatively, including only deaths from the 12 cancers currently formally designated as causally related to smoking by the US Surgeon General, but that the PAF was 31.7% when estimated more comprehensively, including excess deaths from all cancers (Jacobs et al., 2015). The PAF also varies by

 203

Tobacco state and region, reflecting the levels of tobacco use and control. In a recent state-​wide analysis, the PAF estimates for active smoking ranged from 16.6% in Utah to 34% in Kentucky (Lortet-​Tieulent et  al., 2016). Although substantial, these estimates do not include involuntary exposure to tobacco smoke or other types of tobacco use such as cigars, pipes, or smokeless tobacco (Jacobs et  al., 2015). The attributable fraction is smaller for incident cancers than for deaths from cancer, due to the large contribution of breast and prostate cancer to incident cases. Parkin has estimated that smoking caused 60,000 new cancer cases in the United Kingdom in 2010 (19.4% of all new cases) (Parkin, 2011). The only global estimate of smoking-​attributable cancer is by Jha, who estimated that 31% of cancer deaths worldwide in men, and 6% in women, were attributable to smoking (Jha, 2009).

CANCER RISKS FROM OTHER TOBACCO PRODUCTS Both the IARC and the US Surgeon General have concluded that all forms of tobacco use are carcinogenic. However, the evidence based on combustible products other than cigarettes and on non-​combustible tobacco products in relation to cancer is less extensive than that for cigarettes (IARC, 2012b).

Other Combustible Products Pipes and Cigars

Pipes and cigars are the main non-​cigarette smoked tobacco products in high-​and many middle-​income countries. Pipe smoking was designated causally related to lip cancer in 1964 (US Department of Health Education and Welfare, 1964). Exclusive pipe smoking was associated with increased mortality rates from four cancer sites among men in an analysis of the CPS-​II cohort by Henley et al. (2004). Compared to never smokers, men who currently smoked pipes but reported no history of using other tobacco products had increased mortality from cancer of the oral cavity/​pharynx (RR = 3.90; 95% CI: 2.15, 7.08), larynx (RR = 13.1; 95% CI: 5.2, 33.1, lung (RR = 5.00; 95% CI: 4.16, 6.01), and esophagus (RR  =  2.44; 95% CI:  1.51, 3.95), as well as coronary heart disease, stroke, and chronic obstructive pulmonary disease (Henley et al., 2004). The risks were generally smaller than those associated with cigarette smoking and similar to or larger than those associated with cigar smoking. A separate analysis of men who currently and exclusively smoked cigars in CPS-​II reported increased death rates from cancers of the lung (RR = 5.1; 95% CI: 4.0, 6.6), oral cavity/​pharynx (RR = 4.0; 95% CI: 1.5, 10.3), larynx (RR = 10.3; 95% CI: 2.6,41.0), and esophagus (RR  =  1.8; 95% CI:  0.9, 3.7) compared to never smokers (Shapiro et al., 2000). Both cigar and pipe smokers had increased incidence of head and neck cancer in the International Head and Neck Cancer Epidemiology (INHANCE) Consortium (Wyss et  al., 2013). The risk of head and neck cancer was elevated for those who reported exclusive cigar smoking (odds ratio  =  3.49; 95% CI:  2.58, 4.73) or exclusive pipe smoking (odds ratio = 3.71; 95% CI: 2.59, 5.33) compared to never smokers. The associations with cancers of the head and neck and esophagus are predominantly with squamous cell carcinoma rather than adenocarcinoma (Chang et al., 2015; National Cancer Institute, 1998). The INHANCE study separately examined the association of pipe and cigar smoking with oral cavity cancer, based on approximately 4100 cases. The odds ratio (OR) estimates for oral cavity cancer were 2.51 (95% CI: 1.68, 3.75) for exclusive pipe smoking and 2.83 (95% CI: 1.91, 4.17) for cigar smoking, compared to the risk of never smokers (Wyss et al., 2013).

Bidis

As mentioned earlier, bidis are local tobacco products composed of coarse and uncured tobacco, generally smoked without filters by men in India (IARC, 2012a). Epidemiologic studies of bidi smoking and

203

cancer were reviewed by the IARC (IARC, 2012a). Additional cohort and case control studies from India have reported associations between bidi smoking and cancers of the oral cavity (Jayalekshmi et al., 2011; Pednekar et  al., 2011), hypopharynx and larynx (Jayalekshmi et  al., 2013; Pednekar et al., 2011), lung (Pednekar et al., 2011), esophagus, and stomach (Jayalekshmi et al., 2015).

Kreteks, Waterpipes, and Chuttas Kreteks, waterpipes, and chuttas are other smoked tobacco products for which the long-​term effects on health are not well characterized. Kreteks are clove-​flavored cigarettes, manufactured and widely consumed in Indonesia (Palipudi K et al., 2015). They contain eugenol, a natural compound found in high concentration in clove buds, which has an anesthetic effect (Tobacco Atlas, 2015). Waterpipes, also known as “hookah,” “narghile,” and “shisha,” are commonly smoked in North Africa, the Mediterranean, and parts of Asia. They have recently been popularized among students in high-​income countries. A meta-​ analyses of 13 case-​control studies suggest increased risks for waterpipe use with cancers of the lung, esophagus, and possibly other sites (Montazeri et al., 2017). However, most existing studies have methodological limitations, and high-​quality studies with standardized exposure measurements are needed. Chuttas are coarse cheroots, infrequently smoked in South India with the lighted end inside the mouth (reverse smoking). They have been associated with carcinomas of the hard palate in a case series (Reddy, 1974).

SMOKELESS TOBACCO PRODUCTS More than 300 million adults in 70 countries use smokeless tobacco (National Cancer Institute and Centers for Disease Control and Prevention, 2014). The IARC has determined that smokeless tobacco is causally related to cancers of the esophagus, oral cavity, and pancreas (IARC, 2012a). The majority of consumers (89%) are in Southeast Asia, where these products are inexpensive, socially acceptable, and readily available. In most countries, usage is more common in men than women. In some countries, however (e.g., Bangladesh, Thailand, Cambodia, Malaysia, Vietnam, and some African countries), use by adult women is similar to or greater than that by adult men. The chemical composition and levels of free nicotine vary widely among these products. The concentrations of tobacco-​specific carcinogens (TSNAs) such as NNN and NNK can vary by several orders of magnitude (National Cancer Institute and Centers for Disease Control and Prevention, 2014). Products that are widely used in low-​and middle-​income countries are usually handmade or produced by cottage industries, and are consequently less standardized than manufactured smoked or smokeless tobacco products. Studies of these products in relation to cancer are complicated by their diversity and the frequent use of more than one product. They involve exposure to complex and varying mixtures of ingredients that may include other plant materials, such as areca nut and tonka bean, in addition to tobacco (National Cancer Institute and Centers for Disease Control and Prevention, 2014).

Betel Quid Betel quid is commonly chewed in India and throughout much of Southeast Asia and the Western Pacific. As mentioned earlier, it contains a mixture of areca nut, betel leaf (Piper betle), and other ingredients, and can be made with or without tobacco. IARC Working Groups have classified betel quid with tobacco as a Group 1 human carcinogen, based on cancers of the oral cavity (IARC, 2004) and squamous carcinoma of the esophagus (IARC, 2012a). Areca nut is also classified as carcinogenic to humans. A  Taiwanese cohort study reported associations between betel-​quid chewing and laryngeal cancer (Lee et al., 2011), with consumption of > 20 quids daily associated with an adjusted hazard ratio of 1.7 (CI: 1.2, 2.6).

204

204

PART III:  THE CAUSES OF CANCER

Moist Snuff and Chewing Tobacco In high-​income countries, the most commonly used traditional product is moist snuff, followed by chewing tobacco. Fermentation and aging are used to modify the taste of both moist and dry snuff, but these increase the concentration of TSNA. The IARC has designated moist snuff as carcinogenic to humans, based on cancers of the oral cavity and pancreas (IARC, 2007). In high-​income countries, moist snuff is increasingly sold in teabag-​like sachets that reduce the need to spit. A  moist snuff, low-​nitrosamine product called snus is manufactured and widely used in Scandinavia. The levels of TSNA are much higher in commercial brands of moist snuff sold in the United States and in products sold in low-​and middle-​income countries than in snus. Until recently, tobacco control and regulatory efforts have focused primarily on cigarettes, and have paid less attention to the marketing practices and chemical composition of smokeless products. The situation is further complicated in low-​and middle-​income countries by the unorganized nature of a large business sector, which impedes product control and regulation (National Cancer Institute and Centers for Disease Control and Prevention, 2014).

Novel Smokeless Tobacco Products As mentioned previously, tobacco manufacturers have introduced novel smokeless tobacco products using innovations such as portion pouches, dissolvable tobacco, unique flavorings (such as fruit and candy flavors), electronic nicotine delivery systems (ENDS) and varying nicotine levels, which may make products more attractive to consumers, including those who have not previously used tobacco products (National Cancer Institute and Centers for Disease Control and Prevention, 2014). Among these, e-​cigarettes are by far the most popular. E-​cigarettes are overtly marketed to smokers as a source of nicotine in settings where they cannot smoke, and are informally positioned as cessation devices. They also are sold in a remarkable variety of flavors, many of which appeal to youth. The health effects of e-​cigarettes are at this point unclear. With their recent introduction to the marketplace, e-​cigarettes may have long-​ term effects that will not be fully apparent for many years. In addition to directly affecting health, they can indirectly affect disease risks by modifying the use of other tobacco products. For example, adolescents may choose e-​cigarettes as an alternative to cigarette smoking, or they may become addicted to nicotine from e-​cigarettes and then progress to cigarette smoking. Cigarette smokers may use e-​cigarettes to facilitate quitting, or alternatively may use ENDS to supplement their nicotine intake in settings where they cannot smoke. Any product that facilitates continued smoking instead of quitting will increase disease risks. Even smokers who continue to smoke just a few cigarettes a day have measurably higher risks of cancer and other chronic diseases than those who quit (Inoue-​Choi et al., 2016).

INTERACTIONS WITH OTHER EXPOSURES Alcohol The combined effect of tobacco use and alcohol consumption has been examined extensively for cancers of the oral cavity, pharynx, larynx, and esophagus and to a lesser extent for cancers of the liver and pancreas (IARC, 2004). In the larger studies, cancer risk consistently increased more rapidly with the combination of smoking and heavy drinking than with either exposure alone. Case-​control studies that evaluated statistical interaction formally demonstrated a greater than multiplicative relationship with joint exposure.

Infectious Agents Tobacco use is an established cofactor with human papillomavirus infection for anogenital cancers (Chapter 24). There is intriguing, albeit limited, evidence that tobacco use combined with certain infectious

agents may foster transformation of premalignant abnormalities into invasive cancer of the liver, stomach, uterine cervix, and/​or lung. The evidence currently available regarding possible interactions between tobacco use and diet is limited.

Occupational Exposures There are well established interactions between tobacco smoking and several occupational exposures with respect to lung cancer. These include asbestos and radon (National Research Council Committee on Health Risks of Exposure to Radon, 1998). In general, the statistical interaction between smoking and radon appears submultiplicative, but without strong evidence against multiplicative interaction (IARC, 2004).

INVOLUNTARY EXPOSURE TO TOBACCO SMOKE Lung Cancer Many authoritative scientific groups have concluded that involuntary exposure to tobacco smoke causes lung cancer and coronary heart disease in humans (Australian National Health and Medical Research Council, 1987; IARC, 1986, 2004, 2012a; National Research Council, 1986; US Environmental Protection Agency, 1992; US Department of Health and Human Services, 1986, 2006, 2014). The association has long been judged to be causal based on its biologic plausibility, replication in multiple different settings by different investigators using a variety of study designs, and supportive clinical information. Biomarker studies indicate that non-​ smokers exposed to secondhand smoke absorb, metabolize, and excrete toxic constituents of tobacco smoke (US Department of Health and Human Services, 2006). In fact, passive smokers appear to have greater excretion of a metabolite of a tobacco-​ specific carcinogen than active smokers (Vogel et al., 2011). Studies of genotoxicity document a greater prevalence of DNA adducts and strand breaks in non-​ smokers exposed to secondhand smoke (Husgafvel-​ Pursiainen, 2004). The most recent comprehensive review by IARC (2012a) cited more than 50 epidemiologic studies and meta-​analyses published internationally since the first studies in Japan and Greece reported increased lung cancer risk in non-​smoking women married to cigarette smokers (Hirayama, 1981; Trichopoulos et al., 1981). Despite this evidence, the tobacco industry has long sought to maintain the appearance of controversy, as discussed in Chapter 17.

Breast Cancer In contrast to the evidence on secondhand smoke and lung cancer, reviews of breast cancer in relation to involuntary exposure have reached different conclusions (California Environmental Protection Agency, 1997; IARC, 2004, 2012a; Johnson, 2005; Miller et al., 2007; US Department of Health and Human Services, 2006). The IARC and the US Surgeon General have variously described the evidence as “inconsistent” (IARC, 2004)  or as “suggestive but not sufficient” to infer causality, whereas reviews by the California Environmental Protection Agency in (California Environmental Protection Agency, 2005) and by a panel of researchers convened in Canada (Collishaw NE et  al., 2009)  designated the evidence for secondhand tobacco smoke as “consistent with a causal association in younger primarily premenopausal women.” Much of the supportive evidence comes from case-​control studies that have attempted to collect full “lifetime exposure histories” of secondhand smoke (Collishaw NE et al., 2009). These studies observed the strongest associations with breast cancer and were considered by Collishaw et  al. to have the most complete information on lifetime exposure to secondhand tobacco smoke from all sources. An important limitation of these studies, however, is that they are also most susceptible to recall bias, especially for exposures many years in the past, during an era when exposure to secondhand smoke was ubiquitous (IARC, 2012a).

 205

Tobacco According to the IARC, the strongest support for the hypothesis comes from a cohort study in Japan (Hanaoka et al., 2005), which reported significantly increased risk (RR 2.6; 95% CI: 1.3, 5.2) of premenopausal breast cancer in women who previously reported having ever lived with a regular smoker or ever being exposed to secondhand tobacco smoke for at least 1 hour per day in settings outside the home. The referent group in this analysis included only nine unexposed cases, however. A weak association between secondhand tobacco smoke exposure and premenopausal breast cancer was also reported in the California Teachers cohort, when menopausal status was defined by age at diagnosis rather than age at entry into the study (Reynolds et al., 2006). The absence of a strong and convincing relationship between breast cancer and active smoking weakens the case for a causal relationship with secondhand smoke, given that the level of exposure is orders of magnitude lower. Theories have been advanced to explain why secondhand tobacco smoke might have a stronger effect on breast cancer than active smoking (California Environmental Protection Agency, 1997; Collishaw NE et  al., 2009; Johnson, 2005), these assume the existence of a bidirectional effect of tobacco smoke on breast cancer risk, for which there is no direct evidence (IARC, 2012a).

Other Cancer Sites The IARC 2012 review also comprehensively considered the epidemiologic evidence regarding involuntary smoking and cancers of the upper aerodigestive tract, gastrointestinal and genitourinary systems, brain, leukemias, lymphomas, and childhood cancers. The evidence regarding these sites remains inadequate.

OPPORTUNITIES FOR PREVENTION Well-​established “best practices” in tobacco control are discussed in Chapter 62.1. The implementation of these can effectively reduce tobacco use and prevent the devastating effects of tobacco on health (US Department of Health and Human Services, 2014). Long-​term progress depends on the systematic application of primary prevention measures that can reduce the initiation of tobacco use by young people and end the pandemic during the second half of the twenty-​first century. In the near term, substantial reductions in smoking-​attributable cancers and other diseases can be achieved by providing counseling and treatment to facilitate cessation among the 36.5 million Americans and others who currently smoke (Jamal et al., 2016).

Increasing the Age at Initiation A growing public health emphasis has been placed on increasing the minimum age of tobacco purchase to age 21 (Winickoff et al., 2014). As described earlier, most cigarette smokers in economically developed countries begin smoking in their teenage years. Earlier age at initiation has been associated with increased addiction to nicotine and greater risk of nearly all smoking-​related diseases. Furthermore, the main source of cigarettes for minors is from others under the age of 21 (DiFranza and Coleman, 2001). Adolescent brains have also been shown to be particularly sensitive to the addictive properties of nicotine (US Department of Health and Human Services, 2012). For these and other reasons, modeling studies, including those in a comprehensive 2015 report by the Institute of Medicine, estimate that raising the cigarette smoking age to 21 years could substantially decrease the prevalence of cigarette smoking in the population (Institute of Medicine, 2015). At the time of this writing, at least 200 local and two state governments have implemented “tobacco 21” policies, and emerging data suggest that these may indeed affect teenage smoking prevalence (Kessel Schneider et al., 2016).

Tobacco Cessation Many of the adverse effects of tobacco use can be prevented or reversed by cessation (IARC, 2004, 2012a). Paradoxically, the benefits

205

of cessation appear more quickly than does the increase in cancer risk after initiation, presumably because quitting removes late-​stage promoting effects of smoking on carcinogenesis. The benefits of stopping smoking or other forms of tobacco use are best seen by comparing risk among smokers who quit at various ages with the risk of those who continue to smoke (Figure 11–​11). Smoking cessation at any age avoids much of the future risk seen with continued smoking. The relative and absolute benefits are greatest when cessation occurs at an early age, but are substantial even when stopping occurs by age 50 or 60 (Peto et al., 2000). The relative risks among persons who quit smoking compared with those who continue are progressively smaller the earlier the age of cessation and the longer the time that has elapsed since cessation. Only in smokers who quit at younger ages does the relative risk of death from these cancers approach unity when compared with that of persons who have never smoked. Analyses of cessation are more informative when based on the age at cessation than time since quitting, since the relative and absolute benefits vary by age, and individual smokers can more easily relate their personal situation to age at quitting. It is also more informative to compare smokers who have quit smoking to those who continue than to compare them with lifelong non-​smokers. Smokers have the option to quit smoking but not to become lifelong non-​smokers.

FUTURE RESEARCH DIRECTIONS Despite progress in reducing cigarette smoking in many high-​and middle-​income countries and the immense literature that already exists on the harmful health effects of tobacco use, continuing research is needed to strengthen global tobacco control efforts and address unresolved etiologic and mechanistic questions. The significant expansion of efforts to collect nationally representative surveillance data (CDC, 2016), especially in low-​and middle-​income countries, creates opportunities to monitor patterns of tobacco use in countries where the number of smokers is expected to increase most rapidly over the next decade. Studies that track the implementation of “best practices” for comprehensive tobacco control, as mandated by the Framework Convention on Tobacco Control (FCTC) (WHO, 2003), are essential in countries at all levels of economic development. Other types of application research, such as comparative effectiveness studies, can be used to tailor interventions to specific populations and social contexts. Descriptive studies can characterize the temporal and geographic relationships between the implementation of tobacco control measures and changes in knowledge, attitudes, beliefs, and behaviors in populations. Assessment of key indicators, such as tobacco use by medical providers and average age at initiation, are especially informative in this regard. The introduction of novel tobacco products, such as e-​cigarettes, creates both challenges and opportunities for research. For example, much larger randomized clinical trials are urgently needed to determine whether e-​cigarettes are effective for smoking cessation, and, if so, to compare their efficacy to that of established cessation treatments. More studies are needed to characterize the impact of e-​cigarettes on nicotine dependence, intermediate endpoints, and patterns of tobacco usage in populations. Longitudinal studies are needed to determine whether e-​cigarettes as currently marketed deter the uptake of cigarette smoking in young people or simply addict them to nicotine. Continued surveillance research is needed to monitor the shifting patterns of tobacco use, trends in dual or poly use, and increased consumption of roll-​your-​own cigarettes and small filter-​tip cigars to circumvent the excise taxes on cigarettes. Research on marketing must increasingly monitor point-​of-​sale promotions (discounts, single cigarettes) designed to counter tobacco control policies. Important etiologic questions remain unresolved as well, such as whether active cigarette smoking is causally related to breast cancer and aggressive prostate cancer. Larger studies that incorporate detailed information on known risk factors for these cancers and screening are needed to resolve these controversies. It is hypothesized, but not firmly established, that the initiation of smoking between menarche and the time of first childbirth confers increased susceptibility to premenopausal breast cancer, and that tobacco smoke has a bidirectional

206

206

PART III:  THE CAUSES OF CANCER

effect on the metabolism of steroidal sex hormones (US Department of Health and Human Services, 2014). The availability of more accurate high-​throughput measurement of sex hormones and their metabolites (Chapter 22) could be used to test the latter hypothesis. Similarly, measurements of harmful and potentially harmful constituents (HPHCs) among users of different classes of tobacco products have the potential to refine our understanding of tobacco carcinogenesis and may contribute to regulation. Studies based on histologic or molecular characteristics have already demonstrated associations with cigarette smoking for some tumor subtypes but not others. Broader efforts are needed to identify molecular subtypes associated with tobacco use across a range of cancer sites. Advances in molecular and genomic technologies create opportunities to study the effects of tobacco exposure on the genome and transcriptome (methylation, mutational patterns, mosaicism, mRNA expression). Already cigarette smoking has been demonstrated to cause wholesale genomic and epigenetic changes in users. Studies to assess whether smoking affects later stage molecular events in prostate cancer could provide mechanistic information to assess the causality of its relationship with advanced disease. Finally, there is a continuing need to assess the usage and disease consequences of multiple types of tobacco products, including biomarkers of exposure, measures of addictiveness, and transition to manufactured cigarettes. References Agaku IT, and Alpert HR. 2015. Trends in annual sales and current use of cigarettes, cigars, roll-​your-​own tobacco, pipes, and smokeless tobacco among US adults, 2002–​2012. Tob Control. PMID: 25899447. Agaku IT, King BA, Husten CG, et  al. 2014. Tobacco product use among adults:  United States, 2012–​ 2013. MMWR Morb Mortal Wkly Rep, 63(25), 542–​547. PMID: 24964880. Alexandrov LB, Ju YS, Haase K, et al. 2016. Mutational signatures associated with tobacco smoking in human cancer. Science, 354(6312), 618–​622. PMID: 27811275. Apelberg BJ, Hepp LM, Avila-​Tang E, et  al. 2013. Environmental monitoring of secondhand smoke exposure. Tob Control, 22(3), 147–​ 155. PMCID: PMC3639351. Arrazola RA, Singh T, Corey CG, et al. 2015. Tobacco use among middle and high school students:  United States, 2011–​2014. MMWR Morb Mortal Wkly Rep, 64(14), 381–​385. PMID: 25879896. Ashley DL, Beeson MD, Johnson DR, et  al. 2003. Tobacco-​specific nitrosamines in tobacco from U.S. brand and non-​U.S. brand cigarettes. Nicotine Tob Res, 5(3), 323–​331. PMID: 12791527. Asma S, Song Y, Cohen J, et al. 2014. CDC Grand Rounds: Global Tobacco Control. MMWR Morb Mortal Wkly Rep, 63(13), 277–280. PMID: 24699763. Australian National Health and Medical Research Council. (1987). Effects of Passive Smoking on Health. Canbera, AU:  Government Publishing Service. Avila-​Tang E, Al-​Delaimy WK, Ashley DL, et  al. 2013a. Assessing secondhand smoke using biological markers. Tob Control, 22(3), 164–​171. PMCID: PMC3639350. Avila-​Tang E, Elf JL, Cummings KM, et  al. 2013b. Assessing secondhand smoke exposure with reported measures. Tob Control, 22(3), 156–​163. PMCID: PMC3639349. Balfour DJ, and Fagerstrom KO. 1996. Pharmacology of nicotine and its therapeutic use in smoking cessation and neurodegenerative disorders. Pharmacol Ther, 72(1), 51–​81. PMID: 8981571. Baris D, Karagas MR, Verrill C, et al. 2009. A case-​control study of smoking and bladder cancer risk: emergent patterns over time. J Natl Cancer Inst, 101(22), 1553–​1561. PMCID: PMC2778671. Barrington-​Trimis JL, Urman R, Berhane K, et  al. 2016. E-​Cigarettes and future cigarette use. Pediatrics, 138(1). PMCID: PMC4925085. Benowitz N. 2001. Compensatory smoking of low-​yield cigarettes. In: Shopland D, Burns D, Benowitz N, and Amacher R (Eds.), Risks Associated with Smoking Cigarettes with Low Machine-​ Measured Levels of Tar and Nicotine, Smoking and Tobacco Control Monograph 13 (Vol. NIH Publ. No. 02-​5074, pp. 39–​63). Bethesda, MD: US Department of Health and Human Services, National Institutes of Health, National Cancer Institute. Benowitz NL. 1996. Pharmacology of nicotine:  addiction and therapeutics. Annu Rev Pharmacol Toxicol, 36, 597–​613. PMID: 8725403. Bilano V, Gilmour S, Moffiet T, et al. 2015. Global trends and projections for tobacco use, 1990–​2025:  an analysis of smoking indicators from the

WHO Comprehensive Information Systems for Tobacco Control. Lancet, 385(9972), 966–​976. PMID: 25784347. Bjerkaas E, Parajuli R, Weiderpass E, et  al. 2013. Smoking duration before first childbirth: an emerging risk factor for breast cancer? Results from 302,865 Norwegian women. Cancer Causes Control, 24(7), 1347–​1356. PMCID: PMC3675272. Blot WJ, Cohen SS, Aldrich M, et  al. 2011. Lung cancer risk among smokers of menthol cigarettes. J Natl Cancer Inst, 103(10), 810–​ 816. PMCID: PMC3096798. Boffetta P, Clark S, Shen M, et al. 2006. Serum cotinine level as predictor of lung cancer risk. Cancer Epidemiol Biomarkers Prev, 15(6), 1184–​1188. PMID: 16775179. Borgerding MF, Bodnar JA, Curtin GM, and Swauger JE. 2012. The chemical composition of smokeless tobacco: a survey of products sold in the United States in 2006 and 2007. Regul Toxicol Pharmacol, 64(3), 367–​ 387. PMID: 23000415. Botteri E, Iodice S, Bagnardi V, et al. 2008. Smoking and colorectal cancer: a meta-​analysis. JAMA, 300(23), 2765–​2778. PMID: 19088354. Brandt AM. 2007. The Cigarette Century: The Rise, Fall, and Deadly Persistence of the Product That Defined America. New York: Basic Books. Brandon TH, Goniewicz ML, Hanna NH, et al. 2015. Electronic nicotine delivery systems: a policy statement from the American Association for Cancer Research and the American Society of Clinical Oncology. J Clin Oncol, 33(8), 952–963. Brenner H, Arndt V, Bode G, et al. 2002. Risk of gastric cancer among smokers infected with Helicobacter pylori. Int J Cancer, 98(3), 446–​449. PMID: 11920598. Brooks DR, Palmer JR, Strom BL, and Rosenberg L. 2003. Menthol cigarettes and risk of lung cancer. Am J Epidemiol, 158(7), 609–​616; discussion 617–​620. PMID: 14507595. Brown LM, Pottern LM, and Hoover RN. 1987. Testicular cancer in young men:  the search for causes of the epidemic increase in the United States. J Epidemiol Community Health, 41(4), 349–​ 354. PMCID: PMC1052660. Burns D, Lee L, Shen LZ, et  al. 1997. Cigarette smoking behavior in the United States. In: Shopland D, Burns D, Garfinkel L, and Samet J (Eds.), Changes in Cigarette Related Disease Risks and Theit Implication for Prevention and Control, Smoking and Tobacco Control Monograph 8 (pp. 13–​112). Washington, DC: National Cancer Institute. Calafat AM, Polzin GM, Saylor J, et al. 2004. Determination of tar, nicotine, and carbon monoxide yields in the mainstream smoke of selected international cigarettes. Tob Control, 13(1), 45–​51. PMCID: PMC1747810. California Environmental Protection Agency. (2005). Health Effects of Exposure to Environmental Tobacco Smoke:  Final Report. Sacramento, CA. https://​oehha.ca.gov/​air/​report/​health-​effects-​exposure-​environmental-​ tobacco-​smoke-​final-​report Campbell PT, Curtin K, Ulrich CM, et al. 2009. Mismatch repair polymorphisms and risk of colon cancer, tumour microsatellite instability and interactions with lifestyle factors. Gut, 58(5), 661–​667. PMCID: PMC2903215. Cancer Genome Atlas Research Network. 2014a. Comprehensive molecular characterization of gastric adenocarcinoma. Nature, 513(7517), 202–​209. PMCID: PMC4170219. Cancer Genome Atlas Research Network. 2014b. Comprehensive molecular characterization of urothelial bladder carcinoma. Nature, 507(7492), 315–​322. PMCID: PMC3962515. Carter BD, Abnet CC, Feskanich D, et al. 2015. Smoking and mortality: beyond established causes. N Engl J Med, 372(7), 631–​640. PMID: 25671255. Castellsague X, and Munoz N. 2003. Chapter  3:  Cofactors in human papillomavirus carcinogenesis: role of parity, oral contraceptives, and tobacco smoking. J Natl Cancer Inst Monogr, 31, 20–​28. PMID: 12807941. Castillo JJ, Dalia S, and Shum H. 2011. Meta-​analysis of the association between cigarette smoking and incidence of Hodgkin’s lymphoma. J Clin Oncol, 29(29), 3900–​3906. PMID: 21911724. Catsburg C, Miller AB, and Rohan TE. 2015. Active cigarette smoking and risk of breast cancer. Int J Cancer, 136(9), 2204–​2209. PMID: 25307527. Centers for Disease Control and Prevention (CDC). 1994. Cigarette smoking among adults-​-​United States, 1992, and changes in the definition of current cigarette smoking. MMWR Morb Mortal Wkly Rep, 43(19), 342–​346. CDC. 1999. Achievements in public health, 1900–​1999: tobacco use—​United States, 1900–​1999. MMWR Morb Mortal Wkly, 48(43), 986–​993. https://​ www.cdc.gov/​mmwr/​preview/​mmwrhtml/​mm4843a2.htm CDC. 2016. Global Tobacco Surveillance System. www.cdc.gov/​tobacco/​ global/​gtss/​index.htm CDC. 2000. Use of FDA-​ approved pharmacologic treatments for tobacco dependence: United States, 1984–​1998. Morb Mortal Wkly Rep, 49, 4.

 207

Tobacco Chang CM, Corey CG, Rostron BL, and Apelberg BJ. 2015. Systematic review of cigar smoking and all cause and smoking related mortality. BMC Public Health, 15, 390. PMCID: PMC4408600. Chaturvedi AK, Anderson WF, Lortet-​Tieulent J, et al. 2013. Worldwide trends in incidence rates for oral cavity and oropharyngeal cancers. J Clin Oncol, 31(36), 4550–​4559. PMCID: PMC3865341. Chaturvedi AK, Engels EA, Anderson WF, and Gillison ML. 2008. Incidence trends for human papillomavirus-​related and -​unrelated oral squamous cell carcinomas in the United States. J Clin Oncol, 26(4), 612–​619. PMID: 18235120. Chen C, Cook LS, Li XY, et al. 1999. CYP2D6 genotype and the incidence of anal and vulvar cancer. Cancer Epidemiol Biomarkers Prev, 8(4 Pt 1), 317–​321. PMID: 10207635. Chen Z, Peto R, Zhou M, et al. 2015. Contrasting male and female trends in tobacco-​attributed mortality in China: evidence from successive nationwide prospective cohort studies. Lancet, 386(10002), 1447–​ 1456. PMCID: PMC4691901. Chia VM, Newcomb PA, Bigler J, et al. 2006. Risk of microsatellite-​unstable colorectal cancer is associated jointly with smoking and nonsteroidal anti-​ inflammatory drug use. Cancer Res, 66(13), 6877–​ 6883. PMID: 16818666. Chuang SC, Lee YC, Hashibe M, et  al. 2010. Interaction between cigarette smoking and hepatitis B and C virus infection on the risk of liver cancer: a meta-​analysis. Cancer Epidemiol Biomarkers Prev, 19(5), 1261–​1268. PMCID: PMC4170071. Collaborative Group on Epidemiological Studies of Ovarian Cancer, Beral V, Gaitskell K, et  al. 2012. Ovarian cancer and smoking:  individual participant meta-​analysis including 28,114 women with ovarian cancer from 51 epidemiological studies. Lancet Oncol, 13(9), 946–​956. PMCID: PMC3431503. Collishaw N. 2016. Blowing smoke:  the history of tobacco-​specific nitrosamines in Canadian tobacco. Tob Control Published Online First: 07 June 2016. doi: 10.1136/​tobaccocontrol-​2015-​PMID: 27272915. Collishaw NE, Boyd NF, Cantor KP, et  al. 2009. Canadian Expert Panel on Tobacco Smoke and Breast Cancer Risk. OTRU Special Report Series, April 2009. Toronto: Ontario Tobacco Research Unit. Connolly GN, and Alpert HR. 2014. Has the tobacco industry evaded the FDA’s ban on “Light” cigarette descriptors? Tob Control, 23(2), 140–​145. PMCID: PMC3932763. Cumberbatch MG, Rota M, Catto JW, La Vecchina C. The Role of Tobacco Smoke in Bladder and Kidney Carcinogenesis:  A  Comparison of Exposures and Meta-​analysis of Incidence and Mortality Risks. Eur Urol, 70(3), 458–​466. Daling JR, Doody DR, Sun X, et al. 2009. Association of marijuana use and the incidence of testicular germ cell tumors. Cancer, 115(6), 1215–​1223. PMCID: PMC2759698. Daling JR, Madeleine MM, Schwartz SM, et  al. 2002. A population-​based study of squamous cell vaginal cancer:  HPV and cofactors. Gynecol Oncol, 84(2), 263–​270. PMID: 11812085. Daling JR, Sherman KJ, Hislop TG, et al. 1992. Cigarette smoking and the risk of anogenital cancer. Am J Epidemiol, 135(2), 180–​189. PMID: 1311142. Department of the Treasury. 2006. Tax Classification of Cigars and Cigarettes (2006R–​ 276P). Retrieved from Tax Classification of Cigars and Cigarettes website:  https://​www.federalregister.gov/​documents/​2006/​10/​ 25/​06-​8835/​tax-​classification-​of-​cigars-​and-​cigarettes-​2006r-​276p DiFranza JR, and Coleman M. 2001. Sources of tobacco for youths in communities with strong enforcement of youth access laws. Tob Control, 10(4), 323–​328. PMCID: PMC1747607. Doll R. 1998. Uncovering the effects of smoking: historical perspective. Stat Methods Med Res, 7(2), 87–​117. PMID: 9654637. Doll R, and Hill AB. 1964. Mortality in relation to smoking:  ten years’ observations of British doctors. Br Med J, 1(5395), 1399–​ 1410. PMCID: PMC1814562. Doll R, Hill AB, and Kreyberg L. 1957. The significance of cell type in relation to the aetiology of lung cancer. Br J Cancer, 11(1), 43–​48. PMCID: PMC2073933. Doll R, and Peto R. 1978. Cigarette smoking and bronchial carcinoma: dose and time relationships among regular smokers and lifelong non-​smokers. J Epidemiol Community Health, 32(4), 303–​313. PMCID: PMC1060963. Doll R, and Peto R. 1981. The causes of cancer: quantitative estimates of avoidable risks of cancer in the United States today. J Natl Cancer Inst, 66(6), 1191–​1308. PMID: 7017215. Dossus L, Boutron-​Ruault MC, Kaaks R, et al. 2014. Active and passive cigarette smoking and breast cancer risk: results from the EPIC cohort. Int J Cancer, 134(8), 1871–​1888. PMID: 24590452. Dutra LM, Grana R, and Glantz SA. 2016. Philip Morris research on precursors to the modern e-​ cigarette since 1990. Tob Control.

207

pii:  tobaccocontrol-​2016-​053406. doi:  10.1136/​tobaccocontrol-​2016-​ 053406. [Epub ahead of print]. PMID: 27852893. Djordjevic MJ, Hoffmann D, Glynn T, Connally GN. US commercial brands of moist snuff, 1994. I assessment of nicotine, moisture, and pH. NCBI - NIH, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1759381/pdf/v004p00062. pdf. Accessed June 2017. Eriksen M, Mackay J, and Ross H. 2012. The Tobacco Atlas, 4th ed. Atlanta, GA; New York: American Cancer Society, World Lung Foundation. Ezzati M, Lopez AD, Rodgers A, et al. 2002. Selected major risk factors and global and regional burden of disease. Lancet, 360(9343), 1347–​1360. PMID: 12423980. Faber MT, Kjaer SK, Dehlendorff C, et al. 2013. Cigarette smoking and risk of ovarian cancer: a pooled analysis of 21 case-​control studies. Cancer Causes Control, 24(5), 989–​1004. PMCID: PMC3818570. Fagerstrom KO. 1978. Measuring degree of physical dependence to tobacco smoking with reference to individualization of treatment. Addict Behav, 3(3–​4), 235–​241. PMID: 735910. Fagerstrom KO, Kunze M, Schoberberger R, et al. 1996. Nicotine dependence versus smoking prevalence: comparisons among countries and categories of smokers. Tob Control, 5(1), 52–​56. PMCID: PMC1759482. Felix AS, Yang HP, Gierach GL, Park Y, and Brinton LA. 2014. Cigarette smoking and endometrial carcinoma risk:  the role of effect modification and tumor heterogeneity. Cancer Causes Control, 25(4), 479–​489. PMCID: PMC4151521. Fircanis S, Merriam P, Khan N, and Castillo JJ. 2014. The relation between cigarette smoking and risk of acute myeloid leukemia: an updated meta-​ analysis of epidemiological studies. Am J Hematol, 89(8), E125–​E132. PMID: 24753145. Fix BV, O’Connor RJ, Vogl L, et al. 2014. Patterns and correlates of polytobacco use in the United States over a decade: NSDUH 2002–​2011. Addict Behav, 39(4), 768–​781. PMCID: PMC3984759. Flanders WD, Lally CA, Zhu BP, Henley SJ, and Thun MJ. 2003. Lung cancer mortality in relation to age, duration of smoking, and daily cigarette consumption: results from Cancer Prevention Study II. Cancer Res, 63(19), 6556–​6562. PMID: 14559851. Food and Drug Administration. (2012). Harmful and Potentially Harmful Constituents in Tobacco Products and Tobacco Smoke: Established List. Washington, DC: Federal Register. Franceschi S, and Serraino D. 1992. Risk factors for adult soft tissue sarcoma in northern Italy. Ann Oncol, 3 Suppl 2, S85–​S88. PMID: 1622876. Freedman DM, Sigurdson A, Doody MM, Rao RS, and Linet MS. 2003. Risk of melanoma in relation to smoking, alcohol intake, and other factors in a large occupational cohort. Cancer Causes Control, 14(9), 847–​857. PMID: 14682442. Freedman ND, Abnet CC, Caporaso NE, et al. 2016. Impact of changing US cigarette smoking patterns on incident cancer:  risks of 20 smoking-​ related cancers among the women and men of the NIH-​AARP cohort. Int J Epidemiol, 45(3), 846–​856. PMCID: PMC5005940. Freedman ND, Silverman DT, Hollenbeck AR, Schatzkin A, and Abnet CC. 2011. Association between smoking and risk of bladder cancer among men and women. JAMA, 306(7), 737–​745. PMCID: PMC3441175. Gallagher RP, Huchcroft S, Phillips N, et  al. 1995. Physical activity, medical history, and risk of testicular cancer (Alberta and British Columbia, Canada). Cancer Causes Control, 6(5), 398–​406. PMID: 8547537. Gandini S, Botteri E, Iodice S, et  al. 2008. Tobacco smoking and cancer:  a meta-​analysis. Int J Cancer, 122(1), 155–​164. PMID: 17893872. Garcia-​Closas M, Rothman N, Figueroa JD, et al. 2013. Common genetic polymorphisms modify the effect of smoking on absolute risk of bladder cancer. Cancer Res, 73(7), 2211–​2220. PMCID: PMC3688270. Gaudet MM, Gapstur SM, Sun J, et al. 2013. Active smoking and breast cancer risk: original cohort data and meta-​analysis. J Natl Cancer Inst, 105(8), 515–​525. PMID: 23449445. Giovannucci E, Colditz GA, Stampfer MJ, et al. 1994a. A prospective study of cigarette smoking and risk of colorectal adenoma and colorectal cancer in U.S. women. J Natl Cancer Inst, 86(3), 192–​199. PMID: 8283491. Giovannucci E, Rimm EB, Stampfer MJ et al. 1994b. A prospective study of cigarette smoking and risk of colorectal adenoma and colorectal cancer in U.S. men. J Natl Cancer Inst, 86(3), 183–​191. PMID: 8283490. Giovino GA, Mirza SA, Samet JM, et  al. 2012. Tobacco use in 3 billion individuals from 16 countries:  an analysis of nationally representative cross-​ sectional household surveys. Lancet, 380(9842), 668–​ 679. PMID: 22901888. Goniewicz ML, Eisner MD, Lazcano-​Ponce E, et  al. 2011. Comparison of urine cotinine and the tobacco-​specific nitrosamine metabolite 4-​(methylnitrosamino)-​1-​(3-​pyridyl)-​1-​butanol (NNAL) and their ratio to discriminate active from passive smoking. Nicotine Tob Res, 13(3), 202–​208. PMCID: 3045466.

208

208

PART III:  THE CAUSES OF CANCER

Goniewicz ML, Knysak J, Gawron M, et al. 2014. Levels of selected carcinogens and toxicants in vapour from electronic cigarettes. Tob Control, 23(2), 133–​139. PMCID: PMC4154473. Gram IT, Little MA, Lund E, and Braaten T. 2016. The fraction of breast cancer attributable to smoking: The Norwegian Women and Cancer Study 1991–​ 2012. Br J Cancer, 115(5), 616–​623. PMCID: PMC4997535. Gram IT, Lukanova A, Brill I, et al. 2012. Cigarette smoking and risk of histological subtypes of epithelial ovarian cancer in the EPIC cohort study. Int J Cancer, 130(9), 2204–​2210. PMID: 21678398. Gram IT, Park SY, Kolonel LN, et al. 2015. Smoking and risk of breast cancer in a racially/​ethnically diverse population of mainly women who do not drink alcohol: The MEC Study. Am J Epidemiol, 182(11), 917–​925. PMCID: PMC4836396. Grana R, Benowitz N, and Glantz SA. 2014. E-​cigarettes: a scientific review. Circulation, 129(19), 1972–​1986. PMCID: PMC4018182. Gravely S, Fong GT, Cummings KM, et al. 2014. Awareness, trial, and current use of electronic cigarettes in 10 countries: findings from the ITC project. Int J Environ Res Public Health, 11(11), 11691–​11704. PMCID: PMC4245638. Gu F, Wacholder S, Kovalchik S, et al. 2014. Time to smoke first morning cigarette and lung cancer in a case-​control study. J Natl Cancer Inst, 106(6), dju118. PMCID: PMC4072901. Haenszel W, and Shimkin MB. 1956. Smoking patterns and epidemiology of lung cancer in the United States: are they compatible? J Natl Cancer Inst, 16(6), 1417–​1441. PMID: 13320121. Hammond E. 1966. Smoking in relation to the death rates of one million men and women. In:  Haenszel W (Ed.), Epidemiologic Approaches to the Study of Cancer and Other Chronic Diseases (National Cancer Institute Monograph 19, pp. 127–​204). Rockville, MD: US Department of Health, Education, and Welfare. Hanaoka T, Yamamoto S, Sobue T, et al. 2005. Active and passive smoking and breast cancer risk in middle-​aged Japanese women. Int J Cancer, 114(2), 317–​322. PMID: 15540214. Harris JE. 1983. Cigarette smoking among successive birth cohorts of men and women in the United States during 1900–​80. J Natl Cancer Inst, 71(3), 473–​479. PMID: 6577223. Harris JE, Thun MJ, Mondul AM, and Calle EE. 2004. Cigarette tar yields in relation to mortality from lung cancer in the Cancer Prevention Study II prospective cohort, 1982–​8. BMJ, 328(7431), 72. PMCID: PMC314045. Hashibe M. 2010. Risk factors:  tobacco and alcohol. In:  Olshan AF (Ed.), Epidemiology, Pathogenesis, and Prevention of Head and Neck Cancer (pp. 65–​86). New York: Springer. Hashibe M, Brennan P, Chuang SC, et al. 2009. Interaction between tobacco and alcohol use and the risk of head and neck cancer: pooled analysis in the International Head and Neck Cancer Epidemiology Consortium. Cancer Epidemiol Biomarkers Prev, 18(2), 541–​550. PMCID: PMC3051410. Hassan MM, Spitz MR, Thomas MB, et  al. 2008. Effect of different types of smoking and synergism with hepatitis C virus on risk of hepatocellular carcinoma in American men and women: case-​control study. Int J Cancer, 123(8), 1883–​1891. PMCID: PMC2673571. Hecht SS, Carmella SG, Edmonds A, et al. 2008. Exposure to nicotine and a tobacco-​specific carcinogen increase with duration of use of smokeless tobacco. Tob Control, 17(2), 128–​131. PMCID: PMC3889131. Hecht SS, Stepanov I, and Carmella SG. 2016. Exposure and metabolic activation biomarkers of carcinogenic tobacco-​specific nitrosamines. Acc Chem Res, 49(1), 106–​114. PMID: 26678241. Henderson BE, Benton B, Jing J, Yu MC, and Pike MC. 1979. Risk factors for cancer of the testis in young men. Int J Cancer, 23(5), 598–​602. PMID: 37169. Henley SJ, Thun MJ, Chao A, and Calle EE. 2004. Association between exclusive pipe smoking and mortality from cancer and other diseases. J Natl Cancer Inst, 96(11), 853–​861. PMID: 15173269. Hirayama T. 1981. Non-​smoking wives of heavy smokers have a higher risk of lung cancer: a study from Japan. Br Med J (Clin Res Ed), 282(6259), 183–​185. PMCID: PMC1503989. Hoffmann D, and Hoffmann I. 1997. The changing cigarette, 1950–​1995. J Toxicol Environ Health, 50(4), 307–​364. PMID: 9120872. Holford TR, Meza R, Warner KE, et al. 2014. Tobacco control and the reduction in smoking-​related premature deaths in the United States, 1964–​ 2012. JAMA, 311(2), 164–​171. PMCID: PMC4056770. Holford TR, Levy DT, McKay LA, et  al. 2014. Patterns of birth cohort-​ specific smoking histories, 1965–​2009. Am J Prev Med, 46(2), e31–​e37. PMCID: 3951759. Husgafvel-​ Pursiainen K. 2004. Genotoxicity of environmental tobacco smoke: a review. Mutat Res, 567(2–​3), 427–​445. PMID: 15572289. Hussain SK, Madeleine MM, Johnson LG, et  al. 2008. Cervical and vulvar cancer risk in relation to the joint effects of cigarette smoking and genetic

variation in interleukin 2. Cancer Epidemiol Biomarkers Prev, 17(7), 1790–​1799. PMCID: PMC2497438. Huxley RR, Ansary-​Moghaddam A, Clifton P, et al. 2009. The impact of dietary and lifestyle risk factors on risk of colorectal cancer: a quantitative overview of the epidemiological evidence. Int J Cancer, 125(1), 171–​180. PMID: 19350627. International Agency for Research on Cancer (IARC). 1986. Tobacco smoke. IARC Monogr Eval Carcinogenic Risks Hum 83. PMID: 6642935. IARC. 1995. Human papilloma viruses. IARC Monogr Eval Carcinogenic Risks Hum 64. PMID: 6642935. IARC. 2004. Tobacco smoke and involuntary smoking. IARC Monogr Eval Carcinogenic Risks Hum 83. PMID: 6642935. IARC. 2007. Smokeless tobacco and some tobacco-​specific N-​nitrosamines. IARC Monogr Eval Carcinogenic Risks Hum 89. Retrieved from http://​ site.ebrary.com/​lib/​yale/​Doc?id=10252497 IARC. 2012a. A Review of human carcinogens. E. Personal habits and indoor combustions. IARC Monogr Eval Carcinogenic Risks Hum 100E. IARC. 2012b. A review of human carcinogens. F. Chemical agents and related occupations. IARC Monogr Eval Carcinogenic Risks Hum 100E. Inoue-​Choi M, Liao LM, Reyes-​Guzman C, et al. 2017. Association of long-​ term, low-​intensity smoking with all-​cause and cause-​specific mortality in the National Institutes of Health-​AARP Diet and Health Study. JAMA Intern Med. 177(1), 87–​ 95. doi:  10.1001/​ jamainternmed.2016.7511. PMID: 27918784. Institute of Medicine. 2001. Clearing the Smoke: Assessing the Science Base for Tobacco Harm Reduction. Washington, DC: National Academy Press. Institute of Medicine. 2015. Public Health Implications of Raising the Minimum Age of Legal Access to Tobacco Products. Washington, DC:  National Academies Press. International Collaboration of Epidemiological Studies of Cervical Cancer, Appleby P, Beral V, et  al. 2006. Carcinoma of the cervix and tobacco smoking:  collaborative reanalysis of individual data on 13,541 women with carcinoma of the cervix and 23,017 women without carcinoma of the cervix from 23 epidemiological studies. Int J Cancer, 118(6), 1481–​1495. PMID: 16206285. Iodice S, Gandini S, Maisonneuve P, and Lowenfels AB. 2008. Tobacco and the risk of pancreatic cancer: a review and meta-​analysis. Langenbecks Arch Surg, 393(4), 535–​545. PMID: 18193270. Islami F, Moreira DM, Boffetta P, and Freedland SJ. 2014. A systematic review and meta-​analysis of tobacco use and prostate cancer mortality and incidence in prospective cohort studies. Eur Urol, 66(6), 1054–​1064. PMCID: PMC4566150. Jacobs EJ, Newton CC, Carter BD, et  al. 2015. What proportion of cancer deaths in the contemporary United States is attributable to cigarette smoking? Ann Epidemiol, 25(3), 179–​182 e171. PMID: 25487970. Jamal A, King B, Neff L, et  al. 2016. Current cigarette smoking among adults:  United States, 2005–​2015. MMWR Morb Mortal Wkly 65(44), 1205–​1211. Jarvis MJ, Boreham R, Primatesta P, Feyerabend C, and Bryant A. 2001. Nicotine yield from machine-​smoked cigarettes and nicotine intakes in smokers: evidence from a representative population survey. J Natl Cancer Inst, 93(2), 134–​138. PMID: 11208883. Jarvis MJ, Tunstall-​Pedoe H, Feyerabend C, Vesey C, and Saloojee Y. 1987. Comparison of tests used to distinguish smokers from nonsmokers. Am J Public Health, 77(11), 1435–​1438. PMCID: PMC1647100. Jayalekshmi PA, Gangadharan P, Akiba S, Koriyama C, and Nair RR. 2011. Oral cavity cancer risk in relation to tobacco chewing and bidi smoking among men in Karunagappally, Kerala, India: Karunagappally cohort study. Cancer Sci, 102(2), 460–​467. PMID: 21129124. Jayalekshmi PA, Hassani S, Nandakumar A, et  al. 2015. Gastric cancer risk in relation to tobacco use and alcohol drinking in Kerala, India:  Karunagappally cohort study. World J Gastroenterol, 21(44), 12676–​12685. PMCID: PMC4658623. Jayalekshmi PA, Nandakumar A, Akiba S, Gangadharan P, and Koriyama C. 2013. Associations of tobacco use and alcohol drinking with laryngeal and hypopharyngeal cancer risks among men in Karunagappally, Kerala, India:Karunagappally cohort study. PLoS One, 8(8), e73716. PMCID: PMC3756048. Jedrychowski W, Boeing H, Wahrendorf J, et  al. 1993. Vodka consumption, tobacco smoking and risk of gastric cancer in Poland. Int J Epidemiol, 22(4), 606–​613. PMID: 8225732. Jedrychowski W, Popiela T, Drews M, et al. 1999. Effect of Helicobacter pylori infection, smoking and dietary habits on the occurrence of antrum intestinal metaplasia: clinico-​epidemiological study in Poland. Pol J Pathol, 50(4), 289–​295. PMID: 10721269. Jha P. 2009. Avoidable global cancer deaths and total deaths from smoking. Nat Rev Cancer, 9(9), 655–​664. PMID: 19693096.

 209

Tobacco Jha P, Ramasundarahettige C, Landsman V, et al. 2013. 21st-​century hazards of smoking and benefits of cessation in the United States. N Engl J Med, 368(4), 341–​350. PMID: 23343063. Joehanes R, Just AC, Marioni RE, et al. 2016. Epigenetic signatures of cigarette smoking. Circ Cardiovasc Genet, 9(5), 436–​447. PMID: 27651444. Johnson KC. 2005. Accumulating evidence on passive and active smoking and breast cancer risk. Int J Cancer, 117(4), 619–​628. PMID: 15929073. Kahn HA. 1966. The Dorn study of smoking and mortality among U.S. veterans: report on eight and one-​half years of observation. Natl Cancer Inst Monogr, 19, 1–​125. PMID: 5905668. Kamper-​Jorgensen M, Rostgaard K, Glaser SL, et al. 2013. Cigarette smoking and risk of Hodgkin lymphoma and its subtypes: a pooled analysis from the International Lymphoma Epidemiology Consortium (InterLymph). Ann Oncol, 24(9), 2245–​2255. PMCID: PMC3755332. Kessel Schneider S, Buka SL, Dash K, Winickoff JP, and O’Donnell L. 2016. Community reductions in youth smoking after raising the minimum tobacco sales age to 21. Tob Control, 25(3), 355–​359. PMID: 26071428. Kilfoy BA, Zheng T, Lan Q, et al. 2010. Genetic variation in N-​acetyltransferases 1 and 2, cigarette smoking, and risk of non-​Hodgkin lymphoma. Cancer Causes Control, 21(1), 127–​133. PMCID: PMC2972187. Kozlowski L, O’Connor RJ, Sweeney CT. 2001. Cigarette design. In: Shopland D, Burns D, Benowitz N, Amacher R (Eds.), Risks Associated With Smoking Cigarettes With Low Machine-​ Measured Yields of Tar and Nicotine, Smoking and Tobacco Control Monograph 13 (Vol. NIH Publ. No. 02-​5074, pp. 13–​38). Bethesda, MD: U.S. Department of Health and Human Services, National Institutes of Health, National Cancer Institute, pp. 13–​38. Kreyberg L. 1962. Histological lung cancer types:  a morphological and biological correlation. Acta Pathol Microbiol Scand Suppl, Suppl 157, 1–​92. PMID: 14035775. Kuper H, Tzonou A, Kaklamani E, et al. 2000. Tobacco smoking, alcohol consumption and their interaction in the causation of hepatocellular carcinoma. Int J Cancer, 85(4), 498–​502. PMID: 10699921. Kuper HE, Tzonou A, Kaklamani E, et  al. 2000. Hepatitis B and C viruses in the etiology of hepatocellular carcinoma; a study in Greece using third-​generation assays. Cancer Causes Control, 11(2), 171–​ 175. PMID: 10710202. Lacson JC, Carroll JD, Tuazon E, et al. 2012. Population-​based case-​control study of recreational drug use and testis cancer risk confirms an association between marijuana use and nonseminoma risk. Cancer, 118(21), 5374–​5383. PMCID: PMC3775603. Ladeiras-​Lopes R, Pereira AK, Nogueira A, et al. 2008. Smoking and gastric cancer:  systematic review and meta-​analysis of cohort studies. Cancer Causes Control, 19(7), 689–​701. PMID: 18293090. Lee CH, Lee KW, Fang FM, et al. 2011. The use of tobacco-​free betel-​quid in conjunction with alcohol/​tobacco impacts early-​onset age and carcinoma distribution for upper aerodigestive tract cancer. J Oral Pathol Med, 40, 684–​692. PMID: 21385213. Lee YO, Hebert CJ, Nonnemaker JM, and Kim AE. 2014. Multiple tobacco product use among adults in the United States:  cigarettes, cigars, electronic cigarettes, hookah, smokeless tobacco, and snus. Prev Med, 62, 14–​19. PMID: 24440684. Leffondre K, Abrahamowicz M, Siemiatycki J, and Rachet B. 2002. Modeling smoking history: a comparison of different approaches. Am J Epidemiol, 156(9), 813–​823. PMID: 12396999. Leonardi-​Bee J, Ellison T, and Bath-​Hextall F. 2012. Smoking and the risk of nonmelanoma skin cancer: systematic review and meta-​analysis. Arch Dermatol, 148(8), 939–​946. PMID: 22711192. Li Z, Wang Z, Yu Y, Zhang H, and Chen L. 2015. Smoking is inversely related to cutaneous malignant melanoma:  results of a meta-​ analysis. Br J Dermatol, 173(6), 1540–​1543. PMID: 26134854. Liang PS, Chen TY, and Giovannucci E. 2009. Cigarette smoking and colorectal cancer incidence and mortality: systematic review and meta-​analysis. Int J Cancer, 124(10), 2406–​2415. PMID: 19142968. Limsui D, Vierkant RA, Tillmans LS, et  al. 2010. Cigarette smoking and colorectal cancer risk by molecularly defined subtypes. J Natl Cancer Inst, 102(14), 1012–​1022. PMCID: PMC2915616. Lopez AD, Collishaw N, and Piha T. 1994. A descriptive model og the cigarette epidemic in developed countries. Tob Control, 3, 242–​247. Lortet-​Tieulent J, Goding Sauer A, Siegel RL, et al. 2016. State-​level cancer mortality attributable to cigarette smoking in the United States. JAMA Intern Med, 176, 1792–​1798. PMID: 27775761. Lubin JH, Cook MB, Pandeya N, et al. 2012. The importance of exposure rate on odds ratios by cigarette smoking and alcohol consumption for esophageal adenocarcinoma and squamous cell carcinoma in the Barrett’s Esophagus and Esophageal Adenocarcinoma Consortium. Cancer Epidemiol, 36(3), 306–​316. PMCID: PMC3489030.

209

Lynch SM, Vrieling A, Lubin JH, et al. 2009. Cigarette smoking and pancreatic cancer: a pooled analysis from the pancreatic cancer cohort consortium. Am J Epidemiol, 170(4), 403–​413. PMCID: PMC2733861. Madeleine MM, Daling JR, Carter JJ, et al. 1997. Cofactors with human papillomavirus in a population-​based study of vulvar cancer. J Natl Cancer Inst, 89(20), 1516–​1523. PMID: 9337348. Marrero JA, Fontana RJ, Fu S, et al. 2005. Alcohol, tobacco and obesity are synergistic risk factors for hepatocellular carcinoma. J Hepatol, 42(2), 218–​224. PMID: 15664247. Mazurek JM, Syamlal G, King BA, et al. 2014. Smokeless tobacco use among working adults—​United States, 2005 and 2010. MMWR Morb Mortal Wkly Rep, 63(22), 477–​482. PMID: 24898164. McLaughlin JK, Silverman DT, Hsing AW, et  al. 1992. Cigarette smoking and cancers of the renal pelvis and ureter. Cancer Res, 52(2), 254–​257. PMID: 1728398. Miller MD, Marty MA, Broadwin R, et  al. 2007. The association between exposure to environmental tobacco smoke and breast cancer: a review by the California Environmental Protection Agency. Prev Med, 44(2), 93–​ 106. PMID: 17027075. Montazeri Z, Nyiraneza C, El-​Katerji H, and Little J. 2017. Waterpipe smoking and cancer: systematic review and meta-​analysis. Tob Control, 26(1), 92–​97. PMID: 27165994. Morton LM, Slager SL, Cerhan JR, et  al. 2014. Etiologic heterogeneity among non-​Hodgkin lymphoma subtypes: the InterLymph Non-​Hodgkin Lymphoma Subtypes Project. J Natl Cancer Inst Monogr, 2014(48), 130–​ 144. PMCID: PMC4155467. National Cancer Institute. 1998. Cigar Smoking and Cancer. Tobacco Control Monograph 9.  Bethesda, MD:  US Department of Health and Human Services, National Cancer Insitute, National Insitutes of Health. National Cancer Institute and Centers for Disease Control and Prevention. 2014. Smokeless Tobacco and Public Health:  A  Global Perspective. Bethesda, MD: US Department of Health and Human Services, Centers for Disease Control and Prevention, and National Institutes of Health, National Cancer Institute. National Research Council. 1986. Environmental Tobacco Smoke: Measuring Exposures and Assessing Health Effects. Washington, DC:  National Academy Press. National Research Council Committee on Health Risks of Exposure to Radon (BIER VI). 1998. Health Effects of Exposure to Radon. Biological Effects of Ionizing Radiation. Washington, DC: National Academy Press. Ng M, Freeman MK, Fleming TD, et al. 2014. Smoking prevalence and cigarette consumption in 187 countries, 1980–​2012. JAMA, 311(2), 183–​192. PMID: 24399557. Nilsson R. 1998. A qualitative and quantitative risk assessment of snuff dipping. Regul Toxicol Pharmacol, 28(1), 1–​16. PMID: 9784428. Nyante SJ, Gierach GL, Dallal CM, et al. 2014. Cigarette smoking and postmenopausal breast cancer risk in a prospective cohort. Br J Cancer, 110(9), 2339–​2347. PMCID: PMC4007228. O’Keefe AM, and Pollay RW. 1996. Deadly targeting of women in promoting cigarettes. J Am Med Womens Assoc, 51(1–​2), 67–​69. PMID: 8868553. Orellana-​Barrios MA, Payne D, Medrano-​Juarez RM, Yang S, and Nugent K. 2016. Electronic cigarettes for smoking cessation. Am J Med Sci, 352(4), 420–​426. PMID: 27776725. Palipudi K, Mbulo L, Kosen S, et al. 2015. A cross sectional study of kretek smoking in Indonesia as a major risk to public health. Asian Pac J Cancer Prev, 16(16), 6883–​6888. PMID: 26514461. Palipudi KM, Mbulo L, Morton J, et al. 2015. Awareness and current use of electronic cigarettes in Indonesia, Malaysia, Qatar, and Greece: findings from 2011–​2013 Global Adult Tobacco Surveys. Nicotine Tob Res, 18, 501–​507. PMID: 25895951. Parkin DM. 2011. The fraction of cancer attributable to lifestyle and environmental factors in the UK in 2010. Br J Cancer, 105 Suppl 2, S2–​S5. PMCID: PMC3252063. Patel NH, Attwood KM, Hanzly M, et al. 2015. Comparative analysis of smoking as a risk factor among renal cell carcinoma histological subtypes. J Urol, 194, 640–​646. PMID: 25896558. Patrick DL, Cheadle A, Thompson DC, et al. 1994. The validity of self-​reported smoking: a review and meta-​analysis. Am J Public Health, 84(7), 1086–​ 1093. PMCID: PMC1614767. Pednekar MS, Gupta PC, Yeole BB, and Hebert JR. 2011. Association of tobacco habits, including bidi smoking, with overall and site-​specific cancer incidence: results from the Mumbai cohort study. Cancer Causes Control, 22(6), 859–​868. PMCID: PMC3756904. Peto R, Darby S, Deo H, et al. 2000. Smoking, smoking cessation, and lung cancer in the UK since 1950: combination of national statistics with two case-​control studies. BMJ, 321(7257), 323–​329. PMCID: PMC27446.

210

210

PART III:  THE CAUSES OF CANCER

Peto R, Lopez AD, Boreham J, Thun M, and Heath C, Jr. 1992. Mortality from tobacco in developed countries:  indirect estimation from national vital statistics. Lancet, 339(8804), 1268–​1278. PMID: 1349675. Peto R, Lopez AD, Pan H, Boreham J, Thun M. 2017. Mortality from smoking in developed countries, 1950–​2020. Available from: https://​www.ctsu. ox.ac.uk/​.../​mortality-​from-​smoking. Accessed April 24, 2015. Pirie K, Peto R, Reeves GK, et al. 2013. The 21st century hazards of smoking and benefits of stopping: a prospective study of one million women in the UK. Lancet, 381(9861), 133–​141. PMCID: PMC3547248. Poynter JN, Haile RW, Siegmund KD, et al. 2009. Associations between smoking, alcohol consumption, and colorectal cancer, overall and by tumor microsatellite instability status. Cancer Epidemiol Biomarkers Prev, 18(10), 2745–​2750. PMCID: PMC2759847. Purdue MP, Moore LE, Merino MJ, et al. 2013. An investigation of risk factors for renal cell carcinoma by histologic subtype in two case-​control studies. Int J Cancer, 132, 2640–​2647. PMCID: PMC3717609. Purdue MP, and Silverman DT. 2016. Clearing the Air:  Summarizing the smoking-​related relative risks of bladder and kidney cancer. Eur Urol, 70(3), 467–​468. PMID: 27130147. QuickStats. 2016. Cigarette smoking status among current adult e-​cigarette users, by age group:  National Health Interview Survey, United States, 2015. MMWR Morb Mortal Wkly Rep, 65, 1177. Reddy CR. 1974. Carcinoma of hard palate in India in relation to reverse smoking of chuttas. J Natl Cancer Inst, 53(3), 615–​619. PMID: 4412322. Reynolds P, Hurley S, Goldberg D, and California Teachers Study Steering Committee. 2006. Accumulating evidence on passive and active smoking and breast cancer risk. Int J Cancer, 119(1), 239; author reply 240–​231. PMID: 16432834. Rosenberg L, Boggs DA, Bethea TN, et al. 2013. A prospective study of smoking and breast cancer risk among African-​American women. Cancer Causes Control, 24(12), 2207–​2215. PMCID: PMC3855285. Rostron B. 2012. Lung cancer mortality risk for U.S. menthol cigarette smokers. Nicotine Tob Res, 14(10), 1140–​1144. PMID: 22387991. Sergentanis TN, Kanavidis P, Michelakos T, and Petridou ET. 2013. Cigarette smoking and risk of lymphoma in adults: a comprehensive meta-​analysis on Hodgkin and non-​Hodgkin disease. Eur J Cancer Prev, 22(2), 131–​ 150. PMID: 22759975. Serraino D, Franceschi S, Talamini R, Frustaci S, and La Vecchia C. 1991. Non-​ occupational risk factors for adult soft-​tissue sarcoma in northern Italy. Cancer Causes Control, 2(3), 157–​164. PMID: 1873445. Shafey O, Eriksen M, and Ross H. 2009. The Tobacco Atlas, 3rd edition. Atlanta, GA: American Cancer Society, World Lung Foundation. Shapiro JA, Jacobs EJ, and Thun MJ. 2000. Cigar smoking in men and risk of death from tobacco-​related cancers. J Natl Cancer Inst, 92(4), 333–​337. PMID: 10675383. Shields PG. 2002. Tobacco smoking, harm reduction, and biomarkers. J Natl Cancer Inst, 94(19), 1435–​1444. PMID: 12359853. Shimazu T, Asada K, Charvat H, et  al. 2015. Association of gastric cancer risk factors with DNA methylation levels in gastric mucosa of healthy Japanese:  a cross-​sectional study. Carcinogenesis, 36(11), 1291–​1298. PMID: 26354778. Siegel RL, Jacobs EJ, Newton CC, et al. 2015. Deaths due to cigarette smoking for 12 smoking-​related cancers in the United States. JAMA Intern Med, 175(9), 1574–​1576. PMID: 26076120. Siman JH, Forsgren A, Berglund G, and Floren CH. 2001. Tobacco smoking increases the risk for gastric adenocarcinoma among Helicobacter pylori-​infected individuals. Scand J Gastroenterol, 36(2), 208–​ 213. PMID: 11252415. Singh T, Arrazola RA, Corey CG, and et al. 2016. Tobacco use among middle and high school students: United States, 2011–​2015. MMWR Morb Mortal Wkly Rep, 65, 361–​367. Sinha DN, Abdulkader RS, and Gupta PC. 2016. Smokeless tobacco-​associated cancers:  a systematic review and meta-​analysis of Indian studies. Int J Cancer, 138(6), 1368–​1379. PMID: 26443187. Slade J. 1993. Nicotine delivery devices. In:  Orleans C, and Slade J (Eds.), Nicotine Addiction:  Principles and Management (pp. 3–​ 23). New York: Oxford University Press. Srigley JR, Delahunt B, Eble JN, et  al. 2013. The International Society of Urological Pathology (ISUP) Vancouver classification of renal neoplasia. Am J Surg Pathol, 37, 1469–​1489. PMID: 24025519. Steffen MW, Murad MH, Hays JT, et al. 2014. Self-​report of tobacco use status: comparison of paper-​based questionnaire, online questionnaire, and direct face-​ to-​ face interview:  implications for meaningful use. Popul Health Manag, 17(3), 185–​189. PMCID: PMC4442565. Thompson CA, Zhang ZF, and Arah OA. 2013. Competing risk bias to explain the inverse relationship between smoking and malignant melanoma. Eur J Epidemiol, 28(7), 557–​567. PMCID: PMC3864891.

Thun MJ, Carter BD, Feskanich D, et  al. 2013. 50-​year trends in smoking-​ related mortality in the United States. N Engl J Med, 368(4), 351–​364. PMCID: PMC3632080. Thun M, Day-​Lally C, Myers D, et  al. 1997. Trends in tobacco smoking and mortality from cigarette use in Cancer Prevention Studies I (1959–​ 1965) and II (1982–​1988). In:  Shopland D, Burns D, Garfinkel L, and Samet J (Eds.), Changes in Cigarette-​Related Disease Risks and Their Implications for Prevention and Control. National Cancer Institute, Smoking and Tobacco Control, Monograph 8 (Vol. NIH Publication No. 97-​4213, pp. 305–​382). Washington, DC: US Department of Health and Human Services, National Institutes of Health, National Cancer Institute. Thun MJ, and Henley SJ. 2006. Tobacco. In:  Schottenfeld D, and Fraumeni JF (Eds.), Cancer Epidemiology and Prevention (3rd ed., pp. 217–​242). Oxford; New York: Oxford University Press. Tobacco Atlas. 2015. The Tobacco Atlas, 5th edition. Eriksen M, Mackay J, Schluger N, Gomeshtapeh F I, and Drapoe J (Eds.). Atlanta, GA: American Cancer Society, American Lung Association. Tobacco Atlas. 2016. Water pipes. www.tobaccoatlas.org/​topic/​water-​pipes Todd GF. 1978. Cigarette consumption per adult of each sex in various countries. J Epidemiol Community Health, 32(4), 289–​293. PMCID: PMC1060961. Toll BA, and Ling PM. 2005. The Virginia Slims identity crisis: an inside look at tobacco industry marketing to women. Tob Control, 14(3), 172–​180. PMCID: PMC1748044. Tong H, Hu C, Yin X et al. 2013. A Meta-​analysis of the relationship between cigarette smoking and incidence of myelodysplastic syndromes. PLoS One, 8(6), e67537. PMCID: PMC3689714. Torre LA, Siegel RL, and Jemal A. 2016. Lung cancer statistics. Adv Exp Med Biol, 893, 1–​19. PMID: 26667336. Trabert B, Sigurdson AJ, Sweeney AM, Strom SS, and McGlynn KA. 2011. Marijuana use and testicular germ cell tumors. Cancer, 117(4), 848–​853. PMCID: PMC3017734. Trichopoulos D, Kalandidi A, Sparros L, and MacMahon B. 1981. Lung cancer and passive smoking. Int J Cancer, 27(1), 1–​4. PMID: 7251227. Tsoi KK, Pau CY, Wu WK, et  al. 2009. Cigarette smoking and the risk of colorectal cancer:  a meta-​analysis of prospective cohort studies. Clin Gastroenterol Hepatol, 7(6), 682–​688 e681–​685. PMID: 19245853. US Department of Health and Human Services. 1979. Smoking and Health: A Report of the Surgeon General. Vol. DHEW Publ No (PHS) 79-​ 50066. Washington, DC: US Department of Health and Human Services, Public Health Service, Office of the Assistant Secretary for Health, Office on Smoking and Health. US Department of Health and Human Services. 1980. The Health Consequences of Smoking for Women: A Report of the Surgeon General. Washington, DC:  US Department of Health and Human Services, Public Health Service, Office of the Assistant Secretary for Health, Office on Smoking and Health. U.S. Department of Health and Human Services. 1981. The Health Consequences of Smoking:  The Changing Cigarette. A  Report of the Surgeon General. Rockville, MD: US Department of Health and Human Services, Public Health Service, Office on Smoking and Health. US Department of Health and Human Services. 1982. The Health Consequences of Smoking:  Cancer:  A  Report of the Surgeon General. Washington, DC:  US Department of Health and Human Services, Public Health Service, Office on Smoking and Health. US Department of Health and Human Services. 1986. The Health Consequences of Involuntary Smoking: A Report of the Surgeon General. Washington, DC:  US Department of Health and Human Services, Public Health Service, Office on Smoking and Health. US Department of Health and Human Services. 1988. The Health Consequences of Smoking:  Nicotine Addiction. Rockville, MD:  US Department of Health and Human Services, Public Health Service, Centers for Disease Control, Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. US Department of Health and Human Services. 1989. Reducing the Health Consequences of Smoking:  25 Years of Progress. Rockville, MD:  US Department of Health and Human Services, Public Health Service, Centers for Disease Control, Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. US Department of Health and Human Services. 2001. Women and Smoking:  A  Report of the Surgeon General. Rockville, MD:  US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. US Department of Health and Human Services. 2004. The Health Consequences of Smoking:  A  Report of the Surgeon General. Rockville, MD:  US Department of Health and Human Services, Centers for Disease Control

 21

Tobacco and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. US Department of Health and Human Services. 2006. The Health Consequences of Involuntary Exposure to Tobacco Smoke:  A  Report of the Surgeon General. Rockville, MD: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. US Department of Health and Human Services. 2010. How Tobacco Smoke Causes Disease:  The Biology and Behavioral Basis for Smoking-​ Attributable Disease. A  Report of the Surgeon General. Rockville, MD: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. US Department of Health and Human Services. 2012. Preventing Tobacco Use among Youth and Young Adults. A  Report of the Surgeon General. Rockville, MD: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. US Department of Health and Human Services. 2014. The Health Consequences of Smoking:  50 Years of Progress. A  Report of the Surgeon General. Rockville, MD: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. US Department of Health and Human Services. 2016. E-​Cigarette Use Among Youth and Young Adults. Rockville, MD: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. US Department of Health Education and Welfare. 1964. Smoking and Health:  Report of the Advisory Committee to the Surgeon General of the Public Health Service. Washington, DC: US Department of Health, Education and Welfare, Public Health Service, Centers for Disease Control. US Environmental Protection Agency. (1992). Respiratory Health Effects of Passive Smoking:  Lung Cancer and Other Disorders (EPA/​600/​6-​90/​ 006F). Washington, DC: Office of Health and Environmental Assessment, Office of Research and Development, US Environmental Protection Agency. Vardavas CI, Fthenou E, Patelarou E, et al. 2013. Exposure to different sources of second-​hand smoke during pregnancy and its effect on urinary cotinine and tobacco-​specific nitrosamine (NNAL) concentrations. Tob Control, 22(3), 194–​200. PMID: 22253001. Vineis P, Alavanja M, Buffler P, et  al. 2004. Tobacco and cancer:  recent epidemiological evidence. J Natl Cancer Inst, 96(2), 99–​106. PMID: 14734699. Vogel RI, Carmella SG, Stepanov I, Hatsukami DK, and Hecht SS. 2011. The ratio of a urinary tobacco-​specific lung carcinogen metabolite to cotinine is significantly higher in passive than in active smokers. Biomarkers, 16(6), 491–​497. PMCID: PMC3159775. Warren CW, Jones NR, Peruga A, et  al. 2008. Global youth tobacco surveillance, 2000–​2007. MMWR Surveill Summ, 57(1), 1–​ 28. PMID: 18219269.

211

Warren CW, Asma S, Lee J, Lea V, Mackay J. 2009. Global Tobacco Surveillance System: The GTSS Atlas. Brighton, UK: CDC, CDC Foundation, World Lung Foundation. Weir JM, and Dunn JE, Jr. 1970. Smoking and mortality: a prospective study. Cancer, 25(1), 105–​112. PMID: 5410301. Wenbin D, Zhuo C, Zhibing M, et al. 2013. The effect of smoking on the risk of gallbladder cancer:  a meta-​analysis of observational studies. Eur J Gastroenterol Hepatol, 25(3), 373–​379. PMID: 23085578. Wentzensen N, Poole EM, Trabert B, et  al. 2016. Ovarian cancer risk factors by histologic subtype: an analysis from the Ovarian Cancer Cohort Consortium. J Clin Oncol. 34, 2888–​2898. PMID: 27325851. Wiersinga WM. 2013. Smoking and thyroid. Clin Endocrinol (Oxf), 79(2), 145–​151. PMID: 23581474. Winickoff JP, Gottlieb M, and Mello MM. 2014. Tobacco 21: an idea whose time has come. N Engl J Med, 370, 295–​297. World Health Organization (WHO). (2003). WHO Framework Convention on Tobacco Control. Geneva:  World Health Organization. www.who. int/​tobacco/​framework/​WHO_​FCTC_​english.pdf. Accessed April 24, 2017. Wu AH, Crabtree JE, Bernstein L, et  al. 2003. Role of Helicobacter pylori CagA + strains and risk of adenocarcinoma of the stomach and esophagus. Int J Cancer, 103(6), 815–​821. PMID: 12516104. Wynder EL, and Hoffmann D. 1994. Smoking and lung cancer:  scientific challenges and opportunities. Cancer Res, 54(20), 5284–​ 5295. PMID: 7923155. Wyss A, Hashibe M, Chuang SC, et al. 2013. Cigarette, cigar, and pipe smoking and the risk of head and neck cancers: pooled analysis in the International Head and Neck Cancer Epidemiology Consortium. Am J Epidemiol, 178(5), 679–​690. PMCID: PMC3755640. Xue WQ, Qin HD, Ruan HL, Shugart YY, and Jia WH. 2013. Quantitative association of tobacco smoking with the risk of nasopharyngeal carcinoma: a comprehensive meta-​analysis of studies conducted between 1979 and 2011. Am J Epidemiol, 178(3), 325–​338. PMCID: PMC3727336. Yeager DS, and Krosnick JA. 2010. The validity of self-​reported nicotine product use in the 2001–​2008 National Health and Nutrition Examination Survey. Med Care, 48(12), 1128–​1132. PMID: 20940652. Yuan JM, Butler LM, Stepanov I, and Hecht SS. 2014. Urinary tobacco smoke-​ constituent biomarkers for assessing risk of lung cancer. Cancer Res, 74(2), 401–​411. PMCID: PMC4066207. Zahm SH, Heineman EF, and Vaught JB. 1992. Soft tissue sarcoma and tobacco use:  data from a prospective cohort study of United States veterans. Cancer Causes Control, 3(4), 371–​376. PMID: 1617125. Zaridze D, Borisova E, Maximovitch D, and Chkhikvadze V. 2000. Alcohol consumption, smoking and risk of gastric cancer:  case-​control study from Moscow, Russia. Cancer Causes Control, 11(4), 363–​ 371. PMID: 10843447. Zhou B, Yang L, Sun Q, et al. 2008. Cigarette smoking and the risk of endometrial cancer:  a meta-​ analysis. Am J Med, 121(6), 501–​ 508 e503. PMID: 18501231. Zu K, and Giovannucci E. 2009. Smoking and aggressive prostate cancer:  a review of the epidemiologic evidence. Cancer Causes Control, 20(10), 1799–​1810. PMID: 19562492.

21

 213

12

Alcohol and Cancer Risk SUSAN M. GAPSTUR AND PHILIP JOHN BROOKS

OVERVIEW Alcoholic beverages have been consumed by humans for thousands of years. Currently, the alcoholic beverages most frequently consumed are spirits, beer, and wine, respectively. Worldwide, an estimated 47.7% of men and 28.9% of women aged 15 years or older report drinking alcohol, and approximately 16.0% of drinkers engage in heavy, episodic (binge) drinking. In 2010, alcoholic beverage consumption caused an estimated 3.3  million deaths worldwide, and contributed to injuries, violence, liver cirrhosis, social disruption, and at least seven different types of cancer. The International Agency for Research on Cancer (IARC) classifies exposure to both ethanol in alcoholic beverages and acetaldehyde, the primary metabolite of ethanol, as carcinogenic to humans (Group 1) based on “sufficient” evidence that alcoholic beverage consumption is causally related to cancers of the oral cavity, pharynx, larynx, esophagus, liver, colorectum, and female breast. The World Cancer Research Fund/​American Institute of Cancer Research Continuous Update Project (WCRF/​AICR CUP) characterized the evidence as “limited” that heavy alcohol consumption increases pancreatic cancer risk. The IARC also notes lack of carcinogenicity of alcohol consumption with non-​ Hodgkin lymphoma and renal cell carcinoma, and for other cancers the evidence is inconclusive. The biologic mechanisms by which alcohol and its primary metabolite acetaldehyde affect cancer risk appear to vary across anatomic sites. Broadly, these mechanisms involve DNA and protein damage from acetaldehyde and oxidative stress, nutritional malabsorption, and metabolic effects, and for breast cancer, increased estrogen levels. In addition, carcinogenic contaminants can be introduced during alcoholic beverage production. While there are known harmful effects of alcohol consumption, relative to no consumption, light to moderate consumption has been associated with reduced risk of coronary heart disease, although recent Mendelian randomization studies have challenged this finding. The World Health Organization (WHO) has increased global surveillance of alcohol consumption and encourages national efforts to apply evidence-​based policies to reduce consumption.

INTRODUCTION Alcoholic beverages have been consumed for religious, social, and cultural reasons for thousands of years. Archaeological evidence indicates that honey, rice, and hawthorn fruit or grapes were fermented to produce alcohol as early as 7000–​6600 bc in China’s Yellow River Valley (McGovern et  al., 2004). Today, the most common commercially produced alcoholic beverages are spirits distilled from grains, sugars, fruits or vegetables, beer from barley, and wine from grapes (World Health Organization, 2014). In some developing countries, locally or home-​produced alcoholic beverages from, for example, fermented apples (cider) or honey-​water (mead) are important contributors to daily consumption (World Health Organization, 2014). The principal form of alcohol found in alcoholic beverages is ethanol (ethyl alcohol), which is produced by the fermentation of sugars and starches by yeast. Ethanol is a central nervous system depressant that motivates recreational use of alcohol, and engenders a sense of excitement, sociability, pleasure, and intoxication. At progressively higher blood levels, it impairs sensory and motor function, cognition, and judgment; severe intoxication can produce stupefaction, unconsciousness,

and death. The WHO estimates that 5.9% of all deaths (7.4% of males and 4% of females) and 5.1% of disease worldwide is attributed to alcohol, and has identified 60 common alcohol-​related conditions and more than 200 conditions in which alcohol consumption is recognized as a component cause (World Health Organization, 2014). The adverse effects of harmful drinking include, for example, intentional and unintentional injuries, violence, acute alcohol poisoning, liver cirrhosis, social disruption, impoverishment, neuropsychiatric disorders, gastrointestinal and cardiovascular problems, fetal alcohol syndrome, preterm complications, diabetes mellitus, exacerbation of certain infectious diseases, and seven types of cancer (IARC, 2012). Acetaldehyde is the primary metabolite of ethanol from alcoholic beverage consumption; preformed acetaldehyde can be measured in alcoholic beverages. The cytotoxic, mutagenic, and carcinogenic effects of acetaldehyde, including formation of DNA adducts and inhibition of DNA repair, have been shown in eukaryotic cells and animal models (Seitz and Stickel, 2010). Acetaldehyde related to alcohol consumption has been classified as carcinogenic to humans (IARC, 2012).

METABOLISM OF ALCOHOL AND ACETALDEHYDE After ingestion, alcohol undergoes “first pass metabolism” (FPM) in the upper aerodigestive tract (UADT) and liver, which reduces the amount of alcohol in circulation relative to the amount consumed. FPM begins in the oral cavity, where both human cells and microorganisms in saliva oxidize alcohol to acetaldehyde. An estimated 5%–​14% of the oral dose of alcohol undergoes FPM (Ammon et al., 1996), with a slightly lower percent when gastric emptying is rapid or the amount consumed is high. Alcohol that does not undergo FPM is distributed throughout the body; the majority (about 90%) of absorbed alcohol is metabolized in the liver. The enzymes involved in the metabolism of alcohol are expressed in many tissues throughout the human body, albeit at varying levels in different cell types (IARC, 2010), and thus play a role in the toxicity and/​or carcinogenicity of alcohol in specific tissues. In humans, the metabolism of alcohol primarily involves a two-​step process in which ethanol is first oxidized to acetaldehyde by the enzyme alcohol dehydrogenase (ADH), and then acetaldehyde is oxidized to acetate (acetic acid) by the enzyme aldehyde dehydrogenase (ALDH). The activity of these enzymes profoundly influences the ability of a drinker to metabolize alcohol and detoxify acetaldehyde. The human genome contains multiple ADH and ALDH genes that are expressed in different tissues. The ADH1A, ADH1B, and ADH1C genes encode the ADH1 α, β, and γ protein subunits, respectively. These different subunit proteins form dimers that are the active form of the ADH enzyme, and are expressed at high levels in the liver. In contrast, the ADH7 protein is expressed in the stomach and the gastrointestinal tract, and is thought to play a role in FPM in these tissues (Crabb et al., 2004). The ADH and ALDH genes encode proteins that combine to form the functional enzymes and are polymorphic in humans; the prevalence of functional variants varies among different geographic populations (Figure 12–1). Mendelian randomization studies have examined the role of ADH and ALDH gene polymorphisms on alcohol avoidance and consumption, and have informed studies of alcohol-​associated chronic diseases (Holmes et al., 2014). As discussed later in the chapter, the *2 allele of ADH1B and the *2 allele of ALDH2 are particularly relevant to alcohol-​related carcinogenesis.

213

214

214

PART III:  THE CAUSES OF CANCER

The ADH1B*2 allele, which is common in Asian and Ashkenazi Jewish populations, encodes the β2 subunit that forms an enzyme with roughly 40-​ fold greater activity at generating acetaldehyde from ethanol than the β1 subunit, which is encoded by the ADH1B*1 allele. ADH1B*1 is commonly found in European and some African populations. The ADH1B*3 allele, which commonly occurs in Native American and other African populations, encodes a subunit of the enzyme with low affinity for ethanol (Crabb et al., 2004). In humans, ALDH enzymes are coded by 19 different genes (Koppaka et  al., 2012). Based on their affinity for acetaldehyde, the ALDH isoforms that are thought to play a role in ethanol-​derived acetaldehyde metabolism include ALDH1A1, ALDH1B1 and ALDH2. Other ALDH enzymes have low affinity for acetaldehyde and are unlikely to play a significant role in acetaldehyde metabolism under normal physiologic conditions. A polymorphism of ALDH2 (rs671) is of particular relevance for understanding the relationship of alcohol consumption to acetaldehyde production and cancer risk. Whereas the ALDH2*1 allele produces a protein with normal catalytic activity, the ALDH2*2 allele encodes the amino acid LYS instead of GLU at position 487, resulting in a catalytically inactive enzyme subunit. The ALDH2*2 allele is found predominantly in East Asians (Japanese, Chinese, and Koreans) (Goedde and Agarwal, 1987). When individuals carrying the ALDH2*2 allele drink alcohol, they experience a variety of aversive effects, including nausea, headache, and facial flushing, resulting from elevated acetaldehyde concentrations in the body. Individuals who are homozygous for the ALDH2*2 allele have essentially undetectable ALDH2 activity, and they generally cannot tolerate alcohol due to severe acetaldehyde toxicity. Paradoxically, heterozygotes for the ALDH2*2 allele have an enzyme activity roughly 1% of that of ALDH2*1 homozygotes, far less than the 50% value typically expected. The ALDH2 enzyme is a functional tetramer, and the low enzyme activity among heterozygotes suggests that the presence of even one inactive subunit may be sufficient to prevent enzyme activity (Crabb et  al., 2004). However, individuals who are heterozygous for the ALDH2*2 allele can become tolerant to the side effects of alcohol and can become heavy drinkers,

Ethanol

with significantly elevated risk of alcohol-​related esophageal cancer (Yokoyama et al., 1996). In addition to the enzymes in human cells, other important contributors to alcohol metabolism are microbes (particularly bacteria) resident in the human body. Some of these bacteria express ADHs. The microbial metabolism of alcohol to acetaldehyde can result in locally high concentrations of extracellular acetaldehyde in the oral cavity and intestine (Homann et  al., 2000), which are thought to contribute to alcohol-​related carcinogenesis in those tissues. Aside from ADH and ALDH, another enzyme that metabolizes alcohol and affects alcohol-​related carcinogenesis is the ethanol-​inducible cytochrome P4502E1 (CYP2E1) (Cederbaum, 2012). While both ADH and CYP2E1 convert alcohol to acetaldehyde, there are several important differences between these two enzymes. Whereas the level of CYP2E1 protein increases in the presence of high concentrations of circulating alcohol, the level of ADH is generally not affected. This alcohol-​related increase in CYP2E1 occurs because ethanol is both a substrate for the enzyme and, when present, stabilizes the enzyme against degradation. Alcohol metabolism by CYP2E1 significantly increases at alcohol concentrations above ≈ 20 mM (Asai et al., 1996; Salaspuro and Lieber, 1978). For comparison, the legal blood alcohol limit for driving in many countries is 0.08%, which corresponds to an alcohol concentration of approximately 17 mM. CYP2E1 oxidation of ethanol can generate reactive oxygen species that react with cellular lipids to form highly mutagenic DNA adducts (Linhart et al., 2014). Thus, at high blood concentrations, a qualitatively different type of alcohol metabolism takes places in human cells, resulting in an additional mechanism of genotoxicity separate from that involving acetaldehyde (Figure 12–2).

EXPOSURE ASSESSMENT Ethanol Content in Alcoholic Beverages The content (concentration) of pure ethanol in alcoholic beverages is usually expressed either as percent of volume (% vol.), or “proof”

Acetaldehyde

Acetate

ADH

ALDH

Cell death/hyper-regeneration

Mutagenesis carcinogenesis

Acetaldehyde-DNA adducts

O N N HO

O N

NH N

N

O

CH3

N HO

O N

NH N

N

NH

HO

O

O

OH N

N

N H

CH3

H3C OH N2-ethylidenedeoxyguanosine

OH

O

OH

Crotonaldehyde-propanodeoxyguanosine adducts

Figure 12–​1.  Alcohol and acetaldehyde metabolism, acetaldehyde–​DNA adducts. Top: schematic diagram of enzymatic pathway of alcohol and acetaldehyde metabolism. As noted in the text, ALDH2 is the primary ALDH involved in the metabolism of acetaldehyde derived from alcohol metabolism. Other ALDHs may be involved in the acetaldehyde metabolism in certain tissues. Bottom panel: Structures of acetaldehyde-​related DNA adducts. Crotonaldehyde-​derived adducts can exist in either of two forms, ring-​opened (left) and ring-​closed (right). The ring-​opened forms can result in DNA intrastrand crosslinks. ADH: alcohol dehydrogenase; ALDH: aldehyde dehydrogenase.

 215



215

Alcohol and Cancer Risk Ethanol

Acetaldehyde

CYP2E1

O2

H2O

Reactive oxygen species

Lipid peroxidation products

Ethenobase-DNA adducts

N N N HO

N

N N

N HO

OH 1,N6 etheno-deoxyadenosine

O

N N

O

Mutagenesis carcinogenesis

O

O

N HO

N N

N

O

OH

OH 3,N4-etheno-deoxycytidine

1,N2-ethenodeoxyguanosine

Figure 12–​2.  Alcohol metabolism by CYP2E1 and ethenobase DNA adducts. Top: schematic diagram of alcohol metabolism by cytochrome P450 2E1 (CYP2E1). CYP2E1 utilizes molecular oxygen to oxidize ethanol to acetaldehyde, generating H2O in the process. CYP2E1 can also generate reactive oxygen species, including superoxide radical and hydrogen peroxide, which can further react with cellular lipids, resulting in lipid peroxidation and etheno-​ base DNA adducts. The bottom panel shows the molecular structures of three ethenobase DNA adducts.

(which in the United States is twice the percent of alcohol by volume [e.g., 80 proof is 40% vol.]). The ethanol content is typically lowest (4%–​5% vol.) in commercially produced beer, intermediate (about 12% vol.) in wine, and highest (about 40% vol.) in distilled spirits. However, the concentration of ethanol in each beverage type varies widely (IARC, 2010). In beer, the ethanol content ranges from 2.3% vol. to over 10% vol., with home-​or locally produced beverages such as sorghum beer having lower content. The ethanol content in wine ranges from 8% to 15% vol., and in spirits ranges from 20% vol. (aperitifs) to well over 40% vol. (e.g., 80% vol. in certain kinds of absinthe). The absolute amount of pure ethanol contained in a serving of alcoholic beverage is estimated by multiplying the standard volume of an alcoholic drink by its alcohol content. The amount is expressed either as mL or as grams of pure ethanol (1 mL of ethanol = 0.79 g). The volume of a standard alcohol drink in the United States is generally 12 ounces of beer, 5 ounces of wine, or 1.5 ounces of distilled spirits (Centers for Disease Control and Prevention, 2016). Thus, for example, a 12 oz. serving of beer with an ethanol content of 5% vol. contains about 14 grams of ethanol (i.e., 355 mL × 0.05 × 0.79 g/​mL).

Acetaldehyde Content in Alcoholic Beverages Although most exposure to acetaldehyde from consuming alcoholic beverages results from ethanol metabolism, high concentrations of preformed acetaldehyde are present in alcoholic beverages that are commonly consumed in many countries (IARC, 2012). Average levels vary from 9 mg/​L in beer, 34 mg/​L in wine, 66 mg/​L in spirits, and 90 mg/​L in unrecorded alcohol (Lachenmeier et  al., 2012). Concentrations are especially high in distilled beverages from Brazil, the People’s Republic of China, Guatemala, Mexico, and the Russian Federation, as well as in fortified wines such as calvados and spirits distilled from fruit in Europe.

Surveillance and Epidemiologic Assessment of Alcohol Consumption For both surveillance and epidemiologic studies, the assessment of alcohol intake takes into account the volume and ethanol content of beverage(s) consumed. Intake is expressed as the average amount of pure ethanol (in g) consumed per person per unit time. At the population level, surveillance data of alcohol consumption typically reflect annual consumption of pure ethanol per capita, including both recorded data and estimates of unrecorded data. WHO estimates average annual per capita consumption for ages 15+ years to include older adolescents as well as adults (World Health Organization, 2014). Estimates of recorded consumption derive from official statistics on production, trade, taxes, and/​or sales. However, estimates of unrecorded consumption, which comprises nearly 25% of all alcohol consumed, may include formal or informal accounts of domestic production, illegal sales, smuggling and cross-​border shopping, and in some parts of the world a substantial fraction of alcohol consumption is unrecorded (World Health Organization, 2014). This is particularly important when estimating consumption in the poorest regions of Africa, Asia, and South America, and where drinking is prohibited by religion. For example, although the total level of consumption is very low in Islamic countries in the Eastern Mediterranean Region, the proportion contributed by unrecorded consumption is high (i.e., more than 50% of consumption). Consumption by tourists is excluded from these estimates where possible. In large epidemiologic studies of cancer risk, individual alcoholic beverage consumption is often self-​reported on food frequency questionnaires (FFQ). An FFQ queries study participants on the average frequency of consumption (number of days per week or month), the usual quantity consumed (number of drinks), and the types of alcoholic beverages (e.g., spirits, can/​bottle of beer, glass of red or white wine) consumed during a specified time period (usually the previous year). Intake is then characterized in terms of either average consumption of ethanol by weight (expressed as grams [g]‌per day) or the average number of alcoholic beverages consumed per day or per week.

216

216

PART III:  THE CAUSES OF CANCER

The validity of self-​reported information on alcohol consumption has been questioned because of poor recall, inaccurate estimation of the number of drinks consumed on drinking days, inability to report average intake according to standard drink sizes (Stockwell et al., 2004), and variable drinking patterns (Boniface et al., 2014). To address some of these concerns, several large epidemiologic studies compared alcohol intake data estimated from daily diaries or multiple 24-​hour recalls to alcohol intake data obtained from FFQs, and found the correlations to be high (range from 0.7 to 0.9) (Giovannucci et  al., 1991; Kaaks and Riboli, 1997). These findings suggest that the self-​reported data on usual alcohol consumption are reasonably consistent across different methods of collection. However, this comparison does not address all issues of underreporting, such as differences between actual drink size consumed and standard drink sizes. While misreporting does not influence the conclusion that alcohol is a human carcinogen, it can affect the quantitative estimates of the cancer risk associated with a given amount of consumption. Therefore, methods have been developed to correct estimates of relative risk (Rosner, 1995), and population-​attributable risk for errors in the measurement of alcohol consumption (Shield et al., 2013). Another source of error occurs when weekly alcohol consumption is divided by seven to estimate the number of drinks consumed per day. As discussed elsewhere (Brooks and Zakhari, 2013), consuming seven drinks per week might not correspond to consuming one drink per day. An individual who drinks three to four drinks on each weekend night, but nothing during the rest of the week, might report seven drinks per week. However, this pattern of consumption could produce a blood alcohol level sufficient to induce CYP2E1, which would not be induced by actual consumption of one drink per day. As noted earlier, the induction of CYP2E1 has distinct biological and genotoxic consequences, with mechanistic implications for alcohol-​ related carcinogenesis.

GLOBAL PATTERNS OF ALCOHOL CONSUMPTION The Global Information System on Alcohol and Health provides information on 180 alcohol-​related indicators for 194 WHO member states and two associate members. This information is used for the WHO Global Status Report on Alcohol and Health and is a valuable resource for worldwide, regional, and country-​specific estimates of total alcohol consumption (both recorded and unrecorded) (World Health Organization, 2014). Summaries of international statistics on alcohol consumption have been reported for the six WHO regions:  Region for the Americas (AMR), South-​East Asia Region (SEAR), European Region (EUR), Eastern Mediterranean Region (EMR), African Region (AFR) and Western Pacific Region (WPR). These region-​specific summaries provide a useful comparison of alcohol consumption across the world.

Types of Alcoholic Beverages Consumed Globally, distilled spirits comprise 50.1% of total recorded alcohol consumption, beer accounts for 38.8%, wine for 8.0%, and other commercially produced alcoholic beverages for the remainder. Consumption patterns vary regionally, however. Wine consumption accounts for one-​fourth (25.7%) of total consumption in the WHO EUR and one-​ninth (11.7%) of total consumption in the WHO AMR. Beer accounts for over-​half (55.3%) of the consumption of alcoholic beverages in the WHO AMR.

Prevalence of Current Alcohol Drinkers and Abstainers Worldwide, an estimated 38.3% of the population aged 15+ years report current alcohol consumption. However, there are substantial differences in the prevalence of current drinkers by WHO region. The lowest prevalence of current drinkers is in the EMR (5.4%) and the highest prevalence is in the EUR (66.5%), followed closely by the

AMR (61.5%). Interestingly, only 14.7% of the world’s population aged 15+ years lives in the EUR, but residents of this region account for more than 25.7% of global consumption. Males are 1.6 times more likely to be current alcohol drinkers than females (47.7% vs. 28.9%, respectively). This gender difference varies regionally, and is greater in the SEAR (i.e., 21.7% of males vs. 5% of females) than in the EUR or AMR, where the ratio of male to female current drinkers is 1.2 and 1.3, respectively. In the EUR, 73.4% of males and 59.9% of females are current drinkers. Worldwide, 61.7% of the population aged 15+ years did not drink alcohol in the past 12 months; about half of the population report lifelong abstinence. There is wide variation in the prevalence of abstention across WHO Regions, but females are more likely to be abstainers than males.

Per Capita Amount of Alcohol Consumed In 2010, the global average annual consumption per capita of pure ethanol among persons aged 15+ years was estimated to be about 6.2 liters, or about 13.5 g/​day per capita. The average annual per capita consumption was highest in the EUR with 10.9 liters and the lowest in the EMR with 0.7 liters. Since only a subset of the population consumes alcohol, the worldwide average annual alcohol consumption per capita is considerably higher when the estimate is restricted to drinkers. The average for those who drink alcohol is 17.2 liters per year, with a range of 11.3 liters in the EMR to 23.1 liters in the SEAR. Worldwide, men who drink alcohol consume more than twice as much than women; the average annual per capita consumption among male drinkers (aged 15+ years) was 21.2 liters and among female drinkers was 8.9 liters of pure ethanol. The sex difference in average annual alcohol consumption among drinkers varies considerably across WHO regions; in the SEAR annual consumption is 3.2 times greater among male than among female drinkers, and in the AFR annual consumption is 1.7 times greater among male than among female drinkers.

Prevalence of Heavy Episodic Drinking Globally, approximately 7.5% of the total population aged 15+ years and 16.0% of drinkers aged 15+ years engage in heavy episodic drinking (i.e., defined as 60 or more g of pure ethanol on at least one occasion at least monthly).

Trends in Consumption In some parts of the world, alcohol consumption increased from 2005 to 2010, whereas in other parts of the world, consumption decreased or remained stable. For example, the prevalence of current alcohol drinking in those aged 15+ years increased from 58.3% to 61.5% in the AMR, from 3.5% to 5.4% in the EMR, and from 10.7% to 13.5% in the SEAR. Conversely, from 2005 to 2010 the prevalence of current drinking decreased in the EUR from 68.8% to 66.4%, and in the WPR from 56.3% to 45.8%, but remained stable in the AFR at approximately 29%. Worldwide the total average annual per capita alcohol consumption remained stable between 2005 and 2010 at approximately 6.1 and 6.2 liters, respectively. However, there was a slight decrease in consumption in the EUR from 12.2 to 10.9 liters and an increase in consumption in the SEAR from 2.2 to 3.4 liters. Some of the differences in the prevalence of current drinkers and in the average annual amount consumed per capita are due to changes over time in the available data and methods to estimate consumption, and therefore these differences should be interpreted cautiously. Per capita alcohol consumption among those aged 15+ years is expected to increase slightly through 2025. In particular, without effective policies to prevent an increase, the highest increase in average annual per capita consumption is expected to occur in the WPR (i.e., from 6.8 liters to 8.3 liters).

 217



Alcohol and Cancer Risk

Patterns of Consumption by the Economic Wealth of a Country The economic wealth of a country correlates with the prevalence of current drinking, the amount consumed, and the prevalence of heavy episodic drinking. The prevalence of current drinking is 18.3% in low-​income countries, 19.6% in middle-​income countries, 45% in upper middle-​income countries, and 69.5% in high-​income countries. Similarly, average annual per capita alcohol consumption in the total population aged 15+ years ranged from 3.1 liters in low-​income countries to 9.6 liters in high-​income countries. Furthermore, the prevalence of heavy episodic drinking among those who drink alcohol is 11.6% in low-​income countries to 22.3% in high-​income countries.

TYPES OF CANCER CAUSED BY ALCOHOL CONSUMPTION The consumption of alcoholic beverages was first classified as carcinogenic to humans in 1987 by the International Agency for Research on Cancer (IARC, 1988). At that time, the evidence for causality was deemed to be “sufficient” for oral cavity, pharynx, larynx, esophagus, and liver cancer. Although results of some studies had shown that alcohol consumption was also positively associated with the risk of colorectal and female breast cancer, the evidence for these sites was considered inconclusive. Importantly, no conclusions were made about the carcinogenic agent in alcoholic beverages (e.g., ethanol or other constituents of alcoholic beverages) at that time. Another expert Working Group (which included both authors of this chapter) convened by the IARC in 2007 reviewed the epidemiologic and other evidence on the associations of alcoholic beverage consumption and cancer at 27 anatomic sites (IARC, 2010). This Working Group concluded that “ethanol in alcoholic beverages” is carcinogenic to humans (Group  1). The Group found no consistent evidence that the effects of alcoholic beverage consumption on cancer differed by beverage type, and deemed the evidence that ethanol causes cancer in experimental animals to be sufficient. They reaffirmed that alcohol consumption causes cancers of the liver and UADT, including the oral cavity, pharynx, larynx, and esophagus. Moreover, they expanded the list of cancers caused by alcohol consumption to include colorectal and female breast. The 2007 Working Group also concluded that both non-​Hodgkin lymphoma and kidney cancer were not caused by alcohol consumption. However, the evidence linking alcohol consumption to other cancer sites was considered inconclusive. For example, no conclusions were drawn concerning lung or stomach cancer due to potential confounding by tobacco smoking and diet. In 2009, a Working Group convened by the IARC again reviewed the evidence regarding alcohol consumption and updated and confirmed the conclusions of previous Working Groups (IARC, 2012). The evidence that heavy alcohol intake was associated with a small increase in pancreatic cancer risk was considered limited. For the first time, the 2009 Working Group concluded that both ethanol and acetaldehyde associated with the consumption of alcoholic beverages were carcinogenic to humans for cancers of the UADT. As described later in the chapter, the carcinogenic effects of ethanol found in alcoholic beverages and the primary metabolite of ethanol, acetaldehyde, vary by cancer type. Broadly, these mechanisms involve DNA and protein damage from acetaldehyde and oxidative stress, alterations in DNA repair, cell death and hyper-​regeneration, nutritional malabsorption, changes in DNA methylation, and metabolic effects, and for breast cancer, increased estrogen levels (Seitz and Stickel, 2007). In addition, carcinogenic contaminants can be introduced during alcoholic beverage production.

Upper Aerodigestive Tract Alcoholic beverage consumption is a known cause of cancers of the UADT, including the oral cavity, pharynx, larynx, and esophagus (IARC, 1988). The causal relationship is strongest for squamous cell

217

carcinomas (IARC, 2012). Compared to alcohol non-​drinkers, people who consume about 50 g of ethanol per day have about a 2-​to 3-​ fold higher risk of UADT cancers (IARC, 2010). It is worth noting that acetaldehyde, which can be formed during FPM in the oral cavity, has a boiling point of 25ºC, which is well below body temperature. This raises the possibility that acetaldehyde vapor may play a role in alcohol-​related UADT cancers. Experimental studies in animals have shown that chronic exposure to acetaldehyde vapor increases UADT metaplasia (Woutersen et al., 1986). A significant confounder or effect modifier of the association between alcoholic beverage consumption and UADT cancer risk is smoking. Smoking causes multiple types of cancer, and alcohol drinkers, particularly heavy drinkers, are more likely to be smokers than non-​drinkers. Controlling for smoking in statistical models only partly separates the risk due to alcohol consumption from that due to smoking. A more thorough approach is to study the relationship between alcohol and cancer in lifelong non-​smokers. Several early studies showed a significant positive association between alcohol intake and risk, even in never smokers (IARC, 2010). A meta-​analysis of 15 studies found alcoholic beverage consumption (4 or more drinks per day) was most strongly associated with increased risk of pharynx cancer, but was also associated with oropharynx, larynx, and esophageal squamous cell carcinoma among those who did not use tobacco (Zeka et al., 2003). In a pooled analysis of case-​control studies, heavy alcohol consumption (3 or more drinks per day) was associated with an increased risk of cancers of the oropharynx/​hypopharynx and larynx in never users of tobacco (Hashibe et al., 2007). There is extensive evidence of a strong synergistic effect of tobacco smoking and alcohol consumption on the risk of UADT cancer, even at moderate amounts of consumption. In most epidemiologic studies, the interaction is greater than multiplicative (IARC, 2012). In a large prospective study of British women, alcohol consumption was most strongly associated with increased risk of UADT cancers among current smokers, less strongly among former smokers, and not at all among never smokers (Allen et al., 2009). One of the two proposed mechanisms by which alcohol consumption might potentiate the carcinogenicity of tobacco smoking depends on dose. CYP2E1, which is induced by heavy alcohol consumption, can activate procarcinogens in cigarette smoke (Cederbaum, 2012). However, since the interaction between smoking and alcohol consumption has been observed even at low levels of alcohol consumption, where the induction of CYP2E1 is unlikely, alcohol might also act as a “solvent,” enhancing the hydrophobic diffusion of carcinogens present in cigarette smoke into epithelial tissues (Squier et al., 1986). Much of the evidence that acetaldehyde associated with alcoholic beverage consumption is a cause of UADT cancer comes from a series of epidemiologic studies by Yokoyama and colleagues showing increased risk of esophageal squamous cell carcinoma (ESCC) among ALDH2*2 heterozygous-​alcoholics (Yokoyama et al., 2006). As noted earlier, the adverse effects of alcohol in ALDH2*2 homozygous individuals are so severe that they are generally unable to consume significant amounts of alcohol. However, some ALDH2*2 heterozygotes are able to overcome the adverse effects of acetaldehyde related to alcohol drinking, and become heavy drinkers. When matched for alcohol consumption and smoking, ALDH2*2 heterozygotes have a significantly elevated risk of ESCC compared to those with fully active ALDH2 (reviewed by Brooks et al., 2009). Using a Mendelian randomization approach, a meta-​analysis of studies examining ALDH2 genotype and esophageal cancer risk demonstrated a reduced risk for ALDH2*2 homozygous individuals (odds ratio [OR]  =  0.36; 95% confidence interval [CI]:  0.16, 0.80) and a higher risk among ALDH2*2 heterozygotes (OR = 3.19; 95% CI: 1.86, 5.47), compared with ALDH2*1 fully active homozygotes (Lewis and Smith, 2005). The lower risk of esophageal cancer in ALDH2*2 homozygotes likely reflects their avoidance of alcohol. When stratified by alcohol drinking status, ALDH2*2 heterozygosity was not associated with risk of esophageal cancer in non-​drinkers, but there was a 2.5-​fold higher risk among moderate drinkers and a 7-​fold higher risk among heavy drinkers. Similarly, in a meta-​analysis of studies of ALDH2 genotype and head and neck cancer risk, there was a reduced risk for ALDH2*2 homozygous individuals (OR = 0.53; 95% CI: 0.28, 1.00)

218

218

PART III:  THE CAUSES OF CANCER

and a higher risk among ALDH2*2 heterozygotes (OR = 1.83; 95% CI: 1.21, 2.77) as compared with fully active ALDH2*1 homozygotes; the association with ALDH2*2 heterozygosity also varied according to the level of alcohol consumption, with no association among never drinkers and 3.6-​fold higher risk among heavy drinkers (Boccia et al., 2009). Taken together, these studies support the evidence that alcoholic beverage consumption can increase risk of UADT cancers through the carcinogenic effects of acetaldehyde.

Liver The association between alcohol consumption, particularly heavy alcohol consumption, and liver cancer (i.e., hepatocellular carcinoma) is well established (Turati et al., 2014). This association has been observed in studies of individuals infected with hepatitis viruses (HV) B or C, and in unaffected individuals without cirrhosis. A meta-​analysis of 19 prospective studies that excluded cohorts of alcoholics, patients with cirrhosis, and individuals with HBV or HCV infection showed that consumption of 3 or more drinks per day was associated with a 16% increased risk compared to non-​drinkers, although at this level there was considerable heterogeneity among studies (p = 0.002) (Turati et al., 2014). The pooled relative risk for heavy consumption of 6 or more drinks per day was 1.22 (95% CI: 1.10, 1.35) with no evidence of heterogeneity among studies (p = 0.8). However, because most liver cancers occur in the context of cirrhosis, which can motivate individuals to reduce their alcohol consumption, estimates of the dose–​response relationship between alcohol consumption and liver cancer should be interpreted cautiously. Findings from several studies suggest possible synergistic effects of alcohol with other established risk factors for hepatocellular carcinoma, including HBV and HCV infection (Hassan et al., 2002; Yuan et al., 2004), obesity (Loomba et al., 2010, 2013; Marrero et al., 2005), and smoking (Kuper et al., 2000; Marrero et al., 2005). More research from prospective studies is needed to confirm these findings. Most liver cancers occur in the context of viral hepatitis, fibrosis, and/​or cirrhosis (El-​Serag and Rudolph, 2007). Chronic heavy alcohol consumption often leads to fibrosis and ultimately cirrhosis, and alcohol-​related liver cancer in the absence of these conditions is rare (Seitz and Stickel, 2007). Alcohol-​related liver disease not only causes cell death and regeneration, but also results in chronic inflammation (Wang et al., 2012), which can itself generate mutagens and damage DNA (Son et  al., 2008)  beyond the effects of alcohol metabolism. More specifically, inflammation and oxidative stress resulting from elevated levels of CYP2E1 can increase levels of etheno DNA-​base adducts, some of which are highly mutagenic (Pandya and Moriya, 1996). Elevated levels of etheno DNA-​base adducts have been documented in tissue biopsies from patients with alcoholic liver disease (Frank et al., 2004; Wang et al., 2009). Although the liver is the major site of alcohol metabolism, acetaldehyde has not been implicated in liver carcinogenesis. Alcoholics deficient in ALDH2 activity have little or no increase in liver cancer risk (IARC, 2010), despite their dramatically higher risk of UADT cancers. This difference in the effect of acetaldehyde on the liver compared to the UADT may be due, at least in part, to the much lower proliferation rates of hepatocytes compared to epithelial cells in the aerodigestive (and gastrointestinal) tract. In the absence of DNA synthesis and cell replication, mutations resulting from DNA damage are not transmitted to daughter cells, and there is more time for DNA repair to occur. Proliferating cells also have higher levels of polyamines, which can react with acetaldehyde and lead to mutagenic DNA damage (Theruvathu et al., 2005).

Colorectum The IARC first concluded that alcoholic beverage consumption caused colorectal cancer in 2007 (IARC, 2010). The evidence at that time was based partly on a pooled analysis of eight cohort studies (Cho et  al., 2004) and a meta-​analysis of 16 prospective studies (Moskal et  al., 2007). These analyses showed that regular consumption of about 50 g of ethanol per day was associated with an approximately 40% higher

risk of colorectal cancer compared to non-​ drinkers; however, there was uncertainty regarding the nature of the dose–​response relationship (IARC, 2010). A systematic literature review by the WCRF/​AICR CUP designated the evidence for a causal association between ethanol from alcoholic consumption and colorectal cancer risk as convincing in men and probably convincing in women (World Cancer Research Fund International, 2011). More specifically, the CUP meta-​analyses of prospective studies showed that a 10 g per day increase in ethanol was associated with a 10% increased risk of colorectal and rectal cancers, and an 8% increased risk of colon cancer. The association was stronger among men than women for both colorectal and colon cancers. The risk of colorectal cancer increased by 11% among men and 7% among women per 10 g per day increase in ethanol. Reasons for the gender difference are unclear, but may relate to differences in the gut microbiome, or hormone-​related differences in alcohol metabolism (Park et al., 2006). It is unclear whether tobacco use—​an established cause of colorectal cancer—​modifies the association between alcohol consumption and colorectal cancer risk. In the pooled analysis of eight prospective studies mentioned in the preceding paragraph (Cho et al., 2004), the association between alcohol consumption and colorectal cancer risk was statistically significant among past and current smokers, but not among never smokers; however, the test for interaction was not statistically significant (p > 0.2). Similarly, in the European Prospective Investigation into Cancer and Nutrition, alcohol consumption was more strongly associated with colorectal cancer risk among smokers than among non-​smokers, but again there was no evidence of a statistically significant interaction (p = 0.41) (Ferrari et al., 2007). Another study showed a marginally significant interaction between smoking and alcohol consumption for colon cancer (p = 0.051) but not for rectal cancer (p > 0.19) (Tsong et al., 2007). It is plausible that the combined exposure to smoking and alcohol could modify their separate effects on colorectal cancer risk. As noted previously for UADT cancers, alcohol might function as a solvent, enhancing the penetration of other carcinogenic molecules into mucosal epithelium (Squier et al., 1986). There is compelling evidence for a causal relationship between acetaldehyde derived from alcohol metabolism and colorectal cancer. Experiments in animals have shown that alcohol metabolism by colonic bacteria leads to very high (up to mM) concentrations of acetaldehyde (Jokelainen et  al., 1996; Visapaa et  al., 1998). Although no similar measurements have been made in humans, it is not unreasonable to assume that acetaldehyde levels in the human colon could reach the same concentrations after alcohol consumption, at least in some individuals. Variation between individuals might be substantial, depending upon the composition of the colonic microbiome. The mechanisms by which acetaldehyde affects the risk of colorectal cancer are not completely understood. Plausible mechanisms could include DNA adducts from acetaldehyde, and increased cell division in regenerating tissue. Other mechanisms have also been considered (Seitz and Stickel, 2007). While there is some evidence that ALDH2-​deficient individuals are at increased risk of alcohol-​associated colorectal cancer, more data are needed to confirm this finding (IARC, 2010). It is notable that ALDH2 is maximally active at low acetaldehyde concentrations. If the levels of colonic acetaldehyde after alcohol consumption are roughly comparable in humans and animals, the role of ALDH2 compared to other ALDHs expressed in the colon would be limited. Thus, the lack of a large difference in colon cancer risk between ALDH2-​proficient and ALDH2-​deficient alcohol drinkers does not exclude the possibility that acetaldehyde might contribute to alcohol-​related colorectal carcinogenesis. Other mechanisms that could potentially mediate the effects of alcohol on colorectal carcinogenesis include oxidative stress, altered DNA methylation, and nutritional deficiencies, such as reduced levels of folic and retinoic acid as a result of heavy drinking (Seitz et al., 2005).

Breast Based on a large body of evidence, the 2007 IARC Working Group concluded for the first time that alcohol consumption is a cause of breast cancer in humans (IARC, 2010). Included in the IARC review

 219



Alcohol and Cancer Risk

was a large pooled analysis of 53 epidemiologic studies with more than 58,000 women with breast cancer. In that analysis, there was a linear dose–​response association with a 7.1% (95% CI: 5.5, 8.7) increase in breast cancer risk per 10 g per day increase in ethanol consumption (Hamajima et al., 2002). Subsequently, at least 13 additional prospective studies were published. Based on the total evidence, a 2009 IARC Working Group confirmed the causal relationship between alcohol and breast cancer risk (IARC, 2012). In general, most factors do not appear to modify the relationship between alcohol and breast cancer risk. In the aforementioned pooled analysis of 53 case-​control and prospective studies, there was no evidence of effect modification by age at diagnosis, parity, age at first birth, breastfeeding, race, country, education, mother or sister with breast cancer, age at menarche, height, weight, body mass index, ever use of hormonal contraceptives, ever use of hormone replacement therapy, or menopausal status (Hamajima et al., 2002). The possible exception to these null findings is smoking. A  recently published pooled analysis of data from 14 prospective studies with over 934,000 participants and 36,060 invasive breast cancer cases showed a statistically significant interaction between alcohol and smoking status (p-​ heterogeneity = 0.03) (Gaudet et al., 2016). The mechanism(s) underlying the association between alcohol and breast cancer risk are not completely understood. Alcohol consumption typically begins in late adolescence and continues throughout life, and thus it is difficult to isolate the biologically relevant time period of exposure (Brooks and Zakhari, 2013). One proposed mechanism involves the genotoxicity of acetaldehyde in breast cells. ADH enzymes produced by human breast epithelial cells metabolize alcohol, even at low concentrations (Triano et al., 2003), suggesting that acetaldehyde is generated at relatively low levels of consumption. An important unanswered question is whether those cells also express ALDH2, which would metabolize the resulting acetaldehyde. ALDH2 protein is detectable in human breast tissue (Sutton et al., 2010), but the cellular localization has not been reported. Binge drinking can raise blood alcohol levels into the range at which CYP2E1 is induced, resulting in mutagenic DNA damage from free radicals and lipid peroxidation. As noted earlier, this mechanism does not involve acetaldehyde. Relatively few epidemiologic studies have examined binge drinking in relation to breast cancer risk. In the Danish Nurse Cohort Study, binge drinking during the weekend was associated with increased breast cancer risk (relative risk [RR] = 1.49; 95% CI: 1.04, 2.13 for 10–​15 drinks; and RR = 2.51; 95% CI: 1.37, 4.59 for 16–​21 drinks) compared to drinking 1–​3 drinks during the latest weekend (Morch et al., 2007). In the US Nurses’ Health Study, binge drinking, but not the frequency of drinking, was positively associated with breast cancer risk after adjustment for cumulative alcohol intake (Chen et al., 2011). Binge drinking may be particularly deleterious for younger women, who are most likely to binge drink. Breast epithelium is thought to be most susceptible to possible carcinogens between menarche and first full-​term pregnancy, a period of time when breast cells proliferate rapidly but have not yet terminally differentiated. It is well established that a longer duration between menarche and age at first full-​term pregnancy is associated with a higher risk of breast cancer (Li et  al., 2008). The association between alcohol consumption between menarche and first pregnancy was examined in the Nurses’ Health Study (Liu et al., 2013). In analyses adjusting for consumption after first pregnancy, a 10 g per day increase in cumulative average ethanol consumption between menarche and first pregnancy was associated with an 11% increase in risk of breast cancer (RR = 1.11; 95% CI: 1.00, 1.23). This association was comparable to the association with cumulative average consumption after first pregnancy (RR = 1.09; 95% CI: 0.96, 1.23 per 10 g per day increase). Notably, the association between alcohol consumption between menarche and first pregnancy and risk of breast cancer was limited to women with 10 or more years’ duration of consumption (RR = 1.21; 95% CI: 1.08, 1.36 per 10 g per day increase). Similarly, cumulative average ethanol consumption between menarche and first pregnancy, but not consumption after full-​term pregnancy, was

219

associated with higher risk of benign breast disease, another established breast cancer risk factor. Alcohol consumption also increases circulating estrogen levels, which are positively associated with breast cancer risk in both premenopausal (Endogenous et al., 2013) and postmenopausal women (Key et  al., 2002). In a controlled feeding study among premenopausal women, consumption of 30 g/​day of ethanol for three menstrual cycles increased plasma estradiol and estrone levels by 28% and 21%, respectively (Reichman et al., 1993). Similarly, in a randomly assigned cross-​over alcohol feeding study among postmenopausal women, both 15 and 30 g/​day of ethanol increased serum estrone sulfate and dehydroepiandrosterone concentrations compared to placebo, but did not affect other sex hormones, including estradiol and testosterone (Dorgan et al., 2001). In a pooled analysis of 13 prospective studies of postmenopausal women, alcohol consumption was positively associated with both estrogens and androgens, and inversely associated with sex hormone–​binding globulin (Endogenous et  al., 2011). Further evidence that alcohol might act through a hormonal mechanism is provided by epidemiologic studies showing a stronger association between alcohol consumption and risk of hormone-​receptor-​positive breast cancer than with hormone-​receptor-​negative breast cancer. In a 2008 meta-​analysis of three prospective cohort studies and 16 case-​control studies, the summary relative risk for the highest versus the lowest categories of alcohol consumption were 1.27 (95% CI: 1.17, 1.38) for estrogen-​ receptor-​positive (ER+) breast cancer and 1.14 (95% CI: 1.03, 1.26) for estrogen-​receptor-​negative (ER–​) breast cancer (Suzuki et  al., 2008). When further stratified on ER and progesterone-​receptor (PR) status, high versus low alcohol consumption was associated with statistically significantly elevated risks of both ER+/​PR+ (RR = 1.22; 95% CI: 1.11, 1.34) and ER+/​PR–​(RR = 1.28; 95% CI: 1.07, 1.53) breast cancer but not with ER–​/​PR–​breast cancer (RR = 1.10; 95% CI: 0.98, 1.24). Overall, the 2010 IARC Working Group could not conclude that the association between alcohol consumption and breast cancer risk differed by hormonal status (IARC, 2012). An argument against a significant role for estrogen in alcohol-​ related breast cancer is the absence of an association between alcohol consumption and risk of endometrial (Friberg et  al., 2010)  or ovarian cancer (Genkinger et  al., 2006), two other cancer sites strongly related to hormones. Based on evidence from WCRF/​AICR CUP systematic literature reviews, the relative risk per 10 g per day increase in ethanol consumption was 1.01 (95% CI: 0.97, 1.06) for endometrial cancer (World Cancer Research Fund International, 2013), and 1.01 (95% CI: 0.96, 1.06) for ovarian cancer (World Cancer Research Fund International, 2014a). Finally, moderate alcohol consumption has been hypothesized to increase risk of breast cancer by increasing breast density. Breast density reflects fibroglandular breast tissue that appears radiologically dense on mammograms, and is consistently, strongly, and positively associated with breast cancer risk (Boyd et  al., 1998). However, as reviewed by Liu et  al. (2015), the association between alcohol consumption and breast density is inconsistent among epidemiologic studies.

Pancreas Although the 2009 IARC Working Group cited accumulating evidence that high alcohol intake (i.e., ≥ 30 g per day) was associated with a small increased risk for cancer of the pancreas, the evidence was considered “limited.” The available studies had insufficient statistical power to examine the association for heavy consumption (e.g., 3 or more drinks per day), or to eliminate potential confounding by cigarette smoking (IARC, 2012). Subsequently, the association between alcohol intake and pancreatic cancer mortality was examined in the American Cancer Society’s Cancer Prevention Study-​II (CPS-​II), a large prospective study of over 1  million US adults. Based on 24 years of follow-​up and almost 7,000 pancreatic cancer deaths, alcohol consumption of 3 or more drinks per day was associated with an increased risk of death from pancreatic cancer in

20

220

PART III:  THE CAUSES OF CANCER

lifelong never smokers (RR  =  1.36; 95% CI:  1.13, 1.62) (Gapstur et  al., 2011). These results were included in a 2012 WCRF/​AICR CUP systematic literature review, which characterized the evidence for a causal association between heavy alcohol consumption and risk of pancreatic cancer as limited (World Cancer Research Fund International, 2012). A causal relationship between heavy alcohol consumption and pancreatic cancer is biologically plausible, since alcohol causes chronic alcoholic pancreatitis (Dufour and Adamson, 2003), an established risk factor for pancreatic cancer. Go et al. (2005) hypothesized that oxidative and non-​oxidative metabolism of alcohol in the pancreas causes inflammation and promotes carcinogenesis through activation of nuclear transcription factors, increased production of reactive oxygen species, and dysregulation of cell proliferation and apoptosis.

Non-​Hodgkin Lymphoma The 2007 IARC Working Group concluded that for non-​Hodgkin lymphoma (NHL), there was evidence for lack of carcinogenicity for alcohol consumption (IARC, 2010). This conclusion is supported by a substantial body of evidence from more recent prospective studies showing statistically significant inverse, dose–​response relationships with all NHL subtypes combined (Allen et al., 2009; Chang et al., 2010; Chiu et al., 1999; Gapstur et al., 2012; Kanda et al., 2010; Klatsky et al., 2009; Lim et al., 2006, 2007; Troy et al., 2010). Notably, these prospective studies show no clear differences in the associations of alcohol consumption with specific histologic subtypes of NHL. The inverse association between alcohol consumption and NHL might be an artifact of reverse causation, in which the disease influences the exposure. Common symptoms of lymphoma include enlarged lymph nodes, extreme fatigue, unexplained fever, weight loss, and night sweats. These symptoms, as well as several chronic diseases associated with lymphoma, might lead to pre-​diagnostic reductions in alcohol consumption, which could result in an apparent inverse association with NHL risk. The possibility of reverse causation is supported by findings from one prospective study that found no difference in risk for current drinking but a higher risk for former drinking when compared to consistent non-​drinkers (Chang et  al., 2010). However, two other prospective studies found no association with former alcohol consumption (Gapstur et al., 2012; Klatsky et al., 2009). Limited experimental data suggest that alcohol consumption might inhibit the development of NHL. Alcohol drinking has multiple effects on the human immune system, both positive and negative (Hagner et  al., 2009; Romeo et  al., 2007). Empirical evidence that alcohol might inhibit the development of human lymphomas comes from a human lymphoma xenograft mouse model, in which chronic exposure to 5%–​10% alcohol in water (an amount shown to lead to 0.05 to 0.15 g/​dL alcohol in circulation) decreased lymphoma growth; this effect was partially due to an inhibition of mTOR signaling (Hagner et  al., 2009). Similarly, ethanol was also shown to inhibit mTOR signaling in another human lymphoma cell line in vitro (Mazan-​Mamczarz et  al., 2015). Multiple lines of evidence support the development of mTOR pathway inhibitors for the treatment of lymphoma (Arita et al., 2013).

Kidney The 2007 IARC Working Group found a lack of evidence that alcohol was carcinogenic for kidney cancer (IARC, 2010). Indeed, there is some evidence that consumption of alcoholic drinks may be inversely associated with kidney cancer. A 2015 WCRF/​AICR CUP systematic literature review showed that a 10 g per day increase of ethanol was associated with an 8% reduction in risk of kidney cancer (RR  =  0.92; 95% CI:  0.86, 0.97); this association was observed up to 30 g per day of ethanol (World Cancer Research Fund International, 2015).

In a large meta-​analysis of 15 prospective studies, moderate, but not heavy, alcohol consumption was associated with a reduced risk of type 2 diabetes (Koppes et al., 2005). This finding might be relevant to the inverse relationship between alcohol consumption and kidney cancer. Type 2 diabetes was associated with higher risk of kidney cancer in a large well-​designed prospective study that updates diabetes status during follow-​up (Joh et al., 2011). Alternatively, alcohol has diuretic effects that could reduce the exposure of renal cells to carcinogens.

ESTIMATES OF ATTRIBUTABLE FRACTIONS Estimates of the percent of cancer deaths attributable to alcohol consumption have been remarkably stable over the past four decades. Rothman et al. estimated that 3% of all cancer deaths in the United States in 1974 could be attributed to alcohol consumption (Rothman, 1980); Doll and Peto estimated that between 2% and 4% of US cancer deaths in the late 1970s could be attributed to alcohol consumption (Doll and Peto, 1981). A recent comprehensive analysis by Nelson and colleagues (Nelson et  al., 2013)  estimated the population attributable fractions (PAFs) of alcohol-​related deaths for all cancers and for seven cancer sites (oral cavity, pharynx, larynx, esophagus, liver, colorectum, and female breast) designated as causally related to alcohol consumption by the IARC (IARC, 2010, 2012). These estimates were based on two different methods to estimate the prevalence of alcohol consumption and on data from two different population-​based surveys (i.e., the 2009 Behavioral Risk Factor Surveillance System survey and the 2009–​2010 National Alcohol Survey). They also considered alcohol sales data. Overall, the different methods showed similar estimates of the attributable fraction ranging from 3.2% to 3.7% (i.e., 18,178 to 21,284) for all US cancer deaths in 2009. In sex-​specific analyses, the proportion of all US cancer deaths due to alcohol ranged from 2.4% (6970) to 4.0% (11,820) among US men, and from 2.7% (7266) to 4.8% (13,094) among women. Rehm and Shield have estimated that worldwide about 4.2% (i.e., 337,400) of all cancer deaths could be attributed to alcohol consumption in 2010 (Rehm and Shield, 2014). Deaths from cancers of the nasopharynx, oropharynx, oral cavity, larynx, and esophagus are at least twice as likely to be alcohol-​related than deaths from cancers of the liver, colorectum, or female breast. For all cancers combined, the estimates of attributable proportion are higher among men (5.4%) than among women (2.7%). This also is true for most individual cancer sites, and especially for UADT cancers. Among men, 23.5% of deaths from laryngeal cancer and 37.0% of deaths from oral cavity cancer were attributed to alcohol consumption, whereas among women only 7.9% of laryngeal cancer deaths and 12.6% of nasopharynx cancer deaths were attributed to alcohol consumption. Rehm and Shield also reported rates of alcohol-​attributed cancer deaths by WHO Global Burden of Disease regions. The highest alcohol-​ attributed cancer death rate (per 100,000) was in Eastern Europe (8.7), followed closely by East Asia (7.5) and Central Europe (7.2). These rates largely affect men in those regions. The lowest alcohol-​attributed cancer death rates (per 100,000) were in North Africa/​Middle East (0.6), Andean Latin America (1.7) and parts of Sub-​Saharan Africa. Interestingly, men in Andean Latin America have lower alcohol-​attributed deaths rates than women (1.4 vs. 2.0, respectively).

ALCOHOL CONSUMPTION AND PROGNOSIS AMONG CANCER SURVIVORS In 2012, there were over 32.6  million people worldwide alive who have lived at least 5 years after being diagnosed with cancer, other than non-​melanoma skin cancer (Jemal et  al., 2014). Many cancer survivors are highly motivated to learn about and make healthful behavioral changes to reduce their risk of recurrence and/​or disease progression. The most recent American Cancer Society Nutrition and Physical Activity Guidelines for Cancer Survivors advise cancer patients

 21



Alcohol and Cancer Risk

undergoing chemotherapy or radiation therapy to avoid or minimize alcohol consumption during treatment, because alcohol might affect chemotherapeutic clearance and worsen toxicity, compromise healing, or exacerbate treatment associated side effects such as oral mucositis (Rock et  al., 2012). However, whether alcohol consumption affects long-​term cancer survival is not completely understood. Issues to be considered include the following: (1) the correlation between alcohol consumption before and after diagnosis; (2)  the stage of disease at diagnosis (because consumption is more likely to influence early-​than late-​stage disease); (3) confounding by smoking (which is highly correlated with alcohol consumption); and (4) assessment of all-​cause versus site-​specific cancer mortality (since alcohol consumption has been associated with a reduced risk of cardiovascular disease mortality). A WCRF/​ AICR CUP systematic literature review of 18 total studies published through June 2012 examined association of both pre-​and post-​diagnostic alcohol consumption with total-​and breast cancer–​specific mortality, as well as risk of second primary breast cancer among breast cancer survivors (World Cancer Research Fund International, 2014b). Results showed no associations of pre-​diagnostic alcohol intake with total mortality (RR = 1.00; 95% CI: 0.99, 1.00 per 1 drink/​week increase), or breast cancer–​specific mortality (RR = 1.00; 95% CI: 0.97, 1.02 per 1 drink/​week). Similarly, there was no evidence of associations between post-​diagnosis alcohol consumption assessed less than 12 months after diagnosis, or 12 months or more after diagnosis and total-​or breast cancer–​specific mortality. No study examined the association between pre-​diagnostic alcohol consumption and risk of second primary breast cancer. In the two studies that examined the association between consumption less than 12 months after diagnosis and subsequent risk of second primary breast cancer, one reported no association and the other reported increased risk (RR = 1.7). In the CUP meta-​analysis, five studies that investigated the association between alcohol consumption during the period 12 or more months after diagnosis and risk of second primary breast cancer showed a nearly statistically significant increased risk (RR = 1.19; 95% CI: 0.96, 1.47); however, the association was not dose-​ related. Since that review, a large pooled analysis including more than 9300 stage I–​III breast cancer survivors also showed no associations of post-​diagnostic regular alcohol intake with risk of recurrence or with total mortality overall (Kwan et al., 2013). However, postmenopausal women who regularly consumed at least 6 g/​day of ethanol had a statistically significant higher risk of recurrence (RR = 1.19; 95% CI: 1.01, 1.40) compared to non-​drinkers; no association was observed among premenopausal women (RR = 0.91; 95% CI: 0.72, 1.16). Overall, there is no clear evidence that alcohol consumption influences breast cancer prognosis. At least nine epidemiologic studies have examined the influence of alcoholic beverage consumption on outcomes after a diagnosis of colorectal cancer (Asghari-​Jafarabadi et al., 2009; Fung et al., 2014; Park et al., 2006; Patel et al., 2014; Pelser et al., 2014; Phipps et al., 2011, 2016; Walter et  al., 2016; Zell et  al., 2007). The results were inconsistent in studies that examined pre-​diagnostic consumption in relation to mortality; some studies found that light-​to-​moderate, but not heavy, alcohol consumption was associated with lower risk of mortality from all causes and colorectal cancer (Pelser et  al., 2014; Phipps et al., 2011, 2016; Walter et al., 2016; Zell et al., 2007). Others studies showed no association (Asghari-​Jafarabadi et al., 2009; Park et al., 2006). In a study of 175 stage I rectal cancer patients, alcohol drinking at the time of diagnosis was associated with a 2.4-​fold higher risk (p = 0.01) of local recurrence compared to never/​former drinkers (Patel et al., 2014). However, in the Nurses’ Health Study there was no association between post-​diagnostic alcohol consumption and colorectal cancer mortality (Fung et al., 2014). In some studies, stage I–​IV colon and rectal cancer patients were included, potentially obscuring an effect on early-​stage cancers. There is a high prevalence of tobacco and heavy alcohol use among head and neck cancer patients (Hashibe et  al., 2009). Research has shown that continued alcohol consumption after diagnosis is associated with higher risk of both second primary cancers and mortality from head and neck cancers. In a study of nearly 1200 patients with early-​stage head and neck squamous cell carcinoma enrolled in a

221

chemoprevention trial, there was a statistically significant 30% (95% CI: 1.0, 1.7) higher risk of second primary cancer among those who continued to drink alcohol after diagnosis (Do et al., 2003). In another chemoprevention trial of 264 patients with early-​stage cancer of the oral cavity, pharynx, and larynx randomized to β-​carotene, there was a dose–​response increase in total mortality for both pre-​diagnostic alcohol consumption and continuous alcohol consumption after diagnosis; the RR of death from any cause during follow-​up was 2.72 (95% CI: 1.20, 6.14) times higher in those who continued to drink after diagnosis compared to non-​drinkers after diagnosis (Mayne et al., 2009).

OPPORTUNITIES FOR ALCOHOL PREVENTION AND CONTROL Strategies to reduce the harmful use of alcoholic beverages include measures to increase awareness and reduce consumption at both the population and individual level. In 2010, the 63rd World Health Assembly of the WHO endorsed a Global Strategy to Reduce the Harmful Use of Alcohol (World Health Organization, 2010). This strategy was designed to support and complement public health policies in WHO member states, to provide guidance for local, regional, and global action, and to set priorities for global action. The five objectives of the WHO Global strategy are shown in Box 12–1. These objectives parallel population-​level interventions to reduce tobacco use; in particular, increases in excise taxes to decreases sales are among the most effective strategies (Eriksen et al., 2015). There is substantial evidence that increasing the price of alcohol through excise taxes reduces both consumption (Wagenaar et  al., 2009)  and adverse health outcomes caused by alcohol, including motor vehicle crashes, violence, and cirrhosis (Elder et al., 2010; Wagenaar et al., 2010). Despite the evidence that alcohol causes multiple types of cancer, fewer than half of the US Comprehensive Cancer Control Plans developed by the states and funded by the US Centers for Disease Control and Prevention (CDC) specify goals, objectives, or strategies for alcohol control (Henley et al., 2014). In particular, efforts are needed to increase awareness of the long-​term impact of alcohol drinking, including binge drinking, on breast cancer risk in women. A  survey of drinking patterns in the United States found that the prevalence of binge drinking increased more than twice as rapidly among women than men from 2005 to 2012 (Dwyer-​Lindgren et al., 2015). Effective population-​ wide education campaigns are needed to target young women who drink alcohol and encourage them to limit consumption and avoid binge drinking. At the individual level, the treatment of alcohol-​related disorders includes mutual support groups, treatments with medications, behavioral therapies, and/​or combinations of these strategies. In the United States, the National Institute on Alcohol Abuse and Alcoholism list Box 12–​1  OBJECTIVES OF THE WHO GLOBAL STRATEGY

1. Raise global awareness of the magnitude and nature of the health, social, and economic problems caused by harmful use of alcohol, and increase commitment by governments to act to address the harmful alcohol use. 2. Strengthen knowledge base on the magnitude and determinants of alcohol-​related harm and on effective interventions to reduce and prevent such harm. 3. Increase technical support to, and enhance capacity of, member states for preventing the harmful use of alcohol and managing alcohol-​use disorders and associated health conditions. 4. Strengthen partnerships and better coordination among stakeholders and increase mobilization of resources required for appropriate and concerted action to prevent the harmful use of alcohol. 5. Improve systems for monitoring and surveillance at different levels, and more effective dissemination and application of information for advocacy, policy development, and evaluation.

2

222

PART III:  THE CAUSES OF CANCER

numerous educational resources for healthcare professionals for use in diagnosing and treating excessive alcohol intake (National Institute on Alcohol Abuse and Alcoholism, 2016). Despite the known adverse effects of alcohol consumption in the etiology of cancer, several organizations and consensus panels (AHA, 2015; AICR, 2011; Kushi et  al., 2012)  have concluded that low to moderate alcohol consumption can be part of a healthful dietary pattern because of the apparent benefits of moderate alcohol consumption on coronary heart disease (Rimm et al., 1999). However, this conclusion was recently challenged by the results of a recent Mendelian randomization study of a non-​synonymous single nucleotide polymorphism in ADH1B in relation to cardiovascular disease risk factors, and risk of coronary heart disease and stroke (Holmes et al., 2014). In that study, carriers of the genetic variant associated with non-​drinking and lower self-​reported alcohol consumption also had lower systolic blood pressure, circulating interleukin-​6 concentration, waist circumference, body mass index, risk of coronary heart disease (OR  =  0.90; 95% CI: 0.84, 0.96), and risk of ischemic stroke (OR = 0.83; 95% CI: 0.72, 0.95). The authors concluded that reductions of alcohol consumption, even for light to moderate drinkers, may be beneficial for cardiovascular health.

FUTURE RESEARCH Extensive evidence demonstrates the causal relationships of ethanol in alcoholic beverages and its primary metabolite, acetaldehyde, to at least seven different cancer sites in humans. More research is needed, however, to better understand the associations with other cancer sites for which the evidence in considered insufficient, for example gastric and lung cancer. For gastric cancer this might require studying the separate and joint effects of Helicobacter pylori infection and alcohol consumption. For lung cancer, large studies of lifelong non-​smokers will be necessary. In addition, a better understanding of the mechanistic basis of alcohol-​related cancers in different target tissues is needed to help clarify the causal relationship, particularly for cancers where the current evidence is considered inconclusive by the IARC. Technologic advancements in whole-​ genome or exome sequencing, which can detect signatures of carcinogens and carcinogenic mechanisms (Alexandrov and Stratton, 2014), will support these research efforts. Similarly, exploring etiologic heterogeneity based on the molecular phenotype of tumors—​beyond traditional markers such as ER status—​will help to clarify the causal role of alcoholic beverage consumption in the etiology of many cancer sites (Ogino et al., 2013). Further research also is needed to examine whether quitting drinking can reverse the increased risk associated with alcohol consumption for alcohol-​related cancers. There is some evidence that quitting drinking appears to reduce the elevated risk of oral cavity and pharynx cancer (IARC, 2010), but less is known for other cancer sites. In addition, testing interventions and messages that discourage drinking, especially binge drinking among young women, could be useful for developing individual and population strategies to reduce the harmful use of alcohol. References AHA. 2015. Alcohol and Heart Health. Available from:  http://​www.heart. org/​HEARTORG/​HealthyLiving/​HealthyEating/​Nutrition/​Alcohol-​and-​ Heart-​Health_​UCM_​305173_​Article.jsp. Accessed July 25, 2016. AICR. 2011. Recommendations for Cancer Prevention. Available from: http://​ www.aicr.org/​reduce-​your-​cancer-​risk/​recommendations-​for-​cancer-​ prevention/​recommendations_​06_​alcohol.html. Accessed July 25, 2016. Alexandrov LB, and Stratton MR. 2014. Mutational signatures: the patterns of somatic mutations hidden in cancer genomes. Curr Opin Genet Dev, 24, 52–​60. PMCID: PMC3990474. Allen NE, Beral V, Casabonne D, et  al. 2009. Moderate alcohol intake and cancer incidence in women. J Natl Cancer Inst, 101(5), 296–​ 305. PMID: 19244173. Ammon E, Schafer C, Hofmann U, and Klotz U. 1996. Disposition and first-​ pass metabolism of ethanol in humans:  is it gastric or hepatic and does it depend on gender? Clin Pharmacol Ther, 59(5), 503–​513. PMID: 8646821.

Arita A, McFarland DC, Myklebust JH, et al. 2013. Signaling pathways in lymphoma: pathogenesis and therapeutic targets. Future Oncol, 9(10), 1549–​ 1571. PMID: 24106904. Asai H, Imaoka S, Kuroki T, Monna T, and Funae Y. 1996. Microsomal ethanol oxidizing system activity by human hepatic cytochrome P450s. J Pharmacol Exp Ther, 277(2), 1004–​1009. PMID: 8627510. Asghari-​Jafarabadi M, Hajizadeh E, Kazemnejad A, and Fatemi SR. 2009. Site-​ specific evaluation of prognostic factors on survival in Iranian colorectal cancer patients: a competing risks survival analysis. Asian Pac J Cancer Prev, 10(5), 815–​821. PMID: 20104971. Boccia S, Hashibe M, Galli P, et al. 2009. Aldehyde dehydrogenase 2 and head and neck cancer:  a meta-​analysis implementing a Mendelian randomization approach. Cancer Epidemiol Biomarkers Prev, 18(1), 248–​254. PMID: 19124505. Boniface S, Kneale J, and Shelton N. 2014. Drinking pattern is more strongly associated with under-​ reporting of alcohol consumption than socio-​ demographic factors: evidence from a mixed-​methods study. BMC Public Health, 14, 1297. PMCID: PMC4320509. Boyd NF, Lockwood GA, Byng JW, Tritchler DL, and Yaffe MJ. 1998. Mammographic densities and breast cancer risk. Cancer Epidemiol Biomarkers Prev, 7(12), 1133–​1144. PMID: 9865433. Brooks PJ, Enoch MA, Goldman D, Li TK, and Yokoyama A. 2009. The alcohol flushing response:  an unrecognized risk factor for esophageal cancer from alcohol consumption. PLoS Med, 6(3), e50. PMCID: PMC2659709. Brooks PJ, and Zakhari S. 2013. Moderate alcohol consumption and breast cancer in women: from epidemiology to mechanisms and interventions. Alcohol Clin Exp Res, 37(1), 23–​30. PMCID: PMC4551426. Cederbaum AI. 2012. Alcohol metabolism. Clin Liver Dis, 16(4), 667–​685. PMCID: PMC3484320. Centers for Disease Control and Prevention. 2016. Frequently asked questions. Available from: http://​www.cdc.gov/​alcohol/​faqs.htm#standDrink. Accessed June 30, 2016. Chang ET, Clarke CA, Canchola AJ, et  al. 2010. Alcohol consumption over time and risk of lymphoid malignancies in the California Teachers Study cohort. Am J Epidemiol, 172(12), 1373–​1383. PMID: 20952595. Chen WY, Rosner B, Hankinson SE, Colditz GA, and Willett WC. 2011. Moderate alcohol consumption during adult life, drinking patterns, and breast cancer risk. JAMA, 306(17), 1884–​ 1890. PMCID: PMC3292347. Chiu BC, Cerhan JR, Gapstur SM, et al. 1999. Alcohol consumption and non-​ Hodgkin lymphoma in a cohort of older women. Br J Cancer, 80(9), 1476–​1482. PMCID: 2363074. Cho E, Smith-​Warner SA, Ritz J, et  al. 2004. Alcohol intake and colorectal cancer:  a pooled analysis of 8 cohort studies. Ann Intern Med, 140(8), 603–​613. PMID: 15096331. Crabb DW, Matsumoto M, Chang D, and You M. 2004. Overview of the role of alcohol dehydrogenase and aldehyde dehydrogenase and their variants in the genesis of alcohol-​related pathology. Proc Nutr Soc, 63(1), 49–​63. PMID: 15099407. Do KA, Johnson MM, Doherty DA, et  al. 2003. Second primary tumors in patients with upper aerodigestive tract cancers: joint effects of smoking and alcohol (United States). Cancer Causes Control, 14(2), 131–​138. PMID: 12749718. Doll R, and Peto R. 1981. The causes of cancer: quantitative estimates of avoidable risks of cancer in the United States today. J Natl Cancer Inst, 66(6), 1191–​1308. PMID: 7017215. Dorgan JF, Baer DJ, Albert PS, et al. 2001. Serum hormones and the alcohol-​ breast cancer association in postmenopausal women. J Natl Cancer Inst, 93(9), 710–​715. PMID: 11333294. Dufour MC, and Adamson MD. 2003. The epidemiology of alcohol-​induced pancreatitis. Pancreas, 27(4), 286–​290. PMID: 14576488. Dwyer-​Lindgren L, Flaxman AD, Ng M, et al. 2015. Drinking patterns in US counties from 2002 to 2012. Am J Public Health, 105(6), 1120–​1127. PMID: 25905846. El-​Serag HB, and Rudolph KL. 2007. Hepatocellular carcinoma:  epidemiology and molecular carcinogenesis. Gastroenterology, 132(7), 2557–​2576. PMID: 17570226. Elder RW, Lawrence B, Ferguson A, et al. 2010. The effectiveness of tax policy interventions for reducing excessive alcohol consumption and related harms. Am J Prev Med, 38(2), 217–​229. PMCID: PMC3735171. Collaborative Group on Hormonal Factors in Breast Cancer. 2002. Alcohol, tobacco and breast cancer—​collaborative reanalysis of individual data from 53 epidemiological studies, including 58 515 women with breast cancer and 95 067 women without the disease. Br J Cancer, 87, 1234–​1245. Collaborative Group on Hormonal Factors in Breast Cancer. 2011. Circulating sex hormones and breast cancer risk factors in postmenopausal

 23



Alcohol and Cancer Risk

women:  reanalysis of 13 studies. Br J Cancer, 105(5), 709–​ 722. PMCID: PMC3188939. Collaborative Group on Hormonal Factors in Breast Cancer. 2013. Sex hormones and risk of breast cancer in premenopausal women: a collaborative reanalysis of individual participant data from seven prospective studies. Lancet Oncol, 14(10), 1009–​1019. PMCID: PMC4056766. Eriksen M, Mackay J, Schluger N, Gomeshtapeh FI, and Drop J. 2015. The Tobacco Atlas. 5th ed. Atlanta: American Cancer Society. Ferrari P, Jenab M, Norat T, et al. 2007. Lifetime and baseline alcohol intake and risk of colon and rectal cancers in the European prospective investigation into cancer and nutrition (EPIC). Int J Cancer, 121(9), 2065–​2072. PMID: 17640039. Frank A, Seitz HK, Bartsch H, Frank N, and Nair J. 2004. Immunohistochemical detection of 1,N6-​ethenodeoxyadenosine in nuclei of human liver affected by diseases predisposing to hepato-​ carcinogenesis. Carcinogenesis, 25(6), 1027–​1031. PMID: 14742317. Friberg E, Orsini N, Mantzoros CS, and Wolk A. 2010. Alcohol intake and endometrial cancer risk:  a meta-​analysis of prospective studies. Br J Cancer, 103(1), 127–​131. PMCID: PMC2905297. Fung TT, Kashambwa R, Sato K, et  al. 2014. Post diagnosis diet quality and colorectal cancer survival in women. PLoS One, 9(12), e115377. PMCID: PMC4266679. Gapstur SM, Diver WR, McCullough ML, et  al. 2012. Alcohol intake and the incidence of non-​hodgkin lymphoid neoplasms in the cancer prevention study II nutrition cohort. Am J Epidemiol, 176(1), 60–​ 69. PMID: 22562662. Gapstur SM, Jacobs EJ, Deka A, et al. 2011. Association of alcohol intake with pancreatic cancer mortality in never smokers. Arch Intern Med, 171(5), 444–​451. PMID: 21403041. Gaudet MM, Carter BD, Brinton LA, et  al. 2016. Pooled analysis of active cigarette smoking and invasive breast cancer risk in 14 cohort studies. Int J Epidemiol, 2016; doi: 10.1093/​ije/​dyw288. Genkinger JM, Hunter DJ, Spiegelman D, et al. 2006. Alcohol intake and ovarian cancer risk: a pooled analysis of 10 cohort studies. Br J Cancer, 94(5), 757–​762. PMCID: PMC2361197. Giovannucci E, Colditz G, Stampfer MJ, et al. 1991. The assessment of alcohol consumption by a simple self-​ administered questionnaire. Am J Epidemiol, 133(8), 810–​817. PMID: 2021148. Go VL, Gukovskaya A, and Pandol SJ. 2005. Alcohol and pancreatic cancer. Alcohol, 35(3), 205–​211. PMID: 16054982. Goedde HW, and Agarwal DP. 1987. Polymorphism of aldehyde dehydrogenase and alcohol sensitivity. Enzyme, 37(1–​2), 29–​44. PMID: 3106030. Hagner PR, Mazan-​Mamczarz K, Dai B, et al. 2009. Alcohol consumption and decreased risk of non-​Hodgkin lymphoma:  role of mTOR dysfunction. Blood, 113(22), 5526–​5535. PMID: 19293424. Hamajima N, Hirose K, Tajima K, et al. 2002. Alcohol, tobacco and breast cancer-​ -​collaborative reanalysis of individual data from 53 epidemiological studies, including 58,515 women with breast cancer and 95,067 women without the disease. Br J Cancer, 87(11), 1234–​1245. PMCID: PMC2562507. Hashibe M, Brennan P, Benhamou S, et  al. 2007. Alcohol drinking in never users of tobacco, cigarette smoking in never drinkers, and the risk of head and neck cancer:  pooled analysis in the International Head and Neck Cancer Epidemiology Consortium. J Natl Cancer Inst, 99(10), 777–​789. PMID: 17505073. Hashibe M, Brennan P, Chuang SC, et  al. 2009. Interaction between tobacco and alcohol use and the risk of head and neck cancer: pooled analysis in the International Head and Neck Cancer Epidemiology Consortium. Cancer Epidemiol Biomarkers Prev, 18(2), 541–​ 550. PMCID: PMC3051410. Hassan MM, Hwang LY, Hatten CJ, et al. 2002. Risk factors for hepatocellular carcinoma:  synergism of alcohol with viral hepatitis and diabetes mellitus. Hepatology, 36(5), 1206–​1213. PMID: 12395331. Henley SJ, Kanny D, Roland KB, et al. 2014. Alcohol control efforts in comprehensive cancer control plans and alcohol use among adults in the USA. Alcohol Alcohol, 49(6), 661–​667. PMCID: PMC4380007. Holmes MV, Dale CE, Zuccolo L, et al. 2014. Association between alcohol and cardiovascular disease: Mendelian randomisation analysis based on individual participant data. BMJ, 349, g4164. PMCID: PMC4091648. Homann N, Tillonen J, Meurman JH, et  al. 2000. Increased salivary acetaldehyde levels in heavy drinkers and smokers:  a microbiological approach to oral cavity cancer. Carcinogenesis, 21(4), 663–​668. PMID: 10753201. IARC. 1988. Alcohol Drinking. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans (Vol. 44, pp. 15–​378). Lyon: IARC. IARC. 2010. Alcohol Consumption and Ethyl Carbamate. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans (Vol. 96, pp. 7–​ 1280). Lyon: IARC.

223

IARC. 2012. Consumption of Alcoholic Beverages. IARC Working Group (Ed.). IARC Monographs on the Evaluation of Carcinogenic Risks to Humans (Vol. 100e, pp. 373–​499). Lyon: IARC. Jemal A, Vineis P, Bray F, Torre L, and Forman D. 2014. The Cancer Atlas. 2nd ed. Atlanta: American Cancer Society. Joh HK, Willett WC, and Cho E. 2011. Type 2 diabetes and the risk of renal cell cancer in women. Diabetes Care, 34(7), 1552–​ 1556. PMCID: PMC3120193. Jokelainen K, Matysiak-​Budnik T, Makisalo H, Hockerstedt K, and Salaspuro M. 1996. High intracolonic acetaldehyde values produced by a bacteriocolonic pathway for ethanol oxidation in piglets. Gut, 39(1), 100–​104. PMCID: PMC1383240. Kaaks R, and Riboli E. 1997. Validation and calibration of dietary intake measurements in the EPIC project: methodological considerations. European Prospective Investigation into Cancer and Nutrition. Int J Epidemiol, 26(Suppl 1), S15–​S25. PMID: 9126530. Kanda J, Matsuo K, Inoue M, et al. 2010. Association of alcohol intake with the risk of malignant lymphoma and plasma cell myeloma in Japanese: a population-​ based cohort study (Japan Public Health Center-​ based Prospective Study). Cancer Epidemiol Biomarkers Prev, 19(2), 429–​434. PMID: 20086115. Key T, Appleby P, Barnes I, et al. 2002. Endogenous sex hormones and breast cancer in postmenopausal women: reanalysis of nine prospective studies. J Natl Cancer Inst, 94(8), 606–​616. PMID: 11959894. Klatsky AL, Li Y, Baer D, et al. 2009. Alcohol consumption and risk of hematologic malignancies. Ann Epidemiol, 19(10), 746–​753. PMID: 19394864. Koppaka V, Thompson DC, Chen Y, et  al. 2012. Aldehyde dehydrogenase inhibitors:  a comprehensive review of the pharmacology, mechanism of action, substrate specificity, and clinical application. Pharmacol Rev, 64(3), 520–​539. PMCID: PMC3400832. Koppes LL, Dekker JM, Hendriks HF, Bouter LM, and Heine RJ. 2005. Moderate alcohol consumption lowers the risk of type 2 diabetes: a meta-​ analysis of prospective observational studies. Diabetes Care, 28(3), 719–​ 725. PMID: 15735217. Kuper H, Tzonou A, Kaklamani E, et al. 2000. Tobacco smoking, alcohol consumption and their interaction in the causation of hepatocellular carcinoma. Int J Cancer, 85(4), 498–​502. PMID: 10699921. Kushi LH, Doyle C, McCullough M, et  al. 2012. American Cancer Society Guidelines on nutrition and physical activity for cancer prevention: reducing the risk of cancer with healthy food choices and physical activity. CA Cancer J Clin, 62(1), 30–​67. PMID: 22237782. Kwan ML, Chen WY, Flatt SW, et  al. 2013. Postdiagnosis alcohol consumption and breast cancer prognosis in the after breast cancer pooling project. Cancer Epidemiol Biomarkers Prev, 22(1), 32–​ 41. PMCID: PMC3538884. Lachenmeier DW, Przybylski MC, and Rehm J. 2012. Comparative risk assessment of carcinogens in alcoholic beverages using the margin of exposure approach. Int J Cancer, 131(6), E995–E1003. PMID: 22447328. Lewis SJ, and Smith GD. 2005. Alcohol, ALDH2, and esophageal cancer:  a meta-​ analysis which illustrates the potentials and limitations of a Mendelian randomization approach. Cancer Epidemiol Biomarkers Prev, 14(8), 1967–​1971. PMID: 16103445. Li CI, Malone KE, Daling JR, et al. 2008. Timing of menarche and first full-​ term birth in relation to breast cancer risk. Am J Epidemiol, 167(2), 230–​ 239. PMCID: PMC3804121. Lim U, Morton LM, Subar AF, et al. 2007. Alcohol, smoking, and body size in relation to incident Hodgkin’s and non-​Hodgkin’s lymphoma risk. Am J Epidemiol, 166(6), 697–​708. PMID: 17596266. Lim U, Weinstein S, Albanes D, et  al. 2006. Dietary factors of one-​carbon metabolism in relation to non-​Hodgkin lymphoma and multiple myeloma in a cohort of male smokers. Cancer Epidemiol Biomarkers Prev, 15(6), 1109–​1114. PMID: 16775167. Linhart K, Bartsch H, and Seitz HK. 2014. The role of reactive oxygen species (ROS) and cytochrome P-​450 2E1 in the generation of carcinogenic etheno-​DNA adducts. Redox Biol, 3, 56–​62. PMCID: PMC4297928. Liu Y, Colditz GA, Rosner B, et al. 2013. Alcohol intake between menarche and first pregnancy: a prospective study of breast cancer risk. J Natl Cancer Inst, 105(20), 1571–​1578. PMCID: PMC3797023. Liu Y, Nguyen N, and Colditz GA. 2015. Links between alcohol consumption and breast cancer: a look at the evidence. Womens Health (Lond Engl), 11(1), 65–​77. PMCID: PMC4299758. Loomba R, Yang HI, Su J, et al. 2013. Synergism between obesity and alcohol in increasing the risk of hepatocellular carcinoma: a prospective cohort study. Am J Epidemiol, 177(4), 333–​342. PMCID: 3626057. Loomba R, Yang HI, Su J, et al. 2010. Obesity and alcohol synergize to increase the risk of incident hepatocellular carcinoma in men. Clin Gastroenterol Hepatol, 8(10), 891–​898. PMID: 20621202.

24

224

PART III:  THE CAUSES OF CANCER

Marrero JA, Fontana RJ, Fu S, et al. 2005. Alcohol, tobacco and obesity are synergistic risk factors for hepatocellular carcinoma. J Hepatol, 42(2), 218–​224. PMID: 15664247. Mayne ST, Cartmel B, Kirsh V, and Goodwin WJ, Jr. 2009. Alcohol and tobacco use prediagnosis and postdiagnosis, and survival in a cohort of patients with early stage cancers of the oral cavity, pharynx, and larynx. Cancer Epidemiol Biomarkers Prev, 18(12), 3368–​3374. PMCID: PMC2789339. Mazan-​Mamczarz K, Peroutka RJ, Steinhardt JJ, et al. 2015. Distinct inhibitory effects on mTOR signaling by ethanol and INK128 in diffuse large B-​cell lymphoma. Cell Commun Signal, 13, 15. PMCID: PMC4350884. McGovern PE, Zhang J, Tang J, et al. 2004. Fermented beverages of pre-​and proto-​historic China. Proc Natl Acad Sci U S A, 101(51), 17593–​17598. PMCID: PMC539767. Morch LS, Johansen D, Thygesen LC, et al. 2007. Alcohol drinking, consumption patterns and breast cancer among Danish nurses: a cohort study. Eur J Public Health, 17(6), 624–​629. PMID: 17442702. Moskal A, Norat T, Ferrari P, and Riboli E. 2007. Alcohol intake and colorectal cancer risk:  a dose-​response meta-​analysis of published cohort studies. Int J Cancer, 120(3), 664–​671. PMID: 17096321. National Institute on Alcohol Abuse and Alcoholism. 2016. Professional Education Materials. Available at https://​www.niaaa.nih.gov/​publications/​clinical-​guides-​and-​manuals. Accessed July 21, 2016. Nelson DE, Jarman DW, Rehm J, et  al. 2013. Alcohol-​attributable cancer deaths and years of potential life lost in the United States. Am J Public Health, 103(4), 641–​648. PMCID: PMC3673233. Ogino S, Lochhead P, Chan AT, et al. 2013. Molecular pathological epidemiology of epigenetics: emerging integrative science to analyze environment, host, and disease. Mod Pathol, 26(4), 465–​484. PMCID: PMC3637979. Pandya GA, and Moriya M. 1996. 1,N6-​ethenodeoxyadenosine, a DNA adduct highly mutagenic in mammalian cells. Biochemistry, 35(35), 11487–​ 11492. PMID: 8784204. Park SM, Lim MK, Shin SA, and Yun YH. 2006. Impact of prediagnosis smoking, alcohol, obesity, and insulin resistance on survival in male cancer patients:  National Health Insurance Corporation Study. J Clin Oncol, 24(31), 5017–​5024. PMID: 17075121. Patel SA, Chen YH, Hornick JL, et al. 2014. Early-​stage rectal cancer: clinical and pathologic prognostic markers of time to local recurrence and overall survival after resection. Dis Colon Rectum, 57(4), 449–​ 459. PMCID: PMC3954982. Pelser C, Arem H, Pfeiffer RM, et al. 2014. Prediagnostic lifestyle factors and survival after colon and rectal cancer diagnosis in the National Institutes of Health (NIH)-​AARP Diet and Health Study. Cancer, 120(10), 1540–​ 1547. PMCID: PMC4151292. Phipps AI, Baron J, and Newcomb PA. 2011. Prediagnostic smoking history, alcohol consumption, and colorectal cancer survival:  the Seattle Colon Cancer Family Registry. Cancer, 117(21), 4948–​4957. PMCID: PMC3138819. Phipps AI, Shi Q, Limburg PJ, et  al. 2016. Alcohol consumption and colon cancer prognosis among participants in north central cancer treatment group phase III trial N0147. Int J Cancer, 139(5), 986–​ 995. PMCID: PMC4911257. Rehm J, and Shield KD. 2014. Alcohol consumption. Steward B, and Wild C (Eds.). World Cancer Report 2014. Lyon: IARC. Reichman ME, Judd JT, Longcope C, et al. 1993. Effects of alcohol consumption on plasma and urinary hormone concentrations in premenopausal women. J Natl Cancer Inst, 85(9), 722–​727. PMID: 8478958. Rimm EB, Williams P, Fosher K, Criqui M, and Stampfer MJ. 1999. Moderate alcohol intake and lower risk of coronary heart disease: meta-​analysis of effects on lipids and haemostatic factors. BMJ, 319(7224), 1523–​1528. PMCID: PMC28294. Rock CL, Doyle C, Demark-​Wahnefried W, et al. 2012. Nutrition and physical activity guidelines for cancer survivors. CA Cancer J Clin, 62(4), 243–​ 274. PMID: 22539238. Romeo J, Warnberg J, Nova E, et  al. 2007. Moderate alcohol consumption and the immune system: a review. Br J Nutr, 98(Suppl 1), S111–​S115. PMID: 17922947. Rosner B. 1995. Fundamental of Biostatistics. 4th ed. Belmont, CA: Duxbury Press. Rothman KJ. 1980. The proportion of cancer attributable to alcohol consumption. Prev Med, 9(2), 174–​179. PMID: 7383982. Salaspuro MP, and Lieber CS. 1978. Non-​uniformity of blood ethanol elimination:  its exaggeration after chronic consumption. Ann Clin Res, 10(5), 294–​297. PMID: 104651. Seitz HK, Maurer B, and Stickel F. 2005. Alcohol consumption and cancer of the gastrointestinal tract. Dig Dis, 23(3-​4), 297–​303. PMID: 16508294. Seitz HK, and Stickel F. 2007. Molecular mechanisms of alcohol-​mediated carcinogenesis. Nat Rev Cancer, 7(8), 599–​612. PMID: 17646865.

Seitz HK, and Stickel F. 2010. Acetaldehyde as an underestimated risk factor for cancer development:  role of genetics in ethanol metabolism. Genes Nutr, 5(2), 121–​128. PMCID: PMC2885165. Shield KD, Gmel G, Kehoe-​Chan T, et al. 2013. Mortality and potential years of life lost attributable to alcohol consumption by race and sex in the United States in 2005. PLoS One, 8(1), e51923. PMCID: PMC3534703. Son J, Pang B, McFaline JL, Taghizadeh K, and Dedon PC. 2008. Surveying the damage:  the challenges of developing nucleic acid biomarkers of inflammation. Mol Biosyst, 4(9), 902–​908. PMID: 18704228. Squier CA, Cox P, and Hall BK. 1986. Enhanced penetration of nitrosonornicotine across oral mucosa in the presence of ethanol. J Oral Pathol, 15(5), 276–​279. PMID: 3091795. Stockwell T, Donath S, Cooper-​Stanbury M, et  al. 2004. Under-​reporting of alcohol consumption in household surveys:  a comparison of quantity-​ frequency, graduated-​ frequency and recent recall. Addiction, 99(8), 1024–​1033. PMID: 15265099. Sutton CW, Rustogi N, Gurkan C, et al. 2010. Quantitative proteomic profiling of matched normal and tumor breast tissues. J Proteome Res, 9(8), 3891–​ 3902. PMID: 20560667. Suzuki R, Orsini N, Mignone L, Saji S, and Wolk A. 2008. Alcohol intake and risk of breast cancer defined by estrogen and progesterone receptor status: a meta-​analysis of epidemiological studies. Int J Cancer, 122(8), 1832–​1841. PMID: 18067133. Theruvathu JA, Jaruga P, Nath RG, Dizdaroglu M, and Brooks PJ. 2005. Polyamines stimulate the formation of mutagenic 1,N2-​ propanodeoxyguanosine adducts from acetaldehyde. Nucleic Acids Res, 33(11), 3513–​3520. PMCID: 1156964. Triano EA, Slusher LB, Atkins TA, et  al. 2003. Class  I  alcohol dehydrogenase is highly expressed in normal human mammary epithelium but not in invasive breast cancer: implications for breast carcinogenesis. Cancer Res, 63(12), 3092–​3100. PMID: 12810634. Troy JD, Hartge P, Weissfeld JL, et al. 2010. Associations between anthropometry, cigarette smoking, alcohol consumption, and non-​Hodgkin lymphoma in the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial. Am J Epidemiol, 171(12), 1270–​1281. PMID: 20494998. Tsong WH, Koh WP, Yuan JM, et al. 2007. Cigarettes and alcohol in relation to colorectal cancer: the Singapore Chinese Health Study. Br J Cancer, 96(5), 821–​827. PMCID: PMC2360085. Turati F, Galeone C, Rota M, et al. 2014. Alcohol and liver cancer: a systematic review and meta-​analysis of prospective studies. Ann Oncol, 25(8), 1526–​ 1535. PMID: 24631946. Visapaa JP, Jokelainen K, Nosova T, and Salaspuro M. 1998. Inhibition of intracolonic acetaldehyde production and alcoholic fermentation in rats by ciprofloxacin. Alcohol Clin Exp Res, 22(5), 1161–​ 1164. PMID: 9726290. Wagenaar AC, Salois MJ, and Komro KA. 2009. Effects of beverage alcohol price and tax levels on drinking: a meta-​analysis of 1003 estimates from 112 studies. Addiction, 104(2), 179–​190. PMID: 19149811. Wagenaar AC, Tobler AL, and Komro KA. 2010. Effects of alcohol tax and price policies on morbidity and mortality:  a systematic review. Am J Public Health, 100(11), 2270–​2278. PMCID: PMC2951962. Walter V, Jansen L, Ulrich A, et al. 2016. Alcohol consumption and survival of colorectal cancer patients: a population-​based study from Germany. Am J Clin Nutr, 103(6), 1497–​1506. PMID: 27146651. Wang HJ, Gao B, Zakhari S, and Nagy LE. 2012. Inflammation in alcoholic liver disease. Annu Rev Nutr, 32, 343–​368. PMCID: PMC3670145. Wang Y, Millonig G, Nair J, et al. 2009. Ethanol-​induced cytochrome P4502E1 causes carcinogenic etheno-​ DNA lesions in alcoholic liver disease. Hepatology, 50(2), 453–​461. PMID: 19489076. World Cancer Research Fund International. 2011. Food, Nutrition, Physical Activity, and the Prevention of Colorectal Cancer. London: WCRF. World Cancer Research Fund International. 2012. Food, Nutrition, Physical Activity, and the Prevention of Pancreatic Cancer. London: WCRF. World Cancer Research Fund International. 2013. Food, Nutrition, Physical Activity, and the Prevention of Endometrial Cancer. London: WCRF. World Cancer Research Fund International. 2014a. Food, Nutrition, Physical Activity, and the Prevention of Ovarian Cancer. London: WCRF. World Cancer Research Fund International. 2014b. Systematic Review on Diet, Nutrition, Physical Activity and Survival and Second Cancers in Breast Cancer Survivors. Continuous Update Project. London: Amperial College. World Cancer Research Fund International. 2015. Food, Nutrition, Physical Activity, and the Prevention of Kidney Cancer. London: WCRF. World Health Organization. 2010. Global Strategy to Reduce the Harmful Use of Alcohol. Geneva: WHO. World Health Organization. 2014. Global Status Report on Alcohol and Health 2014. World Health Organization (Ed.). Geneva: WHO.

 25



Alcohol and Cancer Risk

Woutersen RA, Appelman LM, Van Garderen-​Hoetmer A, and Feron VJ. 1986. Inhalation toxicity of acetaldehyde in rats. III. Carcinogenicity study. Toxicology, 41(2), 213–​231. PMID: 3764943. Yokoyama A, Muramatsu T, Ohmori T, et al. 1996. Esophageal cancer and aldehyde dehydrogenase-2 genotypes in Japanese males. Cancer Epidemiol Biomarkers Prev, 5(2), 99–102. PMID: 880269. Yokoyama A, Omori T, Yokoyama T et al. 2006. Risk of squamous cell carcinoma of the upper aerodigestive tract in cancer-​free alcoholic Japanese men: an endoscopic follow-​up study. Cancer Epidemiol Biomarkers Prev, 15(11), 2209–​2215. PMID: 17119048.

225

Yuan JM, Govindarajan S, Arakawa K, and Yu MC. 2004. Synergism of alcohol, diabetes, and viral hepatitis on the risk of hepatocellular carcinoma in blacks and whites in the U.S. Cancer, 101(5), 1009–​1017. PMID: 15329910. Zeka A, Gore R, and Kriebel D. 2003. Effects of alcohol and tobacco on aerodigestive cancer risks: a meta-​regression analysis. Cancer Causes Control, 14(9), 897–​906. PMID: 14682447. Zell JA, McEligot AJ, Ziogas A, Holcombe RF, and Anton-​Culver H. 2007. Differential effects of wine consumption on colorectal cancer outcomes based on family history of the disease. Nutr Cancer, 59(1), 36–​ 45. PMID: 17927500.

26

 27

13

Ionizing Radiation AMY BERRINGTON DE GONZÁLEZ, ANDRÉ BOUVILLE, PREETHA RAJARAMAN, AND MARY SCHUBAUER-​BERIGAN

OVERVIEW Ionizing radiation is classified as a universal carcinogen due to its ability to induce cancer in most organs following exposure at any age, including in utero. Several organs are especially radiosensitive, particularly when exposure occurs in childhood, including the female breast, thyroid, brain, and red bone marrow. Very few cancers, notably cervical and Hodgkin lymphoma, do not seem to be related to ionizing radiation, but the reasons are unknown. For most cancers (lung may be an exception) the relative risk decreases with age at exposure or attained age, and time since exposure. Currently the main sources of radiation exposure to the general population are from natural background radiation (e.g., residential radon) and medical (e.g., computed tomography [CT] scans). Natural background exposure varies by location but is generally stable over time. Medical exposure has been increasing in many countries due to the expansion of advanced imaging technologies. Risk projection models suggest that currently 1%–​3% of cancers could be related to medical radiation exposure in countries like the United Kingdom and the United States, and about 1% of cancers could be related to natural background radiation. Occupational radiation exposures have decreased over the last several decades. One exception is the medical workers who are exposed to radiation from performing increasing numbers of interventional fluoroscopy and nuclear medicine procedures. Historically there have been important exposures to large populations due to nuclear weapons detonation and testing and nuclear accidents, particularly Chernobyl. Although ionizing radiation is an established carcinogen, the magnitude of the risk from very low-​dose acute (< 50 mGy) and protracted exposures (< 5 mGy per hour) is still uncertain, and difficult to resolve via epidemiological studies. In the last decade there have been new studies based on large-​scale electronic record linkages of pediatric CT scans, natural background radiation, environmental exposures, and nuclear workers that have provided additional evidence supporting excess cancer risks from low-​dose and/​or low-​dose rate exposures. Developments in molecular epidemiology, particularly tumor sequencing to find a potential “radiation signature,” could also be a novel approach to address this important public health question in the next decade.

INTRODUCTION Most ionizing radiation exposures to the general population today are from very low doses such as CT scans, or residential radon exposure. The so-​called linear no-​threshold (LNT) assumption forms the basis for radiation protection standards for these very low doses—​the assumption being that there is no safe level of radiation exposure (i.e., no threshold) and that the cancer risks from very low doses can be estimated by linear extrapolation of observed risks from studies of populations exposed to higher doses (primarily the atomic bomb survivors). This model has been challenged, however, by both experimental and some epidemiological data, and remains somewhat contentious. Much of the epidemiological research in the last decade

has focused on the LNT question, primarily assessing whether there are excess cancer risks in populations who received very low doses, and quantifying the magnitude of the risk. Another outstanding issue is whether the risk from protracted exposure to ionizing radiation is lower than the risk from the same dose received acutely, due to a greater probability of DNA repair. This is particularly relevant for radiation protection standards for occupational groups, as their radiation exposure is often received over many decades. As cancer survival continues to improve, there has been an expansion of effort to improve and understand the late effects of cancer treatment. Radiotherapy results in exposure to very high (> 5Gy) fractionated doses to the healthy organs surrounding the target organ, and this is associated with an increased risk of developing a second cancer. To balance these potential risks against the benefits of radiotherapy, we need to have good quantitative risk estimates of cancer risks from high doses. Because of the wealth of epidemiological data on the cancer risks from moderate doses, the established carcinogenicity, and the ability to measure biological dose, epidemiological studies of ionizing radiation provide a rich setting for studying interactions with other carcinogens, including gene–​ environment interactions, or genetic susceptibility. These investigations have also been a focus of recent research efforts. The advent of tumor sequencing has provided an exciting opportunity to search for potential “radiation signatures” in populations with very well characterized radiation exposures and high attributable risks, such as the thyroid cancers diagnosed after childhood exposure to iodine-​ 131 (I-​131) from the Chernobyl accident. As there are many types of ionizing radiation exposure, modes of exposure, methods for measuring it, and several different dose units, we start with a review of these basic concepts before reviewing the new epidemiological studies.

NATURE OF THE EXPOSURE Types of Ionizing Radiation Ionizing radiation consists of waves or particles (gamma, X-​rays, neutrons, alpha particles, etc.), usually produced during the radioactive decay of unstable nuclei, that have sufficient energy to ionize atoms in the human body, thus inducing chemical changes that may be biologically important for the functioning of cells. The various forms of radiation are emitted with different energies and penetrating power through materials and potentially different carcinogenic potential. The most common types of ionizing radiation are the following (each of which can cause cancer): • Alpha particles, consisting of helium nuclei, which can be halted by a sheet of paper and thus can hardly penetrate the dead outer layers of the skin; • Beta particles, consisting of electrons, which can penetrate up to 2 cm of soft tissue; and • Gamma radiation, consisting of photons, which traverse the entire human body.

227

28

228

PART III:  THE CAUSES OF CANCER

Modes and Patterns of Irradiation There are many different modes of irradiation that can impact cancer risk and are therefore important to document and take into account in epidemiological studies. The variety of well-​documented exposure patterns provides an important opportunity to evaluate critical exposure windows for carcinogenesis.

Internal Versus External Exposure

External exposures occur when individuals are in proximity to a photon-​ emitting radiation source. Internal exposure occurs from intake of radionuclides, usually by inhalation or ingestion, or sometimes via absorption through intact or damaged skin or from injection for medical reasons.

Acute, Fractionated, or Protracted Exposure

External exposure stops as soon as the source is turned off or removed, or when the individuals move away from the radiation source. External exposures, therefore, can be acute (i.e., delivered within seconds or less), fractionated (consisting of a series of acute or short-​term exposures delivered at relatively long time intervals), or protracted (i.e., occurring over a long time at a relatively constant rate). In contrast, internal exposures, which result from the incorporation of a radionuclide, last as long as the radionuclide has not completely decayed or been excreted from the body, depending on the physical half-​life and biokinetics of the incorporated radionuclide. Internal exposures, therefore, cannot be acute.

Full-​Versus Partial-​Body Exposures

External exposures are usually due to photons, which can irradiate the entire body in a relatively uniform manner, referred to as full-​body exposure; for example the Japanese atomic bombs resulted in relatively uniform full-​body exposure. Medical exposures to external sources, such as diagnostic X-​rays, however, only irradiate the organs in the X-​ ray field and are referred to as partial-​body exposures. Internal exposures can result in the irradiation of a specific organ of the body (for example, the thyroid following exposure to I-​131, with other organs receiving some exposure, e.g., stomach) or in the uniform irradiation of the entire body (e.g., from exposure to cesium-​137).

Units of Exposure The principal radiation quantities used to express dose are presented in Table 13–​1. The most important quantity of radiation dosimetry that is used in epidemiologic studies is the absorbed dose, which is defined as the radiation energy absorbed per unit mass of organ or tissue. The unit of absorbed dose is the gray (Gy), with 1 Gy equal to 1 joule per kg. The absorbed dose is a physical quantity that does not take the type of particle or the radiosensitivity of the organ or tissue into consideration. Different types of radiation have different biological effectiveness because they transfer their energy in different ways that can cause or less more damage. The equivalent dose takes this into account by weighting the absorbed dose in an organ or tissue by a radiation weighting factor, which reflects the relative biological effectiveness (RBE) of the radiation type. The radiation weighting factors that are currently recommended by the International Commission on

Table 13–​1.  Principal Radiation Quantities Used to Express Dose Quantity

Unit

Absorbed dose

Gray (Gy)

Equivalent dose

Sievert (Sv)

Effective dose

Sievert (Sv)

Comment Does not account for the type of radiation exposure Accounts for biological effectiveness of different types of radiation Summary dose often used for partial-​ body exposures that accounts for different radiosensitivity of organs and biological effectiveness

Radiological Protection (ICRP) are 20 for alpha particles and 1 for electrons and photons (ICRP, 2007). These values are based primarily on in vitro experiments, and hence contain a number of uncertainties, especially in their transfer to humans. The unit of equivalent dose is the sievert (Sv). For regulatory purposes, the primary dosimetry quantity is the effective dose, which is obtained as the sum, including all radiosensitive organs and tissues of the body, of the equivalent doses weighted by a tissue-​weighting factor, reflecting the radiosensitivity of the tissue or organ. The current values of the tissue-​weighting factors that are recommended by the ICRP are 0.12 for red bone marrow, lung, stomach, and breasts; 0.08 for gonads; 0.04 for bladder, liver, esophagus, and thyroid; 0.01 for skin, bone surface, salivary glands, and brain; 0.12 for the remainder of body (ICRP, 2007). Because the effective dose takes into account the absorbed doses received by all radiosensitive organs and tissues of the body, weighted according to their radiosensitivity and to the radiation quality, it can be considered to be representative of the radiation impact caused by exposure to a given source of radiation. The effective dose is a regulatory quantity that is directly amenable to the comparison of the radiation impacts of widely different sources of exposure; its unit is also the sievert (Sv). Generally, it is not appropriate to use estimates of effective dose in epidemiological studies.

SOURCES OF EXPOSURE There are three broad sources of ionizing radiation exposure: environmental, medical, and occupational. Currently the general population is primarily exposed to very low doses of radiation from environmental sources from natural background radiation, and medical exposures from diagnostic tests like CT scans (Figure 13–​1). There are also an estimated 23 million radiation workers in the world who receive occupational exposures (UNSCEAR, 2010b). Historically there have been a number of important events that resulted in much higher doses of radiation exposure to the general population and workers, including the atomic bombs dropped in Japan, the Chernobyl accident, and uses of high doses of radiation for the treatment of benign medical conditions, such as ringworm (Tinea capitis). The epidemiological literature includes populations exposed to current sources of exposure as well as these historical sources, and therefore we review both here.

Environmental Exposures Natural Environmental Exposures

Natural background radiation exposure makes up at least half of all radiation exposure to the general population in most countries (Figure 13–​1). Exposure comes from radon, cosmic, terrestital gamma, and internal sources (United Nations Scientific Committee on the Effects of Atomic Radiation [UNSCEAR], 2010b) (Table 13–​ 2). Radon, including thoron, is generally the largest contributor (about 50%, on average, of the world population of natural background exposure) but is also highly variable, depending on where people live and the construction of their houses. There are several documented areas of much higher natural background radiation in Brazil, India, Iran, and China, where annual doses can exceed 10 mSv per year (UNSCEAR, 2010b). Cosmic rays that reach the Earth’s surface originate either from outer space (galactic and extra-​galactic cosmic rays) or from the sun (solar wind) and cause exposures predominantly via external irradiation. Cosmic radiation exposure to the general population varies with latitude and altitude, and represents about 20% of the natural background exposure. Enhanced exposures are due to long-​haul air travel, which is at high altitude; a round-​trip flight from London to Sydney results in an estimated effective dose of 0.16 mSv. Naturally occurring radionuclides in the earth result in effective doses of external gamma radiation that represent about 20% of the natural background exposure. Natural radionuclides in foods such as dairy, cereals, and vegetables contribute the remaining 10% of natural background exposure. Some foods can contain relatively high levels

 29



Ionizing Radiation

229

of radionuclides (although still resulting in small doses), especially mussels and Brazil nuts. Also associated with natural background is a category of exposures called TENORM (technologically enhanced naturally occurring radioactive materials) resulting from natural radionuclides or radiation, exposing a small part of the population because of human activities, such as flying, manufacturing, mineral extraction, combustion of fossil fuels, cigarette smoking, and so on. The resulting exposures are usually < 0.05mSv (National Council on Radiation Protection and Measurements [NCRP], 2009).

The atomic bombs were detonated over Hiroshima and Nagasaki at altitudes of more than 500 meters so that the radiation doses were essentially due to acute whole-​body external irradiation, immediately after the explosion, from photons and neutrons; there was very little radioactive fallout because the explosions occurred at relatively high altitudes, so that the fireballs had little interaction with the ground (Cullings et al., 2006). Radiation dose levels depended primarily on location in the city at the time of detonation and shielding. Those close to the hypocenter who survived the initial blast had whole-​body doses of > 5 Gy, resulting in acute radiation sickness and death within several weeks. Survivors far from the hypocenter (e.g., > 2 km) had very low whole-​body doses (< 5 mGy). In contrast, the nuclear weapons tests were conducted in unpopulated areas at altitudes that were generally less than 100 meters. Consequently, people were not exposed to radiation occurring during or immediately after the explosion. However, radioactive fallout, including a large number of radionuclides, were spread over the entire world (Bennett, 2002). These radionuclides led to external irradiation from the photons produced during their radioactive decay, and internal irradiation when they contaminated food products. Most of the doses were delivered within a few months after the test, with residual radiation, due to long-​lived radionuclides like 90Sr and 137Cs, with radioactive half-​lives of about 30 years, protracted over a century or so (Bouville et  al., 2002). Even longer-​lived radionuclides, like 14C and 239Pu, will remain in the environment for more than 10,000 years. Current estimates for the general population from these nuclear tests are very low, however, at 0.005 mSv per year (Table 13–​2). The manufacture of nuclear weapons was associated with the production of uranium or plutonium in nuclear reactors and the storage of high-​level radioactive waste. At the beginning of the nuclear era, during the 1940s, little attention was paid to the environmental discharges of radioactive material. Large amounts of 131I were released into the atmosphere at Hanford, Washington, in the United States (Shipler et al., 1996) and at Mayak (Eslinger et al., 2014) in the Soviet Union (about 30 PBq at each facility), and substantial amounts of 90Sr, 137Cs, and other short-​lived radionuclides were discharged into the Techa River near Mayak (Degteva et al., 2006).

Man-​made Environmental Exposures

Nuclear Accidents.  Since 1957, nuclear power stations have been

7 Other Medical

6

Environmental 5 4 3 2 1 0

USA

UK

Germany

World

Figure 13–​1.  Comparison of current estimated annual per capita radiation exposure (mSv) for countries with a similar health-​care level.

Atomic Bombs and Nuclear Weapons Tests. The first

nuclear weapons test took place in July 1945 in New Mexico; it was followed by dropping of atomic bombs on the Japanese cities of Hiroshima and Nagasaki in August 1945 and by the detonation of nuclear weapons tests in the atmosphere, mainly by the United States and the Soviet Union between 1946 and 1962 (Beck and Bennett, 2002). The radiation exposures caused by the atomic bombs and the nuclear weapons tests presented different characteristics and very different levels of radiation exposure. Table 13–​2.  Worldwide Averages and Typical Ranges of Annual Effective Doses (mSv) from Environmental Radiation Annual Effective Dose (mSv) Source of Radiation Natural Cosmic Terrestrial Radon in air Radionuclides in food Man-​made Atmospheric nuclear tests Nuclear fuel cycle Source: UNSCEAR (2010b).

Main Type of Exposure

Worldwide Average

Typical Range

External (μ, e-​, n) External (γ) Internal (α) Internal (α, β)

0.39 0.48 1.26 0.29

0.3–​1 0.3–​1 0.2–​10 0.2–​1

External (γ) and internal (β) External (γ)

0.005

0–​0.05

0.002

0–​0.02

used to produce electricity. There are currently 441 nuclear power stations operable in the entire world, including 99 in the United States (International Atomic Energy Agency [IAEA], 2016). Under routine operation, environmental discharges of radioactive material have been constantly decreasing since the 1950s and are presently very low, so that the effective doses received by the populations of the vicinity of these plants are very low as well (0.002 mSv per year) (National Reasearch Council [NRC], 2012; UNSCEAR, 2010b). Environmental discharges of radioactive material and resulting doses are also low for the other facilities in the nuclear fuel cycle, which include the extraction of uranium, the preparation of nuclear fuel, its reprocessing (only in a few countries), and the storage of radioactive wastes (UNSCEAR, 2010b). There have been four reactor accidents that resulted in irreparable damage to the power plant and in substantial radiation exposures as a consequence of the releases of radioactive materials into the environment (Bouville et  al., 2014). The first of those accidents took place in 1957 at Windscale in the United Kingdom and was caused by a fire in the reactor core. The second accident took place at the Three Mile Island (TMI) reactor in the United States in 1979 and was due to both mechanical and human errors. The third, and most severe, reactor accident was at the Chernobyl nuclear power plant in the former Soviet Union in 1986, and resulted from a series of human errors during the conduct of a reactor experiment. Finally, the Fukushima accident, which occurred in northern Japan in 2011, was the consequence of an earthquake-​triggered tsunami that damaged the reactor cooling system. The relative importance of these accidents can be assessed to some extent from the atmospheric discharges of 131I, which were 1 PBq at Windscale (Clarke, 1989), 0.0006 PBq at Three Mile Island, 1,800 PBq at Chernobyl (UNSCEAR, 2011), and 120 PBq at Fukushima (Terada et al., 2012). The best-​documented doses are those related to

230

230

PART III:  THE CAUSES OF CANCER

the Chernobyl accident: in a cohort of about 12,000 children, the mean thyroid dose from intakes of I-​131 was estimated to be 580 mGy, while the mean whole-​body doses was 14 mGy.

Other Man-​made Sources of Environmental Exposure.

Many industrial operations, security inspection systems (such as some airport screening machines), medical facilities, and educational and research institutions use radiation sources or radioactive material. Members of the public may be exposed to low levels of radiation when they are in proximity to such sources, or when they are near a patient who has been administered a radiopharmaceutical. Widespread environmental contamination and exposure of large numbers of people have occurred in a few instances when industrial and medical sources were misplaced or stolen (UNSCEAR, 2011). These include events that occurred in Goiania, Brazil, in 1987 (IAEA, 1998), in Juarez, Mexico, in 1983 (Molina, 1990), and in Taipei, Taiwan, in 1982–​1983 (Chang et al., 1997).

Medical Exposures There have been increases in ionizing radiation exposure to the general population in the last few decades due to increased medical radiation exposure, primarily from CT scans. In the United States the estimated annual per capita effective dose doubled between 1980 and 2006 from 3 to 6 mSv (NCRP, 2009). The increase was due to the nearly 30-​fold increase in number of CT scans over the period from 3 to 80 million per year. Smaller increases in other higher dose diagnostic exposures, nuclear medicine procedures, and fluoroscopically guided interventional procedures also contributed. CT scan use has also increased in other countries also, although less dramatically than in the United States (Berrington de Gonzalez et  al., 2009). Current levels of CT scan use per capita, for example, are three times lower in the United Kingdom than in the United States. Nuclear medicine tests are 30 times less frequent in the United Kingdom than in the United States. Estimated per capita effective dose from medical radiation in the United Kingdom has also increased, but from 0.3 in 1988 to only 0.4 mSv in 2008 (Hart et al., 2010). The total estimated per capita dose from ionizing radiation exposure now varies considerably across countries depending on levels of diagnostic medical exposures (Figure 13–​1). Although the number of procedures has increased, efforts have been made to reduce the dose per procedure without compromising the medical benefits. The range of average effective doses to the patient per procedure varies from 2 to 15 mSv for CT, from 5 to 70 mSv for fluoroscopically guided interventional procedures, and from 0.2 to 40 mSv for nuclear medicine, according to the type of examination (Mettler et  al., 2009). Before 2000, CT scan doses were rarely optimized for the smaller body size and weight of children, resulting in unnecessarily high exposures of up to 30 mGy to the lungs/​breast from a chest CT scan and 40 mGy to the brain from a head CT (Kim et al., 2012). In the United Kingdom, organ doses have been halved for children by optimized protocols (Lee et al., 2016). External-​beam radiotherapy is the use of external irradiation for the treatment of disease. The therapeutic absorbed dose to the target volume in the patient is orders of magnitude greater than the average diagnostic dose due to conventional radiography (e.g., 40+ Gy) (Thierry-​Chef et  al., 2012). Some very high-​dose exposure to surrounding normal tissue is also unavoidable, but technology is continuously evolving to try to minimize these exposures. Currently, radiotherapy is used mostly to treat patients with life-​ threatening diseases such as cancer, and a limited range of benign diseases like hyperthyroidism (McKeown et  al., 2015). During the earlier part of the twentieth century, radiotherapy was used to treat a wide variety of benign conditions including Tinea capitis, tonsillitis, peptic ulcers, and hemangiomas. Once the hazards of ionizing radiation were better recognized, these treatments were stopped, but not before thousands of patients had been exposed. Many long-​term epidemiological studies of these patient populations have documented the excess cancer risks that the patients suffered (Ron, 2003).

Occupational Exposures There are about 23 million workers who are monitored currently for their occupational radiation exposure (UNSCEAR, 2010b). The largest group is those in mining (12 million), followed by medical workers (7 million). In general, occupational doses have been decreasing for many decades, and are now very low (effective dose of 1–​5 mSv per year), which is similar to cumulative annual exposure from natural background radiation. There are a few exceptions where exposures are increasing, particularly for physicians who perform large numbers of fluoroscopy-​guided interventional procedures, and radiologic technologists who administer radionuclides and perform many nuclear medicine scans.

ASSESSMENT OF THE EXPOSURE IN ANALYTIC EPIDEMIOLOGIC STUDIES The ability to estimate a biological measure of dose provides radiation epidemiologists a great advantage in cancer epidemiology. Dose reconstruction methods and the associated uncertainties vary considerably, however, according to the source of exposure. In some situations it is not feasible to estimate doses, in which case exposure is based on proxy measures such as number of years of employment, or number of CT scans. Many studies are retrospective because of the long latency period for radiation-​related cancer (5+ years for solid tumors) and this brings particular challenges due to the use of historical records. The degree to which dose estimates are individualized is important, as is the assessment of uncertainties and their impact on risk.

Environmental Studies Radiation doses from environmental exposures are some of the most challenging and time consuming to estimate. Individual exposure assessment has been conducted for studies of the Japanese atomic bomb survivors (Cullings et al., 2006), and of local populations around nuclear weapons sites in Nevada (Till et al., 2014) and at Semipalatinsk (Gronwald et  al., 2008), and sites of high radioactive discharges at Hanford (Kopecky et al., 2004), the Techa River (Degteva et al., 2006), and around the site of the Chernobyl accident (Drozdovitch et  al., 2013). All of the radiation exposures of the study subjects occurred predominantly in the 1940s and 1950s, with the exception of the Chernobyl accident, which took place in 1986. In order to evaluate the impact of these events, large numbers of radiation measurements on both the environment and population had been conducted or collected in most of the geographic domains of interest prior to the decision to conduct an epidemiologic study. The approaches for estimating doses to study subjects varied according to the type, quality, and amount of information that was available. No two dose reconstructions were alike, but there was a general strategy to estimate doses and uncertainties that includes (Bouville et al., 2014): (1) collection of as many individual-​based radiation measurements as possible for subjects in the target population, (2) collection of questionnaire-​based individual personal and lifestyle information that can influence dose, (3) collection of information on the spatial and temporal patterns and variations of the radiation field, (4) calculation of realistic radiation doses with efforts to minimize sources of bias, (5)  validation of the dose estimates by independent measurements or strategies, and (6) qualitative and quantitative evaluation of the uncertainties associated with dose estimates.

Medical Exposures of Patients Because exposures to patients are given under controlled conditions that are generally well documented, patients exposed to radiation during treatment for malignant and benign diseases and for diagnostic purposes are well suited to be participants in epidemiologic studies (Stovall et al., 2006). For diagnostic exposures, historically the documentation was only a written description of the type of

 231



Ionizing Radiation

examination, with no details of the specific machines or machine parameters. Sometimes information on protocols and machines can be obtained from current staff or institutional archives in the diagnostic radiology department. In the absence of specific information, organ dose is estimated on the basis of typical values reported in the literature for a given examination and time (Stovall et al., 2006), or via reconstruction of typical values using protocols or survey data (Kim et  al., 2012). In the specific case of patients undergoing CT scans, computational methods to readily estimate organ doses for adults and children of various ages were incorporated into a computer program called NCICT (Lee C et al., 2015). Since the widespread introduction of electronic records for examinations like CT, dose reconstruction can now be conducted using the technical parameters directly from these records. As the exposures are usually partial rather than whole body, information on the specific patient anatomy is needed to help determine which organs were likely to be in the exposure field. If patient anatomy is not available, then age-​ and sex-​specific phantoms (models of the human body) can be used, but this will introduce some error. For patients treated with external-​ beam radiation therapy, the quality of the reconstructed dose depends on the degree of completeness and accuracy of the treatment records, which include radiation energy, field size, anatomic location of fields, treatment-​planning details, and daily treatment logs and information on patient anatomy. To estimate the absorbed doses outside the treatment beam, three techniques are used: (1) calculation using a three-​dimensional mathematical computer model based on measurements made in a water or polystyrene phantom, (2)  direct measurements in an anthropomorphic phantom constructed of tissue-​ equivalent material, and (3) calculation of dose using a treatment-​planning computer program (Stovall et  al., 2006). As with diagnostic exposures, if CT images from the treatment-​planning process and patient height and weight are available, this will greatly improve the accuracy of the organ dose estimates. Without these, age-​and sex-​specific phantoms can be used, but this will reduce dose accuracy, and uncertainty assessments should be undertaken.

Occupational Studies Since the 1960s it has been common or required for occupational radiation workers to wear personal dosimeters, which measure external radiation exposures. The dose measurements are monitored and recorded for all workers. This provides an ideal scenario for epidemiological studies, particularly if the records are electronic, and for many nuclear workers personal dosimeters were available from the start of their employment. There are still challenges in estimating organ doses from these badge measurements, however, particularly for medical workers, because of the potential heterogeneity of exposures over the body, differences in where the personal dosimeter is worn on the body (e.g., whether on the collar, belt, pocket, outside a lead apron or inside for a medical worker), and whether or not individual protective devices (e.g., lead aprons) were used by medical workers (Bouville et  al., 2015). Administration of detailed questionnaires to the study subjects can be used in conjunction with badge doses to address these factors (Bouville et al., 2014; Simon et al., 2014). Other considerations are unmonitored dose when a dosimeter was not worn, doses below the limit of detection, and doses received at multiple facilities and the ability to link these records. For exposures predating the 1960s when workers did not have access to personal dosimeters or when the recorded doses cannot be retrieved (e.g., for the earliest medical workers), proxy measures of dose are generally used based on surveys and periods of employment (Simon et  al., 2015). Occupational doses resulting from intakes of radionuclides usually are derived from bioassay measurements. These were available for about 40% of the Mayak workers who received internal plutonium exposure, but are not available at many other nuclear power facilities (Liu et al., 2016). It may be possible to estimate doses for these workers using a job exposure matrix if some measurements as well as a detailed employment history are available (Liu et al., 2016).

231

BIODOSIMETRY In addition to the physical methods of dosimetry, biological markers can theoretically be incorporated into epidemiologic studies. The main methods that have been used are the fluorescence in situ hybridization (FISH) technique for analysis of stable chromosome translocations, and electron paramagnetic resonance (EPR) of tooth enamel and bones. However, these methods are expensive, require extensive effort in collecting, processing, and analyzing the biological specimens, and, currently, cannot capture exposures below 0.1 Gy. Other techniques for lower doses, such as measurement of gamma-​H2AX foci, are also available. None of the approaches currently available is both highly sensitive and highly specific to ionizing radiation exposure, and these methods have not been widely employed in epidemiological studies (Pernot et al., 2015). For further discussion of radiation biomarkers, see the section “Cellular Assays to Measure Biological Changes in Humans” later in this chapter.

Uncertainties Methods for quantifying uncertainties and incorporating them into the risk analysis are more advanced in radiation than in many other fields, but are still at the development stage (Little et al., 2014; Simon et al., 2015; Stram et al., 2015). A single ideal approach to evaluate and account for all dosimetric uncertainties is not available but is an area of active research (Kwon et al., 2016; Land et al., 2015; Li et al., 2007; Simon et  al., 2015; Stayner et  al., 2007; Stram et  al., 2015). Until recent years, the evaluation of the uncertainties consisted of numerical simulations in which variability and lack-​of-​knowledge uncertainties were combined in Monte-​Carlo simulations. In that method, probability density distributions are assigned to the parameter values that are deemed to have a substantial influence on the dose estimate, and multiple realizations of individual doses are estimated (NCRP, 2007). Consideration of whether shared errors and intra-​individual correlations could be large is required. More sophisticated Monte-​ Carlo procedures are now being developed to separate and distinguish between the shared and the unshared components. However, there are always concerns that all sources of uncertainty have not been taken into account. For example, in environmental studies, there may be “unknown” exposure pathways or unsuspected relevant radionuclides, among other factors. As a rule, the quality of the individual doses required in analytical radiation epidemiologic studies is highest when human-​based measurements are reliable, available for all subjects, and representative of the organ dose of interest. This is, for example, the case for patients with therapeutic treatment or workers in nuclear plants. At the other end of the spectrum, the quality of the individual doses is lowest when human-​based measurements are not available and the doses have to be reconstructed on the basis of sparse data and personal interviews conducted long after the exposure occurred. This is the case for some environmental studies. However, it is important to note that, despite the absence of human-​based measurements, the individual doses to the A-​bomb survivors could be reliably reconstructed, but this required several decades of effort and very large resources.

TYPES OF CANCER CAUSED BY THE EXPOSURE From the vast array of epidemiological studies of ionizing radiation exposure, there is evidence that most types of cancer can be caused by radiation, hence, its classification as a universal carcinogen (UNSCEAR, 2006; BEIR VII). Based on standard laboratory tests of carcinogenicity, ionizing radiation is frequently referred to as a weak carcinogen compared to many chemicals (NCRP, 1997). However, this comparison and generalization are misleading. First, many chemicals have tissue-​specific effects and only cause cancer in certain organs (NCRP, 1989). Also there is direct evidence from human studies of high risks in certain organs following radiation exposure in childhood. For example, in the Japanese Life Span Study (LSS), the estimated

23

232

PART III:  THE CAUSES OF CANCER

excess relative risk (ERR)/​Gy for exposure < age 1 year for leukemia is 50 (at attained age 20), 9 for thyroid, and 4.5 for breast cancer (Hsu et al., 2013; Preston et al., 2002a; Veiga et al., 2016). Several national and international radiation protection agencies review the epidemiological data periodically and summarize the evidence regarding cancer risks for each cancer site. Here we provide a meta-​summary of the most recent findings from BEIR VII (2006), UNSCEAR (2006), and the Advisory Group on Ionising Radiation (2011) to categorize each cancer site as related, probably related, or uncertain relation to ionizing radiation. We have also incorporated evidence from recent studies that were published after these reports. The majority of cancer sites can be classified as radiation-​related, generally based on evidence of a significant dose–​response relationship from several rigorous epidemiological studies without major biases. The Life Span Study (LSS) of the Japanese atomic bomb survivors remains the gold-​standard radiation epidemiology study and the primary source of data for the reports described in the preceding paragraph and the basis for the classification of cancer sites here (Table13-​3). The cohort includes about 105,000 individuals and 22,500 incident cancers during the seven decades of follow-​up (Grant et al., 2017. Key features of the study include the whole body exposure, wide dose-​ range to healthy individuals of all ages, accurate dose reconstruction, and follow-​up via cancer registries and national death records. For a detailed description of the study methods and dosimetry, see Preston et  al. (2007) and Cullings et  al. (2016). For some cancer sites such as sarcomas (Henderson et al., 2012), pancreatic cancer (Dores et al., 2014) and rectal cancer (Sakata et al., 2012), higher-​dose therapeutic exposures (> 5Gy) are the primary basis for our classification. Absence of significantly increased risks from lower-​dose studies for these cancers likely reflects lack of power rather than evidence of a threshold for these sites. Cancers where there is some evidence of a dose–​response relationship, but uncertainty about potential confounders or biases, are described as possibly related to radiation (e.g., chronic lymphocytic leukemia [CLL], multiple myeloma, and endometrial cancer). Those in the “unclear” category are sites where there are no high-​quality studies with reasonable power that support a dose–​response relationship with radiation exposure or sites for which there are inconsistent findings across studies: non-​Hodgkin lymphoma (NHL); Hodgkin lymphoma (HL); kidney, cervical, and gallbladder cancers; and melanoma. These classifications are difficult due to conflicting results and the rarity of some of these cancers, and the distinction between possibly related and unclear should be interpreted cautiously. To illustrate the range in the magnitude of the risks across cancer sites, we present estimates of the ERR/​Gy for cancer incidence from the LSS of the Japanese atomic bomb survivors for an individual exposed at age 30 with an attained age of 70 (Table 13–​3) (Preston et al., 2007). Why there are large differences in radiosensitivity of different organs is an interesting question that could provide insights into radiation carcinogenesis. The weighting factors used to estimate the effective dose (described in the section “Units of Exposure” earlier in this chapter) aim to capture these differences in sensitivity for radiation protection standards (ICRP, 2007).

MAJOR NEW STUDIES AND UPDATES We reviewed studies of environmental, medical, and occupational exposures published since 2006. For studies that have not reported significant updates, see the previous edition of this text (Boice, 2006). Summary characteristics of the studies are shown in Table 13–​4.

Environmental Exposures Japanese Atomic Bomb Survivors

As described in the preceding section, the LSS study has demonstrated that most types of cancer can be caused by ionizing radiation exposure, and has provided quantification of the risks per unit dose and variation with gender, attained age, age at and time since exposure. The cohort continues to provide new information about the long-​term effects of

radiation exposure, especially from childhood exposure, as about 80% of those exposed before age 10 are still alive (Grant et al., 2017). The most recent analysis of all solid cancer incidence with follow-​up extended to 2009, which includes 22,538 cancers, incorporates adjustment for smoking for the first time and has updated dose estimates. The dose–​response for females was linear (ERR/​Gy = 0.64; 95% CI: 0.52, 0.77), and similar in magnitude to the previous analysis (Preston et al., 2007). For males, there was evidence of significant upward curvature described by a linear-​quadratic model with an ERR at 1 Gy of 0.2 (0.12, 0.30), which is significantly lower than for females. The EAR estimate for females was 54.7 (95%CI:45, 65), while for males the estimate was 42.9 (95%CI:27, 58) cases per 10,000 PY at 1 Gy. When restricted to doses < 0.1Gy, there was still a statistically significant dose–​response relationship using a sex-​averaged linear model (p = 0.038). The additional follow-​up confirmed that the solid cancer risks remain significantly increased for more than 60 years after radiation exposure. The risk decreased with increasing age at exposure or attained age. There was evidence that a previous finding of an increase in risk for exposure after age 60 (Preston et  al., 2007)  was an artifact due to cases only reported on autopsy (Grant et al., 2017). No simple explanation for the introduction of curvature in the dose–​response relationship for cancer incidence in the males could be identified. The revised dosimetry and increased follow-​up time tended to increase the evidence for curvature, while smoking adjustments and sex-​specific cancers were ruled out. There will be a series of updated site-​specific cancer incidence papers over the next few years that will assess evidence for curvature for each site in more detail, and that will incorporate additional lifestyle factors as potential effect modifiers. The updated analysis of hematopoetic malignancies with cancer incidence data to 2001 reported that elevated leukemia risks, particularly acute myeloid leukemia, persist for more than 50  years after exposure (Hsu et  al., 2013). The dose–​ response relationship was linear-​quadratic for acute myeloid leukemia (AML), but linear for acute lymphoblastic leukemia (ALL) and chronic myeloid leukemia (CML). There is emerging evidence of an increased risk of CLL and NHL in males, but still no evidence that multiple myeloma or Hodgkin lymphoma are related to radiation exposure. For thyroid cancer, additional follow-​up suggests that the excess risk persists for more than 50 years after exposure, but is still largely restricted to exposure before age 20 (Furukawa et al., 2013). An updated analysis of the joint effect of smoking and radiation on lung cancer risk suggested a complex relationship with smoking intensity where the joint effect was multiplicative for lower smoking intensities (< 10 cigarettes/​day) and additive or sub-​additive for higher intensities (Furukawa et al., 2010). For uroethelial cancer, the joint effect of smoking and radiation was best described by a multiplicative model, although the additive model could not be rejected (Grant et al., 2012).

Natural Background Radiation

Most studies of high natural background radiation have used an ecological design, which is susceptible to many biases (Hendry et  al., 2009). The major exception is the case-​control studies of residential radon exposure, which have provided convincing evidence of a link to lung cancer due to relatively high doses, careful dose reconstruction, and pooling of multiple studies (Darby et  al., 2005; Krewski et  al., 2006). The risk estimates from these pooled analyses of residential exposure are compatible with the high-​dose studies of radon-​exposed underground miners (Lubin et al., 1997). The joint effect of smoking and radon was multiplicative, which means that the absolute hazard from radon exposure is much higher for smokers. They estimated that the lifetime lung cancer risk for smokers with high residential radon exposure (e.g., 800 Bq/​m2 usual exposure) could be increased from 10% to 22%, whereas in non-​smokers the lifetime risk is increased from 0.4% to 0.9% (Darby et al., 2005). There also have been several case-​ control studies of natural background radiation and childhood cancer. A  recent record-​based case-​control study in the United Kingdom (n  =  27,447 cases) using radiation exposure from mother’s residence at the child’s birth from national databases found a significant dose–​ response relationship (ERR/​mGy = 0.12, 95%CI:0.03–​0.22) for leukemia in relation to red

 23



233

Ionizing Radiation

Table 13–​3. Cancers Related to Ionizing Radiation and Estimated Excess Relative Risk (ERR) at 1 Gy from the Life Span Study (LSS) of Atomic Bomb Survivors for Age at Exposure 30 and Attained Age 70 Cancer

Radiation Related

ERR at 1 Gy* LSS (90% CI)

N Cases in LSS

Bladder Breast Lung Leukemia (non-​CLL)

Yes Yes Yes Yes

1.23 (0.59, 2.10) 0.87 (0.55, 1.30) 0.81 (0.56, 1.10) 0.79 (0.03, 1.93)

469 1073 1759 416

Brain/​CNS Ovary Thyroid Colon Oesophagus Oral

Yes Yes Yes Yes Yes Yes

0.62 (0.21, 1.20) 0.61 (0.00, 1.5) 0.57 (0.24, 1.1) 0.54 (0.30, 0.81) 0.52 (0.15, 1.00) 0.39 (0.11, 0.76)

281 245 471 1516 352 277

Stomach Liver Non-​melanoma skin Bone

Yes Yes Yes Yes

0.34 (0.22, 0.47) 0.30 (0.11, 0.55) 0.17 (0.003, 0.55) na

4730 1494 330 20

Soft tissue

Yes

na

Pancreas

Yes

0.26 (< 0.07, 0.68)

512

Rectum

Yes

0.19 (−0.04, 0.47)

838

Endometrial

Possibly

0.29 (−0.14, 0.95)

184

Chronic lymphocytic leukemia

Possibly

na

Multiple myeloma

Possibly

0.38 (−0.23 to 1.36)

136

Non-​Hodgkin lymphoma

Unclear

See comment

402

Prostate

Unclear

0.11 (−0.10, 0.54)

387

Renal cell Cervix

Unclear Unclear

0.13 (−0.25-​0.75) 0.06 (−0.14, 0.31)

247 859

Gallbladder Melanoma

Unclear Unclear

−0.05 (< −0.3, 0.3) na

549 17

33

12

Comment

ERR increases with age at exposure. Linear term of linear-​quadratic model provided here.

ERR decreases with age at exposure.

Increase driven by salivary gland, confirmed in high-​dose radiotherapy studies (Boukheris et al., 2013).

ERR decreases with age at exposure. Dose–​response relationship reported after high-​ dose radiotherapy (e.g., Henderson et al., 2012; Rubino et al., 2005). Dose–​response relationship reported after high-​ dose radiotherapy (e.g., Henderson et al., 2012; Rubino et al., 2005). Dose–​response relationship reported after high-​dose radiotherapy (Dores et al., 2014; Hauptmann et al., 2016). Dose–​response relationship reported after high-​dose radiotherapy (e.g., Sakata et al., 2012) and in UK nuclear workers (Muirhead et al., 2009) Possibly increased for exposure < age 20 in LSS and after high-​dose radiotherapy (e.g., Sakata et al., 2012). Dose–​response relationship reported in Chernobyl cleanup workers (Zablotska et al., 2013) but not in nuclear workers (Leraud et al., 2015). Dose–​response relationship for cancer incidence in UK nuclear workers (Muirhead et al., 2009) but not for mortality in INWORKS (Leuraud et al., 2015). ERR at 1 GY = 0.46 (−0.08, 1.29) for males and 0.02 (< −0.44, 0.64) for females. No evidence of increased risks in occupational cohorts (Muirhead et al., 2009). No evidence of increased risks after high-​dose radiotherapy (e.g., Sakata et al., 2012).

*For age at exposure 30, attained age 70 (solid cancers from Preston et al., 2007, and hematopoietic from Hsu et al., 2013). na = not available.

bone marrow dose from gamma radiation (mean RBM dose = 4mSv; range  =  0–​31mSv) (Kendall et  al., 2013). This was also consistent with the dose–​response relationship from higher dose rate studies, such as the LSS (Hsu et  al., 2013). There was a weak but not statistically significant relationship between radon exposure and leukemia, and no association between gamma or radon dose with any other childhood cancers. Previous case-​control studies that utilized a similar design were null, or not statistically significant for gamma radiation, but based on a much smaller sample size (range  =  50 to 860 cases vs. 27,447) (Laurier et al., 2001; Raaschou-​Nielsen, 2008). There is evidence from previous case-​control studies with residential measurements of radon concentration of an increased risk of childhood ALL

(Raaschou-​Nielsen, 2008). The UK study is currently being further expanded and dose estimates refined.

The Chernobyl Accident

The Chernobyl accident in 1986 is the worst nuclear accident to date, and the studies of the exposed children, in particular, are the most important source of information on the risks from internal I-​131 exposure in the general population. There are ongoing follow-​up studies in Ukraine, Belarus, Russia, and other Baltic countries of children exposed to internal radiation, primarily from consumption of contaminated milk and foods, and of the clean-​up workers. A cohort of Ukrainian children (n  =  12,514) with I-​131 exposure estimated from measurement and

234

234

PART III:  THE CAUSES OF CANCER

Table 13–​4. Summary of the Characteristics of Recent Epidemiological Studies Source

Study

Exposure Type

Exposure Pattern

Typical Dose Range

Outcomes

Environmental

Life Span Study Residential radon pooled analysis UK Background Chernobyl UKRAM cohort Chernobyl BELAM cohort Swiss Nuclear Power Residents France Nuclear Power Residents German Nuclear Power Residents Techa River

Gamma, neutrons Radon Gamma I-​131 I-​131 External External External Gamma, Cesium

Acute Chronic Chronic Chronic Chronic Chronic Chronic Chronic Chronic

0–​2 Gy 0–​1400 Bq/​m3 0–​0.015 Gy 0–​4 Gy 0–​10 Gy na na na 0–​0.5 Gy

All cancer incidence Lung cancer incidence Childhood cancer incidence Thyroid cancer incidence Thyroid cancer incidence Childhood cancer incidence Childhood cancer incidence Childhood cancer incidence Cancer incidence

Medical

UK Pediatric CT Australian Pediatric CT US Scoliosis BRCA cohort Kuwait Thyroid case-​control US Brain Cancer case-​control USRT personal diagnostic exposures UKCCS CCSS UKCCS French childhood cancer survivors Hodgkin’s lymphoma survivors Second GI cancers study Retinoblastoma cohort

X-​rays X-​rays X-​rays X-​rays X-​rays X-​rays X-​rays X-​rays X-​rays X-​rays X-​rays X-​rays X-​rays X-​rays

Acute, fractionated Acute, fractionated Acute, fractionated Acute, fractionated Acute, fractionated Acute, fractionated Acute, fractionated Acute, fractionated Acute, fractionated Acute, fractionated Acute, fractionated Acute, fractionated Acute, fractionated Acute, fractionated

0–​0.1 Gy na 0–​1 Gy na na na na na 0–​80 Gy 0–​80 Gy 0–​80 Gy 0–​80 Gy 0–​80 Gy 0–​80 Gy

Cancer incidence Cancer incidence Breast cancer incidence Breast cancer incidence Thyroid cancer incidence Brain cancer incidence Thyroid cancer incidence Childhood cancer incidence Second cancer incidence Second cancer incidence Second cancer incidence Second cancer incidence Second cancer incidence Second cancer incidence

Occupational

Mayak workers 15 Countries Study INWORKS UKNRRW French nuclear workers US nuclear workers Flight crew pooled analysis Scandinavian flight crew US flight crew Chinese medical workers USRT

Gamma, plutonium Gamma Gamma Gamma Gamma Gamma Cosmic Cosmic Cosmic X-​rays X-​rays

Chronic Chronic Chronic Chronic Chronic Chronic Chronic Chronic Chronic Chronic Chronic

0–​4 Gy 0–​0.05 Gy 0–​0.05 Gy 0–​0.05 Gy 0–​0.05 Gy 0–​0.05 Gy na na na 0–​0.4 Gy 0–​0.1 Gy

Cancer incidence and mortality Cancer mortality Cancer mortality Cancer incidence Cancer mortality Cancer mortality Cancer mortality Breast cancer incidence Breast cancer incidence Cancer incidence Cancer incidence

na = not available.

interviews has undergone four rounds of thyroid screening. In the most recent analysis, the linear dose–​response relationship for thyroid cancer (n = 65, ERR/​Gy = 1.91; 95% CI: 0.43,6.34), was consistent with the risk from external acute exposure in the Japanese atomic bomb survivors (Brenner et al., 2011). Similar results were reported in a cohort of children in Belarus (n = 84 thyroid cancers) (Zablotska et al., 2011). Assessment of dose uncertainty did not significantly alter these risk estimates (Little et al., 2014, 2015). More details on the relationship between I-​131 and thyroid cancer are provided in Chapter 44. Interview-​ based case-​ control studies of adulthood exposures received by the liquidators reported an increased risk of thyroid cancer (n  =  107, median thyroid dose  =  69 mGy) (Kesminiene et  al., 2012) and leukemia (n = 137) including both CLL and non-​CLL (controls mean RBM dose = 82 mGy) (Zablotska et al., 2013). Although the non-​ CLL leukemia results were consistent with the Japanese atomic bomb survivors, the excess risks for CLL and thyroid cancer after adulthood exposure were much higher than those observed in the LSS (Hsu et al., 2013; Preston et al., 2007). Prevalent case-​control studies conducted more than 20 years after the accident could be subject to recall bias. In the leukemia/​CLL case-​control study, 50% of the cases compared to 5% of the controls had died, and dosimetry was based on proxy responders. The ERR/​Gy was lower in the subset with

direct interviews (0.88) than proxy responders (3.98), but confidence intervals were wide due to the small number of controls with proxy interviews (Zablotska et al., 2011).

Cancer Near Nuclear Power Stations

Residents living near nuclear power stations remain concerned about potential cancer risks, especially childhood leukemia, since the early reports of potential cancer clusters (Black, 1984). Study design issues include small sample size and ecological design. A recent large cohort study based on the Swiss Childhood Cancer Registry (n  =  2925) using distance from residence at birth from a nuclear power plant and national census data found no evidence of a relationship between distance and childhood leukemia (Spycher et al., 2011). The study could not, however, rule out small risks due to limited statistical power, and distance is a crude and potentially inaccurate proxy for radiation exposure. Two studies with a similar design in France and Germany found some evidence of a relationship between proximity to a power station and leukemia (Kaatsch et al., 2008; Sermage-​Faure et al., 2012). A  review of all the studies and new UK data concluded that there remained suggestive evidence of small risks, but confounding could not be ruled out, especially given the finding of a similar association with distance from sites for proposed nuclear power stations that were

 235



Ionizing Radiation

never built (Stather, 2011). A proposed US study was canceled after feasibility work suggested that the time and costs involved were excessive (Nuclear Regulatory Commission, 2015). An earlier US study of cancer mortality in 107 counties near nuclear facilities and matched control counties was null, but with the caution that the risks may have been too small to detect with the ecological study design (Jablon et al., 1991).

Techa River Residents

The population living near the Techa River received protracted low-​ dose radiation exposure from internal and external sources following radioactive releases into the river from the Mayak nuclear weapons plants. In the cohort of residents followed for cancer incidence (n  =  17,433) with a mean organ dose around 50 mGy, there was a statistically significant linear dose–​response relationship for all solid cancers and leukemia (excluding CLL) (Krestinina et al., 2007, 2013). There is limited power for site-​specific solid cancer risks (n = 2059), but, of 15 sites evaluated, significant dose–​ response relationships were recently reported for cancer of the esophagus and uterus (Davis et al., 2015).

Medical Exposures Diagnostic Medical Radiation

The increase in CT scan use and publications reporting unnecessarily high dose levels in children in 2001 (Paterson et al., 2001) prompted the initiation of a number of retrospective cohort studies of pediatric CT scans and cancer. Collaboration between the investigators ensured that protocols were broadly similar to facilitate future pooling of the data. The UK cohort consists of approximately 180,000 children and young adults (age < 22 years) who underwent CT scans between 1980 and 2002 in about 100 hospitals (Pearce et al., 2012). The exposure data were abstracted from electronic databases in radiology departments and were linked to national cancer registration and vital status databases. Patients with cancer diagnosed before the date of the first CT scan were excluded. Organ-​specific dose estimates were semi-​ individualized, as they were based on body region scanned, date of scan, and age at scan. With a 2-​year lag period, there was evidence of a significant dose–​response relationship for leukemia/​MDS (n = 74) and red bone marrow dose (mean = 12 mGy) (Figure 13–​2). The ERR/​mGy was statistically compatible with the risk estimate from the Japanese

8 7

RR (& 95% CI)

6 5 4 3 2 1 0

0

10

20

30

40

50

60

RBM dose (mGy)

Figure 13–​2.  Relative risk (and 95% CI) for leukemia/MDS according to the estimated absorbed red bone marrow (RBM) dose (Gy) from pediatric CT scans in the UK cohort. Source: Pearce et al. (2012).

235

atomic bomb survivors when restricted to a similar exposure age and follow-​up period. The dose–​response relationship for brain tumors (n = 135) and estimated brain dose (mean = 43 mGy) was also statistically significant, but about four times higher than the comparable risk estimate in the LSS. Sensitivity analyses that assessed additional clinical information suggested that there was some evidence of bias due to unreported previous brain tumors, and when these were excluded the ERR/​mGy was still statistically significant but slightly closer in magnitude to the LSS (Berrington de Gonzalez et al., 2016). The cohort is currently being expanded to include patients who underwent CT scans up to 2010. Additional data have been collected to refine the dosimetry and conduct analyses of the impact of dose uncertainty (Lee et  al., 2016). The cohort also forms one of the centers of the multi-​center EPI-​CT study (Bosch de Basea et al., 2015). With data from nine countries across Europe, the pooled study aims to include approximately one million children, and results are due in the next few years. An Australian cohort used Medicare (universal healthcare scheme) records to identify 680,000 children who had CT scans between 1985 and 2005, aged < 20 years, and approximately 10 million children who were not exposed (Mathews et  al., 2013). Through record linkage, 3150 cancers were recorded in the exposed children. They reported an increase in incidence rate ratio for all cancers of 0.16 (95%CI:0.13–​ 0.19) for each additional CT scan. Site-​specific analyses suggested increased risks of solid cancers in digestive organs, melanoma, soft tissue, female genital, urinary tract, brain, and thyroid cancers, and also leukaemia, myelodysplastic syndromes, and some other lymphoid cancers. The results are difficult to compare with the UK cohort because of the difference in analytic approach (organ dose versus number of CT scans) and different exclusion periods (2–​5 years in the UK versus 1 year in the Australian study). The lack of specificity by cancer site and increased risks for cancers not usually considered to be radiation-​ related (e.g., melanoma and lymphomas) raised concerns about potential bias from preexisting conditions. Dose reconstruction is underway using methods similar to the UK CT study, which should facilitate comparisons. A Canadian cohort is also in progress with an identical design to the UK cohort using electronic medical records from Cancer Care Ontario linked to cancer registrations (Einstein, 2012). This is the only study currently that will also include adulthood exposures. Conventional diagnostic X-​rays continue to be a common source of medical radiation exposure, although doses are typically one-​tenth of the dose of CT scans to the equivalent body region. Direct studies of potential cancer risks face a number of challenges, including low power and exposure misclassification. Cohorts of special populations who received unusually high exposure levels (e.g., women with scoliosis and tuberculosis) based on medical records have, however, provided important information about the cancer risks from repeated low doses of radiation. An updated follow-​up of the US scoliosis cohort (n = 3010) reported a borderline significant dose–​response relationship for breast cancer (ERR/​Gy = 2.86, n = 78, p = 0.058) (Ronckers et al., 2008). The mean breast dose was 120 mGy and the dose–​response relationship was significantly greater for women who reported a family history of breast cancer, although this was based on small numbers. Several recent case-​ control studies reported increased risks for brain and thyroid tumors after dental X-​rays, and breast cancer after chest X-​rays in BRCA mutation carriers (Andrieu et al., 2006; Claus et al., 2012; Memon et al., 2010). However, recall bias is a concern in case-​control studies that rely on self-​reported exposure information. In a cohort of US radiologic technologists there was evidence of an increased risk of thyroid cancer after childhood dental X-​rays based on prospective follow-​up after self-​reporting of X-​ray exposure history at baseline (Neta et al., 2013). Compared to the general population, these medical workers should be able to report their history of diagnostic exposures more reliably. Analyses of other cancer sites and personal diagnostic exposures are ongoing. The UKCCS case-​control study (n = 2690 childhood cancer cases diagnosed 1976–​1996), which used record linkage to evaluate diagnostic X-​rays in utero and in early infancy (< 100 days), reported a non-​statistically significant increased risk from in utero X-​rays for all cancers (OR  =  1.14; 95%CI:0.90, 1.45) and leukemia (OR = 1.36; 95%CI:0.91, 2.02) (Rajaraman et al., 2011). The results are compatible with lower doses compared to the

236

236

PART III:  THE CAUSES OF CANCER

period of the earlier studies in the United Kingdom and northeastern United States (Doll and Wakeford, 1997).

Radiotherapy

With the increasing population of cancer survivors and longer survival, the risk of a treatment-​related second cancer is an increasing clinical and public health concern. Approximately half of all cancer patients in countries like the United States receive radiotherapy as part of their initial treatment. Although there are increasing efforts to minimize the dose to the normal tissue surrounding the tumor, some very high doses (< 40 Gy) to some proximal organs are unavoidable. In the last decade there has been considerable progress in our understanding of the cancer risks from partial-​body, high-​dose fractionated exposure, due to the publication of a plethora of new dose–​response studies for second cancers after radiotherapy (Berrington de Gonzalez et al., 2013; NCRP, 2011). In general, these case-​control studies of second solid cancers (breast, lung, stomach, pancreas, thyroid, brain, sarcoma, skin, esophagus, bladder, colon and rectum) based on medical records and individual dose reconstruction, suggest a linear dose–​response relationship even for very high, fractioned, organ doses (40+ Gy), which was in contrast to the expectation of a flattening or even downturn in risk at high doses due to cell killing. However, the ERR/​Gy from these high-​dose fractionated exposures is lower than the risk from the acute exposure in the Japanese atomic bomb survivors (Figure 13–​3). Thyroid cancer is the only clear exception where there is a clear downturn in risk after about 20 Gy in childhood cancer survivors (Bhatti et al., 2010b; Veiga et al., 2012) (Figure 13–​3). Although the ERR/​Gy is lower than reported in the LSS, the very high doses can result in some high absolute risks for common cancers. For, example an HL patient treated with 40+ Gy to the chest at age 25 is estimated to have a 30% risk of developing radiation-​related breast cancer by age 55, which is comparable to a BRCA2 mutation carrier (Travis et al., 2005). The ERR/​Gy is generally higher for childhood exposure, as observed in the LSS, which also contributes to high absolute risks (Berrington de Gonzalez et al., 2013; Grant et al., 2017. As described earlier, several of these recent case-​control studies of high-​dose radiotherapy provided the first clear evidence that pancreatic cancer (Dores et  al., 2014; Hauptmann et  al., 2016), rectal cancer (Sakata et  al., 2012), and sarcomas (Henderson et al., 2012; Rubino et al., 2005) are radiation-​related. For more details on second cancers, see Chapter 60.

Occupational Exposures Nuclear Workers

Hundreds of thousands of workers worldwide have been exposed to radiation during employment in nuclear weapons and energy facilities.

Studies of workers at these facilities have provided information on risks of cancer following low-​dose, protracted exposure to external ionizing radiation. An important feature of these populations is the routine dose monitoring, which can be used to reconstruct individual long-​term dose histories. These studies generally aim to assess whether the risk of cancer per unit dose at low doses and low-​dose rates is similar to that estimated from acute doses in the Japanese LSS. The major studies published in the past 10 years, which are discussed in the following paragraph, include those of workers at the Mayak facility and of cleanup workers at the Chernobyl nuclear power plant, the 15-​country study by the International Agency for Research on Cancer (IARC), and its successor the International Nuclear Workers Study (INWORKS). Also discussed are major national cohorts of nuclear workers exposed to external ionizing radiation, such as the UK National Registry of Radiation Workers, the Canadian National Dose Registry, and pooled studies of US nuclear workers. Studies of workers at the plutonium production facility in the Russian Federation (former Soviet Union) have been highly informative on the health effects of protracted exposure to gamma radiation exposure, and provide a rare opportunity to study the risks from plutonium exposure. Workers in certain plants at the Mayak Production Association received high plutonium doses (mean lung doses: 0.12 Gy, range 0–​1 Gy), due to several spills and accidents that occurred during periods of heavy production between 1948 and 1958. The key recent studies addressing the effects of plutonium include an updated analysis of lung cancer mortality (Gilbert et al., 2013) and incidence (Labutina et al., 2013), as well as the incidence of liver, bone, and other cancers (Hunter et  al., 2013; Labutina et  al., 2013). Strengths of the Mayak studies include the large cohort size, high exposures, relatively large numbers of exposed female workers, availability of cancer incidence data, and extensive efforts to characterize dose. Challenges of these studies include a lack of complete biomonitoring for plutonium exposure, resulting in large dose uncertainties, and high rates of smoking and heavy alcohol consumption, which have led to concerns about potential confounding and have raised questions about transfer of risks to other exposed populations. Major attention has focused in the Mayak cohort on cancer in lung, liver, and bone, as these are the primary sites associated with plutonium deposition following inhalation exposure. With 486 deaths from lung cancer (most among men), the latest study found substantial elevations in risk with increased plutonium dose: ERR per Gy was 7.4 (95% CI: 5.0, 11) among men with an attained age of 60 and declined with increasing age (Gilbert et al., 2013). Among women, ERR/​Gy at age 60 was 24 (95% CI: 11, 56). Restricting analyses to workers exposed at lower plutonium doses found similar risks per unit dose as in the full cohort, and ruled out anything but very small departures from linearity. Importantly, the study was able to adjust for tobacco smoking and to

Breast cancer after Childhood cancer (n = 107) (Inskip et al., 2009)

Thyroid cancer after Childhood cancer (n = 115) (Bhatti et al., 2010) 40.0

20.0

35.0 RR (& 95% CI)

RR (& 95% CI)

15.0

10.0

30.0 25.0 20.0 15.0 10.0

5.0

5.0 0.0

0.0 0

10

40 50 20 30 Mid-point of dose category (Gy)

60

0

10 20 30 40 Mid-point of dose category (Gy)

50

Figure 13–​3.  Relative risk (and 95% CI) for subsequent cancer according to the estimated absorbed radiation dose (Gy) from fractionated radiotherapy. Dotted black line indicates fitted dose-response for that study. Dashed grey line indicates the RR for similar age at exposure and attained age based on the BEIR VII risk models for low-dose exposure (BEIR VII, 2006). The fitted linear dose-response model for Bhatti et al (2010)b is based on the linear term from the linear-quadratic model.

 237



Ionizing Radiation

evaluate the form of interaction between smoking and plutonium dose. It found a sub-​multiplicative but super-​additive interaction (discussed further in the section on effect modification). Analyses of lung cancer incidence supported the main findings of a steep dose-​response relationship (Hunter et  al., 2013; Labutina et  al., 2013). An analysis by lung tumor subtype found a much higher ERR/​Gy for adenocarcinoma than for squamous cell subtypes, suggesting an important etiologic difference for these two subtypes. Bone and connective tissue cancer incidence was also found to be highly elevated (RR = 13.7; 95% CI: 3.0, 58.5) among a group of unmonitored female workers employed before 1950 at a plutonium-​processing facility with heavy exposure potential (Cardis et al., 2007). Studies of other cancer types have found a highly significant dose–​response relationship for liver cancer and plutonium, but not most other specific cancers (Hunter et al., 2013; Labutina et al., 2013). Efforts are ongoing to assess the impact of the large uncertainties in the plutonium dose estimates (Stram et al., 2015). The IARC has led a number of pooled international studies of nuclear workers. In the first decade of the 2000s, a series of studies was published for solid cancer and leukemia mortality risks in a pooled study of nuclear workers in 15 countries (Cardis et al., 2005, 2007; Vrijheid et al., 2007, 2008). The study, which included 407,000 workers exposed to an average cumulative dose of 19 mSv, found an ERR/​Sv of 0.97 (95% CI: 0.14, 1.97) for all cancers excluding leukemia and 1.93 per Sv (95% CI: < 0, 8.47) for leukemia excluding CLL (Cardis et al., 2005). Although no significant heterogeneity in risk was seen in these results by country or large facility (Cardis et al., 2007), a number of subsequent publications questioned the results of the study, pointing out that findings contributed by Atomic Energy of Canada Ltd seemed highly influential on the statistical significance of the solid cancer results, and questions arose about the quality of the dosimetry data for this cohort (Zablotska et al., 2014; more information is given about this later in this chapter). In addition, it was noted that the exclusion of workers with substantial neutron exposure or radionuclide contamination reduced the overall power of the study, as workers with neutron dose or internal dose contamination also were more likely to have high photon doses (Boice et al., 2006). Despite these criticisms, the 15-​country study remains an important contributor to information about effects of dose protraction on cancer risk from photon radiation exposure, particularly with regard to lymphatic and hematopoietic cancers, in which the influence of the Canadian cohort was minimal. Strengths of the study include the use of a common protocol and a focus on evaluation of workers exposed only to external ionizing radiation, which minimized the influence of poorly characterized internal exposures (IARC, 2012). Since the publication of the 15-​country study, it was recognized that continuing mortality follow-​up of its contributing cohorts would provide an important addition to knowledge on low-​dose cancer risk from photon exposure, as only a small percentage of each cohort was deceased at the time follow-​up ended for the 15-​country study (NRC, 2006). INWORKS was formed by a team from the IARC and international collaborators involved in the 15-​country study, to study radiation-​ related risks of cancer mortality in the nuclear worker cohorts that contributed more than 80% of deaths in the 15-​country study: the UK National Registry of Radiation Workers (NRRW; Muirhead et  al., 2009), the French combined nuclear workers study (Metz-​Flamant et  al., 2013), and the US pooled nuclear workers study (Schubauer-​ Berigan et al., 2015b). All three studies have updated mortality follow-​ up since the publication of the 15-​country study (Hamra et al., 2015), and in some cases (Schubauer-​Berigan et  al., 2015b) included additional cohorts primarily exposed to photon radiation. The Canadian workers were not included, and this avoids the problems with this cohort described earlier. INWORKS includes more than 300,000 workers, with 531 leukemia (other than chronic lymphocytic) deaths (Leuraud et  al., 2015) and nearly 18,000 deaths from solid cancers (Richardson et al., 2015). The pooled analysis found a significantly elevated risk of leukemia (ERR/​Gray [Gy]: 2.96; 90% CI: 1.17, 5.21) and solid cancers (ERR/​ Gy: 0.48; 90% CI: 0.18, 0.79) (Leuraud et  al., 2015; Richardson et al., 2015) (Figure 13–​4). These point estimates changed little when

237

the data were restricted to lower dose ranges (< 100 mGy), or when excluding or adjusting for exposure to neutrons or internal emitters. Indirect evidence suggested little potential for confounding by smoking in the analysis of all solid cancers, as the risk estimates were similar when restricted to non-​smoking-​related cancers. The leukemia subtype showing the highest risk was CML, for which the ERR/​ Gy was 10.45 (90% CI: 4.48, 19.65), and this pattern was consistent across the countries (Leuraud et al., 2015). The ERR/​Gy for AML was 1.29 (90% CI: –​0.82, 4.28). Although the mean cohort dose was only 25 mGy, the risks per unit dose and subtype patterns were similar to the recent analysis in the LSS (Hsu et al., 2013). The size, high-​quality dosimetry, and lengthy follow-​up of the INWORKS cohort suggest that it will remain an important study through which to understand the risk of protracted exposure to low-​dose photons, primarily in adult males. The pooled cohorts will also be informative on site-​specific cancer risks and effect modification by temporal factors, especially as follow-​up is extended for the 78% alive at the end of the recent study. Studies of the component cohorts of INWORKS have also yielded useful information about risk of grouped and individual cancers. The UK NRRW cohort included 174,541 workers monitored for external ionizing radiation (Muirhead et  al., 2009), followed for cancer incidence and mortality through 2001, with 16% deceased. Major strengths of the cohort include its large size, well-​ characterized gamma doses, wide representativeness of the facilities included, and availability of cancer registry linkage. A  limitation of this cohort is the lack of specificity about which workers may have had substantial internal (e.g., plutonium) dose. The ERRs per Sv observed for leukemia (excluding CLL) mortality and incidence (respectively) were 1.71 (90% CI: 0.06, 4.29) and 1.78 (90% CI: 0.17, 4.36). For all cancers excluding leukemia, the ERR/​Sv was 0.28 (90% CI: 0.02, 0.56) and 0.27 (90% CI:  0.04, 0.51), respectively. Lymphoma showed a borderline positive trend with radiation dose (p  =  0.081; 81). Findings for multiple myeloma differed between the mortality and incidence follow-​up:  the ERR/​Gy was 1.20 (90% CI:  –​0.88, 5.96) for mortality (n = 113) and 3.60 (90% CI: 0.77, 8.94) for incidence (n = 149), respectively (Muirhead et al., 2009). The US pooled nuclear worker cohort (Schubauer-​Berigan et  al., 2015b) includes 119,195 workers monitored for external ionizing radiation, from four Department of Energy (DOE) nuclear weapons sites (Hanford, Idaho National Laboratory, Oak Ridge National Laboratory, and Savannah River Site) and the Portsmouth Naval Shipyard. With mortality follow-​up through 2005, 35% of the cohort was deceased, and 10,877 cancers (excluding leukemia) and 369 leukemias (excluding CLL) were observed. Few of these workers received a confirmed deposition of internal radiation, and neutron dose was less than 2% of the total dose for the cohort. The study found ERR/​Sv values of 1.7 (95% CI: –​0.22, 4.7) for leukemia and 0.14 (95% CI: –​0.17, 0.48) for all cancers excluding leukemia. For the latter group, findings differed by cancers classified into those related to smoking (ERR/​Sv: –​0.079; 95% CI: –​0.43, 0.32) and those unrelated to smoking (ERR/​Sv: 0.70; 95% CI:  0.058, 1.5). This heterogeneity suggests that negative confounding may exist with cigarette smoking (where adjustment would shift the risk estimate away from the null), which was also supported by facility-​specific analyses of lung cancer and chronic obstructive pulmonary disease. The ERR/​Sv for mesothelioma of 2.5 (95% CI: –​ 1.3, 10) suggests that positive confounding by asbestos exposure may exist. Other notable findings include significant dose–​response patterns for lymphoma (ERR/​Sv: 1.8; 95% CI: 0.027, 4.4) and multiple myeloma (ERR/​Sv: 3.9; 95% CI: 0.60, 9.6). The French cohorts in INWORKS include 59,021 workers followed for mortality between 1968 and 2004 at three large nuclear energy or weapons cohorts:  Commissariat à l’Energie Atomique (CEA), Electricité de France (EDF), and AREVA Nuclear Cycle (AREVA) (Leuraud et al., 2015). Mean cumulative effective dose was 22.5 mSv. With just 11% of the cohort deceased, the ERR/​Sv was 0.34 (90% CI: –​0.56, 1.4) for all solid cancers, 4.0 (90% CI: < 0, 17) for leukemia excluding CLL, and was non-​calculable for multiple myeloma and lymphoma. The Canadian National Dose Registry (CNDR) is a centralized occupational radiation dose registry, containing records for more than

238

238

PART III:  THE CAUSES OF CANCER (a) Leukemia

(b) Solid cancer 1.7

6

1.6 1.5 RR (& 90% CI)

RR (& 90% CI)

5 4 3 2

1.3 1.2 1.1 1.0

1 0

1.4

0.9 0

50 100 150 200 250 300 350 400 450 Cumulative RBM dose (mGy)

0

100

200

300

400

500

600

700

Cumulative dose (mGy)

Figure 13–​4.  Relative risk (and 90% CI) for leukemia (excluding CLL) and solid cancer according to the cumulative radiation dose (Gy) from occupational radiation exposures in the INWORKS cohort. Source: Leuraud et al. (2015); Richardson et al. (2015).

600,000 workers, including underground miners exposed to radon, nuclear power plant workers, and medical personnel (Ashmore et al., 1998). The cohort or subcohorts within it have been periodically linked to mortality and cancer registry databases (Sont et  al., 2001). Mean cumulative doses were highest in nuclear power workers (20 mSv) and lowest in dental and medical workers (0.3 and 4 mSv, respectively). The study found a significantly elevated ERR/​Sv among male (but not female) registrants, including all leukemias combined, all cancers excluding leukemia, and for many individual cancer sites (Sont et al., 2001). Since then, studies assessing the impact of dose censoring (i.e., doses below certain thresholds were unrecorded during the early registration years) and cancer risk in a subgroup of medically exposed registrants have been published (Shin et al., 2005; Zielinski et al., 2009). No extension of follow-​up has occurred past the late 1980s (cancer incidence) or mid-​1990s (cancer mortality) for the CNDR. Since the publication of the 15-​country study, critical attention has focused on the completeness of the dosimetry and registry information for one of the component cohorts (Atomic Energy of Canada Ltd; Ashmore et  al., 2010). It was noted that risk estimates for this cohort greatly increased in studies based on CNDR data (e.g., Cardis et  al., 2005, 2007; Sont et al., 2001; Zablotska et al., 2004) as compared to earlier studies based directly on the facility data (e.g., Cardis et  al., 1995). Efforts are currently focused on ascertaining that dosimetry and registrants in the CNDR are complete and accurate (Ashmore et al., 2010; Zablotska et al., 2014). A new, large-​scale study, the Million Worker Study (MWS) has been developed with the aim of evaluating cancer risk from protracted radiation exposures of < 100 mGy by combing various groups of US workers, including “atomic veterans” (i.e., military personnel who witnessed atomic test blasts), nuclear weapons workers, nuclear power plant workers, industrial radiographers, and medical workers exposed to radiation (Bouville et al., 2015). While the proposed sample size is impressive, at this stage the epidemiological methods for obtaining complete and accurate follow-​up for each cohort (including confirming who is alive, and cause of death) have not been described and may be prohibitively expensive. Inaccurate follow-​up can be an important, but often overlooked, source of bias in epidemiological studies; non-​differential outcome misclassification will bias risk estimates towards the null. There are also considerable challenges in the dose reconstruction, including substantial dose from radionuclides among some contributing DOE cohorts, which will require extensive dose reconstruction efforts and possibly involve large uncertainties (e.g., Mound and Rocketdyne; Bouville et  al., 2015). For the proposed group of 240,000 medical workers, the dosimetric challenges will include non-​uniform exposures, lack of compliance with badge monitoring, employment in multiple facilities, and variation

in radiation protection such as lead apron usage (Linet et al., 2010; Simon et al., 2014).

Flight Crew

Studies of cancer in flight crew have provided equivocal evidence of cancer risk following repeated exposure to cosmic radiation. Typical doses in these studies are low and have little within-​cohort variability:  for example, a mean absorbed, whole-​body dose of 12 mGy (standard deviation 11 mGy) was observed in a recent breast cancer incidence study among flight attendants (Schubauer-​Berigan et  al., 2015a). Empirical evidence also suggests that the neutron dose quality factor associated with commercial flight conditions is lower than previously thought (e.g., 2–​2.5; Burda et  al., 2013), amplifying the low-​dose limitations of these studies. Recent studies of flight crew, including a pooled international study, found no elevation in breast cancer mortality (compared to the general population) and no association with cosmic radiation (Hammer et al., 2014; Pinkerton et al., 2012). A large, pooled Scandinavian cancer incidence and case-​control study found a 50% higher risk of breast cancer than in the general population, but no association with cosmic radiation or other occupational exposures (Pukkala et al., 2012). Another large study of breast cancer incidence among US flight attendants found that, while the rate of incident breast cancer was 37% higher than among the general US population, the increase was probably due to the much older age at first birth and lower parity in the cohort compared to US women (Schubauer-​ Berigan et al., 2015a). No association was observed between cosmic radiation dose and breast cancer incidence, except among high-​parity women and those with younger age at first birth (Pinkerton et al., 2016; Schubauer-​Berigan et al., 2015a). This pattern, which is the opposite of the expected interaction of radiation with reproductive risk factors (e.g., Ronckers et al., 2005), may be associated with circadian disruption, which was highly correlated with radiation dose in the cohort. In addition to breast cancer, the occurrence of melanoma has been of interest among flight crew. A recent meta-​analysis of 19 studies of cockpit and cabin crew found that, compared to the general population, melanoma incidence rates are approximately doubled for flight crew, with a similar excess observed in both cockpit and cabin crew (Sanlorenzo et  al., 2015). The authors attribute this excess to occupational sources, including potentially ultraviolet-​ A radiation and cosmic radiation, although no direct adjustment for factors such as recreational sun exposure was made.

Medical Workers

Studies of increased leukemia risks in the earliest radiologists provided the first evidence of a link between radiation and cancer (March, 1950). Most of the studies of medical radiation workers have limited

 239



Ionizing Radiation

exposure information, particularly for the early workers, when radiation protection and dosimeters were rare. A  new follow-​up of a cohort of 27,000 Chinese medical radiation workers with individual dose estimates (mean = 0.086 Gy) found a significant dose–​response relationship for solid cancer incidence that was higher than, but statistically compatible with, risk estimates from other low-​dose studies (Sun et al., 2016). Individual dose reconstruction has also recently been completed for the US Radiologic Technologists (USRT) cohort of about 110,000 (mostly female) workers (Simon et al., 2014). There was no evidence of a dose–​response relationship for basal cell skin cancer in these workers, even though some received estimated skin doses of > 1 Gy (Lee T et al., 2015), or for brain cancers, even among those who performed fluoroscopically guided procedures (Kitahara et  al., 2017). A significant dose–​response relationship was found, however, for breast cancer for those workers who started work before 1950 and received the highest doses (estimated mean organ dose = 280 mGy) (Preston et al., 2016). Dose–​response analyses for other cancer sites are ongoing, but power is generally limited for sites other than breast cancer due to low mean doses and relatively small numbers of cases. Some case reports of brain tumors in physicians who perform fluoroscopically guided procedures have raised concerns about current exposure levels in these physicians (Roguin et al., 2013). In a subset of the USRT cohort, technologists who reported performing fluoroscopically guided interventional procedures were at an approximately two-​fold increased relative risk for brain cancer mortality (HR = 2.55; 95% CI: 1.48, 4.40), as well as modestly elevated risk for incidence of breast cancer (HR = 1.16; 95% CI: 1.02, 1.32) and melanoma (HR = 1.30; 95% CI: 1.05, 1.61) compared to those who never performed these procedures. No elevated risk was observed for cancers of the thyroid, non-​melanoma skin, prostate, lung, colon-​rectum, or non-​CLL leukemia in workers who performed these procedures. While exposure to radiation in the workplace is one possible explanation for these findings, the role of chance or unmeasured confounding by non-​radiation risk factors cannot be ruled out until these results are replicated in more studies, and as noted above in subsequent analyses there was no relationship with dose (Rajaraman et al., 2016; Kitahara et al., 2017). The first study to examine this question systematically in physicians, did not find evidence of an increased risk of brain tumors (Linet et al., 2017). Nuclear medicine procedures have also increased dramatically in the United States, and some radiologic technologists who routinely perform these may be exposed above the recommended dose limits. In an analysis of technologists in the USRT cohort who ever (versus never) perform various nuclear medicine procedures, there was an increased risk for squamous cell carcinoma of the skin (HR = 1.29; 95% CI: 1.01, 1.66) with ever performing diagnostic radionuclide procedures and of breast cancer (HR = 2.68; 95% CI: 1.10, 6.51) with ever performing other radionuclide therapy procedures (excluding brachytherapy and radioactive iodine) (Kitahara et al., 2015). Increasing risks were also observed with greater frequency of performing these procedures, particularly before 1980. Reconstruction of individual doses to a group of nuclear medicine technologists is underway, and further follow-​up is needed of the potential long-​term risks in these occupational groups.

EXPOSURE AND HOST PARAMETERS THAT INFLUENCE RISK The focus of many of the current radiation epidemiology studies is the understanding of the effect of dose protraction or fractionation. This has important practical implications for predicting cancer risk from occupational, environmental, and diagnostic medical exposures, which are typically received as low and fractionated (or protracted) doses and also contribute most to the collective dose received in the general population. Theoretically, the risk per unit dose could be lower if the dose is protracted or fractionated because of DNA repair mechanisms. The results from low-​dose epidemiological studies have been mixed. However, it is difficult to find exposed populations that are identical in all ways other than the manner in which they received their radiation exposure. The most authoritative risk assessments for

239

low and protracted doses of radiation have been carried out by the US National Academies as a series of reports on the “Biological Effects of Ionizing Radiation” (BEIR) and by the United Nations Committee on the Effects of Atomic Radiation (UNSCEAR, 2010b). The most recent BEIR report (BEIR VII Phase 2, 2006)  reviewed the animal and epidemiological data and derived a dose-​and-​dose-​rate adjustment factor (DDREF) of 1.5 (95% credible interval), which implies that the risk per unit dose is reduced by one-​third if the dose is protracted or below 100 mGy. However, the recent studies of occupationally exposed individuals, such as INWORKS, natural background radiation, Techa River residents, and the UK CT (computed tomography) study reported dose–​response relationships from these protracted/​fractionated low-​dose exposures that were compatible with the Japanese atomic bomb survivors, implying compatibility with a DDREF of 1. In the next few years there will be additional results from nuclear worker and CT scan cohorts that will provide further information on this question, which is central to radiation protection limits. The impact of fractionation and dose level is also critical in the study of second cancer risks after radiotherapy to aid clinical decision-​ making, and also provides insights into biological mechanisms. Mostly the dose–​response studies have found that the risk per unit dose is 5–​10 times lower following these high-​dose (5+ Gy) highly fractionated (e.g., 40 fractions) exposures compared to the acute exposure of < 2 Gy in the Japanese atomic bomb survivors (Berrington de Gonzalez et al., 2013). The theory is that both cell killing and DNA repair contribute to the lower cancer risk per unit dose. Surprisingly, these studies have also shown that there is little evidence that the dose–​ response curve is nonlinear in the direction of a downturn in risk, even at organ doses of ≥ 60 Gy. It was previously assumed that high levels of cell killing would result in a downturn or plateau in risk at very high doses (e.g., > 5Gy organ dose; Gray, 1965). Thyroid cancer is the only site currently where there is strong evidence of a downturn above doses of 20 Gy to the thyroid from radiotherapy for childhood cancer (Veiga et al., 2012). The lack of downturn in other cancer sites has been hypothesized to be related to cellular repopulation between fractions (Sachs and Brenner, 2005). The type of ionizing radiation exposure also influences the risk of cancer. As described earlier (see section “Units of Exposure”), neutrons may be 20 times more carcinogenic per unit dose than X-​rays. Most of the estimates of the relative biological effectiveness of different types of radiation exposure come from animal or other laboratory studies, as few human studies are available where the populations are comparable other than with respect to the type of radiation. More human studies are needed, however, especially of the RBE of neutrons and protons due to the increasing use of proton therapy to treat cancer patients. Certain types of proton therapy systems expose the patient to a whole-​body scatter dose of neutrons, and the long-​term risks from this are uncertain. The impact of age at radiation exposure has been studied extensively (BEIR VII Phase 2, 2006). These studies have established that childhood exposure to ionizing radiation results in a higher cancer risk than adulthood exposure; the one possible exception is for lung cancer (Preston et al., 2007). Some organs are especially radiosensitive in childhood, including the red bone marrow, thyroid, breast, and brain (Braganza et al., 2012; Hsu et al., 2013; Preston et al., 2002a, 2002b; Veiga et al., 2016). In contrast, the thyroid and brain seem to have very low radiation sensitivity if exposure is in adulthood (Braganza et al., 2012; Preston et al., 2007; see also Chapter 44 in this volume). Long-​term follow-​up of more than 50 years of the Japanese atomic bomb survivors and other studies, including the Canadian fluoroscopy cohort (Howe and McLaughlin, 1996), have demonstrated that the cancer risk from ionizing radiation exposure remains elevated for decades, probably for the rest of a person’s lifetime; that is, unlike smoking, the risk never returns to baseline (BEIR VII Phase 2, 2006). This is consistent with the notion of ionizing radiation being an initiator, as well as potentially a cancer promoter. The risk does decline, however, with time since exposure. In many studies, women have higher cancer risks than men following radiation exposure. For example, the ERR/​Gy in the LSS for all solid cancers (exposure at age 30, attained age 70) is 0.58 for women

240

240

PART III:  THE CAUSES OF CANCER

and 0.35 for men, and differences remain even after removal of sex-​ specific cancers (Preston et al., 2007). However, it is uncertain whether this reflects true biological differences in radiosensitivity, interactions with other cancer risk factors, or differences in background cancer rates. There have been a large number of studies of the joint effect of smoking and radiation exposure. The findings are not consistent, varying from a complicated joint effect model in the Japanese atomic bomb survivors to multiplicative in several studies of radiotherapy and second cancers. Studies of high-​LET radiation, including those of uranium miners exposed to radon progeny and the Mayak plutonium cohort, have found joint effects that appear to be super-​additive but sub-​multiplicative (Gilbert et al., 2013). Reproductive factors and breast cancer risk have also been evaluated in several radiation-​exposed populations, but there is little consistency with findings regarding the joint effects (Ronckers et al., 2005). Studies of genetic susceptibility also have not produced clear-​cut findings, even where there is an obvious potential mechanism such as DNA repair and BRCA mutation (see discussion of radiosensitivity later in this chapter). Given how well the risks from ionizing radiation are known, these difficulties with joint effects analyses are a reminder about how difficult these questions are to address via epidemiological studies.

MECHANISMS OF CARCINOGENESIS Each of the most common types of ionizing radiation exposure in humans (fractionated high-​dose exposures such as those experienced by patients undergoing cancer radiotherapy; acute low-​ moderate doses, such as those experienced by many Japanese atomic bomb survivors; chronic low-​dose exposures received by radiation workers; and fractionated low-​dose exposures from diagnostic medical examinations) has been associated with increased risk of cancer. The development of cancer is characterized by distinct biological abilities acquired during the multistep development of human tumors. These include sustained proliferative signaling, evasion of growth suppressors, resistance of cell death, induction of angiogenesis, activation of invasion and metastasis, reprogramming of energy metabolism, and evasion of immune destruction (Hanahan and Weinberg, 2011). These hallmarks of cancer are driven by genomic instability (which generates genetic diversity leading to the acquisition of hallmark features) and inflammation (which fosters multiple hallmark functions). Cellular damage from exposure to ionizing radiation occurs when radiation is absorbed by biological tissue and interacts either directly or indirectly with atoms of critical targets. As radiation moves through the tissue, energy is deposited, causing ionization (ejection of an electron from an atom) along the track, with some clustering at the ends. Direct action results when the radiation energy itself causes ionization of the critical target. This is the dominant process for radiation with high linear energy transfer (LET), such as neutrons or alpha-​particles (Hall and Giaccia, 2006). Indirect action occurs when radiation interacts with other atoms or molecules in the cell, such as water, to produce highly reactive free radicals that can break chemical bonds and damage the critical target, initiating the chain of biological events that eventually leads to cancer. The carcinogenic effects of ionizing radiation generally result from various types of damage to DNA, including damage to nucleotide bases, single-​strand breaks (SSBs), double-​strand breaks (DSBs), and DNA crosslinks, often forming clustered/​multiple damage sites which represent two or more lesions formed within one or two helical turns of DNA (BEIR VII Phase 2, 2006). Base damage is repaired by mechanisms that involve excision and replacement of individual damaged bases (base-​excision repair) or larger oligonucleotide fragments (nucleotide excision repair). The repair of SSBs uses a similar process to base-​excision repair. The repair of DSBs, on the other hand, usually involves several processes. In some instances, DSBs are rejoined end to end in a process called non-​homologous end joining. An alternative pathway for DSB repair is the homologous recombination process, in which the broken strand is repaired by crossing over with an adjacent identical DNA sequence. In addition

to direct DNA damage, radiation also causes the formation of reactive oxygen species (ROS), which are free radicals involving oxygen, and reactive nitric oxide species (RNOS), which induce stress responses, inflammation, and release of cytokines, growth factors, and chemokines (Barnett et al., 2009). In normal circumstances, when cells detect DNA damage caused by free radicals, they respond by undergoing cell cycle arrest in order to repair the damage, and the majority of the damage is repaired. When this damage cannot be repaired, the cell undergoes death via necrosis or apoptosis within a few cell divisions. Incorrectly repaired or unrepaired DNA damage, on the other hand, can lead to mutation and genomic instability, and cancer can occur after many years. Until recently, research in radiobiology focused mainly on pathways related to the repair of DNA damage. Although the role of repair pathways remains a key area of investigation, there is increasing interest in understanding the mechanisms involved in radiation-​induced inflammatory and stress responses to ROS. These mechanisms are particularly important for understanding the effects of radiation on cells that are not directly irradiated (West and Barnett, 2011).

Types of Evidence It is becoming increasingly apparent that the assessment of disease risk following radiation exposure is a complex process that needs to extend beyond traditional assessment by epidemiologic methods to incorporate biological evaluation of differences in susceptibility between individuals. Biological changes leading to cancer can be studied at different levels, and using various endpoints. At the cellular level, radiosensitivity measures the degree of response of a cell to radiation, with a stronger response indicating higher sensitivity. Typical cellular endpoints include phenomena such as cell killing or measurable chromosomal damage. Radiation response can also be observed at the tissue level: tissues can be more or less sensitive to radiation. Finally, differences in susceptibility can be observed at the level of individuals, with some humans being less tolerant of the effects of a fixed amount of radiation exposure than others (Advisory Group on Ionising Radiation [AGIR], 2013). The path from radiation exposure to the development of cancer can be represented as a theoretical continuum of effects that includes external dose, internal dose, early biological effects, and finally, disease outcome. Potential exposure to other genotoxic agents can also play a role. Measures of different biological changes can be used to quantify various points within the continuum between external radiation dose and final outcome of interest. A biological marker (or biomarker) may be suitable as a measure of dose, or as a measure of early or late effect, or in some scenarios may serve as a measure of both internal dose and effect, whereby some measure (e.g., induction of chromosome aberrations) could tell us not only about the likely dose of ionizing radiation experienced by an individual, but could also be used to predict risk of disease outcome(s) of interest (Pernot et al., 2012; see also Chapter 6 in this volume). Several factors need to be considered in the selection of appropriate biomarkers for characterization of radiation risk. First, natural and meaningful variation of the proposed marker should exist in the human population. A test for a reliable biomarker should be sensitive and specific, and highly reproducible among different laboratories. For use in human populations, the ideal biomarker should be easily obtained with minimal discomfort or risk to the patient. Given that levels of a biomarker often change over time and can differ between cells or tissues, details such as the appropriate timing of obtaining the sample from which the biomarker is to be measured and the biological source are crucial. Finally, a rapid readout of results is preferable. Although a single test may not possess all of these characteristics (e.g., a highly reliable test may not have a rapid readout), and some characteristics may be more important than others depending on the intended use of the biomarker, it is useful to keep all in mind during biomarker selection.

 241



Ionizing Radiation

Cellular Assays to Measure Biological Changes in Humans

The first human cell studies of radiosensitivity were clonogenic assays performed on fibroblasts cultured from the skin of individuals with severe radiation toxicity. In these assays, different levels of radiation are applied to samples of cells that are then plated in a tissue culture vessel and allowed to grow, and the resulting colonies are fixed, stained, and counted. At the conclusion of the experiment, the percentage of cells surviving at each dose is measured, and a cell survival curve is plotted to graphically represent the dose of ionizing radiation versus survival (Franken et al., 2006). These initial studies indicated extreme radiosensitivity in individuals with severe radiation toxicity (Arlett and Priestley, 1983; Smith et al., 1980; Woods et al., 1988). Further studies demonstrated that fibroblast sensitivity significantly differed between cells cultured from individuals with and without known genetic syndromes associated with radiosensitivity, such as patients with ataxia telangectasia (AT) or retinoblastoma (RB) (Deschavanne et  al., 1986). The variation in sensitivity between normal cells and cells with mutations characterized by defects in repair mechanisms seems to be greater when doses are delivered as chronic low dose-​rate exposures versus doses delivered at high dose rates (Kato et al., 2009). Although these observations suggested the possibility of using cellular radiosensitivity to predict a patient’s probable reaction to radiotherapy, studies have not indicated any clear and consistent associations between cellular radiosensitivity and reaction to radiation exposure in patients undergoing radiotherapy for cancers of the breast (Djuzenova et al., 2006; Peacock et al., 2000), cervix (West et al., 2001), or head and neck (Geara et  al., 1993; Rudat et  al., 1999). Furthermore, the relationship between acute toxicity after radiotherapy and late effects (e.g., cancer) remains unclear. Most cellular assays of cancer susceptibility in human populations to date have focused on DNA repair capacity. These assays generally compare DNA damage induced in circulating lymphocytes from cancer patients to circulating lymphocytes from cancer-​free controls, with the subsequent quantification of repair in both groups. Ionizing radiation (or some other agent, such as the radiomimetic chemical bleomycin) is applied to the cell culture, a specified period of time is allowed to pass for repair to occur, and the remaining damage is measured in a variety of ways (e.g., unrepaired SSBs, DSBs, or the incorporation of a radioisotope). Cellular assays of DNA repair can be broadly grouped into the following categories (Berwick and Vineis, 2000): 1. Tests based on DNA damage to cells induced by chemical or physical agents (e.g., bleomycin or ionizing radiation), such as the mutagen sensitivity assay, the G2-​ radiation assay, the micronucleus assay, and the comet assay. 2. Indirect tests of DNA repair, such as assays of unscheduled DNA synthesis, activity of repair enzymes, or measures of gamma-​ H2AX or 53BP1 foci. Levels of enzyme activity or DNA synthesis are measured in radiolabeled cells, usually by scintillation counting or radiography. Gamma-​H2AX or 53BP1 foci are assessed using fluorescence microscopy. 3. Tests based on more direct measures of repair kinetics (e.g., plasmid host cell reactivation assay). Separate sets of fresh or cryopreserved lymphocytes are transfected with both damaged and undamaged plasmids incorporating the chloramphenicol acetyltransferase (cat) or Luciferase gene as a marker. Repair can then be measured as a rate, such as the amount of radiation or fluorescence at a given point in time. Early assays of DNA repair measured chromatid breaks and gaps following the administration of bleomycin (Hsu et al., 1989) or gamma-​ radiation in the G2 phase of the cell cycle (Sigurdson and Stram, 2012) to assess repair efficiency following exposure to various mutagens in a range of cell types (Berwick and Vineis, 2000; Li et al., 2009; Wu et al., 2007). Studies using the G2 assay have demonstrated greater sensitivity to radiation in first-​degree relatives of cancer patients compared to non-​cancer controls (Kato et  al., 2009). Tests of genotoxic response to bleomycin in lymphocytes blood cultures indicate a wide variance in normal individuals (0.20 to more than 2.00 average chromatid breaks per cell). While one study showed distinctly different

241

responses for patients with cancers of the colon, upper aerodigestive tract, and lung from those of the control population, the sensitivity profile of patients with breast cancer was similar to that of the control population, suggesting that mutagen sensitivity response differs between tissues (Hsu et al., 1989). More recent assays for the detection of DNA double-​strand breaks quantify the levels of nuclear foci of DNA damaged proteins, including the gamma-​H2AX and 53BP1 assays, which complement more traditional DNA damage/​repair assays, such as pulsed field gel electrophoretic assessment of DNA double strand breaks and comet assays (AGIR, 2013). The gamma-​H2AX assay uses immunocytochemistry to visualize foci of phosphorylated histone 2AX (H2AX), which accumulates at the site of double-​strand break damage during activation of the DNA damage response signaling pathway (Nikolova et al., 2014). Studies using these newer assays indicate an approximately 1.5-​fold reduction in the repair efficiency of strand breaks among AT heterozygotes and RB family members versus phenotypically normal individuals (Sigurdson and Stram, 2012). Apart from DNA repair assays, other techniques in development include the ability of fibroblasts to undergo radiation-​induced differentiation, telomere length assays, and combinations of various cellular assays (AGIR, 2013), as well as measuring the expression of markers of inflammation (e.g., plasma transforming growth factor [TGF]-​β1) in serum or plasma. Given the rapidly evolving nature of this field, work in this research area is likely to undergo considerable evolution in the next few years.

Evidence for Differing Sensitivity to Radiation Exposure from Epidemiological Studies As illustrated in Table 13–​3, data from the LSS have clearly indicated that different organs and tissues have different sensitivities to the effects of radiation exposure. In atomic bomb survivors, for example, excess relative risks due to radiation are particularly notable for cancers of the esophagus, colon, lung, breast, ovary, and bladder (Preston et al., 2007). The idea that some people are more sensitive to the effects of radiation has been accepted for several decades. Individuals with certain rare hereditary disorders (e.g., ataxia telangiectasia and Nijmegen breakage syndrome) are known to be particularly sensitive to the effects of radiation (Taalman et al., 1983; Taylor et al., 1975). However, these cancer susceptibility syndromes affect only a small proportion of the general population. More relevant for the majority of the population is the theory that some part of the genetic contribution defining radiation susceptibility is likely to follow a polygenic model, whereby inheritance of several low-​penetrance risk alleles can lead to elevated risk of cancer. This theory (the “common-​variant-​common-​disease” model) is supported by the fact that multiple genetic pathways have been implicated in radiosensitivity, including DNA damage repair, radiation fibrogenesis, oxidative stress, and endothelial cell damage (Barnett et al., 2009). Most studies of the effects of ionizing radiation in humans have focused on the DNA repair pathway. A  review of population-​based studies conducted prior to 2000 found that the combined evidence for tests based on DNA damage and direct/​indirect tests of DNA repair indicated mainly positive associations between DNA repair capacity and occurrence of cancer (Berwick and Vineis, 2000; Vineis and Perera, 2000). However, these studies faced a number of limitations, including small sample size, the use of the case-​control study design (which cannot address the question of temporality between exposure and disease), the use of “convenience controls,” the possibility of confounding by factors other than ionizing radiation that could affect DNA repair, and use of cells different from the target organ. Subsequently, a number of population-​based epidemiological studies have examined susceptibility to radiation-​related risk of cancer with respect to genes in the DNA repair and other relevant biological pathways. Results from the “candidate-​SNP” approach, which assumes prior knowledge of one or more functional single nucleotide polymorphisms (SNPs), have not been convincingly replicated to date (Bhatti et  al., 2008,

24

242

PART III:  THE CAUSES OF CANCER

2010a; Bondy et al., 2001; Hu et al., 2002; Liu et al., 2010; Rajaraman et al., 2008; Sigurdson et al., 2007, 2009). An alternate approach to examine genetic susceptibility is to quantify cancer risk in groups of high-​risk individuals. Most (Andrieu et al., 2006; Gronwald et al., 2008; Lecarpentier et al., 2011), but not all (John et al., 2013) interview-​based studies of diagnostic chest X-​ rays in BRCA1 and BRCA2 mutation carriers have reported a slightly elevated risk of breast cancer in at least one subgroup of X-​ray exposure (e.g., exposure at younger age). On the other hand, the association has generally been null for mammograms (Giannakeas et  al., 2014; Goldfrank et al., 2006; Narod et al., 2006). Breast doses from chest X-​ rays are generally very low, and recall bias cannot be ruled out. Studies have also reported some indication of increased risk following radiotherapy in individuals carrying rare mutants in genes in DNA damage repair pathways. A  case-​only study of contralateral breast cancer (CBC) reported increased prevalence of pathogenic germline mutations in DNA repair genes BRCA1, BRCA2, CHEK2, and ATM in women with CBC following radiotherapy for their primary breast cancer compared with women with CBC who were not irradiated for their first breast cancer (Broeks et  al., 2007). A  larger case-​control study of women with CBC (cases) compared to matched controls with unilateral breast cancer reported evidence of greater CBC risk following radiation among women who carried rare ATM missense variants (Bernstein et al., 2010) and individuals who carried a rare haplotype in RAD50 (Brooks et  al., 2012)  than in unexposed carriers, but no increase in risk with radiation dose in carriers of other rare variants, including BRCA1/​2 mutations (Bernstein et al., 2013). The interpretation of these genetic findings (candidate SNP, radiation exposure in high-​risk genetic populations, or rare mutations in patients exposed to radiotherapy) has been complicated by the fact that many of these studies are subject to one or more of the following: lack of replication in an independent population; exposure based on self-​ report; lack of a consistent dose–​response association; subgroup findings that could be due to chance; and overlap of study populations. While earlier genetic studies focused on a handful of candidate genes, the development of technologies that can rapidly analyze thousands of genetic markers at relatively low cost, along with the mapping of linkage disequilibrium between common SNPs across the genome (International HapMap et  al., 2010)  and the definition of functional elements critical for regulation and genomic stability (Encode Project Consortium, 2012), have allowed us to comprehensively examine the approximately 25,000 coding genes and associated functional elements. The genome-​wide association study (GWAS) approach has successfully identified hundreds of risk loci in germline DNA for various cancers (Chung and Chanock, 2011). However, the assessment of gene–​environment interaction for most environmental carcinogens, including radiation, has remained elusive. A GWAS of 100 Hodgkin lymphoma survivors treated with radiotherapy identified two loci in the PRDM1 gene potentially associated with increased risk of secondary malignancy (Best et  al., 2011), but these loci have yet to be consistently confirmed. Genome-​wide association studies have also been completed in adult women with contralateral breast cancer in the WECARE study (Bernstein et  al., 2004)  and in individuals with subsequent malignancies in the Childhood Cancer Survivor Study (Robison et  al., 2009), both studies having detailed radiation doses from radiotherapy. Results from these additional GWAS are expected to be published shortly. For additional details on genetic susceptibilility to second cancers following radiotherapy, see Chapter 60.

Radiation Signatures The pursuit of a “molecular signature” that would be able to discriminate radiation-​induced tumors from “sporadic” tumors has obvious and important implications for radiation protection. The search for ionizing radiation signatures started several decades ago, particularly in the context of sarcomas after radiotherapy and thyroid cancers following head/​neck radiotherapy and after environmental radiation exposure. For sarcomas, most of the studies of molecular signatures use some modification of the 1948 Cahan criteria for classifying tumors as radiation-​induced: the tumor must be in the irradiated field, must

be histologically different from the primary cancer, and must have a latency period of at least 5 years (Cahan et al., 1948). While satisfying these criteria is likely to result in an increased probability that the tumor is radiation-​related, it does not guarantee that it was radiation-​induced. Early work in search of a radiation signature for sarcoma mainly used conventional cytogenetic analysis to identify large-​scale chromosomal abnormalities. These studies were limited by small sample sizes and lack of a comparison group of non-​radiation-​induced tumors. Later studies using polymerase chain reaction followed by direct sequencing were able to examine mutations in specific genes such as TP53, but did not address the probability of multi-​gene involvement, and potential gene–​gene interactions (Berrington de Gonzalez et al., 2012). A more recent study using microarray analysis to identify a signature of 135 genes from a learning/​ training set of 12  “radiation-​ induced” and 12 “sporadic sarcomas” was able to discriminate “radiation-​induced” from “sporadic sarcomas” in an independent set of 36 sarcomas with 96% sensitivity and 62% specificity, but this study remains to be independently replicated (Hadj-​Hamou et al., 2011). For thyroid tumors, several lines of evidence indicate that radiation-​ related tumors may present differently than sporadic tumors. A high frequency of RET/​PTC (RET proto-​oncogene/​papillary thyroid carcinoma) gene rearrangements has been observed in papillary thyroid carcinomas (PTCs) of children exposed to radioactive fallout in Belarus after the Chernobyl accident (Fugazzola et al., 1995; Ito et al., 1994; Leeman-​Neill et  al., 2013)  and in thyroid tumors following external radiotherapy (Bounacer et al., 1997). The relationship between these gene rearrangements and radiation dose, however, remains unclear: no significant association was observed between RET/​PTC activation and individual I-​131 doses in post-​Chernobyl cancers occurring in the Russian Federation (Tuttle et al., 2008); higher doses were associated with higher prevalence of RET/​PTC rearrangements in PTCs of atomic bomb survivors (Hamatani et  al., 2008); and recent data from post-​ Chernobyl thyroid cancer in Ukraine indicate a non-​monotonic relationship between I-​131 dose and RET/​PTC or PAX8/​PPARγ-​positive tumors and dose, with increased risk at low-​to moderate doses and decreased risk at high doses (Leeman-​Neill et al., 2013). A genome-​ wide comparison of somatic copy number alternations (CNAs) between age-​and residence-​ matched exposed and non-​exposed PTC patients from Chernobyl observed a copy number gain of chromosome bands 7q11.22-​11.23 in a higher proportion of radiation-​exposed versus unexposed individuals, with specific m-​RNA overexpression of the CLIP2 gene located within this band (Hess et al., 2011). While a positive dose-​response relationship was observed between I-​131 dose and binary CLIP2 typing for young PTC cases (age at exposure less than 5 years), this relationship was not observed for those exposed at age ≥5  years (Selmansberger et  al., 2015), and these results have yet to be confirmed in other populations. As with studies of germline DNA, advances in DNA sequencing technology in the last decade have yielded many insights into somatic mutations from tumor tissue, largely driven by findings from The Cancer Genome Atlas (TCGA) and the International Cancer Genome Consortium (ICGC). These projects characterize the genome, as well as various aspects of the transcriptome and epigenome, to give a fuller understanding of how genes contribute to tumorigenesis. Over 30 distinct human tumor types have been analyzed to date through large-​ scale genome sequencing and integrated multidimensional analyses, yielding insights into both individual cancer types and across cancers, particularly with respect to the accurate molecular classification of tumors (Chin et al., 2011). While most of these discoveries have focused on the genome rather than associated environmental factors, a recent paper examining 4,938,362 mutations from 7042 cancers identified strong mutational signatures in tumor tissue marking exposure to tobacco carcinogens and ultraviolet (UV) irradiation (Alexandrov et al., 2013). The tobacco signal was most evident in cancers that are known to be tobacco-​related (lung, head and neck, and liver); and the UV irradiation signal was observed mainly in malignant melanoma and squamous carcinoma of the head and neck. A recent study identified potential radiation-​associated mutational signatures in a small number (n = 17) of radiation-​associated second malignancies (Behjati, 2016). Similar studies to detect

 243



Ionizing Radiation

243

a radiation signature, whereby tumors caused by ionizing radiation exposure could be discriminated from their sporadic counterparts, are currently underway in populations environmentally exposed to high doses of ionizing radiation. The best target populations for these studies are those with a high proportion of cancers likely to be attributable to radiation and well-​characterized doses. Results of these studies are expected over the next few years.

Overall, when all sources of radiation were considered in the UK population, Darby and Parkin estimated that about 2% of cancers could be attributable to radiation (Parkin and Darby, 2011). Background radiation/​ radon and diagnostic X-​ rays contributed approximately equally, followed by radiotherapy.

ESTIMATED ATTRIBUTABLE FRACTION

There are a number of approaches to reduce unnecessary radiation exposure to the general population. For example, residential radon exposure can be reduced by a variety of relatively simple and cheap remediation strategies such as improved ventilation, which should also be implemented in schools and workplaces. Diagnostic medical radiation exposure should be limited to clinically justified procedures where the clinical decision-​making depends on the outcome of the test. In addition, doses should be kept as low as reasonably possible. There is increased awareness about dose levels from diagnostic procedures, and new technologies can monitor or automatically control dose, such as automated exposure control for CT scans. Treatment-​planning systems for radiotherapy aim to maximize tumor control and minimize dose to the surrounding normal tissue. This should reduce the amount of normal tissue exposed to very high doses. Many of these systems, such as IMRT, result in increased whole-​body exposure to low levels of radiation due to scatter and the number of monitor units involved. It remains uncertain how this impacts the overall risk–​benefit profile of newer technologies. The current exceptions are flight crew, who are currently unmonitored, and certain medical workers who perform fluoroscopically guided procedures or who handle radionuclides. The length and complexity of these procedures and the need for close proximity to the beam and patient results in radiation exposure to various exposed parts of the operator’s body, including the hands, lens of the eye, and brain. Many of the physicians who perform these procedures have limited training in radiation protection. Heavy protection aprons, gloves, or glasses can make it more difficult to perform the procedures, and there is a need for a careful balance between radiation protection and comfort and clinical performance. The development of personal protective technology that does not impede comfort or clinical performance could help reduce exposures among medical workers. Badge monitors need to be worn routinely to better monitor the cumulative exposures to these physicians.

As ionizing radiation is an established carcinogen, it is appropriate to estimate the potential fraction of cancers that may be attributable to radiation from various sources, including medical, natural background, and nuclear accidents in exposed populations. Some estimates come directly from the study populations, such as the LSS of the Japanese atomic bomb survivors. For low-​dose exposures such as residential radon or diagnostic radiation exposure, attributable risks can be estimated, under certain assumptions, using indirect risk projection methods. Because radiation has been shown to increase cancer risk for at least 50 years after exposure (Preston et al., 2007), attributable risk estimates should be based on studies with very long-​term follow-​up or life-​table methods, with adjustment for competing risks.

Environmental Radiation Exposure Indirect risk modeling estimates suggest that 5% of leukemia in England at any age, and about 15% of childhood leukemias, could be related to background radiation (Kendall et al., 2011). Residential radon is estimated to be responsible for about 3% of lung cancers in the United Kingdom (Parkin and Darby, 2011). The Chernobyl accident is estimated to be responsible for 3%–​4% of cancer deaths in the most highly exposed groups: cleanup workers, residents, and evacuees (WHO, 2006). Radioactive fallout in Europe, by comparison, is estimated to be related to only 0.01% of cancers (Cardis et al., 2006). After 60 years of follow-​up in the LSS of the Japanese atomic bomb survivors, it is estimated that about 10% of the solid cancers were related to the radiation exposure (Preston et al., 2007).

Medical Radiation As described earlier, there are relatively few studies of the cancer risks from diagnostic radiation exposure due to the low doses per procedure. However, diagnostic radiation is the largest man-​made source of radiation, and there are concerns about whether doses are always optimized and procedures always justified. Indirect modeling approaches estimate that in more developed countries between 0.5% (in the UK) and 3% (in Japan) of cancers could be attributable to diagnostic medical radiation (Berrington de Gonzalez and Darby, 2004). With the dramatic increase in CT scans in the United States, the attributable risk could increase from 1% to 3% if current levels continue (Berrington de Gonzalez et al., 2009). Not all of these cancers are avoidable because many of these procedures have important benefits that need to be balanced against the potential risks. The proportion of second primary cancers that may be related to the radiotherapy for the first cancer was estimated in the United States from the SEER cancer registries as 8% overall for adulthood cancer routinely treated with radiation (Berrington de Gonzalez et al., 2011). Estimates for the United Kingdom are similar (6% for males and 8% for females) (Parkin and Darby, 2011). The attributable risk is likely higher for some childhood and young adulthood cancers with good survival such as Hodgkin lymphoma, as children are known to be more radiosensitve, and high relative risks of subsequent cancers have been reported. Estimates for all Hodgkin lymphoma survivors in the United Kingdom suggest that 16% of second cancers for males could be related to radiotherapy, and 19% for females (because of the high risk of second breast cancer) (Parkin and Darby, 2011). For testicular cancer the estimate was 11%, and for cervical cancer 17%; both occur at younger ages, and pelvic radiation exposes a number of organs.

OPPORTUNITIES FOR PREVENTION

FUTURE RESEARCH DIRECTIONS As one of the most extensively studied carcinogens, ionizing radiation is an ideal candidate for cutting-​edge cancer epidemiology, including searching for molecular signatures, risk-​projection modeling, uncertainty analyses, and comprehensive assessment of critical exposure periods. Currently, most of the exposures to the general population are low-​ level doses (effective doses < 10 mSv) from diagnostic medical exposures or background radiation. The magnitude of the cancer risks from these low-​dose exposures, and protracted or fractionated exposures, is more uncertain than that from higher, acute doses that have been studied for many decades in the Japanese atomic bomb survivors. Sample size, measurement error, and the length of follow-​up needed to evaluate cancer risks from low doses make it a difficult problem to study using traditional epidemiological methods. The most promising recent studies have used electronic record linkage with a retrospective study design, such as the UK natural background radiation case-​control study (Kendall et al., 2013), pediatric CT studies (Pearce et al., 2012), and the pooled nuclear workers studies (Schubauer-​Berigan et al., 2015b). This design facilitates large-​ scale populations with individualized dose estimates. In the next decade there will be new publications in all of these areas from large-​scale efforts including EPI-​CT, INWORKS, the Million Workers Study, the USRT cohort, and the UK natural background radiation study. The key limitation in many of these studies, however, is the lack of information on confounding factors, such as

24

244

PART III:  THE CAUSES OF CANCER

smoking in the nuclear worker studies and underlying medical conditions in the CT scan studies. Traditional epidemiology, therefore, may not be able to provide a definitive answer to the question of the risks from low-​dose or protracted radiation exposure. New molecular techniques could, however, provide new insights. In particular, the search for “signatures” that can identify radiation exposure as a causal factor for a particular tumor could technically provide a definitive answer to the low-​dose question if the signature were identified and then shown to be present in tumors after low-​dose exposure with evidence of no other clear cause of cancer. New epidemiologic studies of ionizing radiation and cancer will also likely continue to feature a search for common genetic markers that can identify individuals susceptible to radiation risk effects. These studies still face many challenges, including identifying a population with a high attributable risk, and high-​quality biospesimens, well-​characterized radiation exposure, and meaningful replication sets. The thyroid tumors from the Chernobyl Tissue Bank satisfy many of these requirements, and the first sequencing study is currently in progress. Other populations of interest include the Japanese atomic bomb survivors and second cancers that occur in or near the radiation field. The integrated characterization of germline and somatic alterations as genotyping and analysis methods evolve rapidly in the next decade could be a promising avenue of research. References Advisory Group on Ionising Radiation (AGIR). 2011. Risk of Solid Cancers Following Radiation Exposure:  Estimates for the UK Population. Report of the Independent Advisory Group on Ionising Radiation. London: Health Protection Agency. AGIR. 2013. Human Radiosensitivity. Report of the Independent Advisory Group on Ionising Radiation. London: Health Protection Agency. Alexandrov LB, Nik-​Zainal S, Wedge DC, et  al. 2013. Signatures of mutational processes in human cancer. Nature, 500(7463), 415–​ 421. PMCID: 3776390. Andrieu N, Easton DF, Chang-​Claude J, et al. 2006. Effect of chest X-​rays on the risk of breast cancer among BRCA1/​2 mutation carriers in the international BRCA1/​2 carrier cohort study:  a report from the EMBRACE, GENEPSO, GEO-​HEBON, and IBCCS Collaborators’ Group. J Clin Oncol, 24(21), 3361–​3366. PMID: 16801631. Arlett CF, and Priestley A. 1983. Defective recovery from potentially lethal damage in some human fibroblast cell strains. Int J Radiat Biol Relat Stud Phys Chem Med, 43(2), 157–​167. PMID: 6600729. Ashmore JP, Krewski D, Zielinski JM, et al. 1998. First analysis of mortality and occupational radiation exposure based on the National Dose Registry of Canada. Am J Epidemiol, 148(6), 564–​574. PMID: 9753011. Ashmore JP, Gentner NE, and Osborne RV. 2010. Incomplete data on the Canadian cohort may have affected the results of the study by the International Agency for Research on Cancer on the radiogenic cancer risk among nuclear industry workers in 15 countries. J Radiol Prot, 30(2), 121–​129. PMID: 20530869. Barnett GC, West CM, Dunning AM, et al. 2009. Normal tissue reactions to radiotherapy:  towards tailoring treatment dose by genotype. Nat Rev Cancer, 9(2), 134–​142. PMCID: 2670578. Beck HL, and Bennett BG. 2002. Historical overview of atmospheric nuclear weapons testing and estimates of fallout in the continental United States. Health Phys, 82(5), 591–​608. PMID: 12003011. BEIR VII Phase 2. 2006. Health Risks from Exposure to Low Levels of Ionizing Radiation. Washington, DC:  Board on Radiation Effects Research, National Research Council. Bennett BG. 2002. Worldwide dispersion and deposition of radionuclides produced in atmospheric tests. Health Phys, 82(5), 644–​ 655. PMID: 12003015. Bernstein JL, Langholz B, Haile RW, et  al. 2004. Study design:  evaluating gene-​environment interactions in the etiology of breast cancer:  the WECARE study. Breast Cancer Res, 6(3), R199–​R214. PMCID: 400669. Bernstein JL, Haile RW, Stovall M, et al. 2010. Radiation exposure, the ATM Gene, and contralateral breast cancer in the women’s environmental cancer and radiation epidemiology study. J Natl Cancer Inst, 102(7), 475–​ 483. PMCID: PMC2902825. Bernstein JL, Thomas DC, Shore RE et al. 2013. Contralateral breast cancer after radiotherapy among BRCA1 and BRCA2 mutation carriers: a WECARE study report. Eur J Cancer, 49(14), 2979–​2985. PMCID: PMC3755053. Berrington de Gonzalez A, Curtis RE, Kry SF, et al. 2011. Proportion of second cancers attributable to radiotherapy treatment in adults:  a cohort

study in the US SEER cancer registries. Lancet Oncol, 12(4), 353–​360. PMCID: PMC3086738. Berrington de Gonzalez A, and Darby S. 2004. Risk of cancer from diagnostic X-​rays: estimates for the UK and 14 other countries. Lancet, 363(9406), 345–​351. PMID: 15070562. Berrington de Gonzalez A, Gilbert E, Curtis R, et al. 2013. Second solid cancers after radiation therapy:  a systematic review of the epidemiologic studies of the radiation dose-​response relationship. Int J Radiat Oncol Biol Phys, 86(2), 224–​233. PMCID: PMC3816386. Berrington de Gonzalez A, Kutsenko A, and Rajaraman P. 2012. Sarcoma risk after radiation exposure. Clin Sarcoma Res, 2(1), 18. PMCID: PMC3507855. Berrington de Gonzalez A, Mahesh M, Kim KP, et al. 2009. Projected cancer risks from computed tomographic scans performed in the United States in 2007. Arch Intern Med, 169(22), 2071–​2077. PMID: 20008689. Berrington de Gonzalez A, Salotti JA, McHugh K, et  al. 2016. Relationship between paediatric CT scans and subsequent risk of leukaemia and brain tumours: assessment of the impact of underlying conditions. Br J Cancer, 114(4), 388–​394. PMCID: PMC4815765. Berwick M, and Vineis P. 2000. Markers of DNA repair and susceptibility to cancer in humans: an epidemiologic review. J Natl Cancer Inst, 92(11), 874–​897. PMID: 10841823. Best T, Li D, Skol AD, et al. 2011. Variants at 6q21 implicate PRDM1 in the etiology of therapy-​induced second malignancies after Hodgkin’s lymphoma. Nat Med, 17(8), 941–​943. PMCID: PMC3229923. Bhatti P, Doody MM, Rajaraman P, et al. 2010a. Novel breast cancer risk alleles and interaction with ionizing radiation among U.S. radiologic technologists. Radiat Res, 173(2), 214–​224. PMID: 20095854. Bhatti P, Struewing JP, Alexander BH, et  al. 2008. Polymorphisms in DNA repair genes, ionizing radiation exposure and risk of breast cancer in U.S. Radiologic technologists. Int J Cancer, 122(1), 177–​182. PMID: 17764108. Bhatti P, Veiga LH, Ronckers CM et  al. 2010b. Risk of second primary thyroid cancer after radiotherapy for a childhood cancer in a large cohort study: an update from the childhood cancer survivor study. Radiat Res, 174(6), 741–​752. PMCID: PMC3080023. Black D. 1984. Investigation of the Possible Increased Incidence of Cancer in West Cumbria. Report of the Independent Advisory Group. London: HMSO. Boice JD, Jr. 2006. Ionizing radiation. In:  Schottenfeld D, and Fraumeni JF, Jr. (Eds.), Cancer Epidemiology and Prevention (3rd ed., pp. 259–​293). New York: Oxford University Press. Boice JD, Jr, Leggett RW, Ellis ED, et al. 2006. A comprehensive dose reconstruction methodology for former rocketdyne/​atomics international radiation workers. Health Phys, 90(5), 409–​430. PMID: 16607174. Bondy ML, Wang LE, El-​Zein R, et al. 2001. Gamma-​radiation sensitivity and risk of glioma. J Natl Cancer Inst, 93(20), 1553–​1557. PMID: 11604478. Bosch de Basea M, Pearce MS, Kesminiene A, et  al. 2015. EPI-​CT:  design, challenges and epidemiological methods of an international study on cancer risk after paediatric and young adult CT. J Radiol Prot, 35(3), 611–​628. PMID: 26226081. Bounacer A, Wicker R, Caillou B, et  al. 1997. High prevalence of activating ret proto-​oncogene rearrangements, in thyroid tumors from patients who had received external radiation. Oncogene, 15(11), 1263–​ 1273. PMID: 9315093. Bouville A, Linet MS, Hatch M, Mabuchi K, and Simon SL. 2014. Guidelines for exposure assessment in health risk studies following a nuclear reactor accident. Environ Health Perspect, 122(1), 1–​5. PMCID: PMC3888574. Bouville A, Simon SL, Miller CW, et al. 2002. Estimates of doses from global fallout. Health Phys, 82(5), 690–​705. PMID: 12003019. Bouville A, Toohey RE, Boice JD, Jr, et al. 2015. Dose reconstruction for the million worker study: status and guidelines. Health Phys, 108(2), 206–​ 220. PMCID: PMC4854640. Braganza MZ, Kitahara CM, Berrington de Gonzalez A, et al. 2012. Ionizing radiation and the risk of brain and central nervous system tumors: a systematic review. Neuro Oncol, 14(11), 1316–​1324. PMCID: PMC3480263. Brenner AV, Tronko MD, Hatch M, et al. 2011. I-​131 dose response for incident thyroid cancers in Ukraine related to the Chornobyl accident. Environ Health Perspect, 119(7), 933–​939. PMCID: PMC3222994. Broeks A, Braaf LM, Huseinovic A, et al. 2007. Identification of women with an increased risk of developing radiation-​induced breast cancer:  a case only study. Breast Cancer Res, 9(2), R26. PMCID: PMC1868917. Brooks JD, Teraoka SN, Reiner AS, et  al. 2012. Variants in activators and downstream targets of ATM, radiation exposure, and contralateral breast cancer risk in the WECARE study. Hum Mutat, 33(1), 158–​164. PMCID: PMC3240722.

 245



Ionizing Radiation

Burda O, Sato T, and Wissmann F. 2013. Quality factor of secondary cosmic radiation at flight altitudes. J Radiol Prot, 33(2), 339–​348. PMID: 23480894. Cahan WG, Woodard HQ, and et al. 1948. Sarcoma arising in irradiated bone; report of 11 cases. Cancer, 1(1), 3–​29. PMID: 18867438. Cardis E, Gilbert ES, Carpenter L, et al. 1995. Effects of low doses and low dose rates of external ionizing radiation: cancer mortality among nuclear industry workers in three countries. Radiat Res, 142(2), 117–​ 132. PMID: 7724726. Cardis E, Krewski D, Boniol M, et al. 2006. Estimates of the cancer burden in Europe from radioactive fallout from the Chernobyl accident. Int J Cancer, 119(6), 1224–​1235. PMID: 16628547. Cardis E, Vrijheid M, Blettner M, et al. 2005. Risk of cancer after low doses of ionising radiation:  retrospective cohort study in 15 countries. BMJ, 331(7508), 77. PMCID: PMC558612. Cardis E, Vrijheid M, Blettner M, et al. 2007. The 15-​Country Collaborative Study of Cancer Risk among Radiation Workers in the Nuclear Industry: estimates of radiation-​related cancer risks. Radiat Res, 167(4), 396–​416. PMID: 17388693. Chang WP, Chan CC, and Wang JD. 1997. 60Co contamination in recycled steel resulting in elevated civilian radiation doses: causes and challenges. Health Phys, 73(3), 465–​472. PMID: 9287087. Chin L, Andersen JN, and Futreal PA. 2011. Cancer genomics:  from discovery science to personalized medicine. Nat Med, 17(3), 297–​ 303. PMID: 21383744. Chung CC, and Chanock SJ. 2011. Current status of genome-​wide association studies in cancer. Hum Genet, 130(1), 59–​78. PMID: 21678065. Clarke RH. 1989. Current radiation risk estimates and implications for the health consequences of Windscale, TMI and Chernobyl accidents. In:  Crosbie WA, and Gittus JH (Eds.), Medical Response to Effects of Ionizing Radiation. New York: Elevier Applied Science. Claus EB, Calvocoressi L, Bondy ML, et al. 2012. Dental x-​rays and risk of meningioma. Cancer, 118(18), 4530–​4537. PMCID: PMC3396782. Cullings HM, Fujita S, Funamoto S, et al. 2006. Dose estimation for atomic bomb survivor studies: its evolution and present status. Radiat Res, 166(1 Pt 2), 219–​254. PMID: 16808610. Cullings H, Grant EJ, Egbert SD, et al. 2016. DS02R1: Improvements to atomic bomb survivors’ input data and implementation of Dosimetry System 2002 (DS02) and resulting changes in estimated doses. Health Physics, 112(1), 56–​97. PMID: 27906788. Darby S, Hill D, Auvinen A, et  al. 2005. Radon in homes and risk of lung cancer: collaborative analysis of individual data from 13 European case-​ control studies. BMJ, 330(7485), 223. PMCID: PMC546066. Davis FG, Yu KL, Preston D, et  al. 2015. Solid cancer incidence in the Techa River incidence cohort:  1956–​2007. Radiat Res, 184(1), 56–​65. PMID: 26121228. Degteva MO, Vorobiova MI, Tolstykh EI, et  al. 2006. Development of an improved dose reconstruction system for the Techa River population affected by the operation of the Mayak Production Association. Radiat Res, 166(1 Pt 2), 255–​270. PMID: 16808612. Deschavanne PJ, Debieu D, Fertil B, and Malaise EP. 1986. Re-​evaluation of in vitro radiosensitivity of human fibroblasts of different genetic origins. Int J Radiat Biol Relat Stud Phys Chem Med, 50(2), 279–​293. PMID: 3488286. Djuzenova CS, Muhl B, Fehn M, et al. 2006. Radiosensitivity in breast cancer assessed by the Comet and micronucleus assays. Br J Cancer, 94(8), 1194–​1203. PMCID: PMC2361251. Doll R, and Wakeford R. 1997. Risk of childhood cancer from fetal irradiation. Br J Radiol, 70, 130–​139. PMID: 9135438. Dores GM, Curtis RE, van Leeuwen FE, et  al. 2014. Pancreatic cancer risk after treatment of Hodgkin lymphoma. Ann Oncol, 25(10), 2073–​2079. PMCID: PMC4176454. Drozdovitch V, Minenko V, Khrouch V, et al. 2013. Thyroid dose estimates for a cohort of Belarusian children exposed to radiation from the Chernobyl accident. Radiat Res, 179(5), 597–​609. PMCID: PMC3682838. Einstein AJ. 2012. Beyond the bombs: cancer risks of low-​dose medical radiation. Lancet, 380(9840), 455–​457. PMCID: PMC3674023. Encode Project Consortium. 2012. An integrated encyclopedia of DNA elements in the human genome. Nature, 489(7414), 57–​ 74. PMCID: PMC3439153. Eslinger PW, Napier BA, and Anspaugh LR. 2014. Representative doses to members of the public from atmospheric releases of (131)I at the Mayak Production Association facilities from 1948 through 1972. J Environ Radioact, 135, 44–​53. PMID: 24769389. Franken NA, Rodermond HM, Stap J, Haveman J, and van Bree C. 2006. Clonogenic assay of cells in vitro. Nat Protoc, 1(5), 2315–​ 2319. PMID: 17406473. Fugazzola L, Pilotti S, Pinchera A, et al. 1995. Oncogenic rearrangements of the RET proto-​oncogene in papillary thyroid carcinomas from children

245

exposed to the Chernobyl nuclear accident. Cancer Res, 55(23), 5617–​ 5620. PMID: 7585643. Furukawa K, Preston DL, Lonn S, et al. 2010. Radiation and smoking effects on lung cancer incidence among atomic bomb survivors. Radiat Res, 174(1), 72–​82. PMCID: PMC3857029. Furukawa K, Preston D, Funamoto S, et al. 2013. Long-​term trend of thyroid cancer risk among Japanese atomic-​bomb survivors: 60 years after exposure. Int J Cancer, 132(5), 1222–​1226. PMCID: PMC3910094. Geara FB, Peters LJ, Ang KK, Wike JL, and Brock WA. 1993. Prospective comparison of in vitro normal cell radiosensitivity and normal tissue reactions in radiotherapy patients. Int J Radiat Oncol Biol Phys, 27(5), 1173–​ 1179. PMID: 8262844. Giannakeas V, Lubinski J, Gronwald J, et  al. 2014. Mammography screening and the risk of breast cancer in BRCA1 and BRCA2 mutation carriers:  a prospective study. Breast Cancer Res Treat, 147(1), 113–​118. PMID: 25082516. Gilbert ES, Sokolnikov ME, Preston DL, et al. 2013. Lung cancer risks from plutonium: an updated analysis of data from the Mayak worker cohort. Radiat Res, 179(3), 332–​342. PMCID: PMC3661277. Goldfrank D, Chuai S, Bernstein JL, et al. 2006. Effect of mammography on breast cancer risk in women with mutations in BRCA1 or BRCA2. Cancer Epidemiol Biomarkers Prev, 15(11), 2311–​2313. PMID: 17119064. Grant EJ, Ozasa K, Preston DL, et al. 2012. Effects of radiation and lifestyle factors on risks of urothelial carcinoma in the Life Span Study of atomic bomb survivors. Radiat Res, 178(1), 86–​98. PMID: 22631857. Grant EJ, Brenner A, Sugiyama H, et al. In press. Solid cancer incidence among the Life Span Study of atomic bomb survivors: 1958–​2009. Radiat Res, 187(5), 513–​537. PMID: 28319463. Gray LH. 1965. A Symposium Considering Radiation Effects in the Cell and Possible Implications for Cancer Therapy:  A  Collection of Papers, Cellular Radiation Biology (pp. 8–​ 25). Baltimore, MD:  Williams & Wilkins. Gronwald J, Pijpe A, Byrski T, et  al. 2008. Early radiation exposures and BRCA1-​associated breast cancer in young women from Poland. Breast Cancer Res Treat, 112(3), 581–​584. PMID: 18205043. Hadj-​Hamou NS, Ugolin N, Ory C, et al. 2011. A transcriptome signature distinguished sporadic from postradiotherapy radiation-​induced sarcomas. Carcinogenesis, 32(6), 929–​934. PMCID: PMC3314285. Hall EJ, and Giaccia AJ. 2006. Radiobiology for the Radiologist. Philadelphia: Lippincott Williams & Wilkins. Hammer GP, Auvinen A, De Stavola BL, et al. 2014. Mortality from cancer and other causes in commercial airline crews: a joint analysis of cohorts from 10 countries. Occup Environ Med, 71(5), 313–​322. PMID: 24389960. Hamra GB, Richardson DB, Cardis E, et  al. 2015. Cohort profile:  The International Nuclear Workers Study (INWORKS). Int J Epidemiol, 45(3), 693–​699. PMCID: PMC4703555. Hanahan D, and Weinberg RA. 2011. Hallmarks of cancer: the next generation. Cell, 144(5), 646–​674. PMID: 21376230. Hart D, Wall BF, Hillier MC, and Shrimpton PC. 2010. Frequency and Collective Dose for Medical and Dental X-​Ray Examinations in the UK, 2008. Oxfordshire, UK: Health Protection Agency (HPA). Hauptmann M, Borge Johannesen T, Gilbert ES, et al. 2016. Increased pancreatic cancer risk following radiotherapy for testicular cancer. Br J Cancer, 115(7), 901–​908. PMID: 27599043. Henderson TO, Rajaraman P, Stovall M, et  al. 2012. Risk factors associated with secondary sarcomas in childhood cancer survivors:  a report from the childhood cancer survivor study. Int J Radiat Oncol Biol Phys, 84(1), 224–​230. PMCID: PMC3423483. Hendry JH, Simon SL, Wojcik A, et al. 2009. Human exposure to high natural background radiation: what can it teach us about radiation risks? J Radiol Prot, 29(2A), A29–​A42. PMCID: PMC4030667. Howe GR, and McLaughlin J. 1996. Breast cancer mortality between 1950 and 1987 after exposure to fractionated moderate-​dose-​rate ionizing radiation in the Canadian fluoroscopy cohort study and a comparison with breast cancer mortality in the atomic bomb survivors study. Radiat Res, 145(6), 694–​707. PMID: 8643829. Hsu TC, Johnston DA, Cherry LM, et al. 1989. Sensitivity to genotoxic effects of bleomycin in humans: possible relationship to environmental carcinogenesis. Int J Cancer, 43(3), 403–​409. PMID: 2466800. Hsu WL, Preston DL, Soda M, et al. 2013. The incidence of leukemia, lymphoma and multiple myeloma among atomic bomb survivors:  1950–​ 2001. Radiat Res, 179(3), 361–​382. PMCID: PMC3875218. Hu JJ, Smith TR, Miller MS, Lohman K, and Case LD. 2002. Genetic regulation of ionizing radiation sensitivity and breast cancer risk. Environ Mol Mutagen, 39(2–​3), 208–​215. PMID: 11921191. Hunter N, Kuznetsova IS, Labutina EV, and Harrison JD. 2013. Solid cancer incidence other than lung, liver and bone in Mayak workers: 1948–​2004. Br J Cancer, 109(7), 1989–​1996. PMCID: PMC3790189.

246

246

PART III:  THE CAUSES OF CANCER

International Atomic Energy Agency (IAEA). 1998. Dosimetric and Medical Aspects of the Radiological Accident in Goiania in 1987. IAEA-​ TECDOC-​1109. Vienna: IAEA. IAEA. 2016. Nuclear Power Reactors in the World. Reference Data Series 2/​ 36. Vienna: IAEA. Internatinal Agency for Research on Cancer (IARC). 2012. Radiation. IARC Monogr Eval Carcinog Risks Hum, 100(Pt D). International Commission on Radiological Protection (ICRP). 2007. The 2007 Recommendations of the International Commission on Radiological Protection. ICRP Publication 103. Ann ICRP 37, 2–​4. International HapMap Consortium Altshuler DM, Gibbs RA, et  al. 2010. Integrating common and rare genetic variation in diverse human populations. Nature, 467(7311), 52–​58. PMCID: PMC3173859. Ito T, Seyama T, Iwamoto KS, et  al. 1994. Activated RET oncogene in thyroid cancers of children from areas contaminated by Chernobyl accident. Lancet, 344(8917), 259. PMID: 7913169. Jablon S, Hrubec Z, and Boice JD, Jr. 1991. Cancer in populations living near nuclear facilities: a survey of mortality nationwide and incidence in two states. JAMA, 265(11), 1403–​1408. PMID: 1999880. John EM, McGuire V, Thomas D, et  al. 2013. Diagnostic chest X-​rays and breast cancer risk before age 50  years for BRCA1 and BRCA2 mutation carriers. Cancer Epidemiol Biomarkers Prev, 22(9), 1547–​1556. PMCID: 4015518. Kaatsch P, Spix C, Schulze-​Rath R, Schmiedel S, and Blettner M. 2008. Leukaemia in young children living in the vicinity of German nuclear power plants. Int J Cancer, 122(4), 721–​726. PMID: 18067131. Kato TA, Wilson PF, Nagasaw H, et  al. 2009. Variations in radiosensitivity among individuals: a potential impact on risk assessment? Health Phys, 97(5), 470–​480. PMID: 19820456. Kendall G, Little MP, and Wakeford R. 2011. Numbers and proportions of leukemias in young people and adults induced by radiation of natural origin. Leuk Res, 35(8), 1039–​1043. PMCID: PMC3998761. Kendall GM, Little MP, Wakeford R, et al. 2013. A record-​based case-​control study of natural background radiation and the incidence of childhood leukaemia and other cancers in Great Britain during 1980–​2006. Leukemia, 27(1), 3–​9. PMCID: PMC3998763. Kesminiene A, Evrard AS, Ivanov VK, et al. 2012. Risk of thyroid cancer among chernobyl liquidators. Radiat Res, 178(5), 425–​436. PMID: 22998226. Kim KP, Berrington de Gonzalez A, Pearce MS, et  al. 2012. Development of a database of organ doses for paediatric and young adult CT scans in the United Kingdom. Radiat Prot Dosimetry, 150(4), 415–​ 426. PMCID: PMC3400529. Kitahara CM, Linet MS, Drozdovitch V, et  al. 2015. Cancer and circulatory disease risks in US radiologic technologists associated with performing procedures involving radionuclides. Occup Environ Med, 72(11), 770–​ 776. PMID: 26220810. Kopecky KJ, Davis S, Hamilton TE, Saporito MS, and Onstad LE. 2004. Estimation of thyroid radiation doses for the hanford thyroid disease study: results and implications for statistical power of the epidemiological analyses. Health Phys, 87(1), 15–​32. PMID: 15194919. Krestinina LY, Davis F, Ostroumova E, et al. 2007. Solid cancer incidence and low-​dose-​rate radiation exposures in the Techa River cohort: 1956–​2002. Int J Epidemiol, 36(5), 1038–​1046. PMID: 17768163. Krestinina LY, Davis FG, Schonfeld S, et  al. 2013. Leukaemia incidence in the Techa River cohort: 1953–​2007. Br J Cancer, 109(11), 2886–​2893. PMCID: PMC3844904. Krewski D, Lubin JH, Zielinski JM, et al. 2006. A combined analysis of North American case-​control studies of residential radon and lung cancer. J Toxicol Environ Health A, 69(7), 533–​597. PMID: 16608828. Kwon D, Hoffman FO, Moroz BE, and Simon SL. 2016. Bayesian dose-​ response analysis for epidemiological studies with complex uncertainty in dose estimation. Stat Med, 35(3), 399–​423. PMID: 26365692. Labutina EV, Kuznetsova IS, Hunter N, Harrison J, and Koshurnikova NA. 2013. Radiation risk of malignant neoplasms in organs of main deposition for plutonium in the cohort of Mayak workers with regard to histological types. Health Phys, 105(2), 165–​176. PMID: 23799501. Land CE, Kwon D, Hoffman FO, et  al. 2015. Accounting for shared and unshared dosimetric uncertainties in the dose response for ultrasound-​ detected thyroid nodules after exposure to radioactive fallout. Radiat Res, 183(2), 159–​173. PMCID: PMC4423551. Laurier D, Valenty M, and Tirmarche M. 2001. Radon exposure and the risk of leukemia: a review of epidemiological studies. Health Phys, 81(3), 272–​ 288. PMID: 11513461. Lecarpentier J, Nogues C, Mouret-​Fourme E, et al. 2011. Variation in breast cancer risk with mutation position, smoking, alcohol, and chest X-​ray history, in the French National BRCA1/​2 carrier cohort (GENEPSO). Breast Cancer Res Treat, 130(3), 927–​938. PMID: 21761160.

Lee C, Kim KP, Bolch WE, Moroz BE, and Folio L. 2015. NCICT: a computational solution to estimate organ doses for pediatric and adult patients undergoing CT scans. J Radiol Prot, 35(4), 891–​909. PMID: 26609995. Lee C, Pearce MS, Salotti JA, et al. 2016. Reduction in radiation doses from paediatric CT scans in Great Britain. Br J Radiol, 89(1060), 20150305. PMID: 26864156. Lee T, Sigurdson AJ, Preston DL, et al. 2015. Occupational ionising radiation and risk of basal cell carcinoma in US radiologic technologists (1983–​ 2005). Occup Environ Med, 72(12), 862–​869. PMID: 26350677. Leeman-​Neill RJ, Brenner AV, Little MP, et  al. 2013. RET/​PTC and PAX8/​ PPARgamma chromosomal rearrangements in post-​Chernobyl thyroid cancer and their association with iodine-​131 radiation dose and other characteristics. Cancer, 119(10), 1792–​1799. PMCID: PMC3648615. Leuraud K, Richardson DB, Cardis E, et al. 2015. Ionising radiation and risk of death from leukaemia and lymphoma in radiation-​monitored workers (INWORKS): an international cohort study. Lancet Haematol, 2(7), e276–​e281. PMCID: PMC4587986. Li C, Wang LE, and Wei Q. 2009. DNA repair phenotype and cancer susceptibility:  a mini review. Int J Cancer, 124(5), 999–​ 1007. PMCID: PMC4349428. Li Y, Guolo A, Hoffman FO, and Carroll RJ. 2007. Shared uncertainty in measurement error problems, with application to Nevada Test Site fallout data. Biometrics, 63(4), 1226–​1236. PMID: 18078484. Linet MS, Kim KP, Miller DL, et al. 2010. Historical review of occupational exposures and cancer risks in medical radiation workers. Radiat Res, 174(6), 793–​808. PMCID: PMC4098897. Linet MS, Kitahara CM, Ntowe E, et al. 2017. Mortality in 45,634 U.S. physicians likely to perform fluoroscopically-​guided interventional procedures compared to 64,401 psychiatrists, 1979–​2008. Radiology, 161306. PMID: 28234559. Little MP, Kukush AG, Masiuk SV, et al. 2014. Impact of uncertainties in exposure assessment on estimates of thyroid cancer risk among Ukrainian children and adolescents exposed from the Chernobyl accident. PLoS One, 9(1), e85723. PMCID: 3906013. Little MP, Kwon D, Zablotska LB, et al. 2015. Impact of uncertainties in exposure assessment on thyroid cancer risk among persons in Belarus exposed as children or adolescents due to the Chernobyl accident. PLoS One, 10(10), e0139826. PMCID: PMC4605727. Liu H, Wakeford R, Riddell A, et al. 2016. A review of job-​exposure matrix methodology for application to workers exposed to radiation from internally deposited plutonium or other radioactive materials. J Radiol Prot, 36(1), R1–​R22. PMID: 26861451. Liu Y, Shete S, Wang LE, et al. 2010. Gamma-​radiation sensitivity and polymorphisms in RAD51L1 modulate glioma risk. Carcinogenesis, 31(10), 1762–​1769. PMCID: 2981459. Lubin JH, Tomasek L, Edling C, et al. 1997. Estimating lung cancer mortality from residential radon using data for low exposures of miners. Radiat Res, 147(2), 126–​134. PMID: 9008203. March HC. 1950. Leukemia in radiologists in a 20 year period. Am J Med Sci, 220(3), 282–​286. PMID: 14771053. Mathews JD, Forsythe AV, Brady Z, et  al. 2013. Cancer risk in 680,000 people exposed to computed tomography scans in childhood or adolescence:  data linkage study of 11  million Australians. BMJ, 346, f2360. PMCID: PMC3660619. McKeown SR, Hatfield P, Prestwich RJ, Shaffer RE, and Taylor RE. 2015. Radiotherapy for benign disease; assessing the risk of radiation-​induced cancer following exposure to intermediate dose radiation. Br J Radiol, 88(1056), 20150405. PMCID: PMC4984935. Memon A, Godward S, Williams D, Siddique I, and Al-​Saleh K. 2010. Dental x-​rays and the risk of thyroid cancer: a case-​control study. Acta Oncol, 49(4), 447–​453. PMID: 20397774. Mettler FA, Jr., Bhargavan M, Faulkner K, et al. 2009. Radiologic and nuclear medicine studies in the United States and worldwide: frequency, radiation dose, and comparison with other radiation sources:  1950–​ 2007. Radiology, 253(2), 520–​531. PMID: 19789227. Metz-​Flamant C, Laurent O, Samson E, et al. 2013. Mortality associated with chronic external radiation exposure in the French combined cohort of nuclear workers. Occup Environ Med, 70(9), 630–​638. PMID: 23716722. Molina G. 1990. Lessons learned during the recovery operations in the Ciudad Juarez accident. In: Proceedings of a Symposium on Recovery Operations in the Event of a Nuclear Accident or Radiological Emergency. STI/​PUB/​ 826 (pp. 517–​524). Vienna: IAEA. Muirhead CR, O’Hagan JA, Haylock RG, et  al. 2009. Mortality and cancer incidence following occupational radiation exposure: third analysis of the National Registry for Radiation Workers. Br J Cancer, 100(1), 206–​212. PMCID: PMC2634664.

 247



Ionizing Radiation

Narod SA, Lubinski J, Ghadirian P, et al. 2006. Screening mammography and risk of breast cancer in BRCA1 and BRCA2 mutation carriers:  a case-​ control study. Lancet Oncol, 7(5), 402–​406. PMID: 16648044. National Council on Radiation Protection and Measurements (NCRP). 1989. Comparative Carcinogenicity of Ionizing Radiation and Chemicals. Report No. 96. Bethesda, MD: NCRP. NCRP. 1997. Deposition, Retention and Dosimetry of Inhaled Radioactive Substances. Report No. 125. Bethesda, MD: NCRP. NCRP. 2007. Uncertainties in the Measurement and Dosimetry of External Radiation. Report No. 124. Bethesda, MD: NCRP. NCRP. 2009. Ionizing Radiation Exposure of the Population of the United States. Report No. 160. Bethesda, MD. NCRP. 2011. Second Primary Cancers and Cardiovascular Disease After Radiation Therapy. Report No. 170. Bethesda, MD. National Research Council (NRC). 2006. Review of the Worker and Public Health Activities Program Administered by the Department of Energy and the Department of Health and Human Services. Washington, DC: National Academies Press. NRC. 2012. Analysis of Cancer Risks in Populations Near Nuclear Facilities: Phase 1. Washington, DC: National Academy Press. NRC. 2015. Committee on the Analysis of Cancer Risks in Populations near Nuclear Facilities Study: Phase 2 Pilot Planning. Washington, D.C. The National Academies Press. Neta G, Rajaraman P, Berrington de Gonzalez A, et  al. 2013. A prospective study of medical diagnostic radiography and risk of thyroid cancer. Am J Epidemiol, 177(8), 800–​809. PMCID: PMC3668423. Nikolova T, Dvorak M, Jung F, et al. 2014. The gammaH2AX assay for genotoxic and nongenotoxic agents:  comparison of H2AX phosphorylation with cell death response. Toxicol Sci, 140(1), 103–​117. PMID: 24743697. Parkin DM, and Darby SC. 2011. Cancers in 2010 attributable to ionising radiation exposure in the UK. Br J Cancer, 105 Suppl 2, S57–​S65. PMCID: PMC3252070. Paterson A, Frush DP, and Donnelly LF. 2001. Helical CT of the body:  are settings adjusted for pediatric patients? AJR Am J Roentgenol, 176(2), 297–​301. PMID: 11159060. Peacock J, Ashton A, Bliss J, et al. 2000. Cellular radiosensitivity and complication risk after curative radiotherapy. Radiother Oncol, 55(2), 173–​178. PMID: 10799729. Pearce MS, Salotti JA, Little MP, et  al. 2012. Radiation exposure from CT scans in childhood and subsequent risk of leukaemia and brain tumours:  a retrospective cohort study. Lancet, 380(9840), 499–​ 505. PMCID: PMC3418594. Pernot E, Hall J, Baatout S, et  al. 2012. Ionizing radiation biomarkers for potential use in epidemiological studies. Mutat Res, 751(2), 258–​286. PMID: 22677531. Pinkerton LE, Hein MJ, Anderson JL, et  al. 2016. Breast cancer incidence among female flight attendants:  exposure-​response analyses. Scand J Work Environ Health, 42(6), 538–​546. PMID: 27551752. Pinkerton LE, Waters MA, Hein MJ, et  al. 2012. Cause-​specific mortality among a cohort of U.S. flight attendants. Am J Ind Med, 55(1), 25–​36. PMID: 21987391. Preston DL, Kitahara CM, Freedman DM, et al. 2016. Breast cancer risk and protracted low-​to-​moderate dose occupational radiation exposure in the US Radiologic Technologists Cohort, 1983–​2008. Br J Cancer, 115(9), 1105–​1112. PMID: 27623235. Preston DL, Mattsson A, Holmberg E, et al. 2002a. Radiation effects on breast cancer risk: a pooled analysis of eight cohorts. Radiat Res, 158(2), 220–​ 235. PMID: 12105993. Preston DL, Ron E, Yonehara S, et al. 2002b. Tumors of the nervous system and pituitary gland associated with atomic bomb radiation exposure. J Natl Cancer Inst, 94(20), 1555–​1563. PMID: 12381708. Preston DL, Ron E, Tokuoka S, et al. 2007. Solid cancer incidence in atomic bomb survivors: 1958–​1998. Radiat Res, 168(1), 1–​64. PMID: 17722996. Pukkala E, Helminen M, Haldorsen T, et  al. 2012. Cancer incidence among Nordic airline cabin crew. Int J Cancer, 131(12), 2886–​2897. PMID: 22447246. Raaschou-​Nielsen O. 2008. Indoor radon and childhood leukaemia. Radiat Prot Dosimetry, 132(2), 175–​181. PMID: 19010936. Rajaraman P, Bhatti P, Doody MM, et  al. 2008. Nucleotide excision repair polymorphisms may modify ionizing radiation-​ related breast cancer risk in US radiologic technologists. Int J Cancer, 123(11), 2713–​2716. PMID: 18767034. Rajaraman P, Doody MM, Yu CL, et al. 2016. Cancer risks in U.S. radiologic technologists working with fluoroscopically guided interventional procedures, 1994–​2008. AJR Am J Roentgenol, 206(5), 1101–​1108; quiz 1109. PMID: 26998721.

247

Rajaraman P, Simpson J, Neta G, et al. 2011. Early life exposure to diagnostic radiation and ultrasound scans and risk of childhood cancer: case-​control study. BMJ, 342, d472. PMCID: PMC3037470. Richardson DB, Cardis E, Daniels RD, et al. 2015. Risk of cancer from occupational exposure to ionising radiation: retrospective cohort study of workers in France, the United Kingdom, and the United States (INWORKS). BMJ, 351, h5359. PMCID: PMC4612459. Robison LL, Armstrong GT, Boice JD, et  al. 2009. The Childhood Cancer Survivor Study:  a National Cancer Institute–​ supported resource for outcome and intervention research. J Clin Oncol, 27(14), 2308–​2318. PMCID: 2677920. Roguin A, Goldstein J, Bar O, and Goldstein JA. 2013. Brain and neck tumors among physicians performing interventional procedures. Am J Cardiol, 111(9), 1368–​1372. PMID: 23419190. Ron E. 2003. Cancer risks from medical radiation. Health Phys, 85(1), 47–​59. PMID: 12852471. Ronckers CM, Doody MM, Lonstein JE, Stovall M, and Land CE. 2008. Multiple diagnostic X-​rays for spine deformities and risk of breast cancer. Cancer Epidemiol Biomarkers Prev, 17(3), 605–​613. PMID: 18349278. Ronckers CM, Erdmann CA, and Land CE. 2005. Radiation and breast cancer:  a review of current evidence. Breast Cancer Res, 7(1), 21–​ 32. PMCID: PMC1064116. Rubino C, Shamsaldin A, Le MG, et al. 2005. Radiation dose and risk of soft tissue and bone sarcoma after breast cancer treatment. Breast Cancer Res Treat, 89(3), 277–​288. PMID: 15754127. Rudat V, Dietz A, Nollert J et al. 1999. Acute and late toxicity, tumour control and intrinsic radiosensitivity of primary fibroblasts in vitro of patients with advanced head and neck cancer after concomitant boost radiochemotherapy. Radiother Oncol, 53(3), 233–​245. PMID: 10660204. Sachs RK, and Brenner DJ. 2005. Solid tumor risks after high doses of ionizing radiation. Proc Natl Acad Sci U S A, 102(37), 13040–​ 13045. PMCID: PMC1199000. Sakata R, Kleinerman RA, Mabuchi K, et al. 2012. Cancer mortality following radiotherapy for benign gynecologic disorders. Radiat Res, 178(4), 266–​ 279. PMCID: PMC3471655. Sanlorenzo M, Wehner MR, Linos E, et  al. 2015. The risk of melanoma in airline pilots and cabin crew: a meta-​analysis. JAMA Dermatol, 151(1), 51–​58. PMCID: PMC4482339. Schubauer-​Berigan MK, Anderson JL, Hein MJ, et  al. 2015a. Breast cancer incidence in a cohort of U.S. flight attendants. Am J Ind Med, 58(3), 252–​ 266. PMCID: PMC4566958. Schubauer-​Berigan MK, Daniels RD, Bertke SJ, Tseng CY, and Richardson DB. 2015b. Cancer mortality through 2005 among a pooled cohort of U.S. nuclear workers exposed to external ionizing radiation. Radiat Res, 183(6), 620–​631. PMID: 26010709. Sermage-​Faure C, Laurier D, Goujon-​Bellec S, et al. 2012. Childhood leukemia around French nuclear power plants: the Geocap study, 2002–​2007. Int J Cancer, 131(5), E769–​E780. PMID: 22223329. Shin H, Ramsay T, Krewski D, and Zielinski JM. 2005. The effect of censoring on cancer risk estimates based on the Canadian National Dose Registry of occupational radiation exposure. J Expo Anal Environ Epidemiol, 15(5), 398–​406. PMID: 15592443. Shipler DB, Napier BA, Farris WT, and Freshley MD. 1996. Hanford Environmental Dose Reconstruction Project: an overview. Health Phys, 71(4), 532–​544. PMID: 8830754. Sigurdson AJ, Bhatti P, Chang SC, et al. 2009. Polymorphisms in estrogen biosynthesis and metabolism-​related genes, ionizing radiation exposure, and risk of breast cancer among US radiologic technologists. Breast Cancer Res Treat, 118(1), 177–​184. PMID: 19214745. Sigurdson AJ, Bhatti P, Doody MM, et al. 2007. Polymorphisms in apoptosis-​ and proliferation-​related genes, ionizing radiation exposure, and risk of breast cancer among U.S. Radiologic Technologists. Cancer Epidemiol Biomarkers Prev, 16(10), 2000–​2007. PMID: 17932347. Sigurdson AJ, and Stram DO. 2012. Genetic predisposition to radiation-​related cancer and potential implications for risk assessment. Ann ICRP, 41(3-​4), 108–​116. PMCID: PMC3916824. Simon SL, Hoffman FO, and Hofer E. 2015. The two-​dimensional Monte Carlo: a new methodologic paradigm for dose reconstruction for epidemiological studies. Radiat Res, 183(1), 27–​41. PMCID: PMC4423557. Simon SL, Preston DL, Linet MS, et al. 2014. Radiation organ doses received in a nationwide cohort of U.S. radiologic technologists: methods and findings. Radiat Res, 182(5), 507–​528. PMCID: PMC4406479. Smith KC, Hahn GM, Hoppe RT, and Earle JD. 1980. Radiosensitivity in vitro of human fibroblasts derived from patients with a severe skin reaction to radiation therapy. Int J Radiat Oncol Biol Phys, 6(11), 1573–​1575. PMID: 7462061.

248

248

PART III:  THE CAUSES OF CANCER

Sont WN, Zielinski JM, Ashmore JP, et al. 2001. First analysis of cancer incidence and occupational radiation exposure based on the National Dose Registry of Canada. Am J Epidemiol, 153(4), 309–​318. PMID: 11207146. Spycher BD, Feller M, Zwahlen M, et al. 2011. Childhood cancer and nuclear power plants in Switzerland: a census-​based cohort study. Int J Epidemiol, 40(5), 1247–​1260. PMCID: PMC3204210. Stather JW. 2011. Childhood leukaemia near nuclear sites:  fourteenth report of the Committee on Medical Aspects of Radiation in the Environment (COMARE). Radiat Prot Dosimetry, 147(3), 351–​354. PMID: 21831861. Stayner L, Vrijheid M, Cardis E, et al. 2007. A Monte Carlo maximum likelihood method for estimating uncertainty arising from shared errors in exposures in epidemiological studies of nuclear workers. Radiat Res, 168(6), 757–​763. PMID: 18088178. Stovall M, Weathers R, Kasper C, et al. 2006. Dose reconstruction for therapeutic and diagnostic radiation exposures: use in epidemiological studies. Radiat Res, 166(1 Pt 2), 141–​157. PMID: 16808603. Stram DO, Preston DL, Sokolnikov M, et  al. 2015. Shared dosimetry error in epidemiological dose-​response analyses. PLoS One, 10(3), e0119418. PMCID: PMC4370375. Sun Z, Inskip PD, Wang J, et al. 2016. Solid cancer incidence among Chinese medical diagnostic x-​ray workers, 1950–​1995:  Estimation of radiation-​ related risks. Int J Cancer, 138(12), 2875–​2883. PMCID: PMC4826787. Taalman RD, Jaspers NG, Scheres JM, de Wit J, and Hustinx TW. 1983. Hypersensitivity to ionizing radiation, in vitro, in a new chromosomal breakage disorder, the Nijmegen Breakage Syndrome. Mutat Res, 112(1), 23–​32. PMID: 6828038. Taylor AM, Harnden DG, Arlett CF, et al. 1975. Ataxia telangiectasia: a human mutation with abnormal radiation sensitivity. Nature, 258(5534), 427–​ 429. PMID: 1196376. Terada H, Katata G, Chino M, and Nagai H. 2012. Atmospheric discharge and dispersion of radionuclides during the Fukushima Dai-​ichi Nuclear Power Plant accident. Part II: verification of the source term and analysis of regional-​scale atmospheric dispersion. J Environ Radioact, 112, 141–​ 154. PMID: 22721917. Thierry-​Chef I, Simon SL, Weinstock RM, Kwon D, and Linet MS. 2012. Reconstruction of absorbed doses to fibroglandular tissue of the breast of women undergoing mammography (1960 to the present). Radiat Res, 177(1), 92–​108. PMCID: PMC3876279. Till JE, Beck HL, Aanenson JW, et  al. 2014. Military participants at U.S. Atmospheric nuclear weapons testing: methodology for estimating dose and uncertainty. Radiat Res, 181(5), 471–​484. PMCID: PMC4082990. Travis LB, Hill D, Dores GM, et al. 2005. Cumulative absolute breast cancer risk for young women treated for Hodgkin lymphoma. J Natl Cancer Inst, 97(19), 1428–​1437. PMID: 16204692. United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR). 2006. UNSCEAR 2006 Report to the General Assembly, with Scientific Annexes, Annex A: Epidemiological Studies of Radiation and Cancer. New York: United Nations. UNSCEAR. 2010a. Sources and Effects of Atomic Radiation. UNSCEAR 2010 Report to the General Assembly with Scientific Annexes, Vol. I, Annex A: Medical Radiation Exposures. New York: United Nations. UNSCEAR. 2010b. Sources and Effects of Atomic Radiation. UNSCEAR 2010 Report to the General Assembly with Scientific Annexes, Vol. I, Annex B: Exposures of the Public and Workers from Various Sources of Radiation. New York: United Nations.

UNSCEAR. 2011a. Sources and Effects of Atomic Radiation. UNSCEAR 2008 Report to the General Assembly with Scientific Annexes, Vol. II, Annex C: Radiation Exposures in Accidents. New York: United Nations. UNSCEAR. 2011b. Sources and Effects of Atomic Radiation. UNSCEAR 2008 Report to the General Assembly with Scientific Annexes, Vol. II, Annex D: Health Effects Due to Radiation from the Chernobyl Accident. New York: United Nations. Veiga LH, Holmberg E, Anderson H, et al. 2016. Thyroid cancer after childhood exposure to external radiation:  an updated pooled analysis of 12 studies. Radiat Res, 185(5), 473–​484. PMCID: PMC4893786. Veiga LH, Lubin JH, Anderson H, et al. 2012. A pooled analysis of thyroid cancer incidence following radiotherapy for childhood cancer. Radiat Res, 178(4), 365–​376. PMCID: PMC3488851. Vineis P, and Perera F. 2000. DNA adducts as markers of exposure to carcinogens and risk of cancer. Int J Cancer, 88(3), 325–​328. PMID: 11054658. Vrijheid M, Cardis E, Ashmore P, et al. 2007. Mortality from diseases other than cancer following low doses of ionizing radiation:  results from the 15-​Country Study of nuclear industry workers. Int J Epidemiol, 36(5), 1126–​1135. PMID: 17666424. Vrijheid M, Cardis E, Ashmore P, et  al. 2008. Ionizing radiation and risk of chronic lymphocytic leukemia in the 15-​country study of nuclear industry workers. Radiat Res, 170(5), 661–​665. PMCID: PMC4051434. West CM, and Barnett GC. 2011. Genetics and genomics of radiotherapy toxicity:  towards prediction. Genome Med, 3(8), 52. PMCID: PMC3238178. West CM, Davidson SE, Elyan SA, et al. 2001. Lymphocyte radiosensitivity is a significant prognostic factor for morbidity in carcinoma of the cervix. Int J Radiat Oncol Biol Phys, 51(1), 10–​15. PMID: 11516845. Woods WG, Byrne TD, and Kim TH. 1988. Sensitivity of cultured cells to gamma radiation in a patient exhibiting marked in vivo radiation sensitivity. Cancer, 62(11), 2341–​2345. PMID: 3179949. World Health Organization (WHO). 2006. Health Effects of the Chernobyl Accident and Special Health Care Programmes:  Report of the UN Chernobyl Forum Health Expert Group. Available from: http://​www.who. int/​ionizing_​radiation. Wu X, Gu J, and Spitz MR. 2007. Mutagen sensitivity: a genetic predisposition factor for cancer. Cancer Res, 67(8), 3493–​3495. PMID: 17440053. Zablotska LB, Ashmore JP, and Howe GR. 2004. Analysis of mortality among Canadian nuclear power industry workers after chronic low-​dose exposure to ionizing radiation. Radiat Res, 161(6), 633–​641. PMID: 15161357. Zablotska LB, Bazyka D, Lubin JH, et al. 2013. Radiation and the risk of chronic lymphocytic and other leukemias among chornobyl cleanup workers. Environ Health Perspect, 121(1), 59–​65. PMCID: PMC3553431. Zablotska LB, Lane RS, and Thompson PA. 2014. A reanalysis of cancer mortality in Canadian nuclear workers (1956–​1994) based on revised exposure and cohort data. Br J Cancer, 110(1), 214–​223. PMCID: PMC3887280. Zablotska LB, Ron E, Rozhko AV, et al. 2011. Thyroid cancer risk in Belarus among children and adolescents exposed to radioiodine after the Chornobyl accident. Br J Cancer, 104(1), 181–​187. PMCID: PMC3039791. Zielinski JM, Garner MJ, Band PR, et al. 2009. Health outcomes of low-​dose ionizing radiation exposure among medical workers: a cohort study of the Canadian national dose registry of radiation workers. Int J Occup Med Environ Health, 22(2), 149–​156. PMID: 19546093.

 249

14

Ultraviolet Radiation ADÈLE C. GREEN AND DAVID C. WHITEMAN

OVERVIEW Ultraviolet (UV) radiation is the principal cause of over 95% of keratinocyte cancers (basal cell carcinomas and squamous cell carcinomas of the skin), the most common cancers in white populations worldwide; it also causes the majority (estimated 60%–​90%) of cases of cutaneous melanoma, the cancer of the skin’s pigment-​ producing cells. In addition, UV radiation is the major cause of many eye diseases, including ocular cancers and cataract, the most common cause of blindness, and is responsible for the underlying changes in aged skin on which billions of dollars are spent annually in efforts to repair the damage. Clinicians first recognized the causal role of sun exposure in keratinocyte cancer development around the turn of the twentieth century. The probable role of sunlight in driving high melanoma mortality in populations of European ancestry living at low latitudes was identified 50 years later. The sun is the universal source of human exposure to UV radiation, but artificial sources are also encountered in a wide range of settings, from medical to industrial, and increasingly from commercial tanning outlets. A  century ago, high cumulative UV exposure was the provenance of those who worked in outdoor occupations (and in whom skin cancers were first described), but during the course of the postwar decades, social customs changed dramatically such that people with pale skins sought to spend leisure time outdoors to acquire a suntan, and people living in temperate climates likewise spent their vacations in sunny locations and began to patronize tanning salons. By the late twentieth century, however, the near epidemic increases in skin cancer incidence in white populations, especially in Australia and New Zealand, stimulated the initiation of skin cancer awareness and prevention campaigns, including sun protection and legislation to restrict or ban sunbed use, in an effort to control this growing public health problem.

SOLAR RADIATION AND THE ELECTROMAGNETIC SPECTRUM Solar UV radiation, a ubiquitous environmental carcinogen, is part of the spectrum of electromagnetic radiation emanating from the sun. UV radiation is also generated by artificial sources encountered in a wide range of settings. The ultraviolet spectral region spans wavelengths of 10 nm–​400 nm, and is divided into several bands: UVA (320–​400 nm); UVB (280–​320  nm); UVC (200–​280  nm); far UV (120–​200  nm); extreme UV (10–​120 nm). Although UV radiation is classified as non-​ ionizing, UV photons are sufficiently energetic to destabilize electron configurations within molecules such as DNA and to have a biological effect. In terms of solar UV radiation, only UVB and UVA have biological significance, as shorter wavelengths are absorbed by the atmosphere. UV radiation is typically absorbed over a surface and is measured as a radiant exposure. The term irradiation represents the dose of radiant energy delivered to an area within a given time, and is measured J/​m2. The rate at which UV energy reaches a surface is termed irradiance (units W/​m2). The total irradiance from any given source of UV radiation is derived by summing the wavelength-​specific irradiances across the spectrum of wavelengths emitted. A “biologically effective UV irradiance” (UVeff) for a given source is determined by weighting the irradiance at each emitted wavelength by its ability to cause the

biological effect of interest (e.g., DNA mutation, erythema), and then summing these weighted values across all wavelengths.

METHODS OF MEASUREMENT At the earth’s surface, solar UV radiation is measured by a wide variety of detector instruments (the most common of which are radiometers), each having different attributes and requirements for calibration and maintenance (Godar, 2005). In addition, portable monitors and data-​loggers are increasingly available (Hooke et al., 2014). Broadband radiometers measure UV irradiance over a broad spectral band, integrated over the wavelengths of radiation known to exert a particular biological effect (e.g., erythema). The output from broadband radiometers is a single number, calculated from the voltage generated by filtered UV radiation striking a photodiode. These simple instruments are portable and robust, although relatively insensitive to changes in irradiance at specific wavelengths. Networks of radiometers were first established during the 1970s and 1980s, often in remote locations, and subsequently have provided long-​term information about levels of biologically effective UV radiation at the earth’s surface (Frederick et al., 2000; Henderson et al., 2010; Pearson et al., 2000). Popular early models (e.g., Robertson-​Berger meters) have been largely replaced by newer models that have less technical error and are more stable under extremes of temperature, although some limitations remain (Huber et al., 2002). Spectroradiometers are more sophisticated instruments that measure small changes in UV flux at specific wavelengths across the spectrum such as occurs following depletions in atmospheric ozone from time to time (Roy et  al., 1997). Notwithstanding their cost and the skill required to operate them, networks of spectroradiometers have been established worldwide to monitor changes in surface irradiance. For example, the US National Oceanic and Aerospace Administration (NOAA) and the Environmental Protection Agency jointly manage the NEU-​Brew network of spectroradiometers (http://​esrl.noaa.gov/​gmd/​ grad/​neubrew). Since 2006, these instruments have measured daily UV radiation and total column ozone at six locations in the western, central, and eastern United States (Houston, TX; Table Mountain, CO; Mountain Research Station, CO; Bondville, IL; Raleigh, NC; Ft. Peck, MT). This latest generation of instruments replaced an earlier network of spectroradiometers operated by the EPA from the 1990s until 2004 (UV-​NET; http://​www.epa.gov/​uvnet/​), of which 14 instruments were located in US National Parks as part of the Park Research and Intensive Monitoring of Ecosystems Network (PRIMENet). Similar networks of UV spectroradiometers have been established in many other countries, contributing to a global network coordinated by the World Meteorological Organization to monitor UV irradiance (World Ozone and Ultraviolet Radiation Data Centre, administered by Environment Canada, http://​www.woudc.org/​home.php/​). Historical UV data from global locations can be downloaded from this site for analysis.

DETERMINANTS OF SOLAR IRRADIANCE AND SOLAR DOSE The irradiance at the surface of the earth is much less than in the upper atmosphere due to absorption by stratospheric ozone of all

249

250

250

PART III:  THE CAUSES OF CANCER

UVC and a large proportion of UVB before reaching the earth. The thickness of the ozone mantle varies according to season, latitude, and meteorological conditions and by levels of chlorofluorocarbons and other gases generated by human activities that can destroy ozone (Rowland, 2006). International agreements like the Montreal Protocol aimed to reduce the emission of chlorofluorocarbons (Weatherhead and Andersen, 2006), and monitoring shows that increases in terrestrial UVB have been modest (Williamson et  al., 2014). However, it is uncertain whether subsequent small improvements in ozone levels are due to a decline in the amount of ozone-​depleting substances in the Earth’s atmosphere or if they reflect the high natural variability in ozone levels related to the solar cycle and recent changes in global air transport and temperature that make it unlikely that ozone will stabilize at levels observed before 1980 (Weatherhead and Andersen, 2006; Williamson et  al., 2014). UV irradiance, especially UVB, increases with decreasing latitude, increasing altitude, in summer, and in the middle hours of the day, and decreases with heavy cloud cover. UVA levels are less affected than UVB by atmospheric factors and are more stable throughout the day. By the end of the twenty-​first century relative to the present decade (2010–​2020), it is expected that projected decreases in ozone-​depleting gases and changes in cloud cover will lead to relatively small decreases in UVB, while reductions in reflectivity of melting Arctic sea ice will lead to decreases of up to 10%. On the other hand, expected improvement of air quality and in particular reductions of aerosols over densely populated areas of the Northern Hemisphere are expected to result in increases in UVB of 10%–​20%, with even higher projected increases over China (Bais et al., 2015).

The UV Index The standardization of UV information to the public through the UV Index (UVI) was introduced and endorsed by multiple international agencies in the 1990s. UV Index values are available throughout the world from satellite instruments, ground-​based measurements, and modeled forecasts (Zaratti et al., 2014), and are scaled from 0 to 11+, with 11+ classified as “extreme.” Increasing levels of sun protection are recommended with increasing UVI values to 11+. However, a recent call for modification of the UV Index (Zaratti et al., 2014) was based on the bias that exists toward European skin types in the currently depicted UV Index and its failure to properly account for the most extreme values of the UV Index, over 20, experienced by populations living close to the equator and at high altitudes, such as in the Altiplano region of South America.

MEASURING PERSONAL EXPOSURE TO SOLAR RADIATION Humans are exposed to sunlight on a daily basis, yet accurate and reproducible methods for measuring the dose of solar radiation received by an individual have only been developed in the past few decades. In epidemiologic studies, human sun exposure has been measured in a variety of ways, ranging from dosimetry to personal recall of past exposure.

Dosimetry Dosimeters are portable units designed to measure personal, incident exposure to solar UV radiation. Historically, the two most common dosimeters used in human research have been the Bacillus biological dosimeter and the polysulfone badge, both of which estimate total doses of sun exposure over relatively short periods (Green and Whiteman, 2006). The biological dosimeter comprises Bacillus subtilis spores immobilized on a polyester sheet, which are inactivated in a dose-​dependent manner upon exposure to UV radiation (Quintern et  al., 1992). Dosimeters of this type were used in diverse settings, including schools (Munakata et al., 1998), mountains (Moehrle et al., 2003), and the Arctic (Cockell et al., 2001). Polysulfone is UVR-​sensitive polymer that has an action spectrum closely approximating the

erythemal response of human skin when formulated as a film 30–​45 µm thick (Diffey, 1987). These dosimeters have been useful in field studies as they are inexpensive, stable, and easy to apply, all of which contribute to high rates of compliance in large-​scale studies (Sun et al., 2014). However, they do become saturated at relatively modest doses of UVR, requiring participants in exposure studies to periodically change badges. More recently, electronic dosimeters have been developed that comprise miniature battery-​powered UV detectors that can be worn on the wrist or other exposed body sites (Allen and McKenzie, 2005; Heydenreich and Wulf, 2005). These dosimeters have the advantage of logging UV radiation data at specified intervals, thereby providing investigators with information about time-​dependent aspects of sun exposure, such as dose rates and maximum exposure intensity (Seckmeyer et al., 2012). In addition, electronic dosimeters are able to be reused, do not saturate after short intervals, and their response to UVR exposure is linear rather than logarithmic (Allen and McKenzie, 2005; Seckmeyer et al., 2012). While these various approaches to measuring personal UVB exposure provide a degree of objectivity, there is attendant measurement error when compared with gold standard measures of UV radiation as determined by spectroradiometers. It has been suggested that estimates of UVB exposure from polysulfone badges and electronic dosimeters may deviate from the gold standard by up to 26% and 15%, respectively (Seckmeyer et  al., 2012). In addition to issues of calibration, the physical location and placement of dosimeters on the body also cause measurements to vary. Human field studies comparing measurements obtained from dosimeters placed concurrently on various body sites have shown that maximal readings are obtained from dosimeters located on the head (almost 90% of ambient UVR), with lower readings obtained from dosimeters on the shoulder (55% of ambient) and wrist (43%) (Thieden et al., 2000). Given that the aim of UV radiation measurement in most epidemiologic studies is to rank participants by relative exposure, the wrist is generally the preferred site of placement for reasons of convenience, compliance, and reproducibility (Petersen et al., 2014; Sun et al., 2014; Thieden et al., 2006). Numerous field studies have deployed dosimeters and data-​loggers on humans outdoors. Most studies to date have been relatively small in scale, and of limited duration, and often have been conducted within a single geographic site, so generalizations must be made cautiously. Nevertheless, it appears that most people receive a personal dose of UV radiation less than 5% of daily ambient UV radiation, although there is wide inter-​individual variability, and the measured dose is heavily dependent on environmental, behavioral, and occupational factors, as well as personal characteristics and the anatomic site of the dosimeter. As earlier, higher proportions of ambient radiation may be received on horizontal skin surfaces (such as the vertex of the head, the top of the shoulders), with lower doses at other sites (Weihs et al., 2013). Movement, body position, and orientation with respect to the sun all greatly affect the dose received at any given site (Weihs et al., 2013). The level of ambient UV radiation is the strongest influence on a person’s received dose; however, the influence of ambient UV radiation on average doses when compared across populations is modified by latitude and season. In most non-​equatorial parts of the world, the average daily levels of ambient UV radiation are higher in summer than winter, and the magnitude of seasonal difference increases with latitude. One might therefore predict that seasonal fluctuations in personal UV radiation doses would follow a similar temporal course to ambient UVR, with higher doses being received in summer than in winter months. However, dosimetry studies of “free-​living humans” conducted over a range of latitudes report discordant seasonal trends for different geographic locations. In temperate climates at mid-​to high latitudes, average daily personal UV radiation doses tend to rise in spring and peak in summer in line with ambient UV radiation levels (Sun et al., 2014; Thieden et al., 2004b), whereas in subtropical and tropical locations, higher personal UVR doses are observed in winter than in summer (Sun et al., 2014). The accepted explanation is that the intensity of heat and solar radiation in summer months at low latitudes discourages fair-​skinned residents from outdoor activities, particularly during the middle part of the day. Conversely, the mild temperatures

 251



Ultraviolet Radiation

and less intense levels of ambient UV radiation during winter months in tropical settings are more conducive to outdoor work and recreation, resulting in higher levels of personal sun exposure on average. Few studies have investigated correlations between host characteristics and personal UV radiation; the scant available evidence suggests that people with sun-​sensitive phenotypes (such as red or light-​colored hair, a tendency to sunburn, or poor tanning ability) typically receive lower average daily doses of UV than those without these characteristics (Cahoon et  al., 2013; Thieden, 2008). Studies from the high-​ latitude Nordic countries have reported higher average daily personal UV radiation doses in women than in men (Thieden et  al., 2004a), whereas studies from mid-​to low-​latitude Australian settings report the opposite (Xiang et  al., 2015). Within any given population, the highest doses of UV radiation are experienced during outdoor activities. Very high UV doses have been recorded among outdoor workers in diverse settings, including alpine environments and the Canadian Arctic (Cockell et  al., 2001). As epidemiologic studies often seek information on past exposures to the sun, it is interesting to note that a Danish study observed high levels of intra-​individual concordance for personal UVR doses measured at two time points up to 7  years apart (Thieden et al., 2013), suggesting that individual patterns of sun exposure tend to be relatively stable over time.

Biomarkers The use of biomarkers seeks to overcome the inherent limitation of the dosimeters described earlier, namely that while dosimeters measure the amount of ambient UV radiation to which individuals are personally exposed, they do not measure the mutagenic effects of sun exposure of interest to cancer epidemiologists. UV radiation induces specific types of DNA lesions in the nuclei of skin cells (see later discussion), including thymine dimers, and these lesions tend to accumulate in a dose-​dependent manner. These lesions can be detected readily in DNA harvested from skin biopsies, but such invasive methods of sample collection are unsuitable for most epidemiologic studies. Moreover, such samples provide a measure of UV-​induced DNA damage at specific anatomical sites only, which may not correlate with a global measure of mutagenesis. Recently, non-​invasive urine tests have been developed to measure the burden of UV-​specific mutations in human subjects. During the exquisitely refined process of DNA repair (see later discussion), the damaged segments of nucleic acid containing thymine dimers are excised and are replaced with a new segment of DNA. The excised fragments of damaged DNA are secreted from the skin cells into the bloodstream, to be excreted by the kidney in the urine (Cooke et al., 2013). Although presently a labor-​intensive procedure, it is possible to quantify the amount of thymine dimers in the urine in a reproducible manner using radio-​labeled high-​performance liquid chromatography (Kotova et al., 2005). Kinetic studies demonstrate that urinary thymine dimer levels reach a maximum 3–​4 days post-​UV exposure. In field studies among indoor workers in Sweden, thymine dimers were undetectable in urine samples taken in winter, but following typical working weeks in summer, about 20% of such people exhibited measurable levels of thymine dimers (Liljendahl et al., 2013). Levels of urinary thymine dimers are strongly correlated with increased UV exposure. These effects have been demonstrated in children and adults in outdoor settings in Sweden (Liljendahl et al., 2012), as well as in Danish skiers, and Danish and Spanish sun-​seekers in a subtropical beach setting (Petersen et al., 2014). At the present time, thymine dimers have not been used as measures of UV exposure in epidemiologic studies with skin cancer endpoints.

Personal Recollection Because dosimeters and biomarkers measure current or recent solar exposure, they cannot be used in epidemiologic studies that seek to capture past exposure. Accurate recall of past sun exposure is difficult; nevertheless, epidemiologists have attempted to estimate past sun exposure in two main ways (Whiteman et al., 2001).

251

The first approach has been to estimate an individual’s average exposures to solar UV radiation during defined time periods according to places of residence, since ambient solar UV radiation increases with proximity to the equator (Godar, 2005). Epidemiologists have commonly used these ambient exposure measures when conducting ecological studies, such as comparing the incidence of melanoma among native residents and migrants with contrasting experiences of past sun exposure. Although place of residence is a relatively crude measure of personal sun exposure, it is a good proxy for solar dose when comparing populations (Diffey et  al., 1996; Diffey and Gies, 1998; Godar, 2005) and has the practical advantage of being easy to recall. The other approach has been to ask study participants to recall salient exposures, such as numbers of sunburns, time spent outdoors in summer, or recreational activities associated with high levels of sun exposure (English et al., 1998). An episode of sunburn is an integrated measure of acute solar dose, and thus the number of sunburns a person experiences gives an indication of accumulated acute exposures to intense solar UV radiation (Green et al., 1985; Whiteman and Green, 1994). Recall of past sunburn episodes is reasonable, with most studies suggesting moderate repeatability (kappa scores 0.5–​0.7) (Jennings et al., 2013; Morze et al., 2012; van der Mei et al., 2006; Veierod et al., 2008). The number of sunburns experienced by a person is due not only to the intensity or duration of sun exposure, but also to host characteristics including skin type and pigmentations. Other recalled measures of UV exposure that are not completely independent of host characteristics include time spent outdoors, vacations in sunny locations, use of tanning salons and outdoor occupations, most of which have moderate to high levels of agreement when subjected to test-​retest reliability studies (Jennings et  al., 2013; Morze et al., 2012; van der Mei et al., 2006; Veierod et al., 2008).

Phenotypic Markers of UV Exposure Because of the challenges in measuring past sun exposure by questionnaire-​based methods, and because dosimetry and biomarkers are limited to recent or current UV exposures, researchers have sought “objective” measures of historical sun exposure based on biological changes that correlate with exposure. Photoaging encompasses a range of visible and microscopic changes to the skin associated with chronic exposure to UV radiation. Silicone-​rubber skin casts capture a record of the loss of visible surface architecture. These casts have several epidemiologic virtues: they can be obtained rapidly; they are inexpensive, safe, and painless to participants; and they can be stored indefinitely. Moreover, graders can be trained easily to make reliable assessments of photoaging (kappa values > 0.90) that correlate well with histological measures of photoaging (Battistutta et al., 2006).

ARTIFICIAL SOURCES OF UV EXPOSURE While most people’s only UV exposure comes from sunlight, artificial UV sources may contribute to a material proportion of total UV exposure for some. In particular, the prevalence of exposure to artificial UV radiation for cosmetic tanning has increased markedly in recent decades, with users attaching considerable personal and social importance to tanned skin (Grange et al., 2015; Lake et al., 2014). A systematic review and meta-​analysis in 2013 showed the large extent of ever-​exposure to indoor tanning in Western countries, especially in the young (Wehner et al., 2014), with summary prevalences of 36% for adults, 55% for university students, and 19% for adolescents. Although the indoor tanning industry generally implies to users that their use of tanning parlors complying with regulations is safe and might have health benefits (Autier et al., 2011), several reviews show the high risks of ill effects on eyes and skin, including melanoma and other skin cancers (Boniol et al., 2012; International Agency for Research on Cancer, 2012; Wehner et al., 2012). Surveys have indicated variable compliance by the sunbed industry with guidelines (Chandrasena et al., 2013), and exposure to tanning units can still result in high UV doses (Khazova et al., 2015; Nilsen et al., 2011), similar to exposures received while sunbathing in summer at a Mediterranean resort. Measured UV irradiance from a large number of Norwegian

25

252

PART III:  THE CAUSES OF CANCER

solaria (sunbeds and stand-​up cabinets) in 2008 showed that overall compliance had increased since 1998–​1999 but total UV irradiance had not decreased, mainly because of higher UVA irradiance (Nilsen et al., 2011). A recent survey in the United Kingdom showed that an increase in sunbed emissions had occurred in the last decade, with UVB irradiance of over 0.3 W m–​2 in more than 85% of all tested solaria, though exposure per session appeared not to have increased (Khazova et al., 2015). Other sources of artificial UV exposure are found in occupational and medical settings. In occupational settings, workers are normally shielded from material UV radiation, though prolonged exposure to intense UV radiation from arc welding is hazardous to both eyes and skin (International Agency for Research on Cancer, 1992, 2012). In clinical usage, different UV dose regimens influence the occurrence of long-​term carcinogenic side effects. Applications such as PUVA photochemotherapy that combines UVA with photosensitizing psoralens to treat skin diseases such as psoriasis, have been used widely. A  systematic literature review of the carcinogenic risks of psoralen UVA therapy versus narrowband UVB therapy, also commonly used for chronic plaque psoriasis, concluded that there is an increased risk of skin cancer following PUVA shown by both US and European studies, with a greater risk seen in US studies, likely to be partly explained by higher average UVA doses and lighter phototypes of treated patients (Archier et  al., 2012). There were insufficient prospective studies in psoriasis patients treated with UVB to assess the associated carcinogenic risk.

BIOLOGIC MECHANISMS FOR UVR CARCINOGENESIS DNA Damage The principal mechanism of UV carcinogenesis at the molecular level is DNA damage. DNA damage may arise directly through highly energetic UV photons interacting with chemical elements of the double-​helix, or indirectly through the transfer of energy from photons through endogenous photosensitizers to DNA. Most DNA damage is repaired with very high fidelity by specific enzymes before the cell replicates. If repair fails, then the damaged nucleotide can be replaced by an incorrect nucleotide in the DNA of the daughter cell, resulting in a permanent change in the sequence known as a mutation. Drawing a distinction between DNA damage (including lesions, adducts, and photoproducts) and mutation is critically important in establishing the temporal relationship along the pathway to carcinogenesis (Brash, 2015). The chemical structure of DNA, particularly the density of pyrimidine bases (thymine [T]‌and cytosine [C]), renders the molecule a potent absorber of UV photons (Pfeifer and Besaratinia, 2012). Incident UVB radiation directly damages the integrity of DNA by inducing bulky lesions (cyclobutane pyrimidine dimers [CPDs] and [6–​4] photoproducts) between adjacent pyrimidine nucleotide bases that interfere with essential genomic functions such as transcription and replication. If these lesions are not repaired, then characteristic mutations occur, most frequently base substitutions of C by T at dipyrimidine sites, and less commonly, CC to TT tandem base substitutions (Ikehata and Ono, 2011). Such is the specificity of these mutations for sunlight exposure that they are termed UV signature mutations. Recently there has been renewed appreciation that UV damage causes many other nonspecific DNA mutations. Whether or not UV photodamage results in a signature mutation depends on the precise sequence of nucleotides at the target site, as well as the presence of other chemical elements in close proximity at the time of photon strike (Brash, 2015). It follows that the absence of a UV signature mutation in a cancer gene of interest does not mean that the mutation was not caused by sun exposure. While DNA is a major chromophore for UVB radiation, it absorbs photons in the UVA spectrum poorly. Thus UVA induces DNA damage indirectly. For many years, it was held that UVA interacted with sensitizers, both endogenous (e.g., bilirubin, porphyrins) and exogenous (e.g., medications, cosmetics, sunscreens) to produce reactive oxygen species that result in characteristic lesions such as 8-​oxo-​7,8

dihydroxyguanine (8-​oxodGuo). Recent experiments have demonstrated that erythemal doses of UVA actually induce more CPDs than 8-​oxodGuo in human skin (Mouret et  al., 2006), although in relative terms, the amount of CPD damage induced by UVA is about five-​fold less than an erythemally equivalent dose of UVB (Tewari et al., 2012). For any given dose of UVA, however, the yield of CPDs is greater in the cells of the basal epidermis than in more superficial layers, whereas the potency of UVB diminishes with increasing depth (Tewari et al., 2012). These findings are important, since cells in the basal layer include the stem cells and transitional cells that are likely to be the progenitors of keratinocyte cancers (Halliday and Cadet, 2012). Given that around 95% of the daily UV irradiance falls in the UVA range, the realization that UVA also causes CPDs in measurable quantities in the dividing cells of the basal epidermis has potential ramifications for primary prevention strategies, which have focused mainly on the hazards arising from UVB wavelengths. As yet, it is too early to determine how these laboratory findings might affect public health advice; larger scale field studies are required to address the relative importance of UVA and UVB at a population level. The mechanism through which UVA causes CPDs appears to involve melanin, the complex pigment produced by melanocytes that shields nuclear material from UV radiation (Abdel-​Malek and Cassidy, 2015). Following UVA exposure, various classes of enzymes are induced that generate large intracellular quantities of nitric oxide and superoxide. At the same time, UVA sufficiently solubilizes melanin polymers into small fragments that are able to enter the nucleus of cells. Once exposed to nitric oxide and superoxides, the melanin fragments undergo electrolysis in an electrolytic reaction, releasing similar energy to a UV photon, sufficient to generate CPDs when in close proximity to DNA (Premi et al., 2015). This mechanism is thought to explain the observation that induction of CPDs by UVA in skin cells containing melanin peaks several hours after exposure (with so-​called dark CPDs), unlike UVB, for which maximal photodamage is almost instantaneous. Of note, it underscores the potential hazards to DNA integrity from UVA light sources (such as tanning salons) (Abdel-​ Malek and Cassidy, 2015).

DNA Repair Terrestrial organisms have evolved highly efficient mechanisms to repair DNA damaged by UV radiation. Enzymatic nucleotide excision repair (NER) is the only pathway through which UVB-​induced photoproducts are removed in placental mammals (Vermeulen and Fousteri, 2013). NER operates through two mechanisms, called global genome NER (GG-​ NER) and transcription-​coupled NER (TC-​NER), respectively (Marteijn et al., 2014). Other types of DNA photodamage induced by UVR are repaired by different mechanisms; for example, oxidative lesions are removed by base excision repair. In GG-​NER, the entire genome is scanned for distortions to the double-​helix caused by destabilizing DNA lesions. The XPC protein plays a central role in recognizing DNA damage, and in the specific case of CPDs, recognition is enhanced by UV-​DDB (UVR-​DNA damage binding protein). Once detected, a cascade of proteins acts to verify the lesion (XPB and XPD, encoded by ERCC3 and ERCC2 genes, respectively), excise the damaged DNA fragment of about 22–​30 nucleotides (XPF and XPG), and then synthesize the sequence and ligate it to the preserved single strand. The GG-​NER process is relatively slow, however, and it is not uncommon that photoproducts remain unrepaired at the time of transcription or cell replication. Stalled transcription can have lethal consequences for cells, and so the TC-​NER pathway has evolved to remove stalled complexes and to ensure that damaged segments can be repaired and transcribed in a timely fashion. The precise sequence of events is the subject of intense research scrutiny, but it appears that the key trigger for TC-​NER is when RNA polymerase II reaches a bulky DNA photolesion and is unable to proceed with RNA synthesis. This initiates a complex cascade of verification and excision proteins, which remove the damaged DNA fragment and synthesize

 253



Ultraviolet Radiation

a new sequence. The fidelity of the transcription-​ coupled repair process seems to vary according to conditions, however, with both error-​free and error-​prone repair pathways known to exist (Ikehata and Ono, 2011). Overall, it is estimated that about 90% of CPD damage is repaired by GG-​NER, with the remainder repaired by TC-​NER (Vermeulen and Fousteri, 2013). The importance of DNA repair pathways has been shown in people with xeroderma pigmentosum (XP), who lack enzymes that specifically repair UV-​induced lesions (Kiss and Anstey, 2013). These individuals manifest extreme sun sensitivity, premature development of multiple skin cancers, and early death. Even within the population at large, there is considerable variation in DNA repair efficiency, prompting investigations of whether people with less efficient DNA repair mechanisms are at higher risk of UV-​associated cancers, with conflicting findings to date (Han et al., 2005; Mocellin et al., 2009; Ruczinski et al., 2012). Actively transcribed genes are repaired more quickly than infrequently expressed genes, and since the pattern of transcriptional activity differs by cell type, so too will the distribution of mutations.

UV Radiation Mutagenesis Mutations occur when DNA repair mechanisms fail to restore the original sequence of nucleotides that existed prior to photodamage. There is consensus that certain photolesions confer a higher probability of erroneous repair than others, and hence have higher mutagenicity (Pfeifer and Besaratinia, 2012). CPD photoproducts account for the majority of mutations in mammalian cells, as they are incurred more commonly than 6–​4 photoproducts following UVR exposure, and are repaired much more slowly. Even for CPDs, however, it appears that TT sequences are repaired with higher fidelity than TC sequences and that methylated dipyrimidine sites are particular “hot spots” for UVR mutations. DNA damage arising from oxidative stress is also mutagenic, causing G to A mutations. It has long been known that mutations in key regulatory genes are the hallmarks of cancer. Early work focused on the role of tumor suppressor genes, so named because loss-​of-​function mutations confer increased risks of cancer in experimental models. TP53 is one such gene, whose protein product (p53) plays a central role in cell cycle regulation, apoptosis, and DNA repair (Lane, 1992). When this function is lost through mutation or deletion, cells have a reduced capacity to repair DNA damage, facilitating cell transformation. Early research showed that UVR signature mutations in TP53 were detectable in about 50% of basal cell carcinomas (BCC) and more than 90% of squamous cell carcinomas (SCC) in humans (Brash et al., 1991; Campbell et al., 1993; Rady et al., 1992). Of epidemiological importance has been recent research documenting the high prevalence of UV signature mutations in non-​cancerous “normal” human skin. Careful mapping studies have revealed that up to 30% of skin cells on the eyelid carry non-​silent driver mutations in known skin cancer genes including NOTCH1, NOTCH2, TP53, and FAT1 (Martincorena et  al., 2015), suggesting that “preneoplastic clones” are extremely common in sun-​exposed human epidermis. Understanding how sunlight and genes interact to promote tumor development and establishing how the process can be halted or reversed once initiated are the focus of intense research (Viros et al., 2014).

253

This is the tanning response induced by sun exposure (Chen et  al., 2014). From the perspective of cancer epidemiology, it is important to recognize that there are two types of melanin, the brown/​black eumelanin and the red/​yellow pheomelanin, and the relative abundance of each is determined largely by MC1R genotype (Garcia-​Borron et al., 2014). MC1R is a highly polymorphic gene with more than 200 variants recorded. MC1R variants associated with red hair color represent a loss-​of-​function mutation that results in the production of pheomelanin. In addition to its central role in the tanning response, MC1R is now understood to protect against the adverse effects of UVR through non-​pigmentary mechanisms (Robinson et al., 2010), including DNA repair (Swope et al., 2014).

Immunosuppression UV radiation suppresses immunity in humans and animals; however, the extent, duration, and type of immunsuppression depend on many factors (extrinsic and intrinsic) and are subject to complex regulatory control (Halliday et al., 2012; Hart et al., 2011). Both UVA and UVB induce immune suppression of the skin, probably by different mechanisms (Damian et al., 2011). The molecular mediators of UV-​induced local immunosuppression include DNA damage, cis-​urocanic acid, and reactive oxygen species (Schwarz and Schwarz, 2011). At the cellular level, UV exposure causes Langerhans cells, the dendritic antigen-​presenting cells of the skin, to migrate to draining lymph nodes and activate T and B regulatory cells (Halliday et  al., 2012). Other cell types demonstrated to play a role in UV-​induced immunosuppresion of the skin include mast cells (Byrne et al., 2008), natural killer cells (Fukunaga et al., 2010), and effector and memory T-​cells (Rana et al., 2008). UV radiation can also suppress immune responses systemically at sites distant from the skin, including lower levels of activity of memory T-​cells in bone marrow and spleen. The evidence that immune surveillance plays a role in protecting against skin cancer development is indirect, derived mainly from the high incidence of skin cancers in organ transplant patients treated with immunosuppressive drugs, an effect that is amplified in immunosuppressed patients who experience high sun exposure (Euvrard et  al., 2003; Vajdic and van Leeuwen, 2009).

Vitamin D Synthesis UVB radiation is necessary for the conversion in the skin of 7-​dehydrocholesterol to cholecalciferol (vitamin D3), the precursor to the biologically active form of vitamin D.  When measured by dosimetry, personal UV exposure contributes modestly (about 8% of the variance explained) to serum vitamin D concentrations (Kimlin et  al., 2014)  after controlling for related factors like latitude and seasonal factors, use of vitamin D supplements, race, body mass index, and physical activity (Freedman, 2013; Kimlin, 2014; Millen, 2010). Observational studies have suggested that people with low serum concentrations of vitamin D have increased risks of cancers at various sites (particularly colon and breast) when compared with people who have higher concentrations of vitamin D. At the present time, however, the observational evidence regarding vitamin D and cancer falls short of causality (IARC, 2008; Manson and Bassuk, 2015) and is not supported by the findings of clinical trials (Bolland et al., 2014).

UV Radiation and Gene Expression

CANCERS ASSOCIATED WITH UV EXPOSURE

Exposure to UV radiation invariably leads to DNA damage in keratinocytes. Upon detection of damage, p53 is stabilized to halt keratinocyte replication and allow the damage to be repaired. At the same time, p53 activates the transcription of pro-​opiomelanocortin, which is then cleaved to alpha-​melanocyte stimulating hormone (α-​MSH) and released by keratinocytes. α-​MSH binds to the melanocortin-​1 receptor (MC1R) on the cell membrane of neighboring melanocytes, launching an intracellular cAMP-​signaling cascade, culminating in increased production of melanin pigments for distribution back to keratinocytes.

Keratinocyte Cancers There is a large body of observational and experimental evidence that UV radiation is the principal environmental cause of keratinocyte cancers (basal cell carcinomas and squamous cell carcinomas of the skin) (International Agency for Research on Cancer, 1992, 2012; Olsen et al., 2015) (see Chapter 58). Salient points are that keratinocyte cancers are more common among those living in areas of high solar irradiance, and residents of regions with low solar irradiance

254

254

PART III:  THE CAUSES OF CANCER

develop more lesions when they migrate to areas with higher ambient UV radiation. Both BCC and SCC of the skin occur almost exclusively on sun-​exposed body sites among fair-​skinned populations and are frequent in patients with constitutional genetic deficiencies preventing the repair of UVB-​specific DNA mutations. Field trials have demonstrated that regular use of sunscreen reduces the incidence of actinic keratoses (AK) (Darlington et al., 2003; Thompson et al., 1993) and SCC tumors in humans (Green et al., 1999). Finally recent meta-​analyses have shown that sunbed use significantly increases risks of BCC and SCC (Wehner et al., 2012). The era of high-​throughput genomics has revealed the full extent of UV-​induced mutations for each of the common types of skin cancer. Sequencing studies have shown that the total mutational loads for BCC and SCC far exceed those of other common cancers, with up to 75 mutations per megabase (Mb) of DNA for BCC and 35 per Mb for SCC (Jayaraman et al., 2014; South et al., 2014). Overwhelmingly, the most prevalent mutations in the somatic genome of these tumors are the classic UVR signatures—​C to T substitutions at dipyrimidine sites (Hodis et al., 2012; South et al., 2014). The key challenge, however, has been to separate the small number of critically important mutations that drive tumor development from the thousands of passenger mutations scattered throughout the genome that are acquired as the tumor evolves. For SCC, driver mutations appear restricted to a relatively small subset of genes:  NOTCH1, NOTCH2, TP53, CDKN2A, NRAS, and KRAS (Durinck et al., 2011; South et al., 2014; Wang et al., 2011); these mutations typically carry UVR signatures. Driver mutations appear similarly restricted to a small number of genes for BCC (PTCH, TP53) (Jayaraman et  al., 2014). For both types of keratinocyte cancers, other genes have been identified as candidate drivers at lower frequency, and these await confirmation in larger studies.

Finally, mouse models support observational human studies showing that sun exposure in early life has particularly adverse effects (Noonan et al., 2001, 2012; Viros et al., 2014). Numerous epidemiologic studies have examined the possible association of UV with ocular melanoma, and it appears that occupational sun exposure, especially farming and welding, is associated with higher risks of melanoma of the choroid and ciliary body (Vajdic et al., 2002, 2004).

Other Cancers There has been speculation that exposure to UV radiation may indirectly affect a person’s risk of developing cancers at sites other than skin, but evidence from large prospective studies that have tested this hypothesis directly is scarce. The larger of two relevant cohort studies, the National Institutes of Health (NIH)-​AARP Diet and Health Study, linked ambient erythemal UV exposure estimated from satellite Total Ozone Mapping Spectrometer data to the Census 2000 record of baseline residence for 450,934 white, non-​Hispanic subjects and examined their cancer outcomes after 9  years of follow-​up using cancer registries. After accounting for confounding factors, ambient UV exposure was associated with significant decreases in non-​Hodgkin’s lymphoma, colon, squamous cell lung, pleural, prostate, kidney, and bladder cancers (Lin et al., 2012). On the other hand, a Scandinavian cohort study of 49,261 women found no evidence of an association between any measure of personal cumulative UV exposure at ages 10–​39 and overall cancer occurrence after 15 years follow-​up, though sunbathing at ages 10–​29 was related to reduced breast cancer and cancer overall, ≥ 2 sunburns/​year was related to reduced lung cancer, and solarium use was also inversely associated with breast cancer (Yang et al., 2011). On balance, there is insufficient strong evidence available to draw firm conclusions.

Melanoma Cutaneous melanoma is a cancer of the pigment-​producing cells of the skin. The causal role of UV exposure is complex and is determined by phenotypic, genotypic, and behavioral factors (Whiteman et  al., 2011) (see Chapter 63). The evidence linking UV exposure with cutaneous melanoma is similar to that for keratinocyte cancer, namely that melanomas develop more commonly among fair-​skinned migrants to high UV environments than among the population of origin (Whiteman et al., 2001); and sun-​exposed sites, such as the face and ears, are the most commonly affected (per unit area of skin), but intermittently exposed sites, such as the trunk and proximal limbs, are also often affected (Green et al., 1993). Estimates of the population fraction attributable to sun exposure range from 60% to 95% (Olsen et al., 2015). UV-​ induced mutations are highly prevalent throughout the somatic genome of cutaneous melanomas (at around 14 per Mb), both as drivers and passengers, as described earlier (Hodis et  al., 2012). Key genes affected in melanoma include BRAF, NRAS, PTEN, TP53, CDKN2A, and MAP2K1 (Hodis et  al., 2012). Unlike SCC and BCC, however, not all driver mutations for melanoma have the classical UVR signature. The most prevalent somatic mutation in melanoma, occurring in around 50% of tumors, occurs in BRAF and involves a T to A transversion at position 1799 (resulting in the substitution of valine for glutamine at codon 600, hence the abbreviation BRAFV600E). There is conjecture about whether UVR could be responsible for such a mutation, since BRAFV600E mutations are also commonly found in internal cancers (Davies et al., 2002). The most favored explanation is that this pattern is consistent with oxidative damage in the target cell, for which UVR is one such source (Denat et al., 2014). Converging lines of evidence support this theory, since melanin (particularly pheomelanin) is a potent producer of oxygen radicals when exposed to UVR (Napolitano et  al., 2014), and experimental studies indicate that UVA alone induces melanoma in pigmented but not albino mice through oxidative damage (Noonan et al., 2012). It is also possible that pheomelanin induces melanoma independently of sun exposure (Mitra et  al., 2012), although this remains to be established in human studies.

OPPORTUNITIES FOR PREVENTION Reducing Sun Exposure Many health promotion and health education strategies have been developed to change knowledge, attitudes, and behaviors to reduce susceptible people’s sun exposure. Behavioral interventions aim to improve personal sun protection through use of clothing, hats, sunscreen, and sunglasses, staying in shade, and avoiding exposure during peak UV hours. This section provides a broad overview of approaches to reduce exposure to UV radiation. A more detailed discussion of practical issues regarding the implementation of these approaches is provided in Chapter 62.4. Preventive interventions can be delivered through a variety of means including education, counseling by healthcare providers, policy and environmental changes, and community-​wide engagement through media or social media campaigns (NICE, 2007), or by combining several of these approaches (Saraiya et  al., 2004). A  systematic review of the evidence for behavioral counseling to prevent skin cancer conducted for the US Preventive Services Tasks Force (Lin et  al., 2011)  resulted in recommendations to counsel children, adolescents, and young adults aged 10–​24  years who have fair skin to minimize UV exposure to reduce skin cancer risk (Moyer, 2012), but claimed insufficient evidence to assess the balance of benefits and harms of counseling adults over 24 about skin cancer prevention, despite several successful field trials in adults (Green et  al., 2012). Another systematic review of the evidence for reducing UV radiation exposure among outdoor workers found men more likely to wear hats and protective clothing and women more likely to use sunscreen, but found little evidence concerning education and prevention policies in the outdoor workplace (Glanz et al., 2007). Multiple-​level community-​wide interventions have been developed by cancer agencies in many countries. Australia led the way from the 1980s with mass-​media campaigns to increase the public’s knowledge about the link between sun exposure and skin cancer

 25



Ultraviolet Radiation

and to reduce high levels of recreational sun exposure. Beginning with the “Slip, Slop, Slap” Campaign with a logo that encouraged people to “slip on a shirt, slop on some sunscreen and slap on a hat,” successive campaigns targeted specific groups, for example, adolescents, outdoor workers, and schoolchildren (Iannacone and Green, 2014). These campaigns have been conducted predominantly by state Cancer Councils under a “SunSmart” license, reinforced by policies to adopt supportive environments such as shade in schools and access to quality sun-​protection products (Iannacone and Green, 2014). In New Zealand, campaigns have been similar in their history, with the New Zealand Cancer Society first conducting campaigns with an emphasis on children, parents, and teachers, and later on other sections of the community, informed by qualitative research (Watts, 2002). A  formal school accreditation program for sun protection has had positive results (Reeder, 2012). Countries in the Northern Hemisphere with lower skin cancer rates have also conducted comprehensive community-​based campaigns, but with lower levels of intensity and investment of resources (Hiom, 2010)  and often lacking national coordination (Lazovich et al., 2012). For example, the SunWise campaign was launched in 2000 by the US Environmental Protection Agency, and its school program had modest early success (Geller et al., 2003). Economic analysis indicated that if the SunWise school program continued to 2015 it should avert nearly 11,000 skin cancer cases and 960 quality-​adjusted life-​years (undiscounted) among its participants (Kyle et al., 2008). For every dollar invested in SunWise, between approximately $2 and $4 in medical care costs and productivity losses could be saved, depending on funding levels (Kyle et  al., 2008). In the United Kingdom, primary prevention efforts commenced in the 1990s with the government-​supported national skin cancer prevention campaign Sun Know How. Since 2003, Cancer Research UK, commissioned by the UK Health Departments, has run the SunSmart campaign with varying levels of subsequent funding (Hiom, 2010), supported by campaigns conducted by dermatologists and other charities.

Reducing Exposure to Artificial UV for Tanning A review of the trends in legislation to restrict access to indoor tanning found that the number of countries with such legislation increased from two countries in 2003 to 11 countries in 2011 (Pawlak et  al., 2012). Brazil has subsequently banned tanning salons (in 2009) and Australia in 2015. Legislated restrictions of the tanning industry have been fewer in the United States, but in 2014, based on the accumulated evidence of harmful health effects of indoor tanning for cosmetic reasons, the US Food and Drug Administration reclassified tanning lamps from a class I (low risk) medical device to a class II (moderate risk) device, with the added requirement for black-​box labeling warning minors about the harms of using tanning lamps (Ernst et  al., 2015). Also in 2014, the Surgeon General of the US Department of Health and Human Services issued his “Call to Action to Prevent Skin Cancer,” which inter alia included the call to “reduce harms from indoor tanning” (US Department of Health & Human Services, 2014). It has been noted that some 44 US states now have some kind of law or regulation related to indoor tanning, though with varying strength of state laws and compliance enforcement, and some states have no laws related to indoor tanning (Pan and Geller, 2015; US Department of Health & Human Services, 2014).

FUTURE RESEARCH Use of new technologies, including social media, is now being researched to reduce deleterious sun exposure. Implementation and evaluation programs of behavioral interventions, especially among men, adolescents, children, and their caregivers, are high research priorities.

255

References Abdel-​Malek ZA, and Cassidy P. 2015. Dark CPDs and photocarcinogenesis: the party continues after the lights go out. Pigment Cell Melanoma Res, 28(4), 373–​374. PMID: 25950490. Allen M, and McKenzie R. 2005. Enhanced UV exposure on a ski-​field compared with exposures at sea level. Photochem Photobiol Sci, 4(5), 429–​ 437. PMID: 15875076. Archier E, Devaux S, Castela E, et al. 2012. Carcinogenic risks of psoralen UV-​ A therapy and narrowband UV-​B therapy in chronic plaque psoriasis: a systematic literature review. J Eur Acad Dermatol Venereol, 26 Suppl 3, 22–​31. PMID: 22512677. Autier P, Dore JF, Breitbart E, et al. 2011. The indoor tanning industry’s double game. Lancet, 377(9774), 1299–​1301. PMID: 21324519. Bais AF, McKenzie RL, Bernhard G, et al. 2015. Ozone depletion and climate change: impacts on UV radiation. Photochem Photobiol Sci, 14(1), 19–​52. PMID: 25380284. Battistutta D, Pandeya N, Strutton GM, et  al. 2006. Skin surface topography grading is a valid measure of skin photoaging. Photodermatol Photoimmunol Photomed, 22(1), 39–​45. PMID: 16436180. Bolland MJ, Grey A, Gamble GD, and Reid IR. 2014. The effect of vitamin D supplementation on skeletal, vascular, or cancer outcomes:  a trial sequential meta-​analysis. Lancet Diabetes Endocrinol, 2(4), 307–​320. PMID: 24703049. Boniol M, Autier P, Boyle P, and Gandini S. 2012. Cutaneous melanoma attributable to sunbed use: systematic review and meta-​analysis. Br Med J, 345, e4757. PMCID: PMC3404185. Brash D, Rudolph JA, Simon JA, et al. 1991. A role for sunlight in skin cancer: UV-​induced p53 mutations in squamous cell carcinoma. Proc Natl Acad Sci U S A, 88, 10124–​10128. Brash DE. 2015. UV signature mutations. Photochem Photobiol, 91(1), 15–​26. PMCID: PMC4294947. Byrne SN, Limon-​Flores AY, and Ullrich SE. 2008. Mast cell migration from the skin to the draining lymph nodes upon ultraviolet irradiation represents a key step in the induction of immune suppression. J Immunol, 180(7), 4648–​4655. PMCID: PMC2391302. Cahoon EK, Wheeler DC, Kimlin MG, et al. 2013. Individual, environmental, and meteorological predictors of daily personal ultraviolet radiation exposure measurements in a United States cohort study. PLoS ONE, 8(2), e54983. Campbell C, Quinn A, Ro Y, Angus B, and Rees J. 1993. p53 mutations are common and early evens that precede tumor invasion in squamous cell neoplasia of the skin. J Invest Dermatol, 100, 746–​748. Chandrasena A, Amin K, and Powell B. 2013. Dying for a tan:  a survey to assess solarium adherence to world health organization guidelines in Australia, New Zealand, and the United Kingdom. Eplasty, 13, e62. PMCID: PMC3881211. Chen H, Weng QY, and Fisher DE. 2014. UV signaling pathways within the skin. J Invest Dermatol, 134(8), 2080–​2085. PMCID: PMC4102648. Cockell CS, Scherer K, Horneck G, et al. 2001. Exposure of arctic field scientists to ultraviolet radiation evaluated using personal dosimeters. Photochem Photobiol, 74(4), 570–​578. Cooke MS, Harry EL, Liljendahl TS, and Segerback D. 2013. DNA nucleotide excision repair, where do all the cyclobutane pyrimidine dimers go? Cell Cycle, 12(10), 1642. PMCID: PMC3680545. Damian DL, Matthews YJ, Phan TA, and Halliday GM. 2011. An action spectrum for ultraviolet radiation-​induced immunosuppression in humans. Br J Dermatol, 164(3), 657–​659. PMID: 21375518. Darlington S, Williams G, Neale R, Frost C, and Green A. 2003. A randomized controlled trial to assess sunscreen application and beta carotene supplementation in the prevention of solar keratoses. Arch Dermatol, 139(4), 451–​455. Davies H, Bignell GR, Cox C, et  al. 2002. Mutations of the BRAF gene in human cancer. Nature, 417(6892), 949–​954. PMID: 12068308. Denat L, Kadekaro AL, Marrot L, Leachman SA, and Abdel-​ Malek ZA. 2014. Melanocytes as instigators and victims of oxidative stress. J Invest Dermatol, 134(6), 1512–​1518. PMCID: PMC4418514. Diffey BL. 1987. A comparison of dosimeters used for solar ultraviolet radiometry. Photochem Photobiol, 46(1), 55–​60. PMID: 3615634. Diffey BL, Gibson CJ, Haylock R, and McKinlay AF. 1996. Outdoor ultraviolet exposure of children and adolescents. Br J Dermatol, 134(6), 1030–​1034. Diffey BL, and Gies HP. 1998. The confounding influence of sun exposure in melanoma. Lancet, 351, 1101–​1102. Durinck S, Ho C, Wang NJ, et al. 2011. Temporal dissection of tumorigenesis in primary cancers. Cancer Discov, 1(2), 137–​143. PMCID: PMC3187561. English DR, Armstrong BK, and Kricker A. 1998. Reproducibility of reported measurements of sun exposure in a case-​control study. Cancer Epidemiol Biomarkers Prev, 7(10), 857–​863. PMID: 9796629.

256

256

PART III:  THE CAUSES OF CANCER

Ernst A, Grimm A, and Lim HW. 2015. Tanning lamps:  health effects and reclassification by the Food and Drug Administration. J Am Acad Dermatol, 72(1), 175–​180. PMID: 25458016. Euvrard S, Kanitakis J, and Claudy A. 2003. Skin cancers after organ transplantation. N Engl J Med, 348(17), 1681–​1691. PMID: 12711744. Frederick JE, Slusser JR, and Bigelow DS. 2000. Annual and interannual behavior of solar ultraviolet irradiance revealed by broadband measurements. Photochem Photobiol, 72(4), 488–​496. PMID: 11045720. Fukunaga A, Khaskhely NM, Ma Y, et  al. 2010. Langerhans cells serve as immunoregulatory cells by activating NKT cells. J Immunol, 185(8), 4633–​4640. PMCID: PMC2950871. Garcia-​Borron JC, Abdel-​Malek Z, and Jimenez-​Cervantes C. 2014. MC1R, the cAMP pathway, and the response to solar UV:  extending the horizon beyond pigmentation. Pigment Cell Melanoma Res, 27(5), 699–​720. PMCID: PMC4150834. Geller A, Rutsch L, Kenausis K, and Zhang Z. 2003. Evaluation of the SunWise School Program. J Sch Nurs, 19(2), 93–​99. PMID: 12653621. Glanz K, Buller DB, and Saraiya M. 2007. Reducing ultraviolet radiation exposure among outdoor workers: state of the evidence and recommendations. Environ Health, 6, 22. PMCID: PMC1995198. Godar DE. 2005. UV doses worldwide. Photochem Photobiol, 81(4), 736–​749. PMID: 15819599. Grange F, Mortier L, Crine A, et al. 2015. Prevalence of sunbed use, and characteristics and knowledge of sunbed users: results from the French population-​based Edifice Melanoma survey. J Eur Acad Dermatol Venereol, 29 Suppl 2, 23–​30. PMID: 25639930. Green A, Harwood C, Lear J, et  al. 2012. Skin cancer prevention:  recent evidence from randomized controlled trials. Curr Dermatol Rep, 1(3), 123–​130. Green A, MacLennan R, Youl P, and Martin N. 1993. Site distribution of cutaneous melanoma in Queensland. Int J Cancer, 53, 232–​236. Green A, Siskind V, Bain C, and Alexander J. 1985. Sunburn and malignant melanoma. Br J Cancer, 51, 393–​397. Green A, Williams G, Nèale R, et  al. 1999. Daily sunscreen application and betacarotene supplementation in prevention of basal-​ cell and squamous-​cell carcinomas of the skin: a randomised controlled trial. Lancet, 354(9180), 723–​729. Green AC, and Whiteman DC. 2006. Solar radiation. In: Schottenfeld Joseph F. Fraumeni, Jr. (ed.), Cancer Epidemiology and Prevention (3rd ed., pp. 294–​305). New York: Oxford University Press. Halliday GM, and Cadet J. 2012. It’s all about position:  the basal layer of human epidermis is particularly susceptible to different types of sunlight-​ induced DNA damage. J Invest Dermatol, 132(2), 265–​ 267. PMID: 22241442. Halliday GM, Damian DL, Rana S, and Byrne SN. 2012. The suppressive effects of ultraviolet radiation on immunity in the skin and internal organs: implications for autoimmunity. J Dermatol Sci, 66(3), 176–​182. PMID: 22277701. Han J, Colditz GA, Liu JS, and Hunter DJ. 2005. Genetic variation in XPD, sun exposure, and risk of skin cancer. Cancer Epidemiol Biomarkers Prev, 14(6), 1539–​1544. PMID: 15941969. Hart PH, Gorman S, and Finlay-​Jones JJ. 2011. Modulation of the immune system by UV radiation: more than just the effects of vitamin D? Nat Rev Immunol, 11(9), 584–​596. PMID: 21852793. Henderson S, Javorniczky J, and Gies P. 2010. Spectral measurements of solar UV at several altitudes under Australian conditions. Paper presented at the 2010 UV Workshop, Wellington, New Zealand. https://​www.niwa. co.nz/​sites/​niwa.co.nz/​files/​spectral_​measurements_​of_​uv.pdf. Heydenreich J, and Wulf HC. 2005. Miniature personal electronic UVR dosimeter with erythema response and time-​stamped readings in a wristwatch. Photochem Photobiol, 81(5), 1138–​1144. PMID: 15850424. Hiom S, Naumann L, Cerny C. 2010. CRUK Expert paper 2: National campaigns (UK and worldwide). Cancer Research UK. Hodis E, Watson IR, Kryukov GV, et al. 2012. A landscape of driver mutations in melanoma. Cell, 150(2), 251–​263. PMCID: PMC3600117. Hooke R, Pearson A, and O’Hagan J. 2014. Autonomous portable solar ultraviolet spectroradiometer (APSUS): a new CCD spectrometer system for localized, real-​time solar ultraviolet (280–​400  nm) radiation measurement. Photochem Photobiol, 90(4), 903–​910. PMID: 24483759. Huber M, Blumthaler M, Schreder J, Bais A, and Topaloglou C. 2002. Effect of ambient temperature on Robertson-​berger-​type erythemal dosimeters. Applied Optics, 41(21), 4273–​4277. Iannacone MR, and Green AC. 2014. Towards skin cancer prevention and early detection:  evolution of skin cancer awareness campaigns in Australia. Melanoma Manage, 1(1), 75–​84. Ikehata H, and Ono T. 2011. The mechanisms of UV mutagenesis. J Radiat Res, 52(2), 115–​125. PMID: 21436607.

International Agency for Research on Cancer. 1992. IARC monographs on the evaluation of carcinogenic risks to humans:  solar and ultraviolet radiation. IARC Monogr Eval Carcinog Risks Hum, 55, 1–​316. PMID: 1345607. International Agency for Research on Cancer. 2012. IARC monographs on the evaluation of carcinogenic risks to humans: radiation. IARC Monogr Eval Carcinog Risks Hum, 100D, 1–​68. Jayaraman SS, Rayhan DJ, Hazany S, and Kolodney MS. 2014. Mutational landscape of basal cell carcinomas by whole-​exome sequencing. J Invest Dermatol, 134(1), 213–​220. PMID: 23774526. Jennings L, Karia PS, Jambusaria-​Pahlajani A, Whalen FM, and Schmults CD. 2013. The Sun Exposure and Behaviour Inventory (SEBI): validation of an instrument to assess sun exposure and sun protective practices. J Eur Acad Dermatol Venereol, 27(6), 706–​715. PMID: 22487012. Khazova M, O’Hagan JB, and Robertson S. 2015. Survey of UV emissions from sunbeds in the UK. Photochem Photobiol, 91(3), 545–​552. PMID: 25626375. Kimlin MG, Lucas RM, Harrison SL, et al. 2014. The contributions of solar ultraviolet radiation exposure and other determinants to serum 25-​hydroxyvitamin D concentrations in Australian adults: the AusD Study. Am J Epidemiol, 179(7), 864–​874. PMID: 24573539. Kiss F, and Anstey AV. 2013. A review of UVB-​mediated photosensitivity disorders. Photochem Photobiol Sci, 12(1), 37–​46. PMID: 23023766. Kotova N, Hemminki K, and Segerback D. 2005. Urinary thymidine dimer as a marker of total body burden of UV-​ inflicted DNA damage in humans. Cancer Epidemiol Biomarkers Prev, 14(12), 2868–​2872. PMID: 16365002. Kyle JW, Hammitt JK, Lim HW, et  al. 2008. Economic evaluation of the US Environmental Protection Agency’s SunWise program:  sun protection education for young children. Pediatrics, 121(5), e1074–​ 1084. PMID: 18450850. Lake JR, Thomson CS, Twelves CJ, and Davies EA. 2014. A qualitative investigation of the motivations, experiences and views of female sunbed users under the age of 18 in England. J Public Health (Oxf), 36(1), 56–​64. PMID: 23365261. Lane DP. 1992. p53, guardian of the genome. Nature, 359, 486–​487. Lazovich D, Choi K, and Vogel RI. 2012. Time to get serious about skin cancer prevention. Cancer Epidemiol Biomarkers Prev, 21(11), 1893–​1901. PMID: 22962407. Liljendahl TS, Blomqvist A, Andersson EM, Barregard L, and Segerback D. 2013. Urinary levels of thymine dimer as a biomarker of exposure to ultraviolet radiation in humans during outdoor activities in the summer. Mutagenesis, 28(3), 249–​256. PMID: 23339196. Liljendahl TS, Kotova N, and Segerback D. 2012. Quantification of ultraviolet radiation-​induced DNA damage in the urine of Swedish adults and children following exposure to sunlight. Biomarkers, 17(7), 634–​641. PMID: 22889379. Lin JS, Eder M, and Weinmann S. 2011. Behavioral counseling to prevent skin cancer: a systematic review for the U.S. Preventive Services Task Force. Ann Intern Med, 154(3), 190–​201. Lin SW, Wheeler DC, Park Y et al. 2012. Prospective study of ultraviolet radiation exposure and risk of cancer in the United States. Int J Cancer, 131(6), E1015–​1023. PMCID: PMC3402606. Manson JE, and Bassuk SS. 2015. Vitamin D research and clinical practice: at a crossroads. JAMA, 313(13), 1311–​1312. PMID: 25695911. Marteijn JA, Lans H, Vermeulen W, and Hoeijmakers JH. 2014. Understanding nucleotide excision repair and its roles in cancer and ageing. Nat Rev Mol Cell Biol, 15(7), 465–​481. PMID: 24954209. Martincorena I, Roshan A, Gerstung M, et  al. 2015. Tumor evolution:  high burden and pervasive positive selection of somatic mutations in normal human skin. Science, 348(6237), 880–​886. PMCID: PMC4471149. Millen AE, Wactawski-​ Wende J, Pettinger M, Melamed ML, Tylavsky FA, Liu S, Robbins J, LaCroix AZ, LeBoff MS, Jackson RD. 2010. Predictors of serum 25-​hydroxyvitamin D concentrations among postmenopausal women:  The Women's Health Initiative Calcium plus Vitamin D clinical trial. The American Journal of Clinical Nutrition. 91(5):1324–​1235. Mitra D, Luo X, Morgan A, et al. 2012. An ultraviolet-​radiation-​independent pathway to melanoma carcinogenesis in the red hair/​fair skin background. Nature, 491(7424), 449–​453. PMCID: PMC3521494. Mocellin S, Verdi D, and Nitti D. 2009. DNA repair gene polymorphisms and risk of cutaneous melanoma:  a systematic review and meta-​analysis. Carcinogenesis, 30(10), 1735–​1743. PMID: 19706646. Moehrle M, Dennenmoser B, and Garbe C. 2003. Continuous long-​term monitoring of UV radiation in professional mountain guides reveals extremely high exposure. Int J Cancer, 103(6), 775–​ 778. PMID: 12516097.

 257



Ultraviolet Radiation

Morze CJ, Olsen CM, Perry SL, et al. 2012. Good test-​retest reproducibility for an instrument to capture self-​reported melanoma risk factors. J Clin Epidemiol, 65(12), 1329–​1336. PMID: 22974496. Mouret S, Baudouin C, Charveron M, et  al. 2006. Cyclobutane pyrimidine dimers are predominant DNA lesions in whole human skin exposed to UVA radiation. Proc Natl Acad Sci U S A, 103(37), 13765–​13770. PMCID: PMC1564232. Moyer VA. 2012. Behavioral counseling to prevent skin cancer: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med, 157(1), 59–​65. PMID: 22751761. Munakata N, Ono M, and Watanabe S. 1998. Monitoring of solar-​UV exposure among schoolchildren in five Japanese cities using spore dosimeter and UV-​coloring labels. Jpn J Cancer Res, 89(3), 235–​245. Napolitano A, Panzella L, Monfrecola G, and d’Ischia M. 2014. Pheomelanin-​ induced oxidative stress:  bright and dark chemistry bridging red hair phenotype and melanoma. Pigment Cell Melanoma Res, 27(5), 721–​733. PMID: 24814217. National Institute for Health and Clinical Excellence. NICE public health guidance 6: Behaviour change at population, community and individual levels. London; 2007. Nilsen LT, Aalerud TN, Hannevik M, and Veierod MB. 2011. UVB and UVA irradiances from indoor tanning devices. Photochem Photobiol Sci, 10(7), 1129–​1136. PMID: 21445424. Noonan FP, Recio JA, Takayama H, et al. 2001. Neonatal sunburn and melanoma in mice. Nature, 413(6853), 271–​272. PMID: 11565020. Noonan FP, Zaidi MR, Wolnicka-​Glubisz A, et al. 2012. Melanoma induction by ultraviolet A but not ultraviolet B radiation requires melanin pigment. Nat Commun, 3, 884. PMCID: PMC3621412. Olsen CM, Wilson LF, Green AC, et al. 2015. Cancers in Australia attributable to exposure to solar ultraviolet radiation and prevented by regular sunscreen use. Aust N Z J Public Health, 39(5), 471–​476. PMCID: PMC4606762. Pan M, and Geller L. 2015. Update on indoor tanning legislation in the United States. Clin Dermatol, 33(3), 387–​392. PMID: 25889141. Pawlak MT, Bui M, Amir M, et  al. 2012. Legislation restricting access to indoor tanning throughout the world. Arch Dermatol, 148(9), 1006–​1012. PMID: 22801924. Pearson AJ, Dean SF, Clark IES et al. 2000. NRPB Solar ultraviolet radiation measurement network. Radiat Prot Dosim, 91(1–​3), 169–​172. Petersen B, Wulf HC, Triguero-​Mas M, et  al. 2014. Sun and ski holidays improve vitamin D status, but are associated with high levels of DNA damage. J Invest Dermatol, 134(11), 2806–​2813. PMID: 24844860. Pfeifer GP, and Besaratinia A. 2012. UV wavelength-​dependent DNA damage and human non-​melanoma and melanoma skin cancer. Photochem Photobiol Sci, 11(1), 90–​97. PMCID: PMC3289542. Premi S, Wallisch S, Mano CM, et al. 2015. Photochemistry: chemiexcitation of melanin derivatives induces DNA photoproducts long after UV exposure. Science, 347(6224), 842–​847. PMCID: PMC4432913. Quintern LE, Horneck G, Eschweiler U, and Bucker H. 1992. A biofilm usedas ultra-​violet dosimeter. Photochem Photobiol, 55, 389–​395. Rady P, Scinicariello F, Wagner RF, Jr, and Tyring SK. 1992. p53 mutations in basal cell carcinomas. Cancer Res, 52, 3804–​3806. Rana S, Byrne SN, MacDonald LJ, Chan CY, and Halliday GM. 2008. Ultraviolet B suppresses immunity by inhibiting effector and memory T cells. Am J Pathol, 172(4), 993–​1004. PMCID: PMC2276424. Reeder AI, Jopson JA, Gray A. 2012. Sun protection policies and practices in New Zealand primary schools. N Z Med J, 125(1349), 70–​82. Robinson S, Dixon S, August S, et al. 2010. Protection against UVR involves MC1R-​mediated non-​pigmentary and pigmentary mechanisms in vivo. J Invest Dermatol, 130(7), 1904–​1913. PMID: 20237490. Rowland FS. 2006. Stratospheric ozone depletion. Phil Trans R Soc B, 361(1469), 769–​790. PMID: PMC1609402. Roy CR, Gies HP, and Toomey S. 1997. Monitoring UV-​B at the earth’s surface. Austral Meteorol Mag, 46, 203–​210. Ruczinski I, Jorgensen TJ, Shugart YY et al. 2012. A population-​based study of DNA repair gene variants in relation to non-​melanoma skin cancer as a marker of a cancer-​prone phenotype. Carcinogenesis, 33(9), 1692–​1698. PMCID: PMC3514896. Saraiya M, Glanz K, Briss PA, et al. 2004. Interventions to prevent skin cancer by reducing exposure to ultraviolet radiation: a systematic review. Am J Prev Med, 27(5), 422–​466. Schwarz T, and Schwarz A. 2011. Molecular mechanisms of ultraviolet radiation-​induced immunosuppression. Eur J Cell Biol, 90(6–​7), 560–​564. PMID: 21035896. Seckmeyer G, Klingebiel M, Riechelmann S, et al. 2012. A critical assessment of two types of personal UV dosimeters. Photochem Photobiol, 88(1), 215–​222. PMID: 22017203.

257

South AP, Purdie KJ, Watt SA, et al. 2014. NOTCH1 mutations occur early during cutaneous squamous cell carcinogenesis. J Invest Dermatol, 134(10), 2630–​2638. PMID: 24662767. Sun J, Lucas RM, Harrison SL, et al. 2014. Measuring exposure to solar ultraviolet radiation using a dosimetric technique: understanding participant compliance issues. Photochem Photobiol, 90(4), 919–​ 924. PMID: 24571445. Swope V, Alexander C, Starner R, et  al. 2014. Significance of the melanocortin 1 receptor in the DNA damage response of human melanocytes to ultraviolet radiation. Pigment Cell Melanoma Res, 27(4), 601–​610. PMID: 24730569. Tewari A, Sarkany RP, and Young AR. 2012. UVA1 induces cyclobutane pyrimidine dimers but not 6–​4 photoproducts in human skin in vivo. J Invest Dermatol, 132(2), 394–​400. PMID: 21975824. Thieden E. 2008. Sun exposure behaviour among subgroups of the Danish population. Based on personal electronic UVR dosimetry and corresponding exposure diaries. Dan Med Bull, 55(1), 47–​68. PMID: 18321444. Thieden E, Agren MS, and Wulf HC. 2000. The wrist is a reliable body site for personal dosimetry of ultraviolet radiation. Photodermatol Photoimmunol Photomed, 16(2), 57–​61. PMID: 10823313. Thieden E, Heydenreich J, Philipsen PA, and Wulf HC. 2013. People maintain their sun exposure behaviour in a 5-​7-​year follow-​up study using personal electronic UVR dosimeters. Photochem Photobiol Sci, 12(1), 111–​116. PMID: 22786662. Thieden E, Philipsen PA, Heydenreich J, and Wulf HC. 2004a. UV radiation exposure related to age, sex, occupation, and sun behavior based on time-​ stamped personal dosimeter readings. Arch Dermatol, 140(2), 197–​203. PMID: 14967793. Thieden E, Philipsen PA, Sandby-​Moller J, Heydenreich J, and Wulf HC. 2004b. Proportion of lifetime UV dose received by children, teenagers and adults based on time-​stamped personal dosimetry. J Invest Dermatol, 123(6), 1147–​1150. PMID: 15610527. Thieden E, Philipsen PA, and Wulf HC. 2006. Compliance and data reliability in sun exposure studies with diaries and personal, electronic UV dosimeters. Photodermatol Photoimmunol Photomed, 22(2), 93–​99. PMID: 16606414. Thompson SC, Jolley D, and Marks R. 1993. Reduction of solar keratoses by regular sunscreen use. N Engl J Med, 329(16), 1147–​1151. PMID: 8377777. US Department of Health & Human Services. 2014. Reports of the Surgeon General. Washington, DC: Office of the Surgeon General. Vajdic CM, Kricker A, Giblin M, et  al. 2002. Sun exposure predicts risk of ocular melanoma in Australia. Int J Cancer, 101(2), 175–​182. Vajdic CM, Kricker A, Giblin M, et  al. 2004. Artificial ultraviolet radiation and ocular melanoma in Australia. Int J Cancer, 112(5), 896–​900. PMID: 15386378. Vajdic CM, and van Leeuwen MT. 2009. Cancer incidence and risk factors after solid organ transplantation. Int J Cancer, 125(8), 1747–​1754. van der Mei IA, Blizzard L, Ponsonby AL, and Dwyer T. 2006. Validity and reliability of adult recall of past sun exposure in a case-​control study of multiple sclerosis. Cancer Epidemiol Biomarkers Prev, 15(8), 1538–​1544. PMID: 16896046. Veierod MB, Parr CL, Lund E, and Hjartaker A. 2008. Reproducibility of self-​ reported melanoma risk factors in a large cohort study of Norwegian women. Melanoma Res, 18(1), 1–​9. PMID: 18227701. Vermeulen W, and Fousteri M. 2013. Mammalian transcription-​coupled excision repair. Cold Spring Harb Perspect Biol, 5(8), a012625. PMID: 23906714. Viros A, Sanchez-​Laorden B, Pedersen M et  al. 2014. Ultraviolet radiation accelerates BRAF-​driven melanomagenesis by targeting TP53. Nature, 511(7510), 478–​482. PMCID: PMC4112218. Wang NJ, Sanborn Z, Arnett KL, et  al. 2011. Loss-​of-​function mutations in Notch receptors in cutaneous and lung squamous cell carcinoma. Proc Natl Acad Sci U S A, 108(43), 17761–​17766. PMCID: PMC3203814. Watts C, Reeder AI, Glasgow H. 2002. A cover-​up story: the Cancer Society Melanoma Prevention programme. In: McKenzie RRA, Watts C (eds.), UV Radiation and Its Effects: An Update 2002 (pp. 1–​3). Christchurch: Royal Society of New Zealand. Weatherhead EC, and Andersen SB. 2006. The search for signs of recovery of the ozone layer. Nature, 441(7089), 39–​45. PMID: 16672963. Wehner MR, Chren MM, Nameth D, et al. 2014. International prevalence of indoor tanning: a systematic review and meta-​analysis. JAMA Dermatol, 150(4), 390–​400. PMCID: PMC4117411. Wehner MR, Shive ML, Chren MM, et  al. 2012. Indoor tanning and non-​ melanoma skin cancer: systematic review and meta-​analysis. BMJ, 345, e5909. PMCID: PMC3462818.

258

258

PART III:  THE CAUSES OF CANCER

Weihs P, Schmalwieser A, Reinisch C, et al. 2013. Measurements of personal UV exposure on different parts of the body during various activities. Photochem Photobiol, 89(4), 1004–​1007. PMID: 23587050. Whiteman D, and Green A. 1994. Melanoma and sunburn. Cancer Causes Control, 5, 564–​572. Whiteman DC, Pavan WJ, and Bastian BC. 2011. The melanomas: a synthesis of epidemiological, clinical, histopathological, genetic, and biological aspects, supporting distinct subtypes, causal pathways, and cells of origin. Pigment Cell Melanoma Res, 24(5), 879–​897. Whiteman DC, Whiteman CA, and Green AC. 2001. Childhood sun exposure as a risk factor for melanoma: a systematic review of epidemiologic studies. Cancer Causes Control, 12, 69–​82.

Williamson CE, Zepp RG, Lucas RM et al. 2014. Solar ultraviolet radiation in a changing climate. Nature Clim Change, 4(6), 434–​441. Xiang F, Harrison S, Nowak M et al. 2015. Weekend personal ultraviolet radiation exposure in four cities in Australia: influence of temperature, humidity and ambient ultraviolet radiation. J Photochem Photobiol B, 143, 74-​81. PMID: 25600266. Yang L, Veierod MB, Lof M et al. 2011. Prospective study of UV exposure and cancer incidence among Swedish women. Cancer Epidemiol Biomarkers Prev, 20(7), 1358–​1367. PMID: 21551241. Zaratti F, Piacentini RD, Guillen HA et al. 2014. Proposal for a modification of the UVI risk scale. Photochem Photobiol Sci, 13(7), 980–​985. PMID: 24728468.

 259

15 Electromagnetic Fields MARIA FEYCHTING AND JOACHIM SCHÜZ

OVERVIEW This chapter focuses on the carcinogenic effects of exposure to electromagnetic fields, mainly power frequency and radiofrequency fields/​ microwaves. It discusses current knowledge on interaction mechanisms of relevance for carcinogenicity, describes exposure sources, and reviews the available epidemiological studies. The chapter comments on the strengths and weaknesses of the data available, and discusses sources of bias of concern for the interpretation of the findings. Based on current evidence, electromagnetic fields are not regarded as carcinogenic to humans, but there are open scientific questions that merit further research, as will be outlined.

INTRODUCTION Electromagnetic fields in the non-​ionizing radiation spectrum are an integrated part of modern life, with an ever-​increasing number of applications. The properties of electromagnetic fields and their interaction with the human body are determined by their frequency and wavelength, with the wavelength inversely related to the frequency. Electromagnetic fields generate energy, which is directly proportional to the frequency. In the non-​ionizing part of the electromagnetic spectrum (Figure 15–1), the generated energy is too weak to break chemical bonds, and ions cannot be formed. The low frequency end of the spectrum includes static fields (0 Hz) and fields related to the production, transmission, and use of electricity (usually at 50–​60 Hz). At higher frequencies within the non-​ionizing spectrum are radiofrequency fields (RF) and microwaves, including applications such as radio and television, mobile telephony, radar, and microwave heating. Further up the frequency range are infrared light, visible light, and UV radiation. This chapter focuses on the frequencies from above 0 Hz to 300 GHz, mainly power frequency and RF fields/​microwaves, as this has been the focus of the research on potential carcinogenic effects of electromagnetic fields during the past decades. Extremely low frequency fields (ELF) have a very long wavelength (6000 km at 50 Hz), and electric and magnetic fields are separated. The electric field strength is measured in volts per meter (V/​m), while magnetic fields are usually measured as the magnetic flux density in tesla (T), and in the human environment usually in microtesla (µT). At these low frequencies the energy is too low to cause heating, and the magnetic fields pass through the body. The established mechanism of interaction between ELF fields and the human body is the induction of electric currents. Induced electric currents may cause acute neurological effects, and current exposure guidelines are set to protect against these acute effects. The research on the potential carcinogenicity of ELF magnetic fields has focused on exposures that are orders of magnitude below the current guidelines (ICNIRP, 2010; IEEE, 2002). Radiofrequency fields associated with mobile phone technology have wavelengths of a few centimeters, depending on the frequency used for communication. They are characterized by their power density (i.e., the electromagnetic energy per unit area), which is measured as watts per meter squared (W/​m2). Except for near-​field exposure (i.e., very close to the source), there is a direct relation between the power density, the electric field (V/​m), and the magnetic field (A/​m), and measurement of either the electric or magnetic field is sufficient to estimate the other quantities (AGNIR, 2012). At these frequencies,

energy is deposited and can cause heating (cf. microwaves). Heating of tissue is the established mechanism of interaction with the human body at these frequencies, and current exposure guidelines protect against excessive heating of tissue, at both localized and whole body exposures (ICNIRP, 1998; IEEE, 2005). The energy absorption inside the body is estimated as the specific absorption rate (SAR), expressed in watts per kilogram (W/​kg). It is, however, not possible to directly measure tissue heating inside the body; therefore SAR values are estimated through measurements in phantoms and theoretical modeling (AGNIR, 2012; IARC, 2013). Use of a mobile phone is an example of localized near-​field exposure, and energy is absorbed to a depth of a few centimeters within the head.

Potential Mechanisms of Carcinogenicity Despite considerable research efforts, there is currently no known biological mechanism for a carcinogenic effect of electromagnetic fields at the exposure levels encountered in the general population or in occupational settings. In vitro studies of ELF fields indicate genotoxic effects at exposure levels of 100 µT or higher, which is far above what is encountered in the human environment for longer than short durations, while in vivo studies do not provide evidence that ELF fields cause tumors or enhance tumor growth (SCENIHR, 2015). It is important to note, however, that an appropriate animal model for childhood leukemia has not previously been available, and the recent development of such a model (Hauer et al., 2014; Martin-​Lorenzo et al., 2015) has not yet provided new evidence. For RF fields, studies in cells do not consistently show effects on genotoxicity, cell transformation, or cell proliferation, and occasional positive findings have not been replicated in independent studies. Similarly, studies in experimental animals have not shown compelling evidence of a carcinogenic effect, including several long-​term rodent bioassays, studies in transgenic animals, and studies of co-​carcinogenic effects (AGNIR, 2012; SCENIHR, 2015). Independent studies attempting to replicate a few positive findings for RF field exposure, such as an increased lymphoma incidence in the transgenic Eμ-​Pim1 mouse model (Repacholi et al., 1997), have not been able to confirm the original findings. One exception is a recent replication of an earlier pilot study of a co-​carcinogenic effect of RF fields in mice exposed in utero to ethylnitrosourea (ENU) (Tillmann et al., 2010). The replication study found an increased occurrence of tumors in mice exposed in utero to ENU, and to one of four RF exposure levels (sham, 0.04, 0.4, or 2 W/​kg), but no evidence of dose–​response pattern (Lerchl et al., 2015). Tumor types with increased occurrence in the replication study were lung, liver, and lymphoma, while the original study found an increase in lung tumors. Discovery of a Helicobacter hepaticus infection among the mice made results for liver tumors uninterpretable (Tillmann et al., 2010). None of the studies found increases in brain tumors. Replication of these findings is warranted. The largest and most comprehensive animal study within this area is in the process of being published, and was conducted by US National Toxicology Program (National Toxicology Program, 2014). Selected results have just been made available as a partial report, reporting increased occurrence of two tumor types in male rats (mainly schwannomas of the heart and, somewhat weaker, gliomas in the brain), at exposures of 1.5, 3, and 6 W/​kg compared to non-​exposed sham controls. The partial report indicates that no positive findings were seen

259

260

260

PART III:  THE CAUSES OF CANCER Power lines and electric appliances

AM radio

Visible light

TV

Radar

FM radio

Infrared

UV-light

Gamma rays X-rays

103

10

105

107

109

1011

1013

1015

1017

1019

1021 Hz

0

100

104

Low frequency

Figure 15–1.  The electromagnetic spectrum.

106

108

Radio waves

1010

1012

EXTREMELY LOW FREQUENCY MAGNETIC AND ELECTRIC FIELDS First, notably, it must be stressed that there is no “zero” exposure to ELF magnetic fields; due to the electrification of the modern world, everyone is exposed to ELF fields to various extents, with typical daily average background levels of less than 0.05 µT but possibly slightly higher in urban areas. Sources of exposures to ELF magnetic fields above background levels can be roughly subdivided into three types. Exposures exceeding 0.2 µT are related to either the transmission or distribution of power (fields from high-​voltage power lines, lower-​ voltage distribution lines, underground cables, transformers, indoor wiring) (type 1:  residential exposures), the use of electrical devices (such as hair dryer, vacuum cleaner) (type 2:  exposure from use of electrical appliances), or working in electrical occupations (such as electricians, welders) (type 3:  occupational exposures). Magnetic fields are produced by the flow of the current; thus exposure is determined by the current and distance to the source. Electric devices may emit magnetic fields of several mT in their immediate vicinity, but only during use; therefore they are sources of high but intermittent very short-​term exposure. In certain occupations, when working close to electric motors, for example, average exposures of 1–​10 µT are possible, and exposure may span over the whole working day. Residential exposures are usually at background level, but if elevated, then they result in rather continuous exposure. Magnetic fields of > 0.2 µT to a few µT may be reached within distances of less than 50–​200 m to nearby high-​voltage power lines (> 200 kV, depending on their power load), distances of less than 20–​50 m to medium-​voltage power lines or other installations (> 10 kV), less than 5–​10 m to low-​voltage power lines, their roof connections to buildings, or to transformers (< 1 kV), or in homes with unbalanced currents in indoor wiring or by current produced by ELF fields on closed metallic loops. Even taken together, however, these exposure situations are rather uncommon. It is estimated that only a few percent (1%–​3%) of residences in Europe and potentially slightly more (1%–​10%) in North America have average

1016

1018

1020

1022

Micro waves

Non-ionizing radiation Cellular Telephones (900–2100 Mhz)

for female rats, or male and female mice, or for any other tumor type in any kind of animal. In addition, survival was significantly lower in the sham exposed rats, and tumor occurrence was lower than observed historically in sham exposed rats. While a full assessment of the implications from this major study has to await the publication of the complete set of results, the already shared findings may renew interest in looking into mechanistic pathways. Given the inconsistency across animal sexes and species, the high exposure levels and duration compared to common exposures in humans, the main finding in a rare cancer type, and the lower survival in sham exposed controls, limits the applicability of the findings to what this means for real life exposures in humans. The available epidemiological evidence on potential cancer risks associated with exposure to ELF and RF fields, respectively, is summarized in the two following sections.

1014

Ionizing radiation

residential exposures exceeding 0.2 µT (Maslanyj et al., 2009), and the exposure prevalence of averages > 0.5 µT is below 1%. ELF electric fields are produced wherever there is electric potential, and they are directly related to the voltage, irrespective of whether any current flows. Hence, high exposures are experienced mainly in occupational settings or in the vicinity of high-​voltage installations, such as overhead power lines or substations. In contrast to magnetic fields, which are not perturbed by most materials (and especially not by tissue and therefore penetrating the body), electric fields are shielded by most materials. This is why it is extremely difficult to measure or predict electric field exposures. Distance to the source alone is not a good surrogate measure, as other buildings or trees may be between the source and the home and may distort the electric field. Normally, electric field levels in homes are a few V/​m, but can become as high as several kV/​m in homes in close vicinity and with an unobstructed path to high-​voltage power lines and in certain occupational situations. Most epidemiological studies have investigated magnetic field exposures, and here either residential or occupational exposures. After approximately 40 years of epidemiological research on cancer effects of ELF electric and magnetic field exposures, most notable evidence is an association between residential ELF magnetic fields and a modestly increased risk of leukemia in children (about 1.5-​to 2-​fold), observed in a large number of studies with some degree of consistency. For other exposure and cancer outcome combinations, there are either fewer studies, rather exploratory studies, or the studies are inconsistent or mostly showing no association. Therefore the remainder of this section provides a detailed review of the childhood leukemia evidence and only brief summaries of the other evidence.

Exposure Assessment Exposure assessment of ELF magnetic fields is challenging. Various methods have been used to assess residential exposures. Residential exposures are used as surrogates of the individual’s exposure, assuming that total exposure is mostly determined by exposure at home. This has shown to be reliable in especially young children given how much time they spend at home, but may lead to some exposure misclassification in adults due to exposures and time spent at the workplace and elsewhere (Forssen et  al., 2002). Personal monitors, carried by the individual, which constantly record magnetic field levels and provide the most accurate estimate of contemporary exposure, do exist. However, studies of residential exposures are almost exclusively case-​ control studies requiring an accurate estimate of past exposures. This is better achieved by residential measurements, as exposure tends to be more stable over time in cases of elevated field levels, while personal monitoring is more influenced by personal behavior. A person’s behavior is likely to change over time, especially when having a cancer diagnosis, but also by the aging of the subject, so that contemporary exposure is not a good predictor of past exposures. Exposure assessment methods and their features are summarized in Table 15–1.

 261

261

Electromagnetic Fields

Table 15–1.  Main Exposure Assessment Methods Used in Epidemiological Studies of Residential Exposures to Extremely Low-​Frequency Magnetic Fields Method Personal dosimetry

Long-​term stationary measurement

Spot measurements

Calculated fields

Wire code

Distance to power line

Description

Advantages/​Disadvantages

Device carried by study participant over 1 or several Most precise picture of magnetic field exposure of the individual, especially days, recording the magnetic field level several times when combined with activity diary; rarely used in cancer epidemiology, as per minute too expensive for large cohorts and limited use in case-​control studies, as activity patterns change when having cancer and by the aging of subjects (especially in children) so that contemporary exposure is not a good predictor of past exposures. Device placed in study participant’s home over 1 or Most precise picture of magnetic field exposure in the home environment; several days, recording the magnetic field level surveys show good correlation in repeated measurements so that method several times per minute; either used stationary in appears to be a good surrogate of exposure in retrospective studies; good the bedroom or at several places in the home for exposure surrogate in children spending most of their time at home; in calculation of time-​weighted average according adults, time spent at other places or occupational exposures may influence to how much time the participant spends at each the individual exposure more than the home measurement. measurement location As above, but measurements taken only for usually May be as accurate as longer term measurements with stable magnetic several minutes up to a few hours, at several locations field conditions, but much lesser so when fields vary strongly over time; within the home but sometimes only outside the especially outdoor measurements have shown rather poor correlation in residence repeated measurements. Estimation of magnetic field levels from overhead Accurate retrospective estimation of long-​term magnetic field exposure in high-​voltage power lines based on distance from line the home environment caused by high-​voltage power lines; neglects other to home, historic power load of the line, and other line sources of magnetic field exposures in the home environment and when characteristics not at home; can only be estimated when the respective historic power load data are available, which is often not the case (used almost exclusively in the Nordic countries and the UK). Estimation of magnetic field levels from overhead power Can be used for retrospective assessment; surveys show some variability on lines and other electrical installations similar to the how well the method predicts actual magnetic field exposures in the home preceding method, but rather based on characteristics environment depending on the local conditions, which makes it difficult to of the source and no data on power load judge how well the method worked in the individual studies; neglects other sources of magnetic field exposures in the home environment and when not at home; mostly used in North America. Measuring or estimating the distance between the home Weakest surrogate of magnetic field exposure; works well in very close and nearby overhead high-​voltage power lines or other proximity (50 m) to the power lines in countries where power lines are electrical installations usually run with high power load, but leads to gross misclassification of exposure in the home environment otherwise; may also be surrogate for other living conditions and therefore raises issue of confounding.

Exposure assessment methods for occupational settings range from categorizations of job titles, to expert ratings of job-​related tasks (Hug et al., 2010), to the development of more complex quantitative job-​exposure matrices (JEMs) based on measurement surveys (Bowman et al., 2007). While the latter is believed to be the most reliable approach, exposure misclassification remains a problem, due to the large exposure variation within jobs and the usually relatively small number of available measurements per job category (Gobba et al., 2011; Greenland et al., 2016). Exposure from electrical appliances is usually assessed through interviews, with self-​reported type and frequency of use of devices, and distance to the source (e.g., for television), and exposure estimates are therefore prone to both random and systematic error. Electric field exposures are even more difficult to estimate. Similar approaches to those used with magnetic fields were used in occupational studies, especially using job categorizations by experts. Measurements of residential exposures as reported in a few studies (e.g., McBride et al., 1999) must be interpreted with care, as it is questionable how well they predict past exposures. In the example of Canadian children, most time-​weighted average exposures to electric fields were below 20 V/​m (McBride et al., 1999), although in some instances they exceeded 50 V/​m.

Childhood Cancer Childhood Leukemia

ELF magnetic fields have been studied as a potential risk factor for childhood leukemia since the late 1970s, when Wertheimer and Leeper published a case-​control study from Denver, Colorado, observing a three-​fold increased risk for children living in houses assigned to the highest exposure category using the wire code exposure assessment method, a crude categorization of exposure based on proximity of the residence to certain electrical installations, such as power lines of

different voltages (Wertheimer and Leeper, 1979). At present, more than 30 epidemiological studies have investigated this topic. Most of these were conducted in the late 1980s to early 2000s, and used variable sample sizes and design features. As the studies differ greatly in exposure assessment, exposure indices, and exposure categorizations, as well as analytical strategies, direct comparison is difficult and may even be misleading. Therefore, the most informative evidence comes from systematic reviews or meta-​ analyses. Authoritative hazard identifications or health risk assessments were prepared by the International Agency for Research on Cancer (IARC) program on the evaluation of carcinogenic risks to humans, the World Health Organization (WHO) Environmental Health Criteria, and the evaluation of the European Commission’s Scientific Committee on Emerging and Newly Identified Health Risks (SCENIHR). In 2001, the IARC classified ELF magnetic fields as possibly carcinogenic to humans, based on limited evidence from human (epidemiological) studies and inadequate evidence from studies in experimental animals, including studies published before 2001 (IARC, 2002). WHO confirmed the IARC assessment, stating that evidence from the more recent studies does not alter the classification as possible carcinogen (WHO, 2007). SCENIHR also confirmed that the evidence of possible carcinogenicity remains unchanged, most recently in their update in 2015 (SCENIHR, 2015). What was considered the strongest evidence in all risk assessments stems from the pooled analyses of the original studies, which allowed to some extent the harmonization of the different study features and was therefore more informative than a comparison of the individual studies. Greenland et al. pooled most of the available studies at the time (Greenland et al., 2000), irrespective of exposure assessment method, which is why they had to combine various exposure indices into a single metric. The combined relative risk estimate was 1.7 (95% confidence interval [CI] 1.2–​2.3) in children exposed

26

262

PART III:  THE CAUSES OF CANCER

Table 15–2.  Pooled Analyses of Epidemiological Studies on Residential Exposure to Extremely Low Frequency Magnetic Fields and the Risk of Childhood Leukemia Ahlbom et al., 2000 Study base

Inclusion criteria of studies Exposure assessment Exposure indices Numbers of subjects Main result

Further results

Kheifets et al., 2010a

Greenland et al., 2000

Schüz et al., 2007

9 studies (Canada, Denmark, Finland, Germany, New Zealand, Norway, Sweden, UK, US) Population-​based case-​control studies; exposure either assessed through long-​term stationary measurements or calculated fields Long-​term stationary measurements or calculated fields Average magnetic field level, in µT

6 studies (Brazil, Germany, Japan, Italy (2), UK); not included in Ahlbom et al./​Greenland et al. except partly the German study As Ahlbom et al., 2000

12 studies; 8 as in Ahlbom et al. (except UK), 4 additional ones (2nd from Sweden, other UK, 2nd and 3rd from US) Inclusive; no restriction but delivery of original data

4 studies (Canada, Germany, UK, US); all included in Ahlbom et al. or Kheifets et al. (Germany) Exposure during the night had to be extractable from the exposure measurement

As Ahlbom et al., 2000

Measurements (personal, long-​term Long-​term stationary or spot), calculated fields measurement

As Ahlbom et al., 2000

Average exposure during nighttime (10 pm–​6 am), in µT

3247 cases, 10,400 controls

10,818 cases, 12,806 controls

Average magnetic field level, in µT, albeit from different measurement methods 2656 cases, 7084 controls

Reference 0.1–​≤ 0.2: OR 1.07, CI 0.81–​1.41 >0.2–​≤ 0.3: OR 1.16, CI 0.69–​1.93 > 0.3: OR 1.44, CI 0.88–​2.36 Using ≥ 0.4 µT as highest category gives OR of 1.46 (CI 0.80–​2.68); large overall numbers come from UK study (9665 cases, 9677 controls), but only few are exposed > 0.3 µT (2 cases, 2 controls)

Reference ≤ 0.1 µT > 0.1–​≤ 0.2: OR 1.01, CI 0.84–​1.21 >0.2–​≤ 0.3: OR 1.06, CI 0.78–​1.44 > 0.3: OR 1.68, CI 1.23–​2.31 Included pooling of 8 studies having wire code data (all from Canada, Mexico, or US), with an OR of 1.65 (CI 1.15–​2.35) for highest exposure category, but some heterogeneity across studies

Reference 0.3 µT compared with those exposed ≤ 0.1 µT. Ahlbom et al. pooled studies that fulfilled certain criteria, such as a defined population base for case ascertainment and control recruitment, as well as having used longer term measurements or calculated fields for exposure (Ahlbom et al., 2000); the main finding was a relative risk estimate of 2.0 (CI 1.3–​3.1) for exposures ≥ 0.4 µT compared with exposures < 0.1 µT. While the studies demonstrated a high degree of consistency, a major contribution to the overall excess risk came from a large US study (Linet et al., 1997), and no association was seen in a large study in the United Kingdom (UKCCS, 1999). Despite the large number of studies, the numbers of exposed children were too small to reliably predict the shape of a dose–​response function. Hence, trends are compatible with monotonic increases, thresholds at 0.3/​0.4 µT, plateauing at higher magnetic field levels, as well as no significant increase at all. Kheifets and colleagues (2010a) used the approach of Ahlbom et al. (2000) to update the evidence, pooling only the more recent studies, and overall confirming the modest association (though somewhat weaker than in the earlier studies). In another pooled analysis of four studies, the focus was exposure during the night, driven by hypotheses that there may be less exposure misclassification as the measurement better captures the time when the child is really at home or that nighttime exposure may be of more biological relevance. The result was that using nighttime exposure did not yield relative risk estimates different from the results when using whole-​day exposure (Schüz et  al., 2007). Summaries of these four influential pooled analyses are provided in Table 15–2 and Figure 15–2. In summary, based on a large number of studies, a modest association (1.5-​to 2-​fold) between residential ELF magnetic fields and childhood leukemia was observed, with—​especially when accounting for differences in study design features—​a quite high degree of consistency across studies carried out in different parts of the world, at different times, and by different investigators. On the other

hand, despite the large number of studies, the number of exposed children remained small; for instance, combining the pooled data sets utilized by Greenland et al. (2000) and Kheifets et al. (2010a), only 125 of 13,474 leukemia cases (0.9%) had exposures > 0.3 µT. While the consistency of studies and indication of a dose–​response trend may favor a causal interpretation, chance and bias cannot be ruled out with reasonable confidence as alternative explanations, so that together with the lack of supportive experimental evidence and plausible mechanistic hypotheses, the overall evidence was classified as possibly carcinogenic (IARC, 2002; SCENIHR, 2015). With respect to confounding, no candidate exposure has been identified yet (Schüz, 2007); traffic density, use of pesticides in the vicinity of power lines, and environmental tobacco smoke are among those that were considered. Because of the low prevalence of elevated magnetic field exposure, the confounding factor would have to be a strong risk factor itself and, at the same time, be strongly correlated with magnetic field strengths from all sources. Hence, while confounding by an unknown factor is always a possibility, it seems to be an unlikely explanation. With regard to exposure assessment methods, there is potential for exposure misclassification. However, as it is unlikely that the magnitude of misclassification from most of the exposure assessment methods was associated with disease status, the expected bias would rather be in the direction of diluting an association (Schüz, 2007). With regard to selection bias, the low participation rates in many studies were of concern, especially in those studies using measurements. Data suggested that families with a lower socioeconomic status were particularly under-​represented among controls, resulting in an underestimation of exposure prevalence in controls (Schüz, 2007). The studies using calculated magnetic fields were not affected by selection bias (Feychting et al., 1995; Tynes and Haldorsen, 1997; Verkasalo et al., 1993), as no contact with the families was necessary. While they also show an association, this is based on much smaller numbers of exposed

 263

Electromagnetic Fields Childhood Leukemia studies after 2000 (Kheifets et al., 2010a)

Childhood Leukemia studies before 2000 (Ahlbom et al., 2000)

263

Childhood Brain Tumors All studies (Kheifets et al., 2010b)

10

10

10

1

1

1

0.1

0.1

0.1

By exposure categories of < 0.1 µT, 0.1–0.2 µT, 0.2–0.4 µT, and ≥ 0.4 µT

children (Ahlbom et  al., 2000; Kheifets et  al., 2010a). Taking all these considerations into account, selection bias alone is perhaps not sufficient to explain the entire association, but a combination of selection bias, confounding, and chance cannot be ruled out as an alternative explanation for the observed association. Recently, it was suggested that if ELF magnetic fields would increase the risk of developing leukemia, they may also increase the risk of a relapse after initial treatment, leading to a lower survival in exposed leukemia patients (Foliart et al., 2006). Pooling six studies totaling 3073 children with acute lymphoblastic leukemia, however, did not show any association between ELF magnetic field exposure and overall or event-​free survival from leukemia (Schüz et al., 2012). Some studies also investigated paternal exposure to ELF magnetic fields and the risk of childhood leukemia in their offspring, hypothesizing a genetic alteration of the sperm. Meta-​analysis of those studies showed a combined relative risk estimate of 1.35 (CI 0.95–​1.91), but at the same time there was some evidence of publication bias (Hug et al., 2010). Less data are available for maternal exposures, mainly because occupational exposure is much less common than in fathers.

Other Childhood Cancers

There are much fewer studies on other childhood cancers, including lymphoma, perhaps because those cancers are less frequent in children than leukemia. The only exception is tumors of the central nervous system (CNS)/​brain. Thirteen studies were combined in a literature-​based meta-​analysis, separately analyzing studies based on distance, wire codes, calculated fields, and measurements. For the latter, four studies were available and showed a combined relative risk estimate of 1.14 (CI 0.65–​2.00) for exposures ≥ 0.2 µT compared to < 0.2 µT (Mezei et al., 2008). A recent pooling study having access to original data of 10 studies (in Denmark, Finland, Germany, Japan, Norway, Sweden, UK [2]‌, US [2]) showed relative risk estimates of 0.95 (CI 0.65–​1.41), 0.70 (0.40–​1.22), and 1.14 (CI 0.61–​2.13) for ELF magnetic field exposures of 0.1–​< 0.2 µT, 0.2–​< 0.4 µT, and ≥ 0.4 µT, compared to < 0.1 µT (Kheifets et al., 2010b; Figure 15–2). Taken as a whole, those results were interpreted as providing little evidence for an association between residential ELF magnetic field exposure and childhood brain tumors, suggesting that the observed association with an increased risk may be specific to childhood leukemia.

Cancer in Adults Studies on EMF exposures and cancer risk in adults included residential and more often occupational studies. Especially the latter often did not

Figure  15–2. Pooled analyses of epidemiological studies on residential extremely low frequency magnetic fields and the risk of childhood leukemia (studies published before and after 2000) and the risk of brain tumors; pooled relative risk estimates for three exposure categories compared to the reference category of average magnetic fields < 0.1 µT.

distinguish between electric and magnetic field exposures. Most studies investigated the risks of leukemia, brain tumors, or breast cancer. In 2008, Kheifets et al. published a meta-​analysis of studies of occupational EMF, including 47 studies of brain cancer and 56 of leukemia (Kheifets et al., 2008). For brain cancer, the pooled relative risk estimate was 1.14 (CI 1.07–​1.22), and for all leukemia it was 1.16 (CI 1.11–​1.22). There was little variation by diagnostic subtypes, and the associations seen in the more recent studies were slightly weaker than in those from the more distant past. Although similarly small elevations in risk were seen for both outcomes, the apparent lack of a clear pattern of exposure and risk detracts from a causal interpretation. Studies were of varying quality, especially when it comes to exposure assessment; however, restricting the meta-​analysis to only studies that were considered of better design quality by the authors of the meta-​analyses did not show substantially different results. In a recent multicenter case-​control study in seven countries (InterOcc study) involving 1939 cases of glioma, 1822 cases of meningioma, and 5404 controls, in which a JEM based on workplace EMF surveys was applied (Bowman et al., 2007), no association was seen overall between the ELF magnetic field estimate and brain tumor risk (for cumulative exposure in µT-​years in the highest exposure category, the odds ratios were 0.80 [CI 0.63–​1.00] for glioma and 0.89 [CI 0.70–​ 1.12] for meningioma [Turner et  al.,  2014]). In analyses looking at different time windows of exposure, there was an increased risk of glioma with a dose–​response trend for exposures 1–​4  years before diagnosis, while there was no association in both the 5–​9 years before diagnosis or 10+ years before diagnosis time windows; this may suggest a role of EMF in tumor promotion, but too little data are available yet to draw firm conclusions. Residential exposures were also investigated in some studies, but mainly based on distance to nearby high-​voltage power lines, which for adults may introduce substantial exposure misclassification. In a large UK record-​linkage-​based case-​control study using distance to power lines as well as calculated fields as exposure metrics, cases of leukemia or of brain cancer did not have increased risks compared to cancers not assumed to be related to EMF exposures (used as controls) (Elliott et al., 2013). With the potential to calculate historic exposures from power load of power lines (see Table 15–1), studies from the Nordic countries are particularly informative. No increased risks of either brain tumors or leukemias have been observed in Finland (Verkasalo et al., 1996). In Norway, leukemias showed little evidence of an association (Tynes and Haldorsen, 2003), and for brain tumors the overall risk was 1.6 (CI 0.9–​2.7), but there was an inverse association with occupational exposures (Klaeboe et al., 2005). In Sweden, risks were slightly elevated for two subtypes of leukemia but not others and not for brain tumors (Feychting and Ahlbom, 1994).

264

264

PART III:  THE CAUSES OF CANCER

It has also been hypothesized that exposure to ELF magnetic fields may increase the risk of female breast cancer, assuming that the ELF magnetic fields would affect nocturnal melatonin production (as light at night does) and that melatonin is protective against breast cancer (Stevens, 1987), mainly by acting as potent antioxidant (Reiter, 1995). Several studies addressed the question, looking at occupational exposures, residential exposures, and use of electric bed heating as exposure sources, reviewed in 2006 (Feychting and Forssen, 2006). While some early smaller studies were inconsistent but showed some positive associations, the more recent studies reported were mostly negative. In a meta-​analysis of almost 25,000 cases and > 60,000 controls from 15 studies published between 2000 and 2009, there was no evidence of an association with a combined relative risk estimate of 0.99 (CI 0.90–​ 1.09) (Chen et al., 2010). Combining results from 23 studies published from 1990 to 2010, however, showed an overall combined relative risk estimate of 1.07 (CI 1.02–​1.13), with the association restricted to pre-​ menopausal breast cancer (Chen et al., 2013). New studies continue to appear, such as another well-​conducted occupational study in textile workers in China (Li et al., 2013) showing no increased risk, suggesting the accumulation of evidence against an association. Several of the studies were addressing various cancer types as outcome, but overall much fewer studies are available for other cancers in adults than the ones mentioned earlier. For none of the cancers did any consistent picture emerge (SCENIHR, 2015). Among the most recent cohort studies, no increased risk was seen for leukemia, brain cancer, and breast cancer in Danish electric utility workers (Johansen et al., 2007). In the prospective Netherlands cohort study, among pre-​ selected cancers (namely lung, breast, and brain cancer), as well as hematological malignancies, no increases were seen for lung, breast, and brain cancer, but among hematological malignancies there was an increased risk of acute myeloid leukemia with ever higher exposures and an increased risk of follicular lymphoma with both ever and cumulative exposures (Koeman et al., 2014).

Future Research Needs Whether the observed association between residential ELF magnetic fields and childhood leukemia is causal remains an open question that merits further research. Given the consistency of results from epidemiological studies, combined with their difficulty of further reducing the potential of bias and confounding to a level that would address the concerns of having created a spurious association, it is unlikely that further case-​control studies of similar design would provide any new insights. Promising advancements were recently made in experimental settings with the development of a mouse model that better mimics leukemia development in children (Hauer et  al., 2014; Martin-​Lorenzo et al., 2015); application of this model is underway. If the results are positive, this may provide insights into mechanisms, perhaps suggesting the existence of particularly susceptible children, which could in turn be addressed by epidemiology. If the results are negative, it adds to the evidence that the epidemiological studies may show a non-​causal association and will be a general warning of how cautious the interpretation of epidemiological studies investigating the combination of rare exposures and rare diseases must be. If the association was causal, due to the rarity of relevant exposure levels, the preventive potential is limited; only a small proportion of childhood leukemias would be attributable to ELF magnetic field exposure, recently estimated to be < 1%–​2% in European countries (Grellier et al., 2014) and possibly slightly higher in North America (Greenland and Kheifets, 2006). For adults, any future studies should be based on better exposure measures, alongside testable plausible mechanistic hypotheses. Further studies linking job titles or other EMF surrogates with cancer outcomes will not provide additional insight. “Better” exposure measures need to be validated with respective measurements and comprehensive analyses of their direction and magnitude of potential exposure misclassification. “Testable” hypotheses should be described by plausible or novel scenarios. Candidates not sufficiently addressed are co-​exposures of EMF with known carcinogens (such as chemicals or ionizing radiation). Like for childhood leukemia, if there were

consistent associations observed in experimental animals, future epidemiological studies may be better guided by addressing a particular mechanistic scenario.

RADIOFREQUENCY FIELDS Exposure Sources Currently, the most prevalent sources of radiofrequency (RF) field exposure in the general population are related to wireless communication, and can be grouped into personal (near-​field) and environmental (far-​field) exposures. The so-​called bag-​phones and car phones that were introduced in the early 1980s had the antenna far from the human body, giving rise to far-​field exposure; handheld mobile phones with the antenna in the handset became available in the late 1980s, resulting in near-​field localized exposure to the head. The RF exposure decreases rapidly with distance to the exposure source; thus use of a wired hands-​free device reduces the RF exposure to the head by approximately 90%. The first handheld mobile phones used the analogue system, which had maximum average radiated powers typically at 0.6 W (IARC, 2013). In the early 1990s, digital phones were introduced, which have lower radiated power (maximum average at 0.1–​0.25 W). Development of the technologies for more efficient utilization of the cellular networks has resulted in even lower exposure levels during actual use through the introduction of adaptive power control (APC), by which mobile phones reduce their output power to the level sufficient to maintain a good quality contact with the base station, and discontinuous transmission (DTX), which means that the mobile phone transmits only when the user is talking. Through this, average RF field levels from mobile phones have decreased considerably (AGNIR, 2012). The third generation of mobile phones has the same maximum average output level as the second generation, but the actual output power in real use is about 10 times lower than that of second-​generation mobile phones when in use. Another source of near-​field exposure is cordless phones used within homes, with DECT phones being the most successful technology (i.e., wireless handsets connected to the landline phone through a DECT base station in the home). As the DECT phone only transmits to a base station that is a few meters away, exposure is considerably lower, with average powers 10 or more times smaller than those from mobile phones at their highest power level. Sources of environmental, or far-​field, exposure include, for example, mobile phone base stations and radio and television transmitters, DECT phone base stations, and wireless local area networks (Wi-​Fi), which give rise to exposures that are only a fraction of the exposure guidelines and less than 1% of the exposure to the head from a handheld mobile phone (AGNIR, 2012). Similarly, devices such as baby monitors and smart meters used to measure electricity consumption also generate very low RF fields. These technologies have also changed over time; for example, analog television has been replaced by digital television broadcasting, which in most cases has resulted in decreased total output power (AGNIR, 2012). Occupational settings may be associated with higher exposures than are found in the general population. Workers in occupations involving the use of dielectric heaters or induction heaters (e.g., PVC welders) are among the highest exposed group. Other examples are operators of surgical and other medical diathermy devices or magnetic resonance imaging (MRI), workers near radar equipment, or those involved in the maintenance of transmitters such as base stations and radio and television towers (AGNIR, 2012). Applications with sometimes intentionally relatively high exposures to the general public are anti-​theft devices, such as those used in retail stores, but the time that humans stay in the close vicinity of those sources is usually very short.

Mobile Phone Use During the last 15 years, studies of RF exposure associated with wireless communication have focused on mobile phone use, which is the source that gives the highest localized RF exposure. Consequently,

 265

Electromagnetic Fields the main studied outcomes have been tumors in the head and neck region, as they are located in the areas of the body where most of the energy is absorbed during a mobile phone call (i.e. glioma, meningioma, and acoustic neuroma). Systematic reviews of the scientific evidence have continuously been conducted by international and national bodies such as the SCENIHR, IARC, and the Independent Advisory Group on Non-​ionizing Radiation (AGNIR) of Public Health England (AGNIR, 2012; IARC, 2013; SCENIHR, 2009, 2015). The WHO is currently working on an update of the Environmental Health Criteria document on RF fields. All reviews identify considerable methodological limitations and inconsistencies in the available epidemiological studies, which prevent firm conclusions. In 2011, the IARC classified RF fields as possibly carcinogenic to humans, based on limited evidence in humans from epidemiological case-​control studies of mobile phone use and limited evidence from animal studies (IARC, 2013). AGNIR, however, concluded in 2012 that the “accumulating evidence on cancer risks … is increasingly in the direction of no material effect of exposure” (AGNIR, 2012), also based on the epidemiological studies of mobile phone use, combined with information from incidence trend studies. Based on even more recent data, the SCENIHR review concludes that the evidence for an effect on glioma has become weaker since the IARC evaluation (SCENIHR, 2015). It was noted in all those assessments that conclusions could not be drawn about longer latencies than up to 15  years since first mobile phone use.

Methodological Limitations

Most of the available studies are case-​control studies with retrospective assessment of mobile phone use history through structured personal interviews or self-​administered questionnaires. This method has the advantage that detailed information can be collected, such as the preferred ear when using the phone and use of hands-​free devices, but it is prone to exposure misclassification, both non-​differential and differential, where the former would more likely lead to a dilution of an effect should there be a true association, and the latter would more likely lead to reports of overestimated or even spurious effects. The early studies of mobile phone use and cancer risk were conducted at a time period when use of handheld mobile phones was uncommon in the general population, and the technology had only been available for a few years. Thus, these studies could only inform about effects after short exposure duration, which are not expected for the type of outcomes studied. This review focuses on studies that include induction periods of at least 5  years for glioma and meningioma, and at least 10 years for acoustic neuroma. The changes over time of mobile phone technologies, the considerably reduced costs associated with buying and using the devices, and the development of new applications (e.g., the introduction of smartphones) have all led to a substantial increase in mobile phone use over time; at the same time, as illustrated earlier, the output power of the devices has decreased substantially with each new technology. This combination is a considerable challenge for the retrospective assessment of mobile phone use histories in case-​control studies. Furthermore, studies of brain tumors are faced with the additional complication that the tumor itself may affect memory function. Validation studies conducted as part of the Interphone study, a large international collaborative study including 16 study centers (Cardis et  al., 2007), have shown considerable exposure misclassification when healthy volunteers were asked to report their amount of mobile phone use 6 months earlier (Vrijheid et al., 2006). Moreover, it was found that cases tended to overestimate their mobile phone use to a greater extent the further back in time they were reporting about (Vrijheid et al., 2009a), which was not observed among controls (the study covered recall approximately 4 years back in time). A study of self-​reported year when first starting to use a mobile phone also found considerable misclassification (Pettersson et al., 2015), which was also found in responses to a seemingly simple question like preferred side of the head during mobile phone use (Kiyohara et al., 2015). The few cohort studies available have either collected exposure information from subscriber registers, or prospectively collected questionnaires. When collected prospectively, as in the UK Million Women study

265

(Benson et al., 2013), questionnaire information will not be affected by recall bias, but non-​differential exposure misclassification may be substantial because of the difficulty in reporting accurately about mobile phone use among healthy volunteers as well, as described previously. Information obtained from mobile phone operators, if unequivocally linked to a specific person, will reduce non-​differential exposure misclassification considerably. A  Danish cohort study identified mobile phone users through the subscription registers held by mobile phone operators during a period when mobile phone use was still uncommon in the general population (Frei et al., 2011; Johansen et al., 2001; McCarthy et al., 2011; Schüz et al., 2006b). With this approach, non-​ differential exposure misclassification could be limited, but no information was available about the amount of mobile phone use. Thus, the study would not be able to detect an increased risk in a small subgroup of heavy users. Currently, one cohort study with more elaborated exposure information from mobile phone operators is underway (Schüz et al., 2011a). Selection bias may also occur in case-​control studies caused by non-​ participation related to the exposure. Investigation of non-​ response in the Interphone study showed that mobile phone users among both cases and controls were more likely to participate than subjects who did not use a mobile phone (Vrijheid et al., 2009b). As non-​response was more common among controls than cases, selection bias was introduced, leading to a reduction of the risk estimates by approximately 10% for ever regular use of a mobile phone. The cohort studies have used population-​based cancer registers for case identification, and follow-​up could be made independently of study participants (Benson et  al., 2013; Frei et  al., 2011; Johansen et  al., 2001; Schüz et al., 2006b).

Mobile Phone Use Results

There are now many publications on mobile phone use and brain tumors (e.g., AGNIR, 2012; Ahlbom et  al., 2009; SCENIHR, 2015; Swerdlow et al., 2011). Several studies are published in multiple publications, and have also been included in pooled analyses. Most of the studies have focused on glioma and acoustic neuroma, and somewhat fewer on meningioma. In most studies, the main analyses were of exposure, defined as time since first mobile phone use. Table 15–3 includes the original publication of studies with longer induction periods, and displays results for glioma and acoustic neuroma. Amount of use in cumulative hours or cumulative number of calls were analyzed in several of the studies.

Time Since First Mobile Phone Use.  For glioma, most of the

epidemiological evidence does not indicate an increased risk associated with time since first mobile phone use, regardless of how long a time the phone had been used. None of the cohort studies with prospective information on mobile phone use found indications of an increased glioma risk, examining exposure durations of more than 13 years since first mobile phone subscription in the Danish study (Frei et al., 2011) and more than 10 years in the UK cohort (Benson et al., 2013, 2014). Exceptions to this pattern are two case-​control studies conducted in Sweden by Hardell and coworkers (Hardell et al., 2006, 2013), which reported moderately increased risks of malignant brain tumors already after a short period of mobile phone use, and substantially increased risk estimates after longer exposure durations. The Interphone study, on the other hand, reported risk estimates for glioma that were consistently below unity, which could, at least partly, be explained by selection bias caused by non-​participation (Interphone Study Group, 2010; Vrijheid et al., 2009b). For meningioma, one study reported a doubling of the risk after more than 10 years of use of an analogue mobile phone (Hardell et al., 2005), while most other studies reported risk estimates close to or below unity (e.g., Interphone Study Group, 2010; Lahkola et al., 2008). A recent French study reported a non-​significantly raised risk of meningioma after 10 years of mobile phone use (Coureau et al., 2014). For acoustic neuroma, the pattern is very similar to the glioma findings. The Hardell group reported considerably increased risks already after less than 5 years of mobile phone use (Hardell et al., 2002, 2005), while other studies found no association between acoustic neuroma

26

Table 15–3.  Epidemiological Studies of Mobile Phone Use and Risk of Malignant Brain Tumors and Acoustic Neuroma That Present Results for at Least 10 Years Since First Start of Mobile Phone Usea Reference Schüz et al., 2006

Frei et al., 2011

Schüz et al., 2011b

Benson et al., 2013, 2014

Hardell et al., 2002a, 2002b

Location, Time Period, Design

Source

Outcome

Denmark 1982–​2002 Cohort study

420,095 mobile phone Brain and nervous subscribers; the whole system Danish population

Denmark 1990–​2007 Cohort study

358,403 mobile phone subscribers; subset of previous study with information on socioeconomic status

Denmark 1998–​2006 Cohort study

UK 1999–​2005—​followed through 2011 Cohort study

Sweden 1997–​2000 Case-​control study

2,883,665 Danes included in the CANULI cohort

Glioma

Acoustic neuroma

791,710 women participating in the UK Million Women Study, who answered a base-​ line questionnaire 1999–​2005

Glioma 875 cases

Population-​based

Malignant brain tumors 588 cases

Acoustic neuroma 126 cases

Acoustic neuroma 159 cases

Hardell et al., 2005, 2006

Sweden 2000–​2003 Case-​control study

Population-​based

Malignant brain tumors 359 cases

Acoustic neuroma 84 cases

Exposure Time since first subscription 1–​6 years > 6 years Mobile phone use, analog > 1–​5 years > 5–​10 years > 10 years Mobile phone use, analog > 1–​5 years > 5–​10 years > 10 years

Relative Risk/​SIR (95% CI) 0.90 (0.67–​1.18) 1.03 (0.91–​1.17) 0.96 (0.84–​1.09) 0.66 (0.44–​0.95)

1.20 (0.96–​1.50) 1.05 (0.87–​1.26) 1.04 (0.85–​1.26) 1.06 (0.85–​1.34) 0.98 (0.70–​1.36)

0.87 (0.43–​1.75) 1.02 (0.60–​1.72) 1.04 (0.56–​1.95)

1.0 0.87 (0.52–​1.46) No exposed case, 1.6 expected 0.96 (0.75–​1.23) 0.86 (0.72–​1.02) 0.77 (0.62–​0.96) 0.94 (0.53–​1.66) 1.46 (0.94–​2.27) 1.17 (0.60–​2.27) 1.09 (0.68–​1.75) 1.17 (0.75–​1.81) 3.0 (1.0–​9.3) 3.8 (1.4–​10.2) 3.5 (0.7–​16.8) –​ 1.8 (0.9–​3.5) 3.5 (2.0–​6.4)

Mobile phone use, digital > 1–​5 years > 5–​10 years > 10 years

1.6 (1.1–​2.4) 2.2 (1.4–​3.4) 3.6 (1.7–​7.5)

Mobile phone use, analog > 1–​5 years > 5–​10 years > 10 years

9.9 (1.4–​69) 5.1 (1.9–​14) 2.6 (0.9–​8.0)

Mobile phone use, digital > 1–​5 years > 5–​10 years > 10 years

1.7 (0.9–​3.5) 2.7 (1.3–​5.7) 0.8 (0.1–​6.7)

 267

267

Electromagnetic Fields Table 15–3. Continued Reference Hardell et al., 2013

Location, Time Period, Design Sweden 2007–​2009 Case-​control study

Interphone Study Group, Interphone 2010, 2011 13 countries 2000–​2004 Case-​control study

Source Population-​based

Population-​based

Outcome Malignant brain tumors 593 cases

Glioma 2708 cases

Acoustic neuroma 1105 cases

Lahkola et al., 2007; Schoemaker et al., 2005 (Partly overlapping with the Interphone study)

Han et al., 2012

Coureau et al., 2014

Pettersson et al., 2014

a

Nordic-​UK Interphone centers, broader age range 2000–​2004 Case-​control study

Population-​based

Glioma 1496 cases

Acoustic neuroma 676 cases

Pittsburg, US 1997–​2007 Case-​control study France 2004–​2006 Case-​control study

Hospital based

Sweden 2002–​2007 Case-​control study

Population based

Population-​based

Acoustic neuroma 343 cases Glioma 190 cases

Acoustic neuroma 451 cases

Exposure Mobile phone use, analog > 1–​5 years > 5–​10 years > 10–​15 years > 15–​20 years > 20–​25 years > 25 years Mobile phone use, digital > 1–​5 years > 5–​10 years > 10–​15 years > 15–​20 years > 20–​25 years > 25 years Time since first regular use 1–​1.9 years 2–​4 years 5–​9 years ≥ 10 years Time since first regular use 1–​1.9 2–​4 years 5–​9 years ≥ 10 years Time since first regular use 1.5–​4 years 5–​9 years ≥ 10 years Time since first regular use 1.5–​4 years 5–​9 years ≥ 10 years Years of mobile phone use < 10 years ≥ 10 years Time since first regular use 1–​4 years 5–​9 years ≥ 10 years Time since first regular use 1–​4 years 5–​9 years 10–​12 years ≥ 13 years

Relative Risk/​SIR (95% CI) –​ 0.6 (0.1–​3.1) 1.4 (0.7–​3.0) 1.4 (0.7–​2.7) 2.1 (1.1–​4.0) 3.3 (1.6–​6.9) 1.8 (1.01–​3.4) 1.6 (0.97–​2.7) 1.3 (0.8–​2.2) 2.1 (1.2–​3.6) –​ –​ 0.62 (0.46–​0.81) 0.84 (0.70–​1.00) 0.81 (0.60–​0.97) 0.98 (0.76–​1.26) 0.73 (0.49–​1.09) 0.87 (0.69–​1.10) 0.90 (0.69–​1.16) 0.76 (0.52–​1.11) 0.77 (0.65–​0.92) 0.75 (0.62–​0.90) 0.95 (0.74–​1.23) 0.8 (0.7–​1.0) 0.9 (0.7–​1.2) 1.0 (0.7–​1.5) 0.79 (0.45–​1.37) 1.29 (0.69–​2.43) 1.04 (0.64–​1.69) 1.45 (0.91–​2.33) 1.45 (0.68–​3.08) 1.04 (0.72–​1.52) 1.40 (0.98–​2.00) 1.10 (0.68–​1.76) 1.12 (0.72–​1.73)

Studies with only short-​term mobile phone use are not included in the table (in total 7 studies, published in multiple papers).

risk and time since starting mobile phone use, not even after more than 13 years of mobile phone use (Pettersson et al., 2014). Most studies did not take cordless phone use into consideration in the exposure assessment, which could potentially cause non-​differential exposure misclassification. Hardell et al. report increased risks of brain tumors associated with cordless phone use in the same order of magnitude as for mobile phone use (Hardell et al., 2006, 2013). No associations were observed with cordless phone use in two national Interphone studies (Lonn et al., 2005; Schüz et al., 2006a). In addition, when Hardell and colleagues included cordless phone users in the unexposed category to mimic the analyses in the Interphone study, the results were virtually unchanged (Hardell et al., 2011). Thus, cordless phone use does not appear to explain the differences in results between the studies by Hardell and colleagues and other studies. The few reported risk increases appear implausible for several reasons. First, glioma incidence time trends have been stable since the introduction of mobile phone technologies, as shown in incidence

studies from many parts of the world (Barchana et al., 2012; de Vocht et  al., 2011; Deltour et  al., 2012; Dobes et  al., 2011; Kohler et  al., 2011; Little et al., 2012); this was particularly so for middle-​aged and younger persons, which are the more relevant age groups in relation to mobile phone use, while an incidence increase in the elderly, which was observed to start even before the introduction of mobile phone technologies, is believed to be due to improved and more readily available diagnostic imaging techniques (especially MRI). Simulation studies have demonstrated that risk increases after short duration of mobile phone use are incompatible with the observed incidence trends (Deltour et  al., 2012; Little et  al., 2012). More details are provided later in this chapter (section on “Incidence Studies”). Second, recall bias is indicated by the findings of the strongest effects among persons reporting to have started mobile phone use at least 25 years prior to diagnosis when handheld mobile phones had only been available in Sweden 23 years or less at the end of the data collection (Hardell et al., 2013).

268

268

PART III:  THE CAUSES OF CANCER

Also for acoustic neuroma, incidence time trends do not support an effect of mobile phone use (Benson et al., 2013; Larjavaara et al., 2011a), although fewer data are available. Acoustic neuroma is a slow-​ growing tumor, and one would not expect an effect on the tumor occurrence after a short latency period. It is likely that most of the acoustic neuroma tumors diagnosed within 5 years after first mobile phone use were already present when the person started to use a mobile phone (Thomsen and Tos, 1990).

Amount of  Mobile Phone Use. If high intensity of

mobile phone use is required for effects on tumor development, results based on time since first mobile phone use would be less informative, and estimates of amount of use would be needed. Available studies with analyses of amount of mobile phone use have used different exposure cutoff points, most often based on the exposure distribution among controls. In the Interphone study, cumulative hours of mobile phone use were categorized with cutoff points approximately at the deciles of hours of use among controls. As shown in Figure 15–3, no trends of increasing risk with increasing amount of use were observed for glioma, acoustic neuroma, or meningioma (Interphone Study Group, 2010; 2011). In the highest category of cumulative hours (> 1640 h), a slight risk increase was observed for all three outcomes, 1.40 for glioma, 1.15 for meningioma, and 1.32 for acoustic neuroma (although not statistically significant for the latter two). It is noteworthy that the risk estimates in the 9th decile (735–​1639 h) were among the lowest observed: 0.71 for glioma, 0.76 for meningioma, and 0.48 for acoustic neuroma. With a 5-​year latency period for acoustic neuroma, the risk estimate in the 10th decile was statistically significant, but the risk estimate in the 9th decile was still the lowest of all categories. For glioma and meningioma, the highest risk increase associated with heavy mobile phone use was observed within 1–​4 years after starting to use a mobile phone, for acoustic neuroma among long-​term users. No associations were observed with cumulative number of calls. In the first two studies by the Hardell group (Hardell et al., 1999, 2002), no consistent associations between cumulative hours of mobile phone use and brain tumor risk were found, but the last two studies observed considerable risk increases, also after having accumulated relatively few hours; for example, > 80 hours of analog phone use were associated with a four-​fold risk increase of malignant brain tumor, and > 64 hours of digital phone use with a two-​fold risk increase. Associations were also observed for meningioma and acoustic neuroma. Several other studies have analyzed higher amounts of use than these, for example two early studies from the United States (around 500 h as the highest cutoff point; Inskip et  al., 2010; Muscat et  al., 2000), and national Interphone studies (> 500 h; Lahkola et  al., 2007, 2008; Lonn et  al., 2005; Schoemaker et  al., 2005). None of these studies found increased risks at such low cumulative levels of use. In the most recent study by the Hardell group (Carlberg et al., 2013; Hardell et al., 2013), four categories of cumulative hours of phone use were analyzed, and risk estimates for malignant brain tumors were raised in all four categories, but in some categories of borderline statistical significance. In the highest exposure category (> 2376 h), an almost eight-​fold risk increase was observed for analog and a three-​fold risk increase for digital phone use. Also for meningioma, considerable risk increases were found in the highest category of analog and digital mobile phone use, respectively, while acoustic neuroma was not analyzed separately in this study. A recent French case-​control study also reported raised risk estimates at considerably fewer hours of mobile phone use than the Interphone study (Coureau et  al., 2014). The odds ratio for glioma after 339–​895 hours of phone use was 1.78 (95% CI 0.98–​3.24, and after at least 896 hours it was 2.89 (95% CI 1.41–​5.93). For meningioma, an increased risk was observed only in the highest exposure category, with an odds ratio of 2.57 (95% CI 1.02–​6.44). Only one cohort study with prospectively collected information on amount of mobile phone use is currently available (Benson et al., 2013, 2014). The highest exposure category was “daily use,” and the

risk estimates for daily users did not differ from the overall results of no associations. The very high risk increases at quite low amounts of mobile phone use reported in the studies by Hardell et al. are implausible in the light of the stable incidence time trends. The findings in the French case-​control study would also have resulted in increased incidence trends (Deltour et al., 2012). The results of the Interphone study are difficult to interpret, considering the lack of a dose–​response pattern. For acoustic neuroma, there were indications of a downward trend until the 10th percentile. The cases included in the Interphone study were diagnosed in the early 2000s, at which time approximately 10% of the controls had cumulated at least 1640 hours of mobile phone use. Now, about 15 years later, a considerably larger proportion of the population in many countries would have accumulated this amount of use and more, but so far, no corresponding increase in the brain tumor incidence has been observed. In addition, implausibly high cumulative hours of phone use were more frequent among cases than controls in the Interphone study (Interphone Study Group, 2010). As mentioned earlier, the results from the validation study provide empirical evidence that recall bias is likely to have affected the findings (Vrijheid et al., 2009a), but the question is whether it can explain the small risk increase in heavy users entirely. The cohort study with prospectively collected exposure information, unaffected by recall bias, did not support the hypothesis of an increased tumor risk with more frequent mobile phone use, but the highest exposure category was limited to “daily use” (Benson et al., 2013, 2014).

Laterality of  Phone Use and Tumor Localization.  RF

exposure to the head is highly localized during mobile phone use, and reaches only a few centimeters deep into the brain. Therefore, if the exposure is causally related to brain tumor risk, one would expect to find an increased risk on the same side of the head as the phone was usually held, and a risk close to unity on the opposite side. Self-​ reported laterality of mobile phone use is, however, highly uncertain (Kiyohara et  al., 2015). Information about laterality of phone use has so far only been collected retrospectively in case-​control studies, and there is a possibility that cases are affected by their knowledge about the side of the tumor location when reporting the preferred side used during mobile phone calls prior to diagnosis, and they might tend to more often report ipsilateral use (Schüz, 2009). Controls will of course report independently of the fictive “tumor” side assigned to them by the researchers, as they do not even know which side of their head is the one relevant in the analysis. Results in the case-​ control studies indicate that recall bias may indeed be a problem, as several studies found an increased risk of brain tumors on the same side of the head as the phone was reported to be held, and at the same time a reduced risk on the opposite side and/​or among subjects for whom laterality data were missing (Ahlbom et al., 2009; Schüz, 2009; Swerdlow et  al., 2011). Overall, there are strong indications that recall bias may have affected results where laterality was taken into consideration. Another way to take the localized exposure into consideration is to conduct separate analyses of tumors localized in different lobes of the brain. The temporal lobe absorbs the highest energy during mobile phone calls, but studies have often grouped the different areas of the brain differently, which hampers comparability. Most studies did not find higher risk estimates for glioma located in the temporal lobe than in other locations. One exception is the Interphone combined analysis, where the glioma risk was slightly higher in the temporal lobe among long-​term users and in the highest category of cumulative hours of use, but not overall (Interphone Study Group, 2010). For meningioma, the overall analysis found a lower risk estimate for tumors in the temporal lobe than in other locations. The four studies conducted by Hardell and colleagues grouped tumor locations differently in all studies, and small numbers probably prevented analyses of temporal lobe tumors separately in the two first studies. The fourth study reported somewhat stronger risk estimates for temporal lobe locations combined with overlapping lobes (Hardell et al., 2013). The recent French study reported risk estimates in the same order of magnitude for glioma with temporal

 269

269

Electromagnetic Fields hours

Glioma

cases controls

RR (95% CI)

–4

141

197

0.70 (0.52, 0.94)

5–12

145

198

0.71 (0.53, 0.94)

13–30

189

179

1.05 (0.79, 1.38)

31–60

144

196

0.74 (0.55, 0.98)

61–114

171

193

0.81 (0.61, 1.08)

115–199

160

194

0.73 (0.54, 0.98)

200–359

158

194

0.76 (0.57, 1.01)

360–734

189

205

0.82 (0.62, 1.08)

735–1639 159

184

0.71 (0.53, 0.96)

1640+

154

1.40 (1.03, 1.89)

210

.5 hours

.7

.8

1

1.5

Acoustic neuroma

cases controls

2 RR (95% CI)

–4

58

144

0.77 (0.52, 1.15)

5–12

63

129

0.80 (0.54, 1.18)

13–30

80

136

1.04 (0.71, 1.52)

31–60

66

131

0.95 (0.63, 1.42)

61–114

74

137

0.96 (0.66, 1.41)

115–199

68

128

0.96 (0.65, 1.42)

200–359

50

144

0.60 (0.39, 0.91)

360–734

58

126

0.72 (0.48, 1.09)

735–1639 49

126

0.48 (0.30, 0.78)

1640+

107

1.32 (0.88, 1.97)

77

.5

.7 .8

1

1.5

lobe location as for “other locations,” while risk estimates for frontal lobe tumors were lower (Coureau et al., 2014). Taken together, the evidence does not consistently suggest stronger associations with tumors in the temporal lobe. Two studies have further developed analyses of tumor localization beyond lobe of the brain by using information about the exact localization of the tumor based on radiological images (Cardis et al., 2011; Larjavaara et al., 2011b). The study by Larjavaara and colleagues used a case-​case design based on the assumption that gliomas in mobile phone users on average are located closer to the exposure source (i.e., where the mobile phone handset is held) than tumors in cases who were not regular mobile phone users. This type of design eliminates potential selection bias caused by non-​participation among controls, and if the preferred side for phone use is not included, it may also reduce the effect of recall bias. Analyses were based on 873 glioma cases, and no major difference in distance between the tumor location and a hypothetical mobile phone was found between cases who were regular mobile phone users and those who were non-​users; if anything, the distance was somewhat shorter for non-​users.

2

Figure 15–3.  Results from the Interphone study for cumulative hours of mobile phone use, categorized in deciles. The reference category is non-​regular mobile phone use.

The other study estimated the total RF dose at the tumor location or a corresponding location for the controls (Cardis et  al., 2011). The total RF dose was estimated based on retrospective self-​reported information on cumulative hours of mobile phone use, preferred side of the head, frequency band, communication system, and network characteristics. Self-​reported cumulative hours of use and tumor location were the only significant predictors of RF dose (43% and 13% of the variation, respectively). Complete information for estimation of RF dose were available for a subset of the subjects, and results for glioma were almost identical in this subset when based simply on self-​ reported cumulative hours of use as when the more elaborated exposure estimate was used, contrary to what would have been expected if there were a causal association between RF dose and glioma risk. The study design did not attempt to reduce recall bias, and the addition of more technical details about the mobile communication system did not seem to reduce non-​differential exposure misclassification. The investigators also made a case-​case analysis, similar to the analysis in the Larjavaara study, but based on somewhat fewer cases (N = 556). They found that > 10 years since first mobile phone use were associated with

270

270

PART III:  THE CAUSES OF CANCER

an increased risk of glioma in the brain region with the highest exposure, but they also found a reduced risk for those with 5–​9 years since first mobile phone use, and no consistent dose–​response pattern with cumulative call time.

Children.  Currently there is only one study available on mobile

phone use and brain tumor risk in children and adolescents in the age range 7–​19  years (Aydin et  al., 2011). It used a case-​control design with exposure information collected through interviews with the parents and children. Information from mobile phone operators was available for a small subgroup of participants who were able to report their children’s current and previous mobile phone numbers. Overall, a risk estimate close to unity was observed, and risk did not increase with amount of use or by location of the tumor. Having had a mobile phone subscription > 2.8 years prior to diagnosis was associated with an increased brain tumor risk, but these results were based on only about one-​third of the data. Furthermore, this finding was not compatible with the stable incidence trends observed. Thus, the study does not support the hypothesis of an increased risk of brain tumors, but it also does not provide strong evidence against a risk increase.

Other Tumors.  For other tumor types (e.g., parotid gland tumors,

ocular melanoma, pituitary tumors, leukemia, lymphoma, and testicular cancer), the available data are fewer than for brain tumors. Currently, no consistent evidence suggests that mobile phone use is associated with an increased risk of these tumors (AGNIR, 2012; Ahlbom et al., 2009; SCENIHR, 2015).

Incidence Studies

Normally, incidence time trend studies would not be regarded as informative for inferences about etiology because exposure information at the individual level is not available, and there is no information about potential confounding factors. In addition, changes in clinical practices and new diagnostic procedures may affect incidence time trends without any real change in disease occurrence (as it does indeed for brain tumors, with the detection of smaller tumors following the introduction of computed tomography and later MRI) (Fisher et al., 2007; Helseth, 1995). Nevertheless, the situation with mobile phone use and brain tumor risk is somewhat different. Brain tumors are not related to lifestyle factors like smoking or alcohol consumption; the only established environmental risk factor is ionizing radiation, which explains only a small fraction of the occurrence. During the last decades there has been a rapid increase in the prevalence of mobile phone use in the general population in many countries, from a few percent at the end of the 1980s to near 100% in some age groups by the middle of the first decade of the 2000s. If RF exposure from mobile phone use increases the risk of brain tumors, one would expect to see an increasing brain tumor incidence in many countries, in the age groups that first adopted mobile phone technologies, unless the latency period is longer than the present risk time, or the risk increase is restricted to a small subgroup of the population. As mentioned earlier, brain tumor incidence studies have been published in many countries, including the Nordic countries, the United Kingdom, the United States, and Australia (Ahlbom and Feychting, 2011; de Vocht et  al., 2011; Deltour et  al., 2009, 2012; Dobes et  al., 2011; Kohler et al., 2011; Little et al., 2012). They have all reported on malignant brain tumors or gliomas specifically. The time periods included differ between the incidence studies, but the majority cover the incidence until 2007 or 2008, and one covers the period until 2009. All studies report stable incidence trends in age groups where mobile phone use has become prevalent, with no indications of increasing incidence after the introduction of mobile phones. In the oldest age groups, however, where mobile phone use has not been prevalent, an increasing brain tumor trend can be observed, which started before the introduction of mobile phones (Deltour et al., 2009; Dobes et al., 2011). Two studies used simulations to predict the shape of the glioma incidence trend over time under assumptions of different risk scenarios related to mobile phone use (Deltour et al., 2012; Little et al., 2012). The aim was consistency checks, that is, to investigate whether the risk of malignant brain tumors reported in some of the case-​control

studies would have been detected in incidence time trends. Both studies concluded that strong risk increases, such as reported in the studies by Hardell et al., would have been detected in incidence time trends. The Nordic incidence time trend study also had statistical power to detect smaller risk increases and would most likely also have detected an increased risk after 1640 cumulated hours of mobile phone use (Deltour et  al., 2012), which was the risk increase observed in the Interphone study. No such incidence increase was observed. Several studies reported brain tumor incidence time trends for children and adolescents separately (Aydin et  al., 2011; de Vocht et  al., 2011; Dobes et al., 2011; Inskip et al., 2010; McKean-​Cowdin et al., 2013). None of these studies found increases in the brain tumor incidence after the introduction of mobile phones. A few incidence studies of acoustic neuroma and parotid gland tumors have also been published (Benson et al., 2013; de Vocht, 2011; Larjavaara et al., 2011a; Shu et al., 2012; Stangerup et al., 2010). None of them reports changes in incidence trends that could be linked to the introduction of handheld mobile phones.

Environmental Exposure Studies of cancer risk associated with RF exposure from environmental sources have usually focused on radio and television transmitters or mobile phone base stations. Several studies were conducted as a result of public concern, and used an ecological design with exposure simply measured as distance to the nearest transmitter. These studies are not included in this review, as distance alone is too crude to measure RF exposure accurately. Two studies have assessed RF exposure from radio and television transmitters on an individual level, using elaborated modeling with detailed information about important exposure determinants, such as frequency, field strength, and other relevant information (Bethke et al., 2008; Ha et al., 2007; Merzenich et al., 2008; Schüz et al., 2008), and one study focused on exposure from mobile phone base stations, also with elaborated modeling to assess exposure (Elliott et al., 2010). None of the studies found any indications of increased risk for any types of tumors among children in relation to environmental RF exposure from transmitters. Taken together, the evidence does not suggest that environmental RF exposure from transmitters affects cancer risk, but few data are currently available. There are currently no epidemiological studies available on potential effects on cancer risk from exposures emanating from sources such as Wi-​Fi, smart meters, and other low-​level RF-​emitting devices. Although these exposure sources are now ubiquitous in society, the level of exposure to individuals from these devices is very low, far below current exposure guidelines (AGNIR, 2012; Tell et al., 2013), and considerably lower than the exposure during a mobile phone call, for example.

Occupational Exposure Epidemiological research on occupational exposure to RF fields has not evolved in the same way as for ELF fields. Accurate assessment of occupational RF exposure remains a key concern in these studies, as they have usually been limited to job title alone, with no actual knowledge about the exposure levels for the workers in the particular occupations, to expert assessment, for example by occupational hygienists when no objective measurements were available, or by restricting the exposed group to specific areas, such as militaries in the navy. Exposure assessment has not been conducted on an individual level, and information on potential confounding factors is lacking. Many exposed occupations are also exposed to low frequency fields, such as radio and television repairmen and radio and telegraph operators. Most occupational studies focused on the risk of cancer overall, and leukemia and brain tumors (AGNIR, 2012; Ahlbom et al., 2004). No consistently increased risks were reported, but findings were often based on small numbers, and exposure misclassification is likely to be considerable. A  few risk increases were observed, generally in studies with severe methodological shortcomings. The largest studies

 271

Electromagnetic Fields with the most sophisticated exposure assessment provide little support for an association (Groves et al., 2002; Morgan et al., 2000). Groves et al. reported higher leukemia mortality among groups with the highest potential for radar exposure, with a risk estimate of 1.48 (95% CI 1.01–​2.17), but also a reduced brain cancer mortality in the same group, with a risk estimate of 0.65 (95% CI 0.43–​1.01). Morgan and colleagues reported risk estimates below unity for both leukemia and nervous system tumors, but results were based on small numbers of exposed cases (Morgan et al., 2000). Overall, the data do not suggest an increased cancer risk associated with occupational RF exposure, but the evidence is not sufficient to exclude the possibility of a risk increase.

CONCLUSIONS Overall, the epidemiological evidence does not suggest that exposure to RF fields increases cancer risk, taking into consideration the methodological limitations in the case-​control studies of mobile phone use and brain tumor risk, lack of associations in the cohort studies, and the stable brain tumor incidence time trends in the age groups that have been frequent mobile phone users for a long time. Some uncertainty remains regarding the heaviest mobile phone users, although the proportion of the population that has reached the highest cumulative hours of mobile phone use investigated thus far is likely to be substantial by now. Mobile phone use during the more recent years is, however, likely to be associated with considerably lower average RF exposure levels, as the third-​generation mobile phone technology has become more widespread. Experimental studies, including animal studies with long-​term exposure, have not found consistent evidence of a carcinogenic effect (AGNIR, 2012), and despite the large amount of research performed, a plausible biological mechanism has still not been suggested. Thus, the mounting evidence speaks increasingly against a carcinogenic effect of exposure to low-​level radiofrequency electromagnetic fields. However, neither cohort, case-​control, nor incidence time trend studies performed to date would be able to detect a risk increase after an induction period exceeding 20  years, and limited data are available on potential effects in children.

FUTURE RESEARCH NEEDS Although brain tumor incidence time trends have remained stable since the introduction of mobile phones, and recall and selection bias in available case-​control studies are likely explanations for observed risk increases after moderate and short-​term mobile phone use, the uncertainty regarding the most intensive mobile phone use and the widespread use of mobile phones throughout society and in all age groups, with an increasing amount of time spent using the technology, make further research warranted. A continued monitoring of brain tumor incidence time trends in various age groups is recommended, provided that well-​established high-​quality cancer registries are available. Further case-​control studies with retrospective self-​reported exposure information are unlikely to add useful knowledge. Highly recommended are studies that use a cohort design with prospectively collected exposure information, preferably from independent sources such as mobile phone operator registers, combined with prospective questionnaire information that covers important aspects that may modify exposure levels, such as use of hands-​free devices, preferred side of the head, and DECT phones and other wireless technologies. Access is also needed to population-​based sources for complete follow-​up of cancer occurrence, and studies must cover long enough exposure durations and be of sufficient size to provide statistically stable risk estimates. Such an approach would address the sources of bias identified in currently available epidemiological studies. Focus on mobile phone use is warranted, as it is still the main source of RF exposure in the general population, while exposure levels from mobile phone base stations, smart meters, and other far-​field sources are considerably lower.

271

Few data are available on mobile phone use and cancer risk in children, and although the single available case-​control study of brain tumors in children and adolescents did not indicate an increased risk, and brain tumor incidence time trends in this age group have been stable during the last decades, further studies focused on children are recommended. These must, however, be able to overcome potential bias from differential recall and selection bias. Prospective cohort studies would overcome these limitations and could also enable follow-​up into adult life. To address cancer risk in children, adolescents, and young adults, however, these cohorts need to be extremely large because of the rarity of the diseases, which is an obstacle for the feasibility of such studies. Exposure assessment among children poses additional challenges, as they do not have mobile phone subscriptions in their own name, and may more often share a mobile phone. In addition, exposure patterns have changed considerably with the introduction of smartphones, which are more frequently used for texting and surfing than making phone calls. Studies of occupational RF exposure would allow investigation of higher levels of exposure, but to be informative, exposure assessment needs to be improved considerably, to be on an individual level, and to be properly validated. In addition, information on potential confounding factors must be collected. An obstacle is that exposed occupational groups are small in numbers, and to achieve sufficient statistical power international collaboration is necessary. Numerous studies of genotoxic or carcinogenic effects of RF fields in experimental animals have been conducted, but have not shown consistent effects at exposure levels below guideline levels. Studies of RF fields as a co-​carcinogen have been mostly negative, or have had methodological limitations. The recent finding of increases in tumors of the lung and liver, and lymphoma in ENU-​treated mice exposed to RF fields, but with no dose response, warrants replication. References AGNIR. 2012. Health effects from radiofrequency electromagnetic fields: report from the Independent Advisory Group on Non-​Ionising Radiation. In Documents of the Health Protection Agency R, Chemical and Environmental Hazards. RCE 20, Health Protection Agency, UK (Ed.). Ahlbom A, Day N, Feychting M, et  al. 2000. A pooled analysis of magnetic fields and childhood leukaemia. Br J Cancer, 83(5), 692–​698. PMCID: PMC2363518. Ahlbom A, and Feychting M. 2011. Mobile telephones and brain tumours. BMJ, 343, d6605. Ahlbom A, Feychting M, Green A, et  al. 2009. Epidemiologic evidence on mobile phones and tumor risk: a review. Epidemiology, 20(5), 639–​652. Ahlbom A, Green A, Kheifets L, Savitz D, and Swerdlow A. 2004. Epidemiology of health effects of radiofrequency exposure. Environ Health Perspect, 112(17), 1741–​1754. PMCID: PMC1253668. Aydin D, Feychting M, Schüz J, et al. 2011. Mobile phone use and brain tumors in children and adolescents:  a multicenter case-​control study. J Natl Cancer Inst, 103(16), 1264–​1276. Barchana M, Margaliot M, and Liphshitz I. 2012. Changes in brain glioma incidence and laterality correlates with use of mobile phones: a nationwide population based study in Israel. Asian Pac J Cancer Prev, 13(11), 5857–​5863. Benson VS, Pirie K, Schuz J et  al. 2014. Authors’ response to:  The case of acoustic neuroma: comment on mobile phone use and risk of brain neoplasms and other cancers. Int J Epidemiol. 43(1):275. Benson VS, Pirie K, Schüz J, et al. 2013. Mobile phone use and risk of brain neoplasms and other cancers:  prospective study. Int J Epidemiol, 42(3), 792–​802. Bethke L, Murray A, Webb E, et  al. 2008. Comprehensive analysis of DNA repair gene variants and risk of meningioma. J Natl Cancer Inst, 100(4), 270–​276. Bowman JD, Touchstone JA, and Yost MG. 2007. A population-​based job exposure matrix for power-​frequency magnetic fields. J Occup Environ Hyg, 4(9), 715–​728. Cardis E, Armstrong BK, Bowman JD, et al. 2011. Risk of brain tumours in relation to estimated RF dose from mobile phones: results from five Interphone countries. Occup Environ Med, 68(9), 631–​640. PMCID: PMC3158328. Cardis E, Richardson L, Deltour I, et al. 2007. The INTERPHONE study: design, epidemiological methods, and description of the study population. Eur J Epidemiol, 22(9), 647–​664. Carlberg M, Soderqvist F, Hansson Mild K, and Hardell L. 2013. Meningioma patients diagnosed 2007–​2009 and the association with use of mobile

27

272

PART III:  THE CAUSES OF CANCER

and cordless phones:  a case-​control study. Environ Health, 12(1), 60. PMCID: PMC3733594. Chen C, Ma X, Zhong M, and Yu Z. 2010. Extremely low-​frequency electromagnetic fields exposure and female breast cancer risk: a meta-​analysis based on 24,338 cases and 60,628 controls. Breast Cancer Res Treat, 123(2), 569–​576. Chen Q, Lang L, Wu W, et al. 2013. A meta-​analysis on the relationship between exposure to ELF-​EMFs and the risk of female breast cancer. PLoS One, 8(7), e69272. PMCID: PMC3712018. Coureau G, Bouvier G, Lebailly P, et  al. 2014. Mobile phone use and brain tumours in the CERENAT case-​control study. Occup Environ Med, 71(7), 514–​522. de Vocht F. 2011. Cell phones and parotid cancer trends in England. Epidemiology, 22(4), 608–​609. de Vocht F, Burstyn I, and Cherrie JW. 2011. Time trends (1998–​2007) in brain cancer incidence rates in relation to mobile phone use in England. Bioelectromagnetics, 32(5), 334–​339. Deltour I, Auvinen A, Feychting M, et al. 2012. Mobile phone use and incidence of glioma in the Nordic countries 1979–​2008: consistency check. Epidemiology, 23(2), 301–​307. Deltour I, Johansen C, Auvinen A, et al. 2009. Time trends in brain tumor incidence rates in Denmark, Finland, Norway, and Sweden, 1974–​2003. J Natl Cancer Inst, 101(24), 1721–​1724. Dobes M, Khurana VG, Shadbolt B, et al. 2011. Increasing incidence of glioblastoma multiforme and meningioma, and decreasing incidence of Schwannoma (2000–​2008):  findings of a multicenter Australian study. Surg Neurol Int, 2, 176. PMCID: PMC3263004. Elliott P, Shaddick G, Douglass M, et al. 2013. Adult cancers near high-​voltage overhead power lines. Epidemiology, 24(2), 184–​190. Elliott P, Toledano MB, Bennett J, et al. 2010. Mobile phone base stations and early childhood cancers: case-​control study. BMJ, 340, c3077. Feychting M, and Ahlbom A. 1994. Magnetic fields, leukemia, and central nervous system tumors in Swedish adults residing near high-​voltage power lines. Epidemiology, 5(5), 501–​509. Feychting M, and Forssen U. 2006. Electromagnetic fields and female breast cancer. Cancer Causes Control, 17(4), 553–​558. Feychting M, Schulgen G, Olsen JH, and Ahlbom A. 1995. Magnetic fields and childhood cancer: a pooled analysis of two Scandinavian studies. Eur J Cancer, 31A(12), 2035–​2039. Fisher JL, Schwartzbaum JA, Wrensch M, and Wiemels JL. 2007. Epidemiology of brain tumors. Neurol Clin, 25(4), 867–​890, vii. Foliart DE, Pollock BH, Mezei G, et  al. 2006. Magnetic field exposure and long-​term survival among children with leukaemia. Br J Cancer, 94(1), 161–​164. PMCID: PMC2361064. Forssen UM, Ahlbom A, and Feychting M. 2002. Relative contribution of residential and occupational magnetic field exposure over twenty-​ four hours among people living close to and far from a power line. Bioelectromagnetics, 23(3), 239–​244. Frei P, Poulsen AH, Johansen C, et al. 2011. Use of mobile phones and risk of brain tumours:  update of Danish cohort study. BMJ, 343, d6387. PMCID: PMC3197791. Gobba F, Bravo G, Rossi P, Contessa GM, and Scaringi M. 2011. Occupational and environmental exposure to extremely low frequency-​ magnetic fields: a personal monitoring study in a large group of workers in Italy. J Expo Sci Environ Epidemiol, 21(6), 634–​645. Greenland S, Fischer HJ, and Kheifets L. 2016. Methods to explore uncertainty and bias introduced by job exposure matrices. Risk Anal, 36(1):74–​82. Greenland S, and Kheifets L. 2006. Leukemia attributable to residential magnetic fields:  results from analyses allowing for study biases. Risk Anal, 26(2), 471–​482. Greenland S, Sheppard AR, Kaune WT, Poole C, and Kelsh MA. 2000. A pooled analysis of magnetic fields, wire codes, and childhood leukemia. Childhood Leukemia-​EMF Study Group. Epidemiology, 11(6), 624–​634. Grellier J, Ravazzani P, and Cardis E. 2014. Potential health impacts of residential exposures to extremely low frequency magnetic fields in Europe. Environ Int, 62, 55–​63. Groves FD, Page WF, Gridley G et  al. 2002. Cancer in Korean war navy technicians:  mortality survey after 40  years. Am J Epidemiol, 155(9), 810–​818. Ha M, Im H, Lee M, et  al. 2007. Radio-​frequency radiation exposure from AM radio transmitters and childhood leukemia and brain cancer. Am J Epidemiol, 166(3), 270–​279. Hardell L, Carlberg M, and Hansson Mild K. 2005. Case-​control study on cellular and cordless telephones and the risk for acoustic neuroma or

meningioma in patients diagnosed 2000–​2003. Neuroepidemiology, 25(3), 120–​128. Hardell L, Carlberg M, and Hansson Mild K. 2011. Re-​analysis of risk for glioma in relation to mobile telephone use: comparison with the results of the Interphone international case-​control study. Int J Epidemiol, 40(4), 1126–​1128. Hardell L, Carlberg M, and Mild KH. 2006. Case-​control study of the association between the use of cellular and cordless telephones and malignant brain tumors diagnosed during 2000–​2003. Environ Res, 100(2), 232–​241. Hardell L, Carlberg M, Soderqvist F, and Mild KH. 2013. Case-​control study of the association between malignant brain tumours diagnosed between 2007 and 2009 and mobile and cordless phone use. Int J Oncol, 43(6), 1833–​1845. PMCID: PMC3834325. Hardell L, Hallquist A, Mild KH, et al. 2002. Cellular and cordless telephones and the risk for brain tumours. Eur J Cancer Prev, 11(4), 377–​386. Hardell L, Nasman A, Pahlson A, Hallquist A, and Hansson Mild K. 1999. Use of cellular telephones and the risk for brain tumours: a case-​control study. Int J Oncol, 15(1), 113–​116. Hauer J, Borkhardt A, Sanchez-​Garcia I, and Cobaleda C. 2014. Genetically engineered mouse models of human B-​cell precursor leukemias. Cell Cycle, 13(18), 2836–​2846. Helseth A. 1995. The incidence of primary central nervous system neoplasms before and after computerized tomography availability. J Neurosurg, 83(6), 999–​1003. Hug K, Grize L, Seidler A, Kaatsch P, and Schüz J. 2010. Parental occupational exposure to extremely low frequency magnetic fields and childhood cancer: a German case-​control study. Am J Epidemiol, 171(1), 27–​35. IARC. 2002. Non-​ionizing radiation. Part  1, static and extremely low-​frequency (ELF) electric and magnetic fields. Vol. 80. Lyon: International Agency for Research on Cancer. IARC. 2013. Non-​ionizing radiation. Part 2, Radiofrequency electromagnetic fields. IARC monographs on the evaluation of carcinogenic risks to humans. Vol. 102. Lyon: International Agency for Research on Cancer. ICNIRP. 1998. Guidelines for limiting exposure to time-​ varying electric, magnetic, and electromagnetic fields (up to 300 GHz). International Commission on Non-​Ionizing Radiation Protection. Health Phys, 74(4), 494–​522. ICNIRP. 2010. Guidelines for limiting exposure to time-​varying electric and magnetic fields (1 Hz to 100 kHz). Health Phys, 99(6), 818–​836. IEEE. 2002. Std C95.6–​2002, IEEE standard for safety levels with respect to human exposure to electromagnetic fields, 0–​3 kHz. https://​standards. ieee.org/​findstds/​standard/​C95.6-​2002.html. Accessed June 26, 2016. IEEE. 2005. Std C95.1–​2005, IEEE standard for safety levels with respect to human exposure to radio frequency electromagnetic fields, 3 kHz–​ 300 GHz. https://​standards.ieee.org/​findstds/​standard/​C95.1-​2005.html. Accessed June 26, 2016. Inskip PD, Hoover RN, and Devesa SS. 2010. Brain cancer incidence trends in relation to cellular telephone use in the United States. Neuro Oncol, 12(11), 1147–​1151. PMCID: PMC3098028. Interphone Study Group. 2010. Brain tumour risk in relation to mobile telephone use: results of the INTERPHONE international case-​control study. Int J Epidemiol, 39(3), 675–​694. Interphone Study Group. 2011. Acoustic neuroma risk in relation to mobile telephone use:  results of the INTERPHONE international case-​control study. Cancer Epidemiol, 35(5), 453–​464. Johansen C, Boice J, Jr, McLaughlin J, and Olsen J. 2001. Cellular telephones and cancer-​-​a nationwide cohort study in Denmark. J Natl Cancer Inst, 93(3), 203–​207. Johansen C, Raaschou Nielsen O, Olsen JH, and Schüz J. 2007. Risk for leukaemia and brain and breast cancer among Danish utility workers:  a second follow-​ up. Occup Environ Med, 64(11), 782–​ 784. PMCID: PMC2078420. Kheifets L, Ahlbom A, Crespi CM, et al. 2010a. Pooled analysis of recent studies on magnetic fields and childhood leukaemia. Br J Cancer, 103(7), 1128–​1135. PMCID: PMC2965855. Kheifets L, Ahlbom A, Crespi CM, et al. 2010b. A pooled analysis of extremely low-​ frequency magnetic fields and childhood brain tumors. Am J Epidemiol, 172(7), 752–​761. PMCID: PMC2984256. Kheifets L, Monroe J, Vergara X, Mezei G, and Afifi AA. 2008. Occupational electromagnetic fields and leukemia and brain cancer: an update to two meta-​analyses. J Occup Environ Med, 50(6), 677–​688. Kiyohara K, Wake K, Watanabe S, et al. 2015. Recall accuracy of mobile phone calls among Japanese young people. J Expo Sci Environ Epidemiol. Mar 18. doi: 10.1038/​jes.2015.13. [Epub ahead of print]

 273

Electromagnetic Fields Klaeboe L, Blaasaas KG, Haldorsen T, and Tynes T. 2005. Residential and occupational exposure to 50-​Hz magnetic fields and brain tumours in Norway: a population-​based study. Int J Cancer, 115(1), 137–​141. Koeman T, van den Brandt PA, Slottje P, et al. 2014. Occupational extremely low-​frequency magnetic field exposure and selected cancer outcomes in a prospective Dutch cohort. Cancer Causes Control, 25(2), 203–​214. Kohler BA, Ward E, McCarthy BJ, et  al. 2011. Annual report to the nation on the status of cancer, 1975–​ 2007, featuring tumors of the brain and other nervous system. J Natl Cancer Inst, 103(9), 714–​ 736. PMCID: PMC3086878. Lahkola A, Auvinen A, Raitanen J, et al. 2007. Mobile phone use and risk of glioma in 5 North European countries. Int J Cancer, 120(8), 1769–​1775. Lahkola A, Salminen T, Raitanen J, et al. 2008. Meningioma and mobile phone use: a collaborative case-​control study in five North European countries. Int J Epidemiol, 37(6), 1304–​1313. Larjavaara S, Feychting M, Sankila R et al. 2011a. Incidence trends of vestibular schwannomas in Denmark, Finland, Norway and Sweden in 1987–​ 2007. Br J Cancer, 105(7), 1069–​1075. PMCID: PMC3185956. Larjavaara S, Schüz J, Swerdlow A, et al. 2011b. Location of gliomas in relation to mobile telephone use: a case-​case and case-​specular analysis. Am J Epidemiol, 174(1), 2–​11. Lerchl A, Klose M, Grote K, et  al. 2015. Tumor promotion by exposure to radiofrequency electromagnetic fields below exposure limits for humans. Biochem Biophys Res Comm, 459(4), 585–​590. Li W, Ray RM, Thomas DB, et al. 2013. Occupational exposure to magnetic fields and breast cancer among women textile workers in Shanghai, China. Am J Epidemiol, 178(7), 1038–​1045. PMCID: PMC3783098. Linet MS, Hatch EE, Kleinerman RA, et al. 1997. Residential exposure to magnetic fields and acute lymphoblastic leukemia in children. N Engl J Med, 337(1), 1–​7. Little MP, Rajaraman P, Curtis RE, et al. 2012. Mobile phone use and glioma risk: comparison of epidemiological study results with incidence trends in the United States. BMJ, 344, e1147. PMCID: PMC3297541. Lonn S, Ahlbom A, Hall P, and Feychting M. 2005. Long-​term mobile phone use and brain tumor risk. Am J Epidemiol, 161(6), 526–​535. Martin-​Lorenzo A, Hauer J, Vicente-​Duenas C, et al. 2015. Infection exposure is a causal factor in B-​cell precursor acute lymphoblastic leukemia as a result of Pax5-​inherited susceptibility. Cancer Discov, 5(12), 1328–​1343. Maslanyj M, Simpson J, Roman E, and Schüz J. 2009. Power frequency magnetic fields and risk of childhood leukaemia: misclassification of exposure from the use of the ‘distance from power line’ exposure surrogate. Bioelectromagnetics, 30(3), 183–​188. McBride ML, Gallagher RP, Theriault G, et al. 1999. Power-​frequency electric and magnetic fields and risk of childhood leukemia in Canada. Am J Epidemiol, 149(9), 831–​842. McKean-​Cowdin R, Razavi P, Barrington-​Trimis J, et  al. 2013. Trends in childhood brain tumor incidence, 1973–​2009. J Neurooncol, 115(2), 153–​160. Merzenich H, Schmiedel S, Bennack S, et al. 2008. Childhood leukemia in relation to radio frequency electromagnetic fields in the vicinity of TV and radio broadcast transmitters. Am J Epidemiol, 168(10), 1169–​1178. Mezei G, Gadallah M, and Kheifets L. 2008. Residential magnetic field exposure and childhood brain cancer: a meta-​analysis. Epidemiology, 19(3), 424–​430. Morgan RW, Kelsh MA, Zhao K, et  al. 2000. Radiofrequency exposure and mortality from cancer of the brain and lymphatic/​hematopoietic systems. Epidemiology, 11(2), 118–​127. Muscat JE, Malkin MG, Thompson S, et al. 2000. Handheld cellular telephone use and risk of brain cancer. JAMA, 284(23), 3001–​3007. National Toxicology Program. 2014. Cell Phone Radiofrequency Radiation Studies. http://​www.niehs.nih.gov/​health/​assets/​docs_​a_​e/​cell_​phone_​ radiofrequency_​radiation_​studies_​508.pdf. Accessed December 19, 2015. Pettersson D, Bottai M, Mathiesen T, Prochazka M, and Feychting M. 2015. Validation of self-​reported start year of mobile phone use in a Swedish case-​control study on radiofrequency fields and acoustic neuroma risk. J Expo Sci Environ Epidemiol, 25(1), 72–​79. Pettersson D, Mathiesen T, Prochazka M, et al. 2014. Long-​term mobile phone use and acoustic neuroma risk. Epidemiology, 25(2), 233–​241. Reiter RJ. 1995. Reported biological consequences related to the suppression of melatonin by electric and magnetic field exposure. Integr Physiol Behav Sci, 30(4), 314–​330. Repacholi MH, Basten A, Gebski V, et  al. 1997. Lymphomas in E mu-​Pim1 transgenic mice exposed to pulsed 900 MHZ electromagnetic fields. Radiat Res, 147(5), 631–​640.

273

SCENIHR. 2009. Scientific Committee on Emerging and Newly Identified Health Risks:  Health Effects of Exposure to EMF. http://​ec.europa.eu/​ health/​archive/​ph_​risk/​committees/​04_​scenihr/​docs/​scenihr_​o_​022.pdf. Accessed October 19, 2012. SCENIHR. 2015. Scientific Committee on Emerging and Newly Identified Health Risks:  Potential Health Effects of Exposure to Electromagnetic Fields (EMF). http://​ec.europa.eu/​health/​scientific_​committees/​emerging/​docs/​scenihr_​o_​041.pdf. Accessed August 15, 2015. Schoemaker MJ, Swerdlow AJ, Ahlbom A, et al. 2005. Mobile phone use and risk of acoustic neuroma: results of the Interphone case-​control study in five North European countries. Br J Cancer, 93(7), 842–​848. PMCID: PMC2361634. Schüz J. 2007. Implications from epidemiologic studies on magnetic fields and the risk of childhood leukemia on protection guidelines. Health Phys, 92(6), 642–​648. Schüz J. 2009. Lost in laterality: interpreting “preferred side of the head during mobile phone use and risk of brain tumour” associations. Scand J Public Health, 37(6), 664–​667. Schüz J, Böhler E, Berg G, et al. 2006a. Cellular phones, cordless phones, and the risks of glioma and meningioma (Interphone Study Group, Germany). Am J Epidemiol, 163(6), 512–​520. Schüz J, Elliott P, Auvinen A, et al. 2011a. An international prospective cohort study of mobile phone users and health (Cosmos): design considerations and enrolment. Cancer Epidemiol, 35(1), 37–​43. Schüz J, Grell K, Kinsey S, et al. 2012. Extremely low-​frequency magnetic fields and survival from childhood acute lymphoblastic leukemia:  an international follow-​up study. Blood Cancer J, 2, e98. PMCID: PMC3542478. Schüz J, Jacobsen R, Olsen JH, et al. 2006b. Cellular telephone use and cancer risk: update of a nationwide Danish cohort. J Natl Cancer Inst, 98(23), 1707–​1713. Schüz J, Philipp J, Merzenich H, Schmiedel S, and Bruggemeyer H. 2008. Re: Radio-​frequency radiation exposure from AM radio transmitters and childhood leukemia and brain cancer. Am J Epidemiol, 167(7), 883–​884. Schüz J, Steding-​Jessen M, Hansen S, et al. 2011b. Long-​term mobile phone use and the risk of vestibular schwannoma: a Danish nationwide cohort study. Am J Epidemiol, 174(4), 416-​422. Schüz J, Svendsen AL, Linet MS, et al. 2007. Nighttime exposure to electromagnetic fields and childhood leukemia: an extended pooled analysis. Am J Epidemiol, 166(3), 263–​269. Shu X, Ahlbom A, and Feychting M. 2012. Incidence trends of malignant parotid gland tumors in Swedish and nordic adults 1970 to 2009. Epidemiology, 23(5), 766–​767. Stangerup SE, Tos M, Thomsen J, and Caye-​Thomasen P. 2010. True incidence of vestibular schwannoma? Neurosurgery, 67(5), 1335–​1340; discussion 1340. Stevens RG. 1987. Electric power use and breast cancer: a hypothesis. Am J Epidemiol, 125(4), 556–​561. Swerdlow AJ, Feychting M, Green AC, Leeka Kheifets LK, and Savitz DA. 2011. Mobile phones, brain tumors, and the interphone study: where are we now? Environ Health Perspect, 119(11), 1534–​1538. PMCID: PMC3226506. Tell RA, Kavet R, and Mezei G. 2013. Characterization of radiofrequency field emissions from smart meters. J Expo Sci Environ Epidemiol, 23(5), 549–​553. Thomsen J, and Tos M. 1990. Acoustic neuroma: clinical aspects, audiovestibular assessment, diagnostic delay, and growth rate. Am J Otol, 11(1), 12–​19. Tillmann T, Ernst H, Streckert J, et  al. 2010. Indication of cocarcinogenic potential of chronic UMTS-​modulated radiofrequency exposure in an ethylnitrosourea mouse model. Int J Radiat Biol, 86(7), 529–​541. Turner MC, Benke G, Bowman JD, et  al. 2014. Occupational exposure to extremely low-​frequency magnetic fields and brain tumor risks in the INTEROCC study. Cancer Epidemiol Biomarkers Prev, 23(9), 1863–​ 1872. PMCID: PMC4154968. Tynes T, and Haldorsen T. 1997. Electromagnetic fields and cancer in children residing near Norwegian high-​voltage power lines. Am J Epidemiol, 145(3), 219–​226. Tynes T, and Haldorsen T. 2003. Residential and occupational exposure to 50 Hz magnetic fields and hematological cancers in Norway. Cancer Causes Control, 14(8), 715–​720. UKCCS. 1999. Exposure to power-​frequency magnetic fields and the risk of childhood cancer. UK Childhood Cancer Study Investigators. Lancet, 354(9194), 1925–​1931. Verkasalo PK, Pukkala E, Hongisto MY, et  al. 1993. Risk of cancer in Finnish children living close to power lines. BMJ, 307(6909), 895–​899. PMCID: PMC1679080. Verkasalo PK, Pukkala E, Kaprio J, Heikkila KV, and Koskenvuo M. 1996. Magnetic fields of high voltage power lines and risk of cancer in Finnish adults: nationwide cohort study. BMJ, 313(7064), 1047–​1051. PMCID: PMC2352388.

274

274

PART III:  THE CAUSES OF CANCER

Vrijheid M, Armstrong BK, Bedard D, et al. 2009a. Recall bias in the assessment of exposure to mobile phones. J Expo Sci Environ Epidemiol, 19(4), 369–​381. Vrijheid M, Cardis E, Armstrong BK, et al. 2006. Validation of short term recall of mobile phone use for the Interphone study. Occup Environ Med, 63(4), 237–​243. PMCID: PMC2078087. Vrijheid M, Richardson L, Armstrong BK, et al. 2009b. Quantifying the impact of selection bias caused by nonparticipation in a case-​control study of mobile phone use. Ann Epidemiol, 19(1), 33–​41.

Wertheimer N, and Leeper E. 1979. Electrical wiring configurations and childhood cancer. Am J Epidemiol, 109(3), 273–​284. WHO. 2007. World Health Organization. Environmental Health Criteria (EHC) Document on ELF Fields, Doc. No. 238, downloadable from the WHO EMF Project website, www.who.int/​emf.

 275

16 Occupational Cancer KYLE STEENLAND, SHELIA HOAR ZAHM, AND A. BLAIR

OVERVIEW Occupational cancer was one of the earliest areas of cancer research, as astute clinicians observed clusters of cancer in workers exposed to high levels of workplace toxins. These early observations were followed by some of the earliest epidemiologic cohort studies, beginning after World War II, which followed the leads from the early clusters. Many of today’s recognized carcinogens were discovered via occupational studies, including classic carcinogens such as asbestos and arsenic, as well as more recently established carcinogens, such as silica and diesel fumes. The earlier discoveries were characterized by high exposures and high risks; later discoveries of agents with lower exposure and lower risks have required a multiplicity of large studies to establish causality, and to provide sufficient data for risk assessment to be performed by regulatory agencies. It is likely that this pattern will continue, with cancer risks from more difficult to characterize exposures like shift work, currently considered a probable but not definite carcinogen, requiring a large body of evidence before acceptance. Most occupational exposures recognized in the past were of a chemical or particle nature. A more expansive view of occupational exposures today should consider a broader range of infectious, medical, behavioral, and physical factors that might impact cancer, such as shift work, sedentary behavior, or workplace stress. Occupational carcinogen discovery has always been controversial because such findings have economic as well as public health implications. Consequently, decision-​making and action may involve more political activity than findings in other epidemiologic areas. The example of diesel exhaust, recently declared a definite carcinogen by the International Agency for Research on Cancer (IARC) after 25 years of controversy, is a good example (see later discussion in this chapter). In this chapter we summarize the past history of occupational cancer epidemiology, and indicate which occupational exposures are presently considered to be definite or probable carcinogens. We describe the basic study designs of occupational cancer research, which are similar to the designs of non-​occupational studies, but which have their own peculiarities, particularly in regard to exposure assessment. We discuss the type of evidence that has stimulated occupational cancer studies and how data generated by occupational cancer studies are used in risk assessment for workplace regulations, and the calculation of attributable fractions to quantify the burden of occupational cancer. Finally, we discuss some current controversies and propose likely future directions for occupational epidemiology. These include a focus on “exposures” like shift work and sedentary work habits, which are not traditional toxins. In addition, it will be important to document the carcinogenic effects of established occupational carcinogens in less developed countries where they have not been studied, as a means to affect policy and ensure safe workplaces.

INTRODUCTION Studies of workplace exposures have provided a wealth of information regarding the causes of cancer from early (Ramazzini, 1713) to recent times (Siemiatycki, 2014). Approximately 40% of the factors classified as sufficient, probable, or possible human carcinogens by the IARC were initially investigated as occupational exposures (Siemiatycki et al., 2004). There are many practical and disease prevention benefits

of studies of cancer in the workplace. Exposures in the workplace are often considerably higher than those in the non-​occupational setting. Occupational exposures can often be measured more precisely than non-​occupational exposures because of the availability of production records, personal and area measurements, and operating procedures, which provide information that can be used to characterize exposures. Occupational studies are needed to protect workers from workplace hazards. Many workers have little flexibility regarding how tasks are performed, equipment and ingredients used, and the location of work, and thus have relatively little individual control over their personal exposures. Occupational exposures are, therefore, largely involuntary. Although the excesses of cancer from occupational exposures may occur in worker groups of relatively small size, the high level of exposure can result in large relative risks. Because of this, over three decades ago Doll and Peto (1981) noted that “detection of occupational hazards should therefore have a higher priority in any program of cancer prevention than their proportional importance might suggest.” This conclusion is still true today. It should be noted that many occupational exposures also occur in non-​occupational settings and have adverse impacts beyond the workplace (e.g., arsenic, diesel fumes, polychlorinated biphenyls [PCBs], dioxins, polycyclic aromatic hydrocarbons, and pesticides). Despite the many established links between occupational exposures and cancer, much remains unknown. A starting place for a list of needed studies could be those occupational exposures already classified as possible or probable human carcinogens by the IARC. These often have limited or no epidemiologic information, but are already suspect, and clarity is needed. The selection priority for the agents on this list would depend upon factors such as number of workers exposed, typical levels of exposure, likely magnitude of relative risk, opportunities for high-​quality investigations, and the possibility of exposures outside the workplace. In addition, there are many occupations or industries where cancer excesses have been noted, but specific agents that might be involved have not been clearly identified. Despite the obvious need for additional investigations (Blair et al., 2011; Siemiatycki, 2014; Straif, 2012; Ward et al., 2010), the number of publications on cancer in the workplace is decreasing and is now only about one-​third the number of just a decade ago (Raj et al., 2014). Explanations for this decline are not clear, but may include a change in the political/​social climate that no longer supports this type of health research, a decrease in heavy industry in high-​income countries, a weakening of labor unions which have encouraged research, cycles of focus and popularity in public health research, a belief that control of hazardous occupational exposures has been accomplished because of previous successes, and the failure in recent times to find occupational exposures with such dramatic impacts on cancer as previously identified agents such as asbestos, benzidine, vinyl chloride, benzene, and chromium (Blair et al., 2011; Raj et al., 2014; Siemiatycki, 2014). Most recognized occupational carcinogens have been identified through studies of men in high-​income countries (Hohenadel et al., 2015; Zahm et al., 1994, 1999). The inclusion of women in studies of the workplace has increased since the early 1990s, but studies of men still predominate and typically have more detailed analyses on men than on women, possibly due to small numbers or a lower priority placed on investigating risks among women (Hohenadel et al., 2015). Furthermore, in the past few decades, low-​and middle-​income countries have assumed a larger and larger fraction of

275

276

276

PART III:  THE CAUSES OF CANCER

the world’s production in many industries, but there does not yet appear to have been a comparable increase in investigations of occupational exposures in most low-​income countries (Raj et al., 2014). Studies are particularly needed in low-​and middle-​income countries because exposures may be less well controlled (Siemiatycki, 2014).

HISTORY The era of modern occupational epidemiology, with systematic assembling and follow-​up of occupational cohorts, did not begin until after World War II. Most early identification of the occupational origin of some cancers was done by astute clinicians seeing cancer in clear excess (i.e., clusters) among workers who were exposed at the time. Perhaps the earliest identification of occupational cancer was particularly prescient. Bernardo Ramazzini, the father of occupational medicine, noted in 1713 that nuns, who can be considered members of a specific occupation, were more likely to have breast cancer than other women. He speculated that this might be due to a lack of sexual activity (Olson, 2002). Today we know that Ramazzini was right about the phenomenon but (partly) wrong about the cause, which we now attribute to nulliparity and increased exposure of the breast to estrogen. Ramazzini also described diseases among miners, painters, weavers, and printers. Another famous clinician who identified occupational cancer was Percival Pott, who noted in 1775 that chimney sweepers in London had an excess of scrotal cancer, which he attributed to soot accumulation in the scrotum (Dobson, 1972; Pott, 1775; Waldron, 1983). Subsequently there were other reports of scrotal cancer among other workers exposed to soot, also known as coal tar. Coal tar can be considered the first recognized occupational carcinogen, for which workmen’s compensation was given in England as of 1907 (Waldron, 1983). Another clinician who identified an early occupational carcinogen was Ludwig Rehn, who reported a cluster of bladder cancers among aniline dye workers in 1895 (Dietrich and Dietrich, 2001). In the 1920s, mineral oil was also recognized as a cause of scrotal cancer, and other skin cancers, due to a cluster of cases among cotton spinners (Waldron, 1983). Kuroda and Kawahata identified a cluster of lung cancer among workers exposed to coal gas in steel mills in Japan in 1935 (Takahashi and Ishii, 2013). In one of the earliest examples of calculation of standardized mortality ratios (SMRs) using death certificates and census data by occupation, Kennaway and Kennaway (1936) showed that coal gas workers in England had excess cancer of the lung (coal gas was later replaced by coke in steel mills, which also caused lung cancer) (Lloyd, 1971). Other notable clusters included bone and blood cancer in radium dial painters, identified by Harrison Martland in New Jersey in 1931 (Loutit, 1970), and lung cancer in nickel refinery workers in Wales, identified by Bridge in 1933 (Bridge, 1933). There were also several German case reports of lung cancer among chromate workers in the 1930s (Hayes et al., 1988). After World War II, occupational epidemiologists, primarily in England, systematically began using retrospective cohort studies based on industry records to study cancer mortality. Defining the study population as all workers in a particular job or industry and following them over time, instead of studying case series or clusters among current workers only, allowed the inclusion of cancers occurring after leaving work, and allowed statistical comparisons of cancer rates in exposed workers versus low-​or non-​exposed referents. Most of the previously observed clusters were confirmed in such studies, including bladder cancer in dye workers (Case et al., 1954), arsenic and lung/​skin cancer (Hill and Faning, 1948), chromates and lung cancer (Machle and Gregorius, 1948), nickel refining and lung cancer (Doll, 1958), and coal gas and lung cancer (Doll, 1952). Radium decays into radon gas, and after World War II investigators linked radon in uranium miners to lung cancer (Wagoner et al., 1965). All of these studies were retrospective cohort mortality studies, with two exceptions. In the case of arsenic and nickel, investigators did not have access to plant records, and instead conducted studies of proportional

mortality in the areas where the plants were located, compared with other areas in England. Asbestos had been known to cause asbestosis since before World War II. There were case reports of lung cancer among asbestos workers, and some published literature began to draw links between the two before the war (Greenberg, 1999). However, the first solid epidemiologic evidence came from the landmark cohort mortality study of asbestos workers by Doll in 1955, which showed the lung cancer risk among men employed for 20 or more years to be 10 times that of the general population (Doll, 1955). The first strong epidemiologic evidence for a link between asbestos and mesothelioma followed in 1960 among asbestos miners in South Africa (Wagner et al., 1960). Wagner et al. (1960) described a case series of 33 cases living near or working in crocidolite asbestos mines (i.e., a cluster). The fact that the tumor was so rare and did not occur elsewhere in South Africa provided compelling evidence of a link. The evidence of the relationship is now so strong that mesothelioma is considered a sentinel tumor or marker for asbestos exposure (Friedman, 2011). Following these classic early studies, subsequent cohort studies found excess cancer associated with a large number of occupational exposures, including dioxin, radon in mines, ethylene oxide, silica, vinyl chloride, diesel fumes, cadmium, benzene, beryllium, PCBs, wood dust, acid mists, and others all of which are classified as IARC Group 1 (definite) human carcinogens, as detailed in the next section of this chapter. Ionizing radiation, beyond the radon that caused lung cancer among miners (Darby et  al., 1995; IARC, 2012d), has also been identified as an occupational carcinogen. Long-​term follow-​up of nuclear workers from three countries, exposed primarily to γ-​radiation, showed the risk of developing leukemia, particularly chronic myeloid leukemia, to be linked to the protracted low doses of radiation (Leuraud et  al., 2015). Nuclear weapons workers exposed to plutonium had increased risk of lung cancer, liver cancer, and bone sarcoma (Gilbert, et al., 2000, 2004; IARC, 2012d; Koshurnikova et al., 2000). Workers involved in cleanup after the Chernobyl nuclear accident were exposed to X-​and γ-​radiation and have excess risk of leukemia, including chronic lymphocytic leukemia (Kesminiene et  al., 2008; Zablotska et al., 2013). Several of the carcinogens mentioned in the preceding paragraph are also present in the general environment at lower levels (e.g., dioxin, diesel fumes, PCBs, indoor radon). In addition, two occupations, painting and foundry work, have been identified by the IARC as increasing cancer risk (Group 1), without being able to specify which specific carcinogens are responsible. While most occupational carcinogens have been discovered via retrospective cohort studies, large population-​ based case-​ control studies have often confirmed these findings, using self-​reported job histories and assessment of potential occupational exposures by industrial hygienists. Perhaps the most important model for this approach has been the numerous studies in Montreal by the group led by Jack Siemiatycki (Siemiatycki et al., 1991, 1997). Siemiatycki’s team elicited detailed job histories through a semi-​structured questionnaire with supplementary questionnaires for selected occupations, followed by comprehensive review by chemists and industrial hygienists, who translated each job into a list of potential exposures using a checklist of 294 agents.

ACCEPTED HUMAN OCCUPATIONAL CARCINOGENS Several groups regularly convene expert panels to evaluate the carcinogenicity of agents, with the most long-​standing being the IARC, an agency of the World Health Organization. The IARC Monograph Program assembles interdisciplinary working groups to evaluate epidemiologic, experimental, exposure, and mechanistic data to classify agents, mixtures, or exposure circumstances as carcinogenic (Group 1), probably carcinogenic (Group 2A), possibly carcinogenic (Group 2B), not classifiable (Group 3), or probably not carcinogenic (Group 4) to humans. Although there are some critics of the IARC process who

 27

Table 16–1.  Agents, Occupations, and Industries Classified by IARC as Human Carcinogens (Group 1) That Are Primarily Occupational Exposures Agent

Cancer

Main Industry or Use

Acid mists containing sulfuric acid, strong inorganic 4-​Aminobiphenyl Arsenic and arsenic compounds Asbestos (all forms) Benzene Benzidine Benzo(a)pyrene

Larynx

Chemical

Bladder Lung, skin, bladder Larynx, lung, pleura, ovary Leukemia Bladder N/​A*

Beryllium and beryllium compounds Bis(chloromethyl) ether; chloromethyl methyl ether 1,3-​Butadiene Cadmium and cadmium compounds Chromium (VI) compounds Coal tar pitch 1,2-​Dichloropropane Diesel engine exhaust Erionite Ethylene oxide Fission products, including strontium-​90 Formaldehyde Gallium arsenide Ionizing radiation (all types, including radon-​222 Progeny)

Lung Lung

Rubber Glass, metals, pesticides Insulation, construction, renovation Chemical production, solvent, fuel Dye/​pigment manufacture Coal liquefaction and gasification, coke production, coke ovens, roofing, paving, aluminum production Aerospace industry, metals Chemical intermediate/​byproduct

Leather dust 4,4′-​Methylenebis (2-​chloroaniline) (MOCA) Mineral oils, untreated and mildly treated β-​Naphthylamine Nickel compounds 3,4,5,3′,4′-​Pentachlorobiphenyl (PCB-​126) 2,3,4,7,8-​Pentachlorodibenzofuran Polychlorinated biphenyls Polychlorinated biphenyls (“dioxin-​like”) Shale oils Silica dust, crystalline, in the form of quartz or cristobalite Solar radiation Soot Sulfur mustard 2,3,7,8-​Tetrachlorodibenzo-​p-​dioxin Tobacco, secondhand Ortho-​Toluidine Trichloroethylene Ultraviolet radiation Vinyl chloride Wood dust Occupation or Industry Acheson process Aluminum production Auramine, production of Coal gasification Coal-​tar distillation Coke production Hematite mining, underground Iron and steel founding Isopropyl alcohol manufacture using strong acids Magenta production Painter (occupation as) Rubber manufacturing industry

Leukemia, lymphoma Lung Nasal cavity, lung Skin, lung Biliary tract Lung Pleura N/​A* Leukemia, lymphoma Nasopharynx, leukemia N/​A* Thyroid, leukemia, salivary gland, lung, bone, esophagus, stomach, colon, rectum, skin, breast, kidney, bladder, brain Nasal cavity N/​A* Skin Bladder Nasal cavity, lung N/​A* N/​A* Skin N/​A* Skin Lung

Plastic, rubber Dye/​pigment manufacture, batteries Metal plating, dye/​pigment manufacture Construction, electrodes Chemical, paint stripping, printing Transport, mining Hydrocarbon cracking, agriculture Chemical, sterilant Nuclear workers Plastics, textiles Semiconductors Radiology, nuclear industry, underground mining

Skin Skin, lung Lung N/​A* Lung Bladder Kidney Skin, eye Liver Nasal cavity

Outdoor workers Chimney sweeps, masons, firefighters War gas Chemical Restaurant and bar workers, offices Pigment Solvent, dry cleaning Outdoor workers Plastic Wood, furniture

Leather industry Rubber Lubricant Pigment Metal alloy Electrical components Incineration, smelting, refining Electrical components Electrical components Lubricant, fuel Construction, mining

Cancer

Main Industry or Use

Lung Lung, bladder Bladder Skin, lung, bladder Skin Skin, lung, kidney Lung Lung Nasal cavity

Graphite and silicon carbide production Dye manufacture Dye manufacture Power plants, chemical industry Coke oven plants Coke oven plants Mining Iron and steel founding Chemical

Lung, bladder Lung, bladder Leukemia, lymphoma, bladder, lung, stomach

Painting Painting Rubber

Source: Benbrahim et al., 2014; Boyle and Levin, 2008; Grosse et al., 2014; Guyton et al., 2015; IARC, 2013, 2014a, 2014b, 2015c, 2015d, 2015e; Stewart and Wild, 2014. ** N/​A—​Not applicable (agent classified in Group 1 on the basis of mechanistic evidence).

278

278

PART III:  THE CAUSES OF CANCER

claim the reviews have not sufficiently taken into account potential methodological weaknesses in epidemiologic data in general and may be biased by working group members with vested interests, there is very broad involvement of the scientific community in the IARC Monograph Program, overwhelming support for the scientific rigor of the process, and ongoing international agreement with its decisions (Pearce et al., 2015) The IARC does not systematically indicate which carcinogens are occupational, and thus lists of occupational carcinogens may vary depending on the criteria used. Table 16–1 presents the agents classified by IARC as Group 1 carcinogens that we consider to be mostly occupational exposures. Note that occupational exposures can involve a broad range of infectious, medical, behavioral, and physical factors, in addition to the long-​recognized toxic chemicals. For example, chemotherapeutic medications are an occupational exposure for some healthcare workers. Medical radiation workers have excess leukemia and cancers of the skin and breast related to their occupational exposure (Linet et  al., 2010; Yoshinaga et  al., 2004). Tobacco, in the form of secondhand smoke, is an occupational exposure for bar and restaurant workers, resulting from the failure to ban smoking in these workplaces in some localities. Restrictions on smoking in public indoor areas have recently decreased this exposure in some areas and countries, but not all (Schmidt, 2007). Shift work adversely affects diurnal rhythm (IARC, 2010a). Physical inactivity and sedentary work are the norm in most modern workplaces (Brownson et  al., 2005). Sunlight is an occupational exposure for persons who work outdoors (Carey et al., 2014). Agents classified by the IARC as Group 1 carcinogens must have sufficient evidence for carcinogenicity in humans, or exceptionally limited evidence in humans but sufficient evidence of carcinogenicity in animals, along with strong evidence in exposed humans that the agent acts through a mechanism relevant to human carcinogenicity (IARC, 2015a). Ethylene oxide, which is used as an intermediate in the production of several industrial chemicals, as a fumigant, and as a sterilant for medical equipment and supplies, was classified as a carcinogen based on limited evidence from humans, sufficient evidence from animals, and compelling genotoxicity data from studies of exposed workers (IARC, 2012f). Similarly, mechanistic data on the presence of DNA adducts, sister chromatid exchanges, and micronuclei among workers exposed to 4,4'-​Methylenebis (2-​chloroaniline) (MOCA) resulted in its classification as a known human carcinogen (IARC, 2010b), despite limited human evidence. 2,3,7,8-​ Tetrachlorodibenzo-​p-​dioxin, which is a byproduct of the synthesis of chlorophenols or chlorophenoxy acid herbicides, waste incineration, metal production, and wood combustion, was also classified as a Group 1 carcinogen based on limited human data, sufficient animal data, and evidence that TCDD acts through a mechanism involving the Ah receptor, which is involved in normal cell homeostasis and is present in both humans and animals. Other occupational agents classified as carcinogens based on limited human evidence, but sufficient animal data and mechanistic data, include 2,3,4,7,8-​pentachlorodibenzofuran, “dioxin-​like” polychlorinated biphenyls, dyes metabolized to benzidine, and benzo(a)pyrene. Table 16–2 presents the occupational agents classified as probable carcinogens (Group 2A), generally based on limited evidence of carcinogenicity in humans and sufficient evidence of carcinogenicity in experimental animals. In some cases, agents are classified Group 2A based on lower levels of evidence of carcinogenicity in humans, but with strong evidence that the carcinogenesis is mediated by a mechanism that also operates in humans or that the agent clearly belongs, based on mechanistic considerations, to a class of agents for which one or more members have been classified as Group 1 or Group 2A. Agents on this list have considerable potential for human carcinogenicity, so those with widespread human exposure have a high priority for research to resolve the uncertainty in their classification (Ward et al., 2010). These lists (Tables 16–1 and 16–2) include some exposure scenarios that have been linked to cancer, but for which the specific agents responsible have not been identified. These include exposures associated with the Acheson process (synthesis of graphite and silicon

carbide), iron and steel founding, isopropyl alcohol production by the strong-​acid process, magenta production, working as a painter, frying (emissions from high temperatures), working as a hairdresser or barber, spraying and applying non-​arsenical insecticides, petroleum refining, and shift work that involves circadian disruption. The exposures associated with these processes and occupations could change over time or by geographic location, and change the cancer risk previously associated with them. For example, if the cancer risk associated with occupation as a hairdresser were due to exposure to hair dyes, changes in the formulations of hair dyes might cause the risk to change over calendar time or geographic region. Table 16–3 presents the most common types of occupational cancer; these are lung cancer, bladder cancer, mesothelioma, and leukemia, although occupational exposures have been linked to many cancer sites (IARC, 2015b). In 2008–​2009, the IARC held six working group meetings during which expert panels reviewed the evidence for substances and factors that the IARC had classified as known (Group  1) carcinogens in the past. This re-​review reaffirmed the earlier Group 1 classifications, and added new cancer sites for some agents. For example, formaldehyde had originally been classified in 2006 as a human carcinogen based on excess risk of nasopharyngeal cancer, but the recent review in 2009 added leukemia, particularly myeloid leukemia, as having sufficient evidence of carcinogenicity due to formaldehyde (IARC, 2012f). To date, the accumulation of additional evidence has never caused the IARC to downgrade an agent that was previously classified as a known carcinogen, although some agents classified as possible carcinogens (Group 2B, possibly carcinogenic to humans) have been downgraded to Group 3 (inadequate evidence), such as amitrole, atrazine, and ethylene thiourea.

EVIDENCE STIMULATING EPIDEMIOLOGIC STUDIES OF OCCUPATIONAL CANCER There are several types of evidence that by themselves have suggested the presence of occupational cancer hazards and have stimulated epidemiologic studies of occupational exposures or groups. These informative approaches include investigations of cancer clusters, searches for explanations of unusual geographic patterns, whole animal cancer bioassays, Ames testing, and animal and human biomarker and mechanistic studies. A cancer cluster is the apparent excess of cancer cases among a relatively small group of people. As noted earlier, occupational cancer clusters have led to the identification of carcinogens and in-​depth epidemiologic studies dating back centuries. In the modern era after World War II, a plant physician noted three cases of angiosarcoma of the liver among workers exposed to vinyl chloride monomer in a polymerization operation at a chemical plant (Creech and Johnson, 1974). The extremely rare nature of angiosarcoma, and the fact that three cases came from the same workplace, contributed to the ability of the physician to recognize the excess cancer incidence. Epidemiologic investigations confirmed the excess of angiosarcoma of the liver and discovered that vinyl chloride also caused hepatocellular carcinoma, a more common cancer that would be less likely to be noted as excessive among an occupational group without a formal study (IARC, 2012f). A similar story of discovery occurred in 2013 with the observation of 11 male patients with intrahepatic or extrahepatic biliary tract cancer (cholangiocarcioma) among workers exposed to 1,2-​dichloropropane among 62 male workers at a printing company in Japan (Kumagai et al., 2013). The rarity of cholangiocarcioma, the very high relative risk, the young ages of the patients, the absence of non-​occupational risk factors, and the intensity of the exposure, along with sufficient evidence of carcinogenicity among experimental animals, led the IARC to classify 1,2-​dichloropropane as a known human carcinogen (Benbrahim-​Tallaa et al., 2014; Kubo et al., 2014; Kumagai et al., 2013). Geographic patterns of cancer occurrence have provided clues to occupational carcinogens. Publication of the first US cancer mortality atlas in 1975 revealed a string of counties along the southeast Atlantic

 279

279

Occupational Cancer

Table 16–2.  Agents, Occupations, Industries, and Occupational Circumstances Classified by IARC as Probable Human Carcinogens (Group 2A) That Are Primarily Occupational Exposures Agent Acrylamide Bitumens (combustion products during roofing) Captafol α-​Chlorinated toluenes 4-​Chloro-​o-​toluidene Cobalt metal with tungsten carbide Creosotes Diazinon Dibenz[a,j]acridine Dichloromethane (methylene chloride) Diethyl sulfate Dimethycarbamoyl carbide 1.2-​Dimethylhydrazine Dimethysulfate Epichlorhydrin Ethylene dibromide Glycidol Glyphosate Indium phosphide Lead compounds, inorganic Malathion Methyl methanesulfonate 6-​Nitrochrysene 1-​Nitropyrene 2-​Nitrotoluene Non-​arsenical insecticides Polybrominated biphenyls 1,3-​Propane sultone Silicon carbide whiskers Styrene-​7,8-​oxide Tetrachloroethylene Tetrafluoroethylene 1,2,3-​Trichloropropane Tris(2,3-​dibromopropyl)phosphate Vinyl bromide Vinyl fluoride Occupation, Industry, or Occupational Circumstance Art glass, glass containers, and pressed ware, manufacture Carbon electrode manufacture Food frying at high temperature Hairdresser or barber Petroleum refining Shift work that involves circadian disruption

Cancer

Main Industry or Use

–​ Lung –​ –​ Bladder Lung Skin Lymphoma, lung –​ Biliary tract, lymphoma –​ –​ –​ –​ –​ –​ –​ Lymphoma –​ Lung, stomach Lymphoma, prostate –​ –​ –​ –​ Leukemia –​ –​ Lung –​ Liver, lymphoma –​ –​ –​ –​ –​ Cancer Lung, stomach Lung Lung Bladder, lung Leukemia, skin Breast

Plastics Roofing Pesticide Pigment, chemical Pigment, textile Hard metal production Wood Agriculture, pest control Mastic asphalt worker Plastics, solvent, paint stripping Chemical Chemical Research Chemical Plastic Fumigant Pharmaceutical industry Agriculture, forestry Semiconductors Metals, pigments Agriculture, pest control Chemical Transportation, underground mining Transportation, underground mining Dye production Agriculture Flame retardants, plastics Chemical, batteries Plastics Plastic Solvent, dry cleaning Chemical Solvent Plastic, textile Plastic, textile Chemical Main Industry or Use Glass manufacture Carbon electrode manufacture Restaurant Cosmetology Petroleum refining Health care, manufacturing, custodial work

Source: Benbrahim et al., 2014; Boyle and Levin, 2008; Grosse et al., 2014; Guyton et al., 2015; IARC, 2013, 2014a, 2014b, 2015c, 2015d, 2015e; Stewart and Wild, 2014.

coast with markedly elevated rates of lung cancer (Mason et al., 1975). A series of case-​control studies conducted in the areas found that the excess lung cancer was due to asbestos exposures associated with short-​term shipyard work that took place largely during World War II (Blot et al., 1978, 1979b, 1980; Tagnon et al., 1980). Many of the shipyards had been closed for years. In recent time, a common motivation for an epidemiologic study of a suspect occupational carcinogen is demonstration of cancer among experimental animals. Animal bioassays have characteristics that must be considered when extrapolating the results to humans, such as animal/​human species metabolic differences, typically greater doses to animals than are experienced by humans, different routes of exposure, fewer concomitant exposures that could affect cancer risk, and some target organs that may not exist in humans (e.g., the rodent Zymbal glands) or that rarely develop cancer among humans (e.g., the pituitary gland). However, despite these limitations, the results of long-​term cancer bioassays are highly predictive for identifying human carcinogens,

and the overwhelming majority of established human carcinogens have also been shown to cause cancer in animals (Huff, 1999, 2008). There are many chemicals that were found to cause cancer in animals before epidemiologic studies demonstrated effects in humans. Some notable carcinogens first identified in animal studies and later found to cause cancer among exposed workers are dioxin (IARC, 1997, 2012f), 1,3-​ butadiene (Melnick and Huff, 1992), and formaldehyde (IARC, 2006). Concordance overall between animal and human carcinogens, as well as concordance of the specific affected sites, has improved with better bioassay methods, such as improved histopathology with more detailed examination of more organs per test animal, use of various routes of administration, and different forms of the compounds (Maronpot et  al., 2004). For example, arsenic was recognized as a human carcinogen in 1987, while animal bioassays were negative or provided limited evidence until many years later, when organic metabolites of arsenic were studied and found to be positive (Huff et  al., 2000; IARC, 2012c).

280

280

PART III:  THE CAUSES OF CANCER

Table 16–3.  IARC Human Carcinogens (Group 1) That Are Primarily Occupational Exposures by Cancer Site Cancer Site Bladder

Bone

Table 16–3. Continued Cancer Site

Occupational Exposure

Occupational Exposure Aluminum production 4-​Aminobiphenyl Arsenic and inorganic arsenic compounds Auramine production Benzidine Magenta production 2-​Naphthylamine Painter (occupational exposure as) Rubber production industry Ortho-​Toluidine X-​ and γ-​radiation Plutonium Radium-​224 and its decay products Radium-​226 and its decay products Radium-​228 and its decay products X-​ and γ-​radiation

Breast

X-​ and γ-​radiation

Eye

Welding

Gallbladder and biliary tract

1,2-​Dichloropropane

Kidney

Trichloroethylene X-​ and γ-​radiation

Larynx

Acid mists, strong inorganic Asbestos

Leukemia/​Lymphoma

Benzene 1,3-​Butadiene Fission products, including strontium-​90 Formaldehyde Rubber production industry X-​ and γ-​radiation

Liver

Plutonium

Lung

Acheson process Aluminum production Arsenic and inorganic arsenic compounds Asbestos Beryllium and beryllium compounds Bis(chloromethyl)ether; chloromethyl methyl ether (technical grade) Cadmium and cadmium compounds Chromium (VI) compounds Coal gasification Coal-​tar pitch Coke production Diesel engine exhausts Hematite mining Iron and steel founding Nickel compounds Painter (occupational exposure as) Plutonium Radon-​222 and its decay products Silica dust, crystalline Soot Sulfur mustard Tobacco smoke, secondhand X-​ and γ-​radiation

Nasal cavity and paranasal sinus

Isopropyl alcohol manufacture using strong acids Leather dust Nickel compounds Radium-​226 and its decay products Radium-​228 and its decay products Wood dust

Nasopharynx

Formaldehyde Wood dust

Ovary

Asbestos

Pleura

Asbestos Erionite Painter (occupation as)

Skin

Arsenic and inorganic arsenic compounds Coal-​tar distillation Coal-​tar pitch Mineral oils, untreated or mildly treated Polychlorinated biphenyls Shale oils Solar radiation Soot X-​ and γ-​radiation

Stomach

Rubber production industry

Source: IARC, 2015b, 2015f.

Measures of exposure and biomarkers of effects in humans can also be critical in assessing and understanding carcinogenesis. They have also stimulated occupational epidemiologic investigations. Early reports of chromosomal damage from ethylene oxide (Ehrenberg and Hallstrom, 1967), a sterilant gas, were instrumental in leading to two epidemiologic studies, which documented excess leukemia (Hogstedt et  al., 1979a; Hogstedt et al., 1979b). Animal studies subsequently showed that inhaled ethylene oxide caused leukemia in rodents (Snellings et al., 1984).

DESIGN OF HUMAN STUDIES The role of occupational exposures in the development of cancer has been investigated using various epidemiologic designs, including cohort, case-​control, cross-​sectional, and ecological studies (Checkoway et al., 2007; Steenland and Moe, 2016). The type of study design has a major impact on the type and source of information available on occupational exposures as well as potential bias, and thus on the quality of the data generated by the study and the conclusions that can be drawn.

Cohort Studies In a cohort study, typically exposed and non-​exposed groups are identified and followed over time to ascertain disease outcomes (although sometimes highly exposed workers are compared to lower exposed workers in an internal analysis). The occurrence of disease among the two groups is compared. For many occupational studies, the “non-​ exposed” group has been the general population of the country or some other political entity in which the exposed group is located. Such a comparison group is typically selected for practical reasons (i.e., information on the incidence or mortality experience is already available, and no specific unexposed population can be assembled without considerable effort and cost). Although use of a non-​worker referent population has practical advantages, it has a methodological limitation known as the healthy worker effect. This occurs when the exposed population (i.e., workers) is made up of relatively healthy individuals, while the referent group includes non-​working people who are not employed because of health problems, so the two groups are not comparable (Fox and Collier 1976). The healthy worker effect can take two forms; one involves the selection of healthy workers into the workforce, and a second involves healthy workers staying in the workforce while others who develop illnesses leave the workforce early (the latter is called the healthy worker survivor effect and is important when considering cumulative exposures). The healthy worker effect can obscure an increase in disease among workers due to an occupational exposure, which may be offset by the generally healthy nature of the working population. The healthy worker effect is a particular problem for diseases that are disabling and which may occur relatively early in life, such as cardiovascular disease, but it is not entirely absent for cancer. However, the healthy worker effect wears off

 281

Occupational Cancer with further follow-​up, when more of the cohort is no longer working, which is when most cancers occur. When a referent group of workers is available, its use not only avoids the healthy worker effect, but also has the advantage that it is likely to be similar to the exposed population in other often unmeasured ways (e.g., sharing lifestyle factors such as smoking habits). Thus it can help limit unmeasured confounding. A cohort study can assess multiple outcomes (e.g., all types of cancer), but unless the cohort is extremely large, the number of subjects with rare diseases will be small. The cohort design is particularly useful for the study of occupational exposures that would otherwise be uncommon in most samples of the general population. An increasing challenge, particularly in the United States, is gaining access to workplaces and to the historical occupational and exposure information necessary to conduct cohort studies. In the United States, until recently most occupational cohort studies ascertained mortality only, not cancer incidence, which means that some associations with cancers that have good survival (e.g., bladder, prostate, and breast cancer) may have been missed. With better coverage of the United States by cancer registries, more incidence-​ based occupational cohorts are appearing, such as the study of US firefighters (Daniels et al., 2014) and the Agricultural Health Study (Blair et al., 2015b). Countries with national cancer registries, such as Scandinavian countries, have long been able to study cancer incidence among workers, such as the studies of dry cleaners in Denmark, Norway, Sweden, and Finland (Lynge et al., 2006). A retrospective cohort study identifies the exposed and unexposed groups (either some general population or a specific unexposed worker population) at some point in the past and follows them to the present, while a prospective cohort study identifies the two groups in the present time and follows them into the future. In general, most occupational studies have been retrospective in nature, which must rely upon historically collected exposure information, but provide results more quickly than prospective studies, which must wait for cancers to develop in the population. There are some notable exceptions to the almost exclusive use of retrospective cohort studies to assess occupational carcinogens (Blair et al., 2015a), including the prospective cohort study of cancer among Swedish construction workers that began in the mid-​1960s (Bergdahl and Järvholm, 2003; Järvholm and Englund, 2014), the US Agricultural Health Study (Alavanja et al., 1996), the French Agricultural and Cancer (AGRICAN) study (Levêque-​Morlias et al., 2015), studies of Chernobyl cleanup workers (Kesminiene et al., 2002; Romanenko et al., 2008), and the recently launched World Trade Center Rescue and Recovery Workers Study (Solan et  al., 2013)  and the GuLF Study (Sandler et  al., 2014). Some prospective cohort studies that were initiated primarily to investigate non-​occupational factors have been utilized to evaluate cancer and occupational cancers as well, such as the Nurses’ Health Study, Shanghai Women’s Study, EPIC study, and Sister Study (Blair et al., 2015a). A limitation of most retrospective cohort studies is the lack of information on important potential confounders, such as occupational exposures from other jobs besides the one under study and lifestyle factors such as smoking and alcohol consumption and diet. However, for smoking, which is usually the potential confounder of most concern, a considerable amount of research based on studies with smoking data, and based on hypothetical simulations, has been published which shows that smoking rarely confounds disease risks in occupational studies, and changes of effect measures (e.g., rate ratios) greater than 20% are extremely unlikely (Axelson and Steenland, 1988; Blair et al., 2007). It is still preferable to collect smoking data, to control for possible confounding, but also to assess if smoking modifies the effect of an occupational exposure, as it does for asbestos (IARC, 2012e) and diesel exhaust (Silverman et al., 2012). For example, smokers who are also exposed to asbestos have a risk of developing lung cancer that is greater than the individual risks from asbestos and smoking added together, but less than the risks multiplied together (IARC, 2012e).

Proportional Mortality Studies A variation on the cohort design in which a defined population is followed over time is a proportional mortality study, which includes

281

observations on deaths only. Full information on the occupational group from which the deaths arose is not available. Without enumeration of the population at risk, it is not possible to calculate incidence or mortality rates and compare them to an unexposed population. Instead, the proportion of deaths due to a specific cause out of the total number of deaths among the workers is compared to the proportion in a comparison population, often the general population. The statistic used to express the results, the proportional mortality ratio, is the origin of the common name for this design, the PMR study. This approach is most common when a company, union, government agency, or insurance company may have death certificates available for study, but no records on the identity or work histories for all persons who worked in the job of interest, regardless of their current employment or vital status. Examples include the NIOSH National Occupational Mortality Surveillance (NOMS) system, which has occupation and industry-​ coded death data from 30 US states from over a 20-​year time period (Robinson et  al., 2015), a study of deaths among unionized plumbers, pipefitters, and allied trades (Lehman et al., 2008), and a study of deaths among cement workers recorded by a Greek insurance company (Rachiotis et al., 2012). Proportional mortality studies are generally inexpensive and quick to conduct; they have been used as a surveillance tool for routine detection of any occupational excess and as an inexpensive initial approach to provide some information on a new issue. However, they may be biased if the death certificates on hand are not representative of all deaths that occurred in the occupational group. Often such studies are based on deaths among active workers or retirees from the company or union under study, but miss deaths among persons who left employment with that company or union before death or before becoming eligible for retirement. Also, because the proportion of deaths for all causes must total 100%, a higher proportion of deaths from one cause must be offset by a lower proportion of deaths from another cause, or vice versa, which can distort results. For example, if aerial pesticide applicators have a large real excess of deaths due to accidents, the mathematical requirement that the PMR for all causes combined equal 1.00 may cause some one or other causes of death to appear deficient, even if their absolute mortality rates are excessive compared to an unexposed population.

Case-​Control Studies In a case-​control study, persons with the disease of interest (“cases”) and persons without the disease (“controls”) are identified and past exposures are ascertained. The relative occurrence of exposures among cases and controls is compared. Case-​control studies are of three types: population-​based, clinic or hospital-​based, and nested within a cohort. A population-​based case-​control study is the optimal approach for the study of rare cancers, which would be infrequent in a cohort study, and allows for ascertainment of important lifestyle factors, such as smoking, that are not typically recorded in work records, upon which most occupational cohort studies rely. Case-​control studies can assess multiple exposures (e.g., all past occupations). Generally, these studies include subjects with a very wide range of jobs, each held by a small number of persons (except for a few common occupations), which can limit the power to evaluate rare exposures and hence limit the usefulness of this approach for studies of occupational exposures. Case-​control studies can sometimes restrict the study area to enhance the proportion of the population exposed, for example, studies of pesticides have been located in rural states, or portions of states, to increase the proportion of farmers who use pesticides (e.g., Miligi et al., 2003; Zahm et al., 1990). Most case-​control studies conduct interviews with study subjects, and rely upon self-​reported occupational histories. Workers are usually able to report job title, industry, and dates of employment with high validity, but often not the actual exposures sustained on the job (Teschke et  al., 2002). Workers in some occupations do better than others. Farmers, for example, can provide quite accurate information about pesticides used and conditions of handling and application (Blair et al., 2002). However, for many case-​control studies, specific

28

282

PART III:  THE CAUSES OF CANCER

exposures must be inferred from the job history through assessment by experts, use of a job-​exposure matrix, or, rarely, biological measurements (Teschke et al., 2002) (see the section “Exposure Assessment” later in this chapter). Nested case-​control studies of occupational carcinogens are case-​ control studies conducted within a cohort, and cases are compared to a set of non-​cases in the cohort. More extensive information on the occupational exposure or important potential confounders or effect modifiers can be more easily obtained for the cases and a set of controls than for the full cohort (Rothman et al., 2008).

Cross-​Sectional Studies In cross-​sectional studies, the exposure and cancer come from approximately the same point in time. The major limitation of this approach is that the temporal sequence of the exposure and the disease may be difficult, if not impossible, to determine. For cancer, which has a long latency, the assumption that current exposure status reflects exposures when the disease process was initiated may be particularly problematic. Despite this limitation, a series of informative studies of the correlation between the concentration of various industries with cancer mortality in US counties yielded clues about occupational carcinogens, and were followed up with case-​control and cohort studies (Blot et al., 1979a). Cross-​sectional studies are often used in the study of biomarkers related to occupational exposures. For example, cadmium in the urine of smelter workers has been measured at the same time as urinary proteins beta-​2-​microglobulin and retinol binding protein showing kidney damage (Thun, 1992). Benzene levels in air and urine of exposed workers were correlated with measures of hematotoxicity, demonstrating its impact at air levels below the US occupational standard of 1 part per million, and among genetically susceptible subpopulations (Lan et al., 2004). A cross-​sectional study that showed formaldehyde-​ associated hematotoxicity among exposed workers was cited as a key justification for the IARC’s classification of formaldehyde as a human leukemogen (IARC, 2012f; Zhang et al., 2010). Cross-​sectional studies of biomarkers can shed light on potential mechanisms, identify early events in the natural history, improve exposure assessment, and, in the case of molecular biomarkers, identify individuals susceptible to disease (Mayeux, 2004; Schulte, 1993). However, a cross-​sectional study may be affected by reverse causation or disease bias, that is, the biomarker may be the effect of the disease rather than a cause (Checkoway et al., 2007).

Exposure Assessment Although studies relating occupation and cancer are informative in suggesting leads to occupational hazards, eventually investigations must focus on specific exposures to provide solid information for hazard and risk assessment. Quantitative occupational exposure assessment is important in evaluating potential carcinogens and critical in risk assessment. The data sources and approaches to develop such quantitative assessments differ in industry-​based and population-​based epidemiologic studies (Friesen et al., 2015). Industry-​ based studies (generally cohort studies) are typically focused on one or a few agents in one or a few types of occupational settings. It is important to remember, however, that although the focus of the research may be on “one or a few” exposures, there are very few workplaces where is there is such a small number of exposures. The subjects’ job histories are usually obtained from preexisting company or union records. With the notable exception of radiation workers, personal exposure measurements for a large proportion of the subjects in retrospective cohort studies are not usually available. There may be some measurements for a few past workers in some settings and area measurements at various work locations, but generally exposure measurements for the time period critical for cancer initiation are limited. To overcome this limitation, researchers collect information from a variety of sources to reconstruct historical exposures by task, job,

department, and time period, and then use that information to estimate exposures experienced by individuals in the study (Friesen et al., 2015). Data sources for such estimates include any existing personal and area industrial hygiene sampling data (noting if the samples were taken for routine monitoring or in response to a known or suspected high-​exposure excursion), descriptions of the industrial process and changes over time, production levels, exposure control devices and efforts, and interviews with longtime workers with knowledge of historical conditions. Experts then use these data to develop qualitative or quantitative estimates of exposure by job, department, and time. The exposure data may incorporate level and probability of the exposure, as well as the expert’s confidence in the assessment. Reliability and validity assessments are sometimes performed to provide an indication of the reliability and accuracy of the exposure assessment, such as in studies of diesel exposure (Stewart et al., 2010) and of ethylene oxide (Hornung et al., 1994). Population-​ based studies (either case-​ control or cohort studies) typically include a wide variety of occupations and industries, with a distribution representative of the general population. With the exception of a few common jobs, the number of subjects holding any specific job can be small. In population-​based case-​control studies using interviews, self-​reported occupational histories have good reliability for job, industry, and dates of employment, but poorer reliability for reporting specific exposures (Friesen et al., 2015). Checklists of exposures may help improve reporting of the job history itself, as well as exposures of interest, but the data are prone to false positives and false negatives. Reporting is somewhat better for exposures that are easily sensed and recognizable, or if the person was involved with the decision-​making about the exposure, such as a farmer who buys, applies, and tracks pesticide use. For many exposures, however, self-​reported exposure histories can have serious limitations and may be affected by case recall bias in case-​control studies. Even more problematic than self-​reports are proxy reports. Spouses and children provide less accurate job and exposure histories than the subjects themselves (Brown et al., 1991). Coworkers have provided job and exposure information for deceased subjects in some studies (e.g., Chernobyl cleanup workers [Romanenko et  al.,  2008]). Generic JEMs (vs. industry-​specific JEMs described earlier) can be used to assign exposure status based on the occupational histories, although one must consider possible differences across geographic areas before applying a JEM developed in one country to a study conducted in another (Friesen et al., 2015; Hoar et al., 1980). In some studies, detailed sets of questions are triggered if the subjects have held jobs for which additional data are needed to assess the potential for exposures of interest. These “occupational modules” gather information on materials and equipment used, sensory descriptions (e.g., smell), dermal exposure, work practices, engineering controls, and personal protective equipment used to improve the accuracy of the assessment of exposure potential and amount at the individual subject level (Siemiatycki et al., 1997; Stewart et al., 1998) Expert reviews and exposure assignments are time-​ consuming, expensive, not transparent, and may not always be reproducible, so more systematic approaches are being developed (Fritschi et al., 2009; Russ et al., 2014, 2016). Current efforts that aim to overcome these limitations include the development of algorithm-​and measurement-​ based decision rules, applying machine learning methods to classification tree models, computerized decision rules (e.g., OccIDEAS: http://​ www.occideas.org/​#!whousing/​c1ykh) (Fritschi et  al., 2009), and computer-​assisted occupational coding software (e.g., SOCcer: http://​ soccer.nci.nih.gov) (Russ et al., 2014, 2016). In many settings, it may be important to consider multiple exposures simultaneously to assess interaction among exposures, to evaluate possible confounding, and to clarify the credibility of associations. De Roos et al. (2003) performed an integrative assessment of multiple pesticides as risk factor for non-​Hodgkin’s lymphoma. Consideration of mixtures is important to evaluate if there are enhanced effects from a mixture, which is greater than that expected from single exposures; this has been reported in a few studies for pesticides (De Roos et al., 2003; Hohenadel et  al., 2011; Lerro et  al., 2015). In contrast, there

 283

Occupational Cancer might be diminished effects from combinations of exposures, such as the observation that the risk of adenocarcinoma of the lung associated with diesel exposure was ameliorated in presence of endotoxin exposure (Tual et al., 2015). Attention must also be given to consideration of gender differences that may affect exposure assessment (Friesen et al., 2012; Locke et al., 2014). Men and women with the same job title do not always have the same duties and exposures. There are also gender differences in self-​ reports of exposures, work practices, personal hygiene, and the effectiveness of protective equipment and engineering controls that affect the accuracy of exposure assessment (Zahm and Blair, 2003). It would be ideal if it were possible to use a biomarker of exposure, which would characterize the total internal dose rather than external exposure. However, biomarkers of exposure are often short-​lived in the body. To fully characterize exposure over the years would require multiple biologic collections, or there would need to be (1) limited intra-​individual variation of the biomarker over time, (2) an accurate laboratory assay, and (3) long enough persistence in the body so that the measure could reflect an exposure that occurred during the time period when carcinogenesis was initiated. Furthermore, for case-​control studies, the biomarker would need to be not affected by disease status or treatment (Baris et  al., 2000; Steenland and Moe, 2016). Nonetheless, biomarkers can often help in the assessing the relationship between historical exposures and dose.

Genetic Susceptibility Because occupational studies are likely to have better assessment of work-​related exposures than studies of many other types of exposures, they offer excellent opportunity to explore potential interactions with genetic susceptibility. For example, using the candidate gene approach, it was established that the slow N-​acetylation phenotype, caused by variation in the NAT2 gene, confers increased risk for bladder cancer among workers exposed to aromatic amines (Gu and Wu, 2011). Technological advances have enabled researchers to look at extremely large numbers of genetic variants in an agnostic fashion in genome-​wide association studies (GWAS), accelerating the search for gene-​environment interactions. A small genome-​wide study of gene–​environment interaction for asbestos exposure in lung cancer susceptibility has been conducted with some promising results, particularly when a pathway approach was used, but more research with a larger number of subjects is needed (Wei et al., 2012). Some recent examples of possible gene-​exposure interactions include a large GWAS study of bladder cancer that found evidence of interaction for rs798766 (TMEM129-​TACC3-​FGFR3) with exposure to straight metalworking fluids, with the interaction more apparent among patients with tumors positive for FGFR3 expression (Figueroa et  al., 2015). Gene–​environment interactions between prostate cancer susceptibility loci identified via GWAS and occupational exposure to pesticides have been identified among applicators in the Agricultural Health Study (Koutros et  al., 2010, 2013). In addition to genomics, the occupational setting can be a fertile ground for the application of epigenomics, transcriptomics, proteomics, and metabolomics and other “-​omics” technologies (Vlaanderen et al., 2010). In summary, occupational cancer epidemiology is a powerful tool for (1) protecting workers’ health through identifying and reducing exposure to carcinogens; (2) protecting the public by identifying and reducing exposure to occupational carcinogens that otherwise would enter the non-​workplace environment or consumer products; and (3) elucidating mechanisms of carcinogenesis.

QUANTITATIVE RISK ASSESSMENT Evaluation of the carcinogenicity of occupational exposures consists of not only identification of carcinogens (qualitative risk assessment), such as that done by the IARC Monograph Program, but also of quantitative risk assessment, which is the characterization of the probability of human cancer resulting from specific levels of exposures to

283

environmental hazards. Quantitative risk assessment is typically conducted by government regulatory agencies. Quantitative risk assessment may be based on either animal data or human data. The former has an advantage in that exposures in observational human studies are not assigned as in animal experiments, and thus there can be considerable uncertainty in exposure estimates. However, animal studies require extrapolation from animals to humans and hence involve much more uncertainty. For this reason, human (epidemiological) data are generally preferred, but are not always available. When human data are available, occupational quantitative risk assessment, aimed at setting permissible exposure levels, is based on exposure–​response data from one or more occupational studies. These results provide information on the increased rate of disease per unit of exposure (exposure–​response data) for an exposed population that must typically be converted to the excess risk of disease over a lifetime for specific exposures to identify a permissible exposure level. The US Occupational Safety and Health Administration (OSHA) usually seeks to limit excess risk to 1 in 1000 deaths (or serious disease) for exposed populations. For example, if the background lifetime risk for lung cancer in the non-​exposed general population is 5% (0.05), then OSHA will seek the level of exposure that will not increase lifetime risk for an exposed population beyond 5.1%. Other countries have adopted similar approaches for carcinogens that are genotoxic (Nielsen and Ovrebø, 2008). There are two major issues of concern for quantitative risk assessment based on epidemiological exposure–​response models. The first is the shape of the exposure–​response curve. When data are sparse, and sometimes even when they are not, it may be difficult to choose among competing models used for setting permissible limits, and different models can have very different consequences. Typical questions involving model selection might be whether the exposure–​response relationship shows a linear increase in disease rate per unit of exposure, whether there is a threshold below which there is no risk followed by an increase, or conversely, whether there is a cut-​point above which disease risk begins to flatten out or even decrease. In general, one expects that greater exposure to a carcinogen would lead to a greater occurrence of cancer. However, risk for some established carcinogens plateaus at higher doses; examples include ethylene oxide, dioxin, beryllium, diesel fumes, and nickel (Stayner et  al., 2003; Steenland et al., 2011). Plateaus in disease rate at higher exposure levels in occupational studies may occur due to (1) bias introduced by the healthy worker survivor effect; (2) depletion of the number of susceptible people in the population at high exposure levels; (3) a natural limit on the relative risk for diseases with a high background rate; (4) inaccurate measurements or misclassification of exposure; (5) influence of other risk factors that vary by the level of the main exposure (confounding); and (6) saturation of key enzyme systems or other processes involved with the development of disease (Stayner et al., 2003). As an example of a biological process that affects the shape of the dose–​response curve, high doses of radiation result in cell death, while lower doses damage cells and result in malignant transformation, so risk for thyroid cancer, but not other sites, plateaus and may even decrease with increasing dose (Berrington de González et al., 2013). A second question typical of quantitative risk assessment is the nature of the exposure–​response relationship in the low-​dose region, where there may be few data. This question often arises when occupational epidemiological studies (high exposure) are used for risk assessment for general environmental exposures (lower exposure), such as diesel fumes, dioxin, or asbestos. Traditionally, when data are sparse in the low-​dose region, a linear extrapolation of risk down to zero exposure is used by regulatory agencies, starting from a point of departure that is a function of the lowest level of risk for which observed data are considered reliable. This procedure may also provide the best fit to data in which the exposure–​response relationship plateaus at higher exposures (Steenland et al., 2011). By way of example of quantitative risk assessment, consider a recent occupational cancer risk assessment for lung cancer and diesel fumes (Vermeulen et  al., 2014). These authors took three occupational studies of workers exposed to diesel fumes (two truck driver studies, one miner study), which were the only epidemiologic studies

284

284

PART III:  THE CAUSES OF CANCER

with quantitative exposure–​response data using the same exposure metric (cumulative exposure to elemental carbon to characterize diesel exhaust), therefore permitting a joint analysis. They combined the three studies to derive a common exposure–​response curve, which indicated that for each 1µg/​m3-​year of elemental carbon exposure, the log RR (relative risk) for lung cancer would increase by 0.00098; this translates to an RR of about 2 for 700 µg/​m3-​years, which in turn to an average exposure of about 35 µg/​m3 elemental carbon for someone exposed for 20 years (the miners had an average exposure of about this level). Next, the authors translated this finding into an estimate of excess lifetime risk for death from lung cancer (above background lifetime risk for an individual not occupationally exposed to diesel fumes, which is about 5%), at different exposure levels over a hypothetical 45-​year working lifetime. US OSHA uses the 45-​year working lifetime as a default assumption for setting permissible levels for workers. Exposure for 45 years at 1 µg/​m3 (45 ug/​m3-​years) would be expected to lead to an excess lifetime risk of 1.7/​1000 (i.e., increasing a background risk of 0.0500 to 0.0517). Given that OSHA seeks to keep excess lifetime risks below 1/​1000, this would argue for a permissible exposure level for workers of slightly less than 1 µg/​m3. Risk assessors must then weigh the optimal standard from a health standpoint versus economic feasibility. Currently, with the exception of miners, exposed workers have diesel exhaust exposures in the range of 3–​13 µg/​m3, while members of the public in urban areas are exposed to about 0.8 µg/​m3 (Vermeulen et al., 2014).

ATTRIBUTABLE FRACTION To estimate the impact of occupational cancer, we can calculate the attributable fraction (AF) of cases or deaths for a specific cancer (or all cancers) due to occupation; this is sometimes also called the attributable risk. The AF is calculated by AF = [(RR – ​1)p]/​[(RR – ​1) •  p + 1], where RR is the relative risk of disease for the exposed versus non-​exposed, and p is the proportion of the population exposed. The proportion of the population exposed includes those exposed currently and those exposed in the past. An overall RR for exposed versus non-​exposed and an overall proportion of the population exposed are commonly used. However, sometimes the population exposed is broken down into different levels and/​or duration of exposure, and RRs specific by level/​duration of exposure are used to calculate separate AFs by exposure level or duration. There is considerable judgment involved in these calculations; for example, if one wants to calculate the AF for all cancers due to occupation, one must choose which cancer/​occupations associations are to be included. RRs for some specific cancers and occupations/​exposures may be based on one large and very well-​conducted study, but more often use a meta-​analysis of existing studies to determine an average RR across studies. Often, authors restrict themselves to occupational carcinogens considered definite or probable according to IARC (Group 1 or 2A). A common difficulty is that estimates of the proportion of the population exposed are not routinely available. AFs can be calculated for both incident cancers and for fatal cancers. Further calculations can be done to estimate disability-​adjusted life years lost (DALYs). DALYs attempt to quantify years lost due to premature death, or years lived with lower quality of life following a disease. They are a weighted estimate of the number of years lived with disability, where the weighting is based on the severity of the disability. There have been a number of estimates of the AF for occupational cancer in the United States and other countries. One of the first and perhaps the best known was published by Doll and Peto in 1981, in which they estimated that 4% (1.2% for females, 7% for males) of US cancer deaths were due to occupational cancer, primarily from lung cancer and with the principal occupational exposure being asbestos (Doll and Peto, 1981). Doll and Peto noted that this estimate could be off by a factor of two in either direction. It is important to put the estimate of AF of 4% for occupational exposures in perspective with other cancer risk factors. Doll and Peto (1981) estimated that tobacco had an AF of 30% and diet of 35%, but all others were generally in the range estimated for occupational exposures, for example, infections (10%), reproductive/​sexual

behavior (7%), geophysical factors (3%), alcohol (3%), and medicines (1%). In addition, the risk from occupational exposures is borne largely by blue-​collar workers who represented only about 25% of the US population in 2000 (those in the construction, extraction, maintenance, transportation, and material moving industries; see https://​www.census.gov/​prod/​2003pubs/​c2kbr-​25.pdf); thus the AF among blue-​collar workers must be considerably larger than 4%. Doll and Peto’s original estimate has stood up relatively well to the test of time (Blot and Tarone, 2015). Since then, there have been a number of other estimates in the United States and other countries, as more information accumulated on the relative risks of cancer from specific agents, and as the prevalence of occupational exposure has changed. Steenland et al. (2003) estimated that 2.3%–​4.8% of US cancer deaths were attributable to occupational factors (0.8%–​1.0% among females, 3.3%–​7.3% among males), which accounted for 13,000 to 26,000 cancer deaths annually. The highest AFs were for lung cancer, estimated to be between 6% and 17% for men, and to be about 2% for women. To give a more recent example of an AF for a specific agent and a specific cancer, we can use the example of diesel fumes and lung cancer discussed earlier (Vermeulen et al., 2014). These authors estimated that at current exposure levels to diesel fumes, the occupational exposure is responsible for about 1% of lung cancer deaths in the United States and England, and another 5% of lung cancer is due to environmental exposures to diesel fumes. AFs from other countries and worldwide are not dissimilar. For example, Rushton et al. (2012) estimated that 5.3% (8.2% men, 2.3% women) of cancer deaths in England in 2005 were due to occupational exposures. The principal cancer sites considered were mesothelioma, sinonasal, lung, nasopharynx, breast, non-​ melanoma skin cancer, bladder, esophagus, soft tissue sarcoma, and stomach. The principal carcinogens were asbestos, mineral oils, solar radiation, silica, diesel engine exhaust, coal tars and pitches, dioxins, environmental tobacco smoke, radon, tetrachloroethylene, and arsenic, as well as occupational circumstances such as shift work and occupation as a painter or welder. Recent estimates on a global scale were made by the Global Burden of Disease (GBD) 2013 group (GBD 2013 Risk Factors Collaborators, 2015), which estimated that IARC Group  1 (definite) occupational carcinogens accounted for about 3.7% of worldwide cancer deaths in 2013 (304,000 deaths). More detailed information available from the GBD website (https://​vizhub.healthdata.org/​gbd-​compare/​), estimates that the principal cancers caused by occupational exposures were lung cancer (269,000 deaths), mesothelioma (25,000 deaths), laryngeal cancer (4600 deaths), and leukemia (3200 deaths). Similarly, the GBD 2103 group estimated that there were 5.8 million occupational cancer DALYs out of a total of 197 million all-​cause cancer DALYs, so cancer DALYs attributable to occupation were 2.9% (GBD 2013; Risk Factors Collaborators, 2015). The principal occupational cancers contributing to DALYS were lung cancer (5.0  million), mesothelioma (514,000), leukemia (108,000), and laryngeal cancer (102,000) (https://​vizhub.healthdata.org/​gbd-​compare/​). The case of mesothelioma is particularly interesting, because it is (1)  only caused by asbestos, largely an occupational exposure, and (2) a highly fatal cancer that has a very long latency (e.g., typically 40  years instead of approximately 20  years for most solid tumors). Thus while the burden of many occupational cancers is gradually decreasing in developed countries, where exposures have been lowered and heavy industry has decreased (which is not the case in less developed countries), the burden from mesothelioma in the developed countries may have not yet peaked. Figure 16–1 shows the increasing mortality rate of mesothelioma in England over time among older men. The number of male mesothelioma deaths per year in England is expected to peak at around 2500/​year in the year 2020 (Health and Safety Executive, 2014). It should be remembered that these estimates of the burden of cancer associated with occupational exposures do not account for secondary exposures. In some circumstances, occupational exposures have been observed to increase the risk of cancer among the spouses and children of workers. Asbestos fibers carried home on workers’ clothing, skin, and hair pose a risk for family members (Ferrante et al.,

 285

Occupational Cancer 1000

Rate per 1,000,000 person-years

100

10

1 Age group 85+ 75–84 65–74 55–64 45–54 35–44 0.1 1970 1980 1990 2000 2010 Year of death

Figure 16–1.  Age-​adjusted mortality rate for mesothelioma among males in the United Kingdom, 1969–​2013.

2007; IARC, 1987, 2012c). In fact, secondary exposure to asbestos is thought to be the major risk factor for mesothelioma among women. Family members on farms where pesticides are applied may be exposed via contact with contaminated clothing of the applicator as well as environmental exposure to airborne pesticides drifting over residential areas, residential dust, and ingestion (Deziel et al., 2015). Spouses and children of farmers had a doubling of the urinary levels of a metabolite of the pesticide carbaryl following application by the farmer on the family farm, documenting secondary exposure (Shealy et al., 1997). In the Agricultural Health Study, pesticide levels in the homes of farmers have exceeded those in non-​farm homes (Curwin et al., 2005). Children of the Agricultural Health Study participants have excess lymphoma (Flower et al., 2004). Other studies of parental exposure to pesticides have suggested that the children’s secondary exposures play a role in the risk of leukemia and brain cancer, and less consistently with Wilms tumor, Ewing sarcoma, soft-​tissue sarcoma, and Hodgkin’s lymphoma (Bailey et  al., 2014; Daniels et al., 1997; Rodvall et al., 2003; Zahm and Ward, 1998).

CONTROVERSY Occupational studies often generate considerable controversy. Cancer is a devastating disease with special emotional overtones in

285

all situations, and especially where individuals have little control over exposures that may cause the disease. Studies linking workplace exposures to cancer may lead government regulators to control exposure. Because regulation has economic consequences, studies of cancer and occupational exposures receive intense scrutiny. They are often reviewed and dissected beyond what may occur for other cancer risk factors. Because of these pressures, intense debates and disagreements regarding the quality of individual studies and the weight of evidence are standard for each occupational exposure–​cancer relationship when associations are new and the amount of scientific information is less robust than it will eventually be. Eventually, however, sufficient scientific information may accumulate to allow general agreement among scientists, stakeholders, and the public regarding an issue. All widely accepted occupational carcinogens (e.g., arsenic, asbestos, benzene, and silica) went through this period—​sometimes a lengthy period—​of debate and controversy. Even after general acceptance that a substance is a carcinogenic hazard to humans by the public and scientific community, there may be disagreement regarding the exposure–​response relationship and other risk assessment issues. Examples of established occupational carcinogens where disagreements continue include arsenic (where risk from exposure at low levels is debated [Tsuji et al., 2014]), for chrysotile asbestos (although classified as a human carcinogen by the IARC [2012c], there is continuing debate about whether the cancer risk associated with chrysotile asbestos is actually due to concomitant amphibole asbestos exposure [Bernstein et al., 2013; Kanarek, 2011; van der Bij et  al., 2013; Yarborough,  2006]), and benzene (where there is disagreement on the range of lymphohematopoietic cancers affected and the risks at lower levels of exposure [Hayes et  al., 1997; Schnatter et al., 2012]). The debate regarding the adequacy of the data to conclude that a substance is a carcinogenic hazard is still ongoing for substances of more recent concern, for example, diesel exhaust (Gamble et al., 2012; Sun et al., 2014), formaldehyde (Checkoway et al., 2012; Gentry et al., 2013; McLaughlin and Tarone, 2014), trichloroethylene (Alexander et al., 2007), and shift work (Brudnowska and Peplonska, 2011; Erren et al., 2010). The example of diesel exhaust, now an established lung carcinogen, is instructive regarding the controversy and special interests often associated with studies of occupational cancer. The National Cancer Institute (NCI) and the National Institute for Occupational Safety and Health (NIOSH) initiated a study of diesel-​exposed miners in 1992, which was not published until 2012. The study faced numerous legal challenges from the mining industry, through its lobbying group The Methane Awareness Research Group (MARG), and was subject to Congressional oversight throughout protocol development, data collection, analysis, peer review, and journal publication. These pressures delayed the study for years (Monforton, 2006; Morris, 2012; Kean, 2012).

FUTURE DIRECTIONS There are many occupational exposures that require further evaluation. An obvious starting point for selection of a few high-​priority candidates could be the substances in the probable (2A) and possible (2B) categories of IARC classification that affect large populations. For these, there is already some information suggesting an association, and this information can provide direction for new investigations. There are an even larger number of occupational exposures that have been linked to cancer in animal bioassays for which epidemiologic data are largely nonexistent. Even for those exposures that are already classified as human carcinogens, additional studies could provide important information. The recent IARC re-​evaluations in Volumes 100 A through F of factors in Group 1 (sufficient evidence for human carcinogenicity) show that even studies of known carcinogens can be beneficial in that they reveal links to specific cancers not previously documented, as well as new mechanistic and toxicological understanding, and data on exposure–​response relationships. It appears, however, that, despite numerous leads and the clear need for study of occupational exposures, research efforts

286

286

PART III:  THE CAUSES OF CANCER

on occupational cancer have been diminishing in recent years (Raj et al., 2014). Future studies should take advantage of the new technologies that allow characterization of mechanistic pathways, gene–​exposure interactions, and epigenetic influences. Also, new techniques in exposure assessment are available that reduce exposure misclassification and result in more accurate estimates of exposure–​response relationships. Effective use of these new opportunities will require less reliance on the historical cohort design, which has been the design of choice in occupational investigations in the past, and more use of prospective designs (Blair et al., 2015a) that allow the incorporation of biologic and mechanistic components. The changing dynamics and geographical location of various industries also indicate that future investigations should expand beyond studies that are composed of white men in high-​income countries, which have predominated in the past (Raj et al., 2014). Although it is difficult to precisely predict specific exposures or workplace conditions that are most important for study in the future, following are a few that deserve consideration, based on some evidence of hazard, number of people exposed, and possible magnitude of population burden.

Sedentary Behavior Lack of physical activity contributes to a number of diseases, including cancer (US Department of Health and Human Services, 1996). Amount of time spent in sedentary behaviors appears to be associated with an additional disease burden beyond that from low levels of physical activity (Matthews et  al., 2012). In the past, most physical activity was associated with work, but currently, many occupations make a negligible contribution to overall activity, and this leads to health threats (Straker et al., 2009). The high level of physical activity in earlier times is dramatically demonstrated in an Old Order Amish community whose lifestyle is similar to that common 150 years ago (Bassett et al., 2004). Their physical activity may play a role in their low overall cancer rate (Westman et al., 2010). In general, over time there has been a drastic decrease in the proportion of jobs that require a moderate intensity of physical activity, with a corresponding increase in jobs with largely sitting and sedentary behavior (Church et  al., 2011). Consequently, many individuals must rely almost entirely on leisure time activities to achieve appropriate levels of physical activity, especially in developed countries. However, there are certainly many occupations in developed countries that still require considerable physical activity.

Shift Work Shift work has been classified as probably carcinogenic in humans (2A) based on limited evidence in humans, sufficient evidence in experimental animals, and mechanistic actions that may be relevant in humans (IARC, 2010a). In humans the strongest association was with breast cancer, but links with other sites (prostate, endometrium, and colon) were observed in some studies. Night work suppresses melatonin, which inhibits tumors, although other mechanisms have also been suggested (Fritschi et al., 2011). The overall evidence regarding shift work and ill health is still in flux. A  recent study provided some evidence for a relationship with lung cancer (Gu et al., 2015). This suggests that additional studies are needed for breast cancer, which is most strongly associated with shift work at the present time, but perhaps studies are also needed to consider links with other cancers. Studies should also evaluate different forms of shift work, characterize exposure more accurately, evaluate timing issues regarding exposure to light at night, explore possible mechanistic actions, and assess possible interaction with other lifestyle and occupational factors. It should be noted that this is a common exposure. For example, Purdue et  al. (2015) have estimated that if shift work causes breast cancer, an estimated 6% of US incident breast cancers in 2010 would have been attributable to shift work.

Pesticides Pesticides provide many important benefits in food production and protection, public health, and aesthetics, but some may present risk of cancer and other diseases (Blair et  al., 2015b). Experimental investigations have been conducted for most pesticides, but epidemiologic data are much more limited. The IARC, for example, has only evaluated a few specific pesticides, but has recently described plans for a series of monographs on pesticides currently in use, and has conducted two. The initial IARC effort concerned commonly used organophosphate insecticides and a very common herbicide (glyphosate, commercial name Round-​Up). The monograph classified tetrachlorvinphos and parathion as possibly carcinogenic to humans (2B) and malathion, diazinon, and glyphosate as probably carcinogenic to humans (2A) (Guyton et  al., 2015). In a second evaluation, the IARC classified DDT as a probable carcinogen (2A), lindane (an organophosphate largely banned) as a definite carcinogen (1), and 2-​4-​5-​T (a widely used herbicide) as a possible carcinogen (2B) (Loomis et al., 2015). DDT use remains common in Africa and India for malaria control, although it was banned as an insecticide in most countries in the late 1970s and early 1980s. Many pesticides currently in use, however, have not been adequately evaluated in epidemiologic studies. In addition to inadequate studies on many pesticides currently in use, new pesticides are periodically developed and introduced to overcome pest resistance that develops over time. Consequently, there is a growing need to initiate new epidemiologic studies to evaluate the cancer hazard that may emanate from established and new pesticides.

Diesel Exhaust Diesel exhaust has recently been classified by the IARC as a human carcinogen (IARC, 2014a). Epidemiologic studies that made major contributions to this decision (Attfield et  al., 2012; Garshick et  al., 2008; Silverman et al., 2012; Steenland et al., 1990) evaluated exposures largely from an older engine technology. These older engines will eventually be replaced by newer, cleaner-​ burning engines designed to reduce exhaust exposures. There is a need for new epidemiologic investigations to evaluate the potential hazard from exhausts from these new engines.

Nanoparticles Naturally occurring nanoparticles are found in engine exhausts and burning carbon materials, but most of the future concern is in relation to engineered nanoparticles. Human exposure to engineered nanoparticles can occur during creation in the laboratory, large-​scale manufacture, use, and environmental contamination (Kumar and Dhawan, 2013; Rim et al., 2013). NIOSH recently conducted an industry-​wide study and exposure assessment (Dahm et al., 2015; Schubauer-​Berigan et al., 2011) and found that to date there are only about 1000 workers in the United States judged as exposed. However, many nanoparticles are composed of known or suspect carcinogenic agents, and the number of exposed individuals is expected to grow rapidly, particularly in the chemical industry, photochemical electricity generation, and imaging and molecular diagnosis (Stark et al., 2015). The IARC evaluated nanoparticles in 2014; most were judged not classifiable in humans (Group 3), but one type was found to have sufficient animal evidence of carcinogenicity, and was classified Group 2B (possible carcinogen) (Grosse et al., 2014).

ACKNOWLEDGMENT The authors would like to thank David Check for graphics contributions. References Alavanja MC, Sandler DP, McMaster SB, et al. 1996. The Agricultural Health Study. Environ Health Perspect, 104(4), 362–​369. PMCID: PMC1469343.

 287

Occupational Cancer Alexander DD, Kelsh MA, Mink PJ, et  al. 2007. A meta-​analysis of occupational trichloroethylene exposure and liver cancer. Int Arch Occup Environ Health, 81(2), 127–​143. PMID: 17492303. Axelson O, and Steenland K. 1988. Indirect methods of assessing the effects of tobacco use in occupational studies. Am J Ind Med, 13(1), 105–​118. PMID: 3344750. Bailey HD, Fritschi L, Infante-​Rivard C, et al. 2014. Parental occupational pesticide exposure and the risk of childhood leukemia in the offspring: findings from the childhood leukemia international consortium. Int J Cancer, 135(9), 2157–​2172. PMID: 24700406. Baris D, Kwak LW, Rothman N, et al. 2000. Blood levels of organochlorines before and after chemotherapy among non-​Hodgkin’s lymphoma patients. Cancer Epidemiol Biomarkers Prev, 9(2), 193–​197. PMID: 10698481. Bassett DR, Schneider PL, and Huntington GE. 2004. Physical activity in an Old Order Amish community. Med Sci Sports Exer, 36(1), 79–​85. PMID: 14707772. Benbrahim-​Tallaa L, Lauby-​Secretan B, Loomis D, et al. 2014. Carcinogenicity of perfluorooctanoic acid, tetrafluoroethylene, dichloromethane, 1,2-​ dichloropropane, and 1,3-​propane sultone. Lancet Oncol, 15(9):  924–​ 925. PMID: 25225686. Bergdahl IA, and Järvholm B. 2003. Cancer morbidity in Swedish asphalt workers. Am J Ind Med, 43(1), 104–​108. PMID: 12494427. Bernstein D, Dunnigan J, Hesterberg T, et al. 2013. Health risk of chrysotile revisited. Crit Rev Toxicol, 43(2), 154–​183. PMCID: PMC3581056. Berrington de González A, Gilbert E, Curtis R, et al. 2013. Second solid cancers after radiation therapy:  a systematic review of the epidemiologic studies of the radiation dose-​response relationship. Int J Radiat Oncol Biol Phys, 86(2), 224–​233. PMCID: PMC3816386. Blair A, Hines CJ, Thomas K, et  al. 2015. Investing in prospective cohort studies on occupational exposures. Am J Ind Med, 58(2), 113–​ 122. PMCID: PMC4516175. Blair A, Marrett L, and Beane Freeman L. 2011. Occupational cancer in developed countries. Environ Health, 10(Suppl 1):S9. PMCID: PMC3073201 Blair A, Ritz B, Wesseling C, et al. 2015. Pesticides and human health. Occup Environ Med, 72:81–​82. PMID: 25540410. Blair A, Stewart P, Lubin JH, et  al. 2007. Methodological issues regarding confounding and exposure misclassification in epidemiological studies of occupational exposures. Am J Ind Med, 50(3), 199–​ 207. PMID: 17096363 Blair A, Tarone R, Sandler D, et al. 2002. Reliability of reporting on lifestyle and agricultural factors by a sample of participants in the Agricultural Health Study from Iowa. Epidemiology, 13(1), 94–​99. PMID: 11805592. Blot WJ, Fraumeni JF Jr, Mason TJ, et al. 1979a. Developing clues to environmental cancer: a stepwise approach with the use of cancer mortality data. Environ Health Perspect, 32(Oct.), 53–​58. PMCID: PMC1637941. Blot WJ, Harrington JM, Toledo A, et  al. 1978. Lung cancer after employment in shipyards during World War II. N Engl J Med, 299(12), 620–​624. PMID: 683235. Blot WJ, Morris LE, Stroube R, et  al. 1980. Lung and laryngeal cancers in relation to shipyard employment in coastal Virginia. J Natl Cancer Inst, 65(3), 571–​575. PMID: 6931936 Blot WJ, Stone BJ, Fraumeni JF Jr et al. 1979b. Cancer mortality in U.S. counties with shipyard industries during World War II. Environ Res, 18(2), 281–​290. PMID: 510250 Blot WJ, and Tarone RE. 2015. Doll and Peto’s quantitative estimates of cancer risks:  holding generally true for 35  years. J Natl Cancer Inst, 107(4). PMID: 25739419 Boyle P, and Levin B (Eds.). 2008. World Cancer Report 2008. Lyon, France: International Agency for Research on Cancer Bridge JC. 1933. Annual Report of the Chief Inspector of Factories and Workshops for the Year 1932 (pp. 103–​104). London: Her Majesty’s Stationery Office. Brown LM, Dosemeci M, Blair A, and Burmeister L. 1991. Comparability of data obtained from farmers and surrogate respondents on use of agricultural pesticides. Am J Epidemiol, 134(4), 348–​355. PMID: 1877595. Brownson RC, Boehmer TK, and Luke DA. 2005. Declining rates of physical activity in the United States: what are the contributors? Ann Rev Public Health, 26, 421–​443. PMID: 15760296. Brudnowska J, and Peplonska B. 2011. Night shiftwork and cancer risk: a literature review. Med Pr, 62(3), 323–​338. PMID: 21870422. Carey RN, Glass DC, Peters S, et  al. 2014. Occupational exposure to solar radiation in Australia: who is exposed and what protection do they use? Aust N Z J Public Health, 38(1), 54–​59. PMID: 24494947. Case R, Hosker M, McDonald DB, and Pearson JT. 1954. Tumours of the urinary bladder in workmen engaged in the manufacture and use of certain dyestuff intermediates in the British chemical industry. I. The role of aniline, benzidine, alpha-​naphthylamine, and beta-​naphthylamine. Br J Ind Med, 11(2), 75–​104. PMCID: PMC1037533.

287

Checkoway H, Boffetta P, Mundt DJ, and Mundt KA. 2012. Critical review and synthesis of the epidemiologic evidence on formaldehyde exposure and risk of leukemia and other lymphohematopoietic malignancies. Cancer Causes Control, 23(1), 1747–​1766. PMCID: PMC3465649. Checkoway H, Pearce N, and Kriebel D. 2007. Selecting appropriate study designs to address specific research questions in occupational epidemiology. Occup Environ Med, 64(9), 633–​638. PMCID: PMC2092571. Church TS, Thomas DM, Tudor-​Locke C, et al. 2011. Trends over 5 decades in the U.S. occupation-​related physical activity and their associations with obesity. PLoS One, 6(5), e19657. PMCID: PMC3102055. Creech JL, Jr, and Johnson MN. 1974. Angiosarcoma of liver in the manufacture of polyvinyl chloride. J Occup Med, 16(3), 150–​151. PMID: 4856325. Curwin BD, Hein MJ, Sanderson WT, et  al. 2005. Pesticide contamination inside farm and nonfarm homes. J Occup Environ Hyg, 2(7), 357–​367. PMID: 16020099. Dahm M, Schubauer-​Berigan M, Evans D, Birch E, Fernback J, and Deddens J. 2015. Carbon nanotube and nanofiber exposure assessments: an analysis of 14 site visits. Ann Occup Hyg, 59(6), 705–​723. PMCID: PMC4507369. Daniels JL, Olshan AF, and Savitz DA. 1997. Pesticides and childhood cancers. Environ Health Perspect, 105(10), 1068–​1077. PMCID: PMC1470375. Daniels RD, Kubale TL, Yiin JH, et al. 2014. Mortality and cancer incidence in a pooled cohort of US firefighters from San Francisco, Chicago and Philadelphia (1950–​2009). Occup Environ Med, 71(6), 388–​397. PMCID: PMC4499779. Darby SC, Whitley E, Howe GR, et al. 1995. Radon and cancers other than lung cancer in underground miners:  a collaborative analysis of 11 studies. J Natl Cancer Inst, 87(5), 378–​384. PMID: 7853419. De Roos AJ, Zahm SH, Cantor KP, et al. 2003. Integrative assessment of multiple pesticides as risk factors for non-​Hodgkin’s lymphoma among men. Occup Environ Med, 60:E11. PMCID: PMC1740618. Deziel NC, Friesen MC, Hoppin JA, Hines CJ, Thomas K, and Beane Freeman LE. 2015. A review of nonoccupational pathways for pesticide exposure in women living in agricultural areas. Environ Health Perspect, 123(6), 515–​524. PMCID: PMC4455586. Dietrich H, and Dietrich B. 2001. Ludwig Rehn (1849–​1930): pioneering findings on the aetiology of bladder tumours. World J Urol, 19(2), 151–​153. PMID: 11374319. Dobson J. 1972. Percivall Pott. Ann R Coll Surg Engl, 50(1), 54–​ 65. PMCID: PMC2388056. Doll R. 1952. The causes of death among gas-​workers with special reference to cancer of the lung. Br J Ind Med, 9(3), 180–​185. PMCID: PMC1037402. Doll R. 1955. Mortality from lung cancer in asbestos workers. Br J Ind Med, 12(2), 81–​86. PMCID: PMC1037613. Doll R. 1958. Cancer of the lung and nose in nickel workers. Br J Ind Med, 15(4), 217–​223. PMCID: PMC1037926. Doll R, and Peto R. 1981. The causes of cancer: quantitative estimates of avoidable risks of cancer in the United States today. J Natl Cancer Inst, 66(6), 1191–​1308. PMID: 7017215. Ehrenberg L, and Hallstrom T. 1967. Haematologic studies on persons occupationally exposed to ethylene oxide. In:  Radiosterilization of Medical Products (pp. 327–​334). Vienna:  International Atomic Energy Agency SM 92/​96. Erren TC, Falaturi P, Morfeld P, Knauth P, Reiter R, and Piekarski C. 2010. Shiftwork and cancer: the evidence and the challenge. Dtsch Arztebl Int, 107(38), 657–​662. PMCID: PMC2954516. Ferrante D, Bertolotti M, Todesco A, Mirabelli D, Terracini B, and Magnani C. 2007. Cancer mortality and incidence of mesothelioma in a cohort of wives of asbestos workers in Casale Monferrato, Italy. Environ Health Perspect, 115(10), 1401–​1405. PMID: 17938727. Figueroa J, Koutros S, Colt JS, et al. 2015. Modification of occupational exposures on bladder cancer risk by common genetic polymorphisms. J Natl Cancer Inst, 107(11), Epub Sep 14 (Print Nov 2015). PMID: 26374428. Flower KB, Hoppin JA, Lynch CF, et al. 2004. Cancer risk and parental pesticide application in children of Agricultural Health Study participants. Environ Health Perspect, 112(5), 631–​635. PMCID: PMC1241933 Fox AJ, and Collier PF. 1976. Low mortality rates in industrial cohort studies due to selection for work and survival in the industry. Br J Prev Soc Med, 30 (4), 225–​230. PMCID: PMC478970. Friedman GK. 2011. Malignant diseases attributable to asbestos exposure. In:  Dodson FR, and Hammar SP (Eds.), Asbestos:  Risk Assessment, Epidemiology, and Health Effects (2nd ed., p.  575). Boca Raton, FL: CRC Press. Friesen M, Lavoue J, Teschke K, and Van Tongeren M. 2015. Occupational exposure assessment in industry-​and population-​based epidemiologic studies. In:  Nieuwenhuijsen MJ (Ed.), Environmental Exposure Assessment for Epidemiology Studies (2nd ed., pp. 139–​ 162). Oxford:  Oxford University Press.

28

288

PART III:  THE CAUSES OF CANCER

Friesen MC, Zahm SH, Ward MH, and Silverman DT. 2012. Occupation and cancer. In:  Goldman MB, Troisi R, and Rexrode K (Eds), Women and Health (2nd ed., pp. 629–​645). San Diego, CA: Academic Press/​Elsevier. Fritschi L, Friesen MC, Glass D, Benke G, Girschik J, and Sadkowsky T. 2009. OccIDEAS: retrospective occupational exposure assessment in community-​based studies made easier. J Environ Public Health, 2009, Article ID 957023. doi: 10.1155/​2009/​957023. PMCID: PMC2778192. Fritschi L, Glass DC, Heyworth JS et  al. 2011. Hypotheses for mechanisms linking shiftwork and cancer. Med Hypotheses, 77(3), 430–​436. PMID: 21723672. Gamble JF, Nicolich MJ, and Boffetta P. 2012. Lung cancer and diesel exhaust: an updated critical review of the occupational epidemiology literature. Crit Rev Toxicol, 42(7), 549–​598. PMCID: PMC3441149. Garshick E, Laden F, Hart JE, et al. 2008. Lung cancer and vehicle exhaust in trucking industry workers. Environ Health Perspect, 116(10), 1327–​1332. PMCID: PMC2569090. Gentry PR, Rodricks JV, Tumbull D, et al. 2013. Formaldehyde exposure and leukemia: Critical review and reevaluation of the results from a study that is the focus for evidence of biological plausibility. Crit Rev Toxicol, 43(8), 661–​670. PMID: 23902349. Gilbert ES, Koshurnikova NA, Sokolnikov M, et  al. 2000. Liver cancers in Mayak workers. Radiat Res, 154(3), 246–​522. PMID: 10956429. Gilbert ES, Koshurnikova NA, Sokolnikov ME, et  al. 2004. Lung cancer in Mayak workers. Radiat Res, 162(5), 505–​516. PMID: 15624305. GBD 2013 Risk Factors Collaborators. 2015. Global, regional and national comparative risk assessment of 79 behavioral, environmental/​occupational and metabolic risks or clusters of risks in 188 countries 1990–​ 2013:  a systematic analysis for the GBD 2013. Lancet, 386 (10010), 2287–​2323. PMID: 26364544. Greenberg M. 1999. A study of lung cancer mortality in asbestos workers: Doll, 1955. Am J Ind Med, 36(3), 331–​347. PMID: 10469998. Grosse Y, Loomis D, Guyton KZ, et al. 2014. Carcinogenicity of fluoro-​edenite, silicon carbide fibres and whiskers, and carbon nanotubes. Lancet Oncol, 15(13), 1427–​1428. PMID: 25499275. Gu F, Han J, Laden F et  al. 2015. Total and cause-​ specific mortality of U.S.  nurses working rotating night shifts. Am J Prev Med, 48(3), 241–​ 252. PMCID: PMC4339532. Gu J, and Wu X. 2011. Genetic susceptibility to bladder cancer risk and outcome. Per Med, 8(3), 365–​374. PMID: PMCID: PMC3172962. Guyton KZ, Loomis D, Grosse Y, et  al. 2015. Carcinogenicity of tetrachlorvinphos, parathion, malathion, diazinon, and glyphosate. Lancet Oncol, 16(5), 490–​491. PMID: 25801782. Hayes RB. 1988. Review of occupational epidemiology of chromium chemicals and respiratory cancer. Sci Total Environ, 71(3), 331–​339. PMID: 3043655. Hayes RB, Dosemeci M, Wacholder S, et  al. 1997. Benzene and the dose-​ related incidence of hematologic neoplasms in China. J Natl Cancer Inst, 89(14), 1065–​1071. PMID: 9230889. Health and Safety Executive England. 2014. Mesothelioma in Great Britain 2014. Available from:  www.hse.gov.uk/​statistics/​causdis/​mesothelioma. Accessed September 2, 2015. Hill AB, and Faning EL. 1948. Studies in the incidence of cancer in a factory handling inorganic compounds of arsenic; mortality experience in the factory. Br J Ind Med, 5(1), 1–​6. PMCID: PMC1035980. Hoar SK, Morrison AS, Cole P, and Silverman DT. 1980. An occupation and exposure linkage system for the study of occupational carcinogenesis. J Occup Med, 22(11), 722–​726. PMID: 7441390. Hogstedt C, Malmqvist N, and Wadman B. 1979b. Leukemia in workers exposed to ethylene oxide. JAMA, 241(11), 1132–​1133. PMID: 105157. Hogstedt C, Rohlén O, Berndtsson BS, Axelson O, and Ehrenberg L. 1979b. A cohort study of mortality and cancer incidence in ethylene oxide production workers. Br J Ind Med, 36(4), 276–​ 280. PMCID: PMC1008604. Hohenadel K, Raj P, Demers PA, Zahm SH, and Blair A. 2015. The inclusion of women in studies of occupational cancer:  a review of the epidemiologic literature from 1991–​2009. Am J Ind Med, 58(3), 276–​281. PMID: 25678456. Hohenadel K, Harris SA, McLaughlin JR, et  al. 2011. Exposure to multiple pesticides and risk of non-​Hodgkin lymphoma in men from six Canadian provinces. Int J Environ Res Public Health, 8(6), 2320–​2330. PMCID: PMC3138027. Hornung RW, Greife AL, Stayner LT, et al. 1994. Statistical model for prediction of retrospective exposure to ethylene oxide in an occupational mortality study. Am J Ind Med, 25(6), 825–​836. PMID: 8067360. Huff J. 1999. Long-​term chemical carcinogenesis bioassays predict human cancer hazards: issues, controversies, and uncertainties. Ann N Y Acad Sci, 895, 56–​79. PMID: 10676409.

Huff J, Chan P, and Nyska A. 2000. Is the human carcinogen arsenic carcinogenic to laboratory animals? Toxicol Sci, 55(1), 17–​23. PMID: 10788555. Huff J, Jacobon MF, and Davis DL. 2008. The limits of two-​year bioassay exposure regimens for identifying chemical carcinogens. Environ Health Perspect, 116(11), 1439–​1442. PMCID: PMC2592260. IARC. 1987. Overall evaluations of carcinogenicity:  an updating of IARC monographs volumes 1 to 42. IARC Monogr Eval Carcinog Risks Hum, 7, 1–​440. PMID: 3482203. IARC. 1997. Polychlorinated dibenzo-​ para-​ dioxins and polychlorinated dibenzofurans. IARC Monogr Eval Carcinog Risks Hum, 69, 1–​631. PMID: 9379504. IARC. 2006. Formaldehyde, 2-​butoxyethanol and 1-​tert-​butoxypropan-​2-​ol. IARC Monogr Eval Carcinog Risks Hum, 88, 1–​478. PMID: 17366697. IARC. 2010a. Painting, firefighting, and shiftwork. IARC Monogr Eval Carcinog Risks Hum, 98, 9–​764. PMID: 21381544. IARC. 2010b. Some aromatic amines, organic dyes, and related exposures. IARC Monogr Eval Carcinog Risks Hum, 99, 1–​658. PMID: 21528837. IARC. 2012a. Pharmaceuticals. Volume 100 A.  A  review of human carcinogens. IARC Monogr Eval Carcinog Risks Hum, 100(Pt A), 1–​401. PMID: 23189749. IARC. 2012b. Biological agents. Volume 100 B.  A  review of human carcinogens. IARC Monogr Eval Carcinog Risks Hum, 100(Pt B), 1–​441. PMID: 23189750. IARC. 2012c. Arsenic, metals, fibres, and dusts. Volume 100 C. A review of human carcinogens. IARC Monogr Eval Carcinog Risks Hum, 100(Pt C), 11–​465. PMID: 23189751. IARC. 2012d. Radiation. Volume 100 D. A review of human carcinogens. IARC Monogr Eval Carcinog Risks Hum, 100(Pt D), 7–​303. PMID: 23189752. IARC. 2012e. Personal habits and indoor combustions. IARC Monogr Eval Carcinog Risks Hum, 100(Pt E), 1–​538. PMID: 23193840. IARC. 2012f. Chemical agents and related occupations. IARC Monogr Eval Carcinog Risks Hum, 100(Pt F), 9–​562. PMID: 23189753. IARC. 2013. Diesel and gasoline engine exhausts and some nitroarenes. IARC Monogr Eval Carcinog Risks Hum, 105, 9–​699, PMID: 26442290. IARC. 2014a. Diesel and gasoline engine exhausts and some nitroarenes. IARC Monogr Eval Carcinog Risks Hum, 105, 9–​699. PMID: 26442290. IARC. 2014b. Trichloroethylene, tetrachloroethylene, and some other chlorinated agents. IARC Monogr Eval Carcinog Risks Hum, 106, 1–​512. PMID: 26214861. IARC. 2015a. Preamble. Available from:  http://​monographs.iarc.fr/​ENG/​ Preamble/​currentb6evalrationale0706.php. Accessed April 21, 2015. IARC. 2015b. Organ Site Poster, Updated March 2015. Available from: http://​ monographs.iarc.fr/​ENG/​Publications/​OrganSitePoster.pdf. Accessed July 6, 2015. IARC. 2015c. Polychlorinated biphenyls and polybrominated biphenyls. IARC Monogr Eval Carcinog Risks Hum, 107, 1–​513. Available from:  http://​ monographs.iarc.fr/​ENG/​Monographs/​vol107/​mono107.pdf. Accessed March 28, 2016. IARC. 2015d. Some drugs and herbal products. IARC Monogr Eval Carcinog Risks Hum, 108, 1–​438. Available from: http://​monographs.iarc.fr/​ENG/​ Monographs/​vol108/​mono108.pdf. Accessed March 28, 2016. IARC. 2015e. Outdoor air pollution. IARC Monogr Eval Carcinog Risks Hum, 109, 1–​454. Available from: http://​monographs.iarc.fr/​ENG/​Monographs/​ vol109/​mono109.pdf. Accessed March 28, 2016. IARC. 2015f. List of classifications by cancer sites with sufficient or limited evidence in humans, Volumes 1 to 113. Available from: http://​monographs. iarc.fr/​ENG/​Classification/​Table4.pdf. Accessed September 2, 2015. Järvholm B, and Englund A. 2014. The impact of asbestos exposure in Swedish construction workers. Am J Ind Med, 57(1), 49–​55. PMID: 24108505. Kanarek MS. 2011. Mesothelioma from chrysotile asbestos:  update. Ann Epidemiol, 21(9), 688–​697. PMID: 21820631. Kean S. 2012. Daylight at last for study of diesel lung cancer risks. Available from:  http://​news.sciencemag.org/​health/​2012/​03/​daylight-​last-​study-​ diesel-​lung-​cancer-​risks. Accessed August 29, 2015. Kennaway NM, and Kennaway EL. 1936. A study of the incidence of cancer of the lung and larynx. J Hyg (Lond), 36(2), 236–​267. PMCID: PMC2170982. Kesminiene A, Cardis E, Tenet V, et  al. 2002. Studies of cancer risk among Chernobyl liquidators:  materials and methods. J Radiol Prot, 22(3A), A137–​A141. PMID: 12400962. Kesminiene A, Evrard AS, Ivanov VK, et  al. 2008. Risk of hematological malignancies among Chernobyl liquidators. Radiat Res, 170(6), 721–​735. PMCID: PMC2904977. Koshurnikova NA, Gilbert ES, Sokolnikov M et  al. 2000. Bone cancers in Mayak workers. Radiat Res, 154(3), 237–​245. PMID: 10956428. Koutros S, Beane Freeman LE, Berndt SI, et al. 2010. Pesticide use modifies the association between genetic variants on chromosome 8q24 and prostate cancer. Cancer Res, 70(22), 9224–​9233. PMCID: PMC2982856.

 289

Occupational Cancer Koutros S, Berndt SI, Hughes Barry K, et al. 2013. Genetic susceptibility loci, pesticide exposure and prostate cancer risk. PLoS One, 8(4), e58195. PMCID: PMC3617165. Kubo S, Nakanuma Y, Takemura S, et  al. 2014. Case series of 17 patients with cholangiocarcinoma among young adult workers of a printing company in Japan. J Hepatobiliary Pancreat Sci, 21(7), 479–​488. PMID: 24420816. Kumagai S, Kurumatani N, Arimoto A, and Ichihara G. 2013. Cholangiocarcinoma among offset colour proof-​printing workers exposed to 1,2-​dichloropropane and/​or dichloromethane. Occup Environ Med, 70(7), 508–​510. PMID: 23493378. Kumar A, and Dhawan A. 2013. Genotoxic and carcinogenic potential of engineered nanoparticles:  an update. Arch Toxicol, 87(11), 1883–​1990. PMID: 24068037. Lan Q, Zhang L, Li G, et al. 2004. Hematotoxicity in workers exposed to low levels of benzene. Science, 306(5702), 1774–​1776. PMCID: PMC1256034. Lehman EJ, Hein MJ, and Estill CF. 2008. Proportionate mortality study of the United Association of Journeyman and Apprentices of the Plumbing and Pipe Fitting Industry. Am J Ind Med, 51(12), 950–​963. PMID: 18942099. Lerro CC, Koutros S, Andreotti G, et al. 2015. Use of acetochlor and cancer incidence in the Agricultural Health Study. Int J Cancer, 137(5), 1167–​ 1175. PMCID: PMC4492894. Leuraud K, Richardson DB, Cardis E, et al. 2015. Ionising radiation and risk of death from leukaemia and lymphoma in radiation-​monitored workers (INWORKS): an international cohort study. Lancet Haematol, 2(7), e276–​e281. PMID: 26436129. Levêque-​Morlais N, Tual S, Clin B, Adjemian A, Baldi I, and Lebailly P. 2015. The AGRIculture and CANcer (AGRICAN) cohort study:  enrollment and causes of death for the 2005–​2009 period. Int Arch Occup Environ Health, 88(1), 61–​73. PMID: 24599726. Linet MS, Kim KP, Miller DL, Kleinerman RA, Simon SL, and Berrington de González A. 2010. Historical review of occupational exposures and cancer risks in medical radiation workers. Radiation Res, 174(6), 793–​808. PMCID: PMC4098897. Lloyd JW. 1971. Long-​term mortality study of steelworkers. V. Respiratory cancer in coke plant workers. J Occup Med, 13(2), 53–​68. PMID: 5546197. Locke SJ, Colt JS, Stewart PA, et  al. 2014. Identifying gender differences in reported occupational information from three US population-​based case-​control studies. Occup Environ Med, 71(12), 855–​864. PMCID: PMC4177972. Loomis D, Guyton K, Grosse Y, et al. 2015. Carcinogenicity of lindane, DDT, and 2,4-​ dichlorophenoxyacetic acid. Lancet Oncol, 16(8), 891–​ 892. PMID: 26111929. Loutit JF. 1970. Malignancy from radium. Br J Cancer, 24(2), 195–​ 207. PMCID: PMC2008571. Lynge E, Andersen A, Rylander L, et al. 2006. Cancer in persons working in dry cleaning in the Nordic countries. Environ Health Perspect, 114(2), 213–​219. PMCID: PMC1367834. Machle W, and Gregorius F. 1948. Cancer of the respiratory system in the United States chromate producing industry. Public Health Rep, 63(35), 1114–​1127. PMCID: PMC1995382. Maronpot RR, Flake G, and Huff J. 2004. Relevance of animal carcinogenesis findings to human cancer predictions and prevention. Toxicol Pathol, 32(Suppl 1), 40–​48. PMID: 15209402. Mason TJ, McKay FW, Hoover R, Blot WJ, and Fraumeni JF, Jr. 1975. Atlas of cancer mortality for U.S.  counties:  1950–​1969. [DHEW Publ No. (NIH)75–​780]. Washington, DC: US Government Printing Office. Matthews CE, George SM, Moore SC, et al. 2012. Amount of time spent in sedentary behaviors and cause-​specific mortality in US adults. Am J Clinical Nutr, 95(2), 437–​445. PMCID: PMC3260070. Mayeux R. 2004. Biomarkers:  potential uses and limitations. NeuroRx, 1(2), 182–​188. PMCID: PMC534923. McLaughlin JK, and Tarone RE. 2014. Mortality from solid tumors in the updated formaldehyde cohort. Am J Ind Med, 57(4), 486–​ 487. PMID: 24501017. Melnick RL, and Huff J. 1992. 1,3-​Butadiene: toxicity and carcinogenicity in laboratory animals and in humans. Rev Environ Contam Toxicol, 124, 111–​144. PMID: 1732994. Nielsen GD, and Ovrebø S. 2008. Background, approaches and recent trends for setting health-​based occupational exposure limits: a minireview. Regul Toxicol Pharmacol, 51(3), 253–​69. PMID: 18502550. Miligi L, Costantini AS, Bolejack V, et al. 2003. Non-​Hodgkin’s lymphoma, leukemia, and exposures in agriculture: results from the Italian multicenter case-​control study. Am J Ind Med, 44(6), 627–​636. PMID: 14635239. Monforton C. 2006. Weight of the evidence or wait for the evidence? Protecting underground miners from diesel particulate matter. Am J Public Health, 96(2), 271–​276. PMCID: PMC1470492.

289

Morris J. 2012. Diesel report’s publication delayed as industry demands to see documents first, Feb 6, 2012. Available from:  http://​www.washingtonpost.com/​national/​health-​science/​diesel-​reports-​publication-​delayed-​ as-​industry-​demands-​to-​see-​documents-​first/​2012/​02/​01/​gIQA5wrFtQ_​ story.html. Accessed August 29, 2015. Olson JS. 2002. Bathsheba’s Breast: Women, Cancer and History (pp. 32–​33). Baltimore, MD: The Johns Hopkins University Press. Pearce N, Blair A, Vineis P, et al. 2015. IARC monographs: 40 years of evaluating carcinogenic hazards to humans. Environ Health Perspect, 123(6), 507–​514. PMCID: PMC4455595. Pott P. 1775. Chirurgical Observations Relative to the Cataract, the Polypus of the Nose, the Cancer of the Scrotum, the Different Kinds of Ruptures and the Mortification of the Toes and Feet. London: TJ Carnegy. Purdue MP, Hutchings SJ, Rushton L, and Silverman DT. 2015. The proportion of cancer attributable to occupational exposures. Ann Epidemiol, 25(3), 188–​192. PMID: 25487971. Rachiotis G, Drivas S, Kostikas K, Makaropoulos V, and Hadjichristodoulou C. 2012. Respiratory tract mortality in cement workers: a proportionate mortality study. BMC Pulm Med, 12, 30. doi: 10.1186/​1471-​2466-​12-​30. PMCID: PMC3425084. Raj P, Hohenadel K, Demers P, Zahm SH, and Blair A. 2014. Recent trends in published occupational cancer epidemiology research: results from a comprehensive review of the literature. Am J Ind Med, 57(3), 259–​264. PMID: 24488816. Ramazzini B. 1713. Diseases of Workers. Translated from the Latin text: De Morbis Artificum of 1713 by WC Wright. New  York Academy of Medicine, History of Medicine Series No. 23, 1964. New York: Hafner. Rim KT, Song SW, and Kim HY. 2013. Oxidative DNA damage from nanoparticle exposure and its application to workers’ health: a literature review. Saf Health Work, 4(4), 177–​186. PMCID: PMC3889076. Robinson CF, Walker JT, Sweeney MH, et al. 2015. Overview of the National Occupational Mortality Surveillance (NOMS) system:  leukemia and acute myocardial infarction risk by industry and occupation in 30 US states 1985–​1999, 2003–​2004, and 2007. Am J Ind Med, 58(2), 123–​137. PMCID: PMC4573401. Rodvall Y, Dich J, and Wiklund K. 2003. Cancer risk in offspring of male pesticide applicators in agriculture in Sweden. Occup Environ Med, 60(10), 798–​801. PMCID: PMC1740388. Romanenko A, Bebeshko V, Hatch M et  al. 2008. The Ukrainian-​American study of leukemia and related disorders among Chornobyl cleanup workers from Ukraine: I. Study methods. Radiat Res, 170(6), 691–​697. PMCID: PMC2856482. Romanenko AY, Finch SC, Hatch M et  al. 2008. The Ukrainian-​American study of leukemia and related disorders among Chornobyl cleanup workers from Ukraine:  III. Radiation risks. Radiat Res, 170(6), 711–​720. PMCID: PMC2856603. Rothman K, Greenland S, and Lash T. 2008. Modern Epidemiology (pp. 122–​ 124). New York: Lippincott, Williams and Wilkins. Rushton L, Hutchings SJ, Fortunato L, et  al. 2012. Occupational cancer burden in Great Britain. Br J Cancer, 107(Suppl 1), S3–​S7. PMCID: PMC3384015. Russ DE, Ho KY, Colt JS, et al. 2016. Computer-​based coding of free-​text job descriptions to efficiently identify occupations in epidemiologic studies. Occup Environ Med, 73(6), 417–​424. PMCID: PMC4871757. Russ DE, Ho KY, Johnson CA, and Friesen MC. 2014. Computer-​based coding of occupation codes for epidemiological analyses. Proc IEEE Int Symp Comput Based Med Syst, 2014, 347–​350. PMCID: PMC4161468. Sandler DP, Kwok RK, Engel LS, et al. 2014. GuLF STUDY: Gulf Long-​Term Follow-​ Up Study. Study Protocol, Version 24.0, December 6, 2014. Available from: http://​www.niehs.nih.gov/​research/​atniehs/​labs/​epi/​studies/​gulfstudy/​publications/​gulf_​study_​protocol_​1262014.pdf. Accessed September 2, 2015. Schmidt CW. 2007. A change in the air:  smoking bans gain momentum worldwide. Environ Health Perspect, 115(8), A412–​A415. PMCID: PMC1940108. Schnatter AR, Glass DC, Tang G, Irons RD, and Rushton L. 2012. Myelodydplastic syndrome and benzene exposure among petroleum workers:  an international pooled analysis. J Natl Cancer Inst, 104(22), 1724–​1737. PMCID: PMC3502195. Schubauer-​Berigan M, Dahm M, and Yencken M. 2011. Engineered carbonaceous nanomaterials manufacturers in the United States: workforce size, characteristics, and feasibility of epidemiologic studies. J Occup Environ Med, 53(6 Suppl), S62–​S67. PMID: 21654420. Schulte PA. 1993. A conceptual and historical framework for molecular epidemiology. In:  Schulte PA, and Perera FP (Eds.), Molecular Epidemiology:  Principles and Practices (pp. 3–​ 44). San Diego, CA: Academic Press.

290

290

PART III:  THE CAUSES OF CANCER

Shealy DB, Barr JR, Ashley DL, Patterson DG Jr, Camann DE, and Bond AE. 1997. Correlation of environmental carbaryl measurements with serum and urinary 1-​naphthol measurements in a farmer applicator and his family. Environ Health Perspect, 105(5), 510–​513. PMCID: PMC1469864. Siemiatycki J. 1991. Risk Factors for Cancer in the Workplace. CRC Press, Boca Raton, FL: CRC Press. Siemiatycki J. 2014. Historical overview of occupational cancer research. In:  Anttila S, and Boffetta P (Eds.), Occupational Cancer (pp. 1–​20). London: Springer-​Verlag. Siemiatycki J, Fritschi L, Nadon L, and Gérin M. 1997. Reliability of an expert rating procedure for retrospective assessment of occupational exposures in community-​based case-​control studies. Am J Ind Med, 31(3), 280–​286. PMID: 9055950. Siemiatycki J, Richardson L, Straif K, et  al. 2004. Listing occupational carcinogens. Environ Health Perspect, 112(15), 1447–​ 1459. PMCID: PMC1247606. Silverman D, Samanic C, Lubin JH, et  al. 2012. The diesel exhaust in miners study: a nested case-​control study of lung cancer and diesel exhaust. J Natl Cancer Inst, 104(11), 855–​868. PMCID: PMC3369553. Snellings WM, Weil CS, and Maronpot RR. 1984. A two-​year inhalation study of the carcinogenic potential of ethylene oxide in Fischer 344 rats. Toxicol Appl Pharmacol, 75(1), 105–​117. PMID: 6464016. Solan S, Wallenstein S, Shapiro M, et  al. 2013. Cancer incidence in world trade center rescue and recovery workers, 2001–​2008. Environ Health Perspect, 121(6), 699–​704. PMCID: PMC3672914. Stark WJ, Stoessel PR, Wohlleben W, and Hafner A. 2015. Industrial applications of nanoparticles. Chem Soc Rev, 44(16), 5793–​5805. PMID:25669838. Stayner L, Steenland K, Dosemeci M, and Hertz-​Picciotto I. 2003. Attenuation of exposure-​response curves in occupational cohort studies at high exposure levels. Scand J Work Environ Health, 29(4), 317–​324. PMID: 12934726. Steenland K, Burnett C, Lalich N, Ward E, and Hurrell J. 2003. Dying for work: the magnitude of US mortality from selected causes of death associated with occupation. Am J Ind Med, 43(5), 461–​482. PMID: 12704620. Steenland K, and Moe C. 2016. Environmental and occupational epidemiology. In: Frumkin H (Ed.), Environmental Health: From Global to Local (2nd ed., pp. 83–​109). San Francisco: John Wiley & Sons. Steenland K, Seals R, Klein M, Jinot J, and Kahn HD. 2011. Risk estimation with epidemiologic data when response attenuates at high-​exposure levels. Environ Health Perspect, 119(6), 831–​837. PMID: 21220221. Steenland NK, Silverman DT, and Hornung RW. 1990. Case-​control study of lung cancer and truck driving in the Teamsters Union. Am J Public Health, 80(6), 670–​674. PMCID: PMC3114819. Stewart B, and Wild C (Eds.). 2014. World Cancer Report 2014. Lyon, France: International Agency for Research on Cancer. Stewart PA, Stewart WF, Siemiatycki J, Heineman EF, and Dosemeci M. 1998. Questionnaires for collecting detailed occupational information for community-​based case control studies. Am Ind Hyg Assoc J, 59(1), 39–​44. PMID: 9438334. Stewart PA, Coble JB, Vermeulen R, et al. 2010. The diesel exhaust in miners study: I. Overview of the exposure assessment process. Ann Occup Hyg, 54(7), 728–​746. PMCID: PMC2953554. Straif K. 2012. Estimating the burden of occupational cancer as a strategic step to prevention. Brit J Cancer, 107(Suppl 1), S1–​S2. PMCID: PMC3384019. Straker L, and Mathiassen SE. 2009. Increased physical work loads in modern work: a necessity for better health and performance? Ergonomics, 52(10), 1215–​1225. PMID: 19787501. Sun Y, Bochmann F, Nold A, and Mattenklott M. 2014. Diesel exhaust exposure and risk of lung cancer:  a review of the epidemiologic evidence. Int J Environ Res Public Health, 11(2), 1312–​1340. PMCID: PMC3945540. Tagnon I, Blot WJ, Stroube RB, et al. 1980. Mesothelioma associated with the shipbuilding industry in coastal Virginia. Cancer Res, 40(11), 3875–​3879. PMID: 7471040. Takahashi K, and Ishii Y. 2013. Historical developments of administrative measures for occupational diseases in Japan, a report commissioned by the ILO, 10 July 2013. Available from:  http://​www.ilo.org/​wcmsp5/​groups/​ public/​ - ​ - ​ - ​ e d_​ p rotect/ ​ - ​ - ​ - ​ p rotrav/ ​ - ​ - ​ - ​ s afework/ ​ d ocuments/ ​ p ublication/​ wcms_​234221.pdf. Accessed September 2, 2015. Teschke K, Olshan AF, Daniels JL, et al. 2002. Occupational exposure assessment in case-​ control studies:  opportunities for improvement. Occup Environ Med, 59(9), 575–​593. PMCID: PMC1740358.

Thun M. 1992. Kidney dysfunction in cadmium workers. In:  Steenland K (Ed.), Case Studies in Occupational Epidemiology. Oxford:  Oxford University Press. Tsuji JS, Alexander DD, Perez V, and Mink PJ. 2014. Arsenic and bladder cancer: quantitative assessment of studies in human populations to detect risks at low doses. Toxicology, 317, 17–​30. PMID: 24462659. Tual S, Silverman DT, Koutros S, et  al. 2015. Use of dieselized farm equipment and incident lung cancer:  findings from the Agricultural Health Study cohort. Environ Health Perspect, 124 (5), 611-​ 618. PMCID: PMC4858397. US Department of Health and Human Services. 1996. Physical Activity and Health: A Report of the Surgeon General. Atlanta, GA: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion. van der Bij S, Koffijberg H, Lenters V, et  al. 2013. Lung cancer risk at low cumulative asbestos exposure: meta-​regression of the exposure-​response relationship. Cancer Causes Control, 24(1), 1–​12. PMID: 23187858. Vermeulen R, Silverman DT, Garshick E, Vlaanderen J, Portengen L, and Steenland K. 2014. Exposure-​response estimates for diesel engine exhaust and lung cancer mortality based on data from three occupational cohorts. Environ Health Perspect, 122(2), 172–​177. PMCID: PMC3915263. Vlaanderen J, Moore LE, Smith MT, et  al. 2010. Application of OMICS technologies in occupational and environmental health research; current status and projections. Occup Environ Med, 67(2), 136–​ 143. PMCID: PMC2910417. Wagner J, Sleggs C, and Marchand P. 1960. Diffuse pleural mesothelioma and asbestos exposure in the North Western Cape Province. Br J Ind Med, 17(Oct), 260–​271. PMCID: PMC1038078. Wagoner J, Archer V, Lundin F, Holday D, and Lloyd J. 1965. Radiation as the cause of lung cancer among uranium miners. N Engl J Med, 273(4), 181–​188. PMID: 14306333. Waldron HA. 1983. A brief history of scrotal cancer. Br J Ind Med, 40(4), 390–​ 401. PMCID: PMC1009212. Ward EM, Schulte PA, Straif K et  al. 2010. Research recommendations for selected IARC-​ classified agents. Environ Health Perspect, 118(10), 1355–​1362. PMCID: PMC2957912. Wei S, Wang LE, McHugh MK et al. 2012. Genome-​wide gene-​environment interaction analysis for asbestos exposure in lung cancer susceptibility. Carcinogenesis, 33(8), 1531–​1537. PMCID: PMC3499061. Westman JA, Ferketich AK, Kauffman RM, et  al. 2010. Low cancer incidence rates in Ohio Amish. Cancer Causes Control, 21(1), 69–​ 75. PMCID: PMC4308689. Yarborough CM. 2006. Chrysotile as a cause of mesothelioma:  an assessment based on epidemiology. Crit Rev Toxicol, 36(2), 165–​ 187. PMID: 16736942. Yoshinaga S, Mabuchi K, Sigurdson AJ, Doody MM, and Ron E. 2004. Cancer risks among radiologists and radiologic technologists:  review of epidemiologic studies. Radiology, 233(2), 313–​321. PMID: 15375227. Zablotska LB, Bazyka D, Lubin JH et al. 2013. Radiation and the risk of chronic lymphocytic and other leukemias among Chornobyl cleanup workers. Environ Health Perspect, 121(1), 59–​65. PMCID: PMC3553431. Zahm SH, and Blair A. 2003. Occupational cancer among women:  where have we been and where are we going? Am J Ind Med, 44(6), 565–​575. PMID: 14635233. Zahm SH, Pottern LM, Lewis DR, Ward MH, and White DW. 1994. Inclusion of women and minorities in occupational-​cancer epidemiologic studies. J Occup Med, 36(8), 842–​847. PMID: 7807263. Zahm SH, Ward MH, and Silverman DT. 1999. Occupational cancer. In: Goldman MB, and Hatch MC (Eds.), Women and Health (pp. 493–​502). San Diego, CA: Academic Press. Zahm SH, and Ward MH. 1998. Pesticides and childhood cancer. Environ Health Perspect, 106(Suppl 3), 893–​908. PMCID: PMC1533072. Zahm SH, Weisenburger DD, Babbitt PA et  al. 1990. A case-​control study of non-​ Hodgkin’s lymphoma and the herbicide 2,4-​ dichlorophenoxyacetic acid (2,4-​D) in eastern Nebraska. Epidemiology, 1(5), 349–​56. PMID: 2078610. Zhang L, Tang X, Rothman N, et al. 2010. Occupational exposure to formaldehyde, hematotoxicity, and leukemia-​specific chromosome changes in cultured myeloid progenitor cells. Cancer Epidemiol Biomarkers Prev, 19(1), 80–​88. PMCID: PMC2974570.

 291

17

Air Pollution JONATHAN M. SAMET AND AARON J. COHEN

OVERVIEW A wide variety of man-​made and naturally occurring air pollutants are known to cause cancer. Diverse exposures such as tobacco smoke, radionuclides (radon), chemicals (benzene, mustard gas, and volatile organic compounds), fibers (asbestos), and metals and metalloids (chromium, nickel, and arsenic) have long been classified as carcinogenic to humans. Historically, the evidence base for these classifications predominantly concerned high levels of exposure in occupational settings. Over the last 30–​40 years, scientific attention has focused on quantifying the adverse health effects of indoor and outdoor air pollutants at exposure levels several orders of magnitude lower than those studied initially. These include secondhand smoke, household exposure to radon, residential and environmental exposure to asbestos, soot from diesel powered engines, ambient exposures to small particles (PM2.5), and indoor air pollution from the combustion of biomass and coal. This chapter provides an overview of recent epidemiologic studies of air pollutants and cancer. It is necessarily selective, because the evidence on air pollution and lung cancer in particular is now extensive. Issues fundamentally important to policy development concern dose–​response relationships and risk assessment. Of great concern is that exposures in economically developing countries rival and often dramatically exceed those found in high-​income Western countries.

INTRODUCTION In the early decades of the twentieth century, lung cancer was a rare disease. By mid-​century, however, it was evident that an epidemic of lung cancer was occurring among males in the United States and a number of European countries, and a parallel epidemic followed at mid-​century among women. By then, there were already several established and suspect causes of lung cancer. The evidence was strongest for radon, linked to lung cancer in underground miners based on epidemiological and clinical observations in mines in Eastern Europe and the measurement of high levels of radon in these mines. Epidemiological research was implemented to search for causal factors, targeting specific groups of workers at high risk as well as the general population. By the 1920s and 1930s, clinicians were already beginning to make the link between smoking and lung cancer, and the ubiquitous and serious air pollution in cities was also considered as a potential cause, a hypothesis supported by the presence of an urban–​ rural gradient in risk (Hoffman, 1929, 1931; Macklin and Macklin, 1940; Overholt and Rumel, 1940; Stocks and Campbell, 1955). These two non-​exclusive hypotheses were given equal weight by Doll and Hill (Doll and Hill, 1952)  as a rationale for their case-​control study of lung cancer in London. After smoking was quickly identified as a powerful cause of lung cancer, concern persisted that air pollution may cause lung and other cancers, and research was ongoing on this hypothesis. Additionally, by the 1950s, other respiratory carcinogens had been identified, including radon and asbestos. The hypothesis that air pollution causes lung cancer had a strong basis in the known release of carcinogens into outdoor air from industrial sources, power plants, and motor vehicles, and in the recognition that indoor air is contaminated by respiratory carcinogens. Historically, epidemiological studies have shown for decades that

air pollution increases lung cancer risk. By 2013, the totality of the evidence on outdoor air pollution and cancer had become sufficiently cohesive to support a conclusion by the International Agency for Research on Cancer (IARC) that outdoor air pollution is a Group  1 carcinogen (IARC, 2015). During recent decades, a number of airborne carcinogens, viewed as potential threats to public health, have been particularly controversial. The energy crisis of the early 1970s led to increased manufacture of diesel-​ powered vehicles; recognition that diesel-​ soot particles were mutagenic raised concern that the increasing numbers of diesel-​powered vehicles would increase lung cancer risk for the population. Three indoor carcinogens received widespread attention from the scientific community and the public during the 1980s and 1990s: tobacco smoke inhaled by non-​smokers, radon, and asbestos fibers. Policies and programs for reducing the risks to the population of each of these carcinogens became extremely controversial. Critics questioned whether the risks were exaggerated and whether the high costs of control, particularly for radon and asbestos, were justified. The scientific evidence on secondhand smoke was repeatedly challenged by the tobacco industry as it sought to maintain controversy, even as review panels concluded that secondhand smoke caused lung cancer (IARC, 1986; National Research Council and Committee on Passive Smoking, 1986). Now, definitive and incontrovertible conclusions on the carcinogenicity of secondhand smoke have been made by both the IARC (2004) and the Surgeon General (2006). In the United States, there has also been extensive litigation around indoor asbestos in the past. For these and other inhaled carcinogens, substantial research has been motivated by concern for the public’s health, and by the need to develop the scientific foundation of evidence for control strategies to address carcinogens for which exposure is widespread. This chapter provides an overview of the evidence that outdoor and indoor air pollution are causally related to lung and other types of cancer. The topic of secondhand smoke and lung cancer is addressed in Chapter 11 of this volume as well. The evidence on air pollution and lung cancer is now extensive and this review is necessarily selective, emphasizing the most recent findings, primarily from the epidemiologic literature. A 1990 monograph provides a more complete review of the earlier literature on outdoor air pollution (Tomatis, 1990). The IARC has published a review monograph that comprehensively covers combustion products, including indoor and outdoor air pollution, and cancer, as well as completing a series of volumes on the carcinogenicity of these agents (IARC, 2010a, 2010b, 2013b, 2013c, 2015). This chapter focuses on outdoor, primarily urban, air pollution and indoor air contaminants in relation to lung cancer in high-​income countries. Mounting evidence indicates that people in developing countries have exposures to indoor and outdoor environments that rival and often dramatically exceed those found in high-​ income Western countries (Gordon et al., 2014). Indoor air pollution from coal combustion and cooking fumes are thought to account for the high lung cancer risk among women in China and Hong Kong, despite the low prevalence of smoking (Bruce et al., 2015). Rising pollution of outdoor air in the mega-​cities of the developing world also poses a risk for lung cancer. These topics are addressed elsewhere (Brauer et al., 2012, 2016; Katsouyanni, 2013; Krzyzanowski et al., 2014).

291

29

292

PART III:  THE CAUSES OF CANCER

EXPOSURES TO INHALED CARCINOGENS

Table 17–​1. Summary Concentrations of Air Toxics “Pollutants of Interest” in the United States for 2010

Factors That Affect Exposure

Compound

Exposures to inhaled carcinogens take place in a variety of settings, including the home, the workplace, other public and commercial locations, and outdoors. The concept of total personal exposure provides a useful framework for conceptualizing exposures to inhaled carcinogens and evaluating the contributions of outdoor and indoor air pollution (National Research Council et al., 2012). Total personal exposure represents the integrated exposure to an agent accumulated in multiple microenvironments (i.e., environments having a relatively homogeneous concentration of the agent of interest during a specified time period). For an inhaled carcinogen such as benzo[a]‌pyrene, relevant microenvironments over the day might include outdoor air contaminated by vehicle exhaust and industrial emissions, workplace exposures, and air contaminated by tobacco smoke in a bar. For each microenvironment, the exposure received depends on the concentration of the carcinogen and the time spent in the microenvironment. Total personal exposure is the sum of the products of concentration with time spent in each microenvironment. Also relevant is the total volume of inhaled air, which for adults is about 10,000 liters per day. Because of the large volume inhaled, meaningful doses of carcinogens may reach the lung at relatively low concentrations. The actual lung dose of the carcinogen will further depend on the exposed person’s ventilation rate and pattern, physical characteristics of the agent, and other factors that affect the location and amount of deposition in the lung (National Research Council and Panel on Dosimetric Assumptions Affecting the Application of Radon Risk Estimates, 1991). Inhaled carcinogens may be absorbed into the systemic circulation. Doses to other organs (e.g., the urinary bladder) depend on uptake, metabolism, distribution, and excretion. The relevant microenvironments vary depending on the carcinogen of concern. Because most time is typically spent at home, exposures to contaminants in household air dominate for certain pollutants such as radon. Based on time-​activity patterns and policies regarding smoking, the workplace can be a significant source of exposure to secondhand smoke as well as industrial carcinogens. In high-​income countries, adults generally spend little time outdoors, but pollutants in ambient air can diffuse into homes. The microenvironments in which exposures occur influence the choice of control strategies. Some environments are shared and public; initiatives to control exposure in these settings require action at the societal level, often involving regulations. Other environments are private, and control lies with individuals. Whereas regulations are often essential to control toxic exposures in public settings, education is usually the mainstay of initiatives to reduce exposures to carcinogens in the home.

OUTDOOR AIR POLLUTION Exposures to Carcinogens in Outdoor Air Ambient air, particularly in densely populated urban environments, contains a variety of known human carcinogens, including organic compounds such as benzo[a]‌ pyrene and benzene, inorganic compounds such as arsenic and chromium, and radionuclides (Table 17–​1) (IARC, 2015). These substances are present as components of complex mixtures, which may include carbon-​based particles to which the organic compounds are adsorbed, oxidants such as ozone, and aerosols of sulfuric acid. The combustion of fossil fuels for power generation or transportation is the main source of organic and inorganic compounds, oxidants, and acids; combustion byproducts contribute heavily to particulate air pollution in most urban settings. Radionuclides also result from fuel combustion, as well as from mining operations. The plausibility that outdoor air pollutants cause cancer is supported by studies that have addressed its genotoxicity using in vitro assay systems and biomarkers. Urban air samples are mutagenic in the Salmonella assay, and much of the activity is attributable to polycyclic aromatic hydrocarbons (PAHs) (DeMarini, 2013; IARC, 2015).

Median

25th Percentile

75th Percentile

2.04 0.02 2.91 66.4

0.983 0 1.75 36.6

4.30 0.1 5.39 117

Metalsb Arsenic (PM10) Beryllium (PM10) Cadmium (PM10) Lead (PM10) Manganese (PM10) Nickel (PM10) Hexavalent chromium

0.415 0.002 0.084 2.32 4.03 0.845 0.018

0.240 0.0003 0.050 1.45 2.15 0.594 0

0.700 0.004 0.176 3.74 7.97 1.22 0.032

Carbonylsc Acetaldehyde Formaldehyde

0.893 1.66

0.597 1.09

1.29 2.47

VOCsd Acrylonitrile Benzene 1,3-​Butadiene Carbon tetrachloride Chloroform p-​Dichlorobenzene 1,2-​Dichloroethane Ethylbenzene Tetrachloroethylene Trichlororethylene Vinyl chloride

0 0.245 0.026 0.037 0.020 0.007 0 0.053 0.016 0 0

0 0.173 0.012 0.013 0.014 0 0 0.03 0.008 0 0

0 0.373 0.048 0.272 0.031 0.019 0 0.1 0.03 0 0

PAHsa Acenaphthene Benzo(a)pyrene Fluorene Naphthalene

PAHs: polycyclic aromatic hydrocarbons; PM10: particulate matter with particles of aerodynamic diameter, 10μm; SD: standard deviation; VOCs: volatile organic compounds. a PAHs: unit is ng/​m3. b Metals: unit is ng/​m3. c Carbonyls: unit is ppbv. d VOCs: unit is ppbv. Source: IARC, 2015

Incinerator emissions are also mutagenic; their activity reflects PAH content and nitroarenes (DeMarini et al., 1998). The IARC has summarized the findings of hundreds of studies of the mutagenicity of particles collected in outdoor air (IARC, 2015). Investigators have used biomarkers of exposure to ambient air pollutants to better characterize patterns of exposure. These studies have been primarily focused on polycyclic aromatic hydrocarbons and markers of oxidative stress. The results have been mixed. In a 1992 study in Poland (Perera et al., 1992), Perrera and colleagues measured PAH-​DNA adducts in lymphocytes as markers of molecular and genetic damage. Residents of a highly industrialized and polluted area had higher mean levels of PAH-​DNA adducts than residents of a comparison area. Autrup and colleagues (Autrup et  al., 1999)  compared measurements of bulky carcinogen-​ DNA adducts and 2-​amino-​apidic semialdehyde, a marker of oxidative stress, in non-​smoking Danish bus drivers and postal workers. The levels of adducts, but not the biomarker of oxidative stress, were higher in the bus drivers in central Copenhagen. In a study in Germany, markers of exposure to benzo[a]‌pyrene tended to be higher in city dwellers, compared with suburban dwellers. In the AULIS project in Greece, the mean levels of bulky DNA adducts were higher in residents of the the rural town of Halkida than in Athens (Georgiadis et al., 2001; Kyrtopoulos et  al., 2001), although the levels were also higher in those exposed to secondhand smoke. The findings in relation to residence location were unanticipated, given that prior studies tended to show higher levels of markers of genotoxicity in urban workers and residents (Kyrtopoulos et  al., 2001). By contrast, in a study of students in Copenhagen, measured personal exposure to small particles was associated with 7-​hydro-​8-​oxo-​2′-​deoxyguanosine, a marker of

 293



Air Pollution

DNA damage in lymphocytes; other associations between exposure and biomarker levels were not found (IARC, 2015). The differences in findings may reflect the differing levels of exposure and the markers considered. Unfortunately, there are few databases that record long-​term measurements of carcinogens and other products from fossil fuel combustion that could be used in epidemiologic studies. Available data indicate that there have been improvements in some indices of air quality in the United States and some other developed countries in recent decades (US Environmental Protection Agency [EPA], 2016). Particulate matter with an aerodynamic diameter < 2.5 μm, PM2.5, has been the airborne pollutant of greatest current interest with respect to lung cancer because particles of size 2.5 μm or less can be inhaled and deposited in the lung. These generally originate from combustion and may transport carcinogens, such as polycylic aromatic hydrocarbons, on their surfaces. Comprehensive monitoring of PM2.5 has only been in place since about 2000; previously, the US Environmental Protection Agency (EPA) monitored particles with diameters up to PM10; before 1988, it monitored only total suspended particulates (TSP), some of which are too large to be inhaled into the lung. Progressive declines in these indicators of combustion product pollution have been documented, following emissions reductions for these pollutants and their precursors (US EPA, 2016). The data discussed in the preceding refer to ambient air pollution over relatively large geographic areas. However, the exposure of human populations to carcinogens in ambient air may result from local sources, such as small businesses (e.g., automotive body or chrome plating shops), municipal facilities (e.g., waste incinerators), or areas with high vehicular traffic (Health Effects Institute, 2010). Studies in Harlem show that diesel particle exposures to pedestrians can be substantial (Kinney et al., 2000) and that high school students are exposed to a variety of toxic air pollutants throughout the day (Kinney et al., 2002). Levy and colleagues collected spot air samples along sidewalks in a section of Boston with heavy diesel bus traffic, and detected “hot spots” for exposure to benzo-​(a)-​pyrene, a marker for combustion emissions (IARC, 2013b; Levy et al., 2000).

Combustion Products As noted earlier, the combustion of fossil fuels for transportation and power generation generates many known or suspected carcinogens. The following section discusses some of the more significant pollutants in terms of exposure prevalence and/​or lung carcinogenicity. The IARC’s Outdoor Air Pollution (Volume 109) provides more detailed information on these pollutants (IARC, 2015).

Polycyclic Organic Matter

Polycyclic organic matter (POM), comprises a large and varied class of chemical compounds that include polycyclic aromatic hydrocarbons (PAH) and nitro-​PAHs. The IARC classifies both categories as carcinogenic and mutagenic (IARC, 2010b, 2013a). These POM have common chemical features (one or more benzene rings and a boiling point > 100°C), and can be found in both the solid and gas phases contaminants of ambient air, depending on their exact chemical structure. Those with > 5 benzene rings tend to be associated with the particle phase. In addition to the compounds released directly into the environment by combustion, others (such as the byproducts of diesel combustion) are created secondarily by chemical and photochemical reactions in the environment (Greenberg, 1987; Natusch, 1978; Winer and Busby Jr, 1995; Zielinska et al., 2010). Although the combustion of fossil fuels is a ubiquitous source of POM in the urban air, it is not the only source of human exposure to POM, and for some individuals it may not be the main source. Other exposure comes from inhaling tobacco or wood smoke, and from diet (e.g., from the consumption of grilled meat). Benzo-​(a)-​pyrene is a constituent of POM that has been extensively studied and is known to be carcinogenic. It is frequently used as a surrogate or marker for combustion sources in epidemiologic studies and

293

risk assessment (see later in this chapter). Armstrong and colleagues have reviewed the literature on cancer risk in relation to occupational and environmental exposure to PAHs (Armstrong et al., 2004). They concluded that PAHs are associated with increased lung cancer risk in occupational and urban environments. Workers exposed to mixtures of polycyclic compounds in coke-​ovens and coal gasification plants (Doll et al., 1972; Redmond, 1983) have increased lung cancer risk (IARC, 2012b). The levels of POM encountered in urban air are substantially less than in these settings.

Particles

The IARC’s Outdoor Air Pollution (Volume 109) classified particulate matter in outdoor air pollution as a Group 1 carcinogen (IARC, 2015). Like POM, particulate air pollution is not a single entity, but rather a chemically and physically diverse group of pollutants. Particles derive from sea spray and crystal dust, as well as from combustion of diesel fuel. Carbonaceous particles produced by burning fossil fuels are in the respirable range (generally < 1.0 μm in aerodynamic diameter) and can transport carcinogens, such as PAHs, adsorbed to their surfaces. Carcinogenicity appears to be modified by the rate of exposure. Experiments in animals demonstrate that even relatively pure carbon particles can induce lung cancer in rats when administered at high concentrations, suggesting that particles per se might, under some conditions, be carcinogenic (Mauderly, 1997). The relevance of these findings for humans is controversial, since the tumors in rats were observed at doses that far exceeded the exposures to diesel exhaust and overwhelmed lung clearance mechanisms (Winer and Busby Jr, 1995). Gaseous pollutants, such as sulphur dioxide (SO2) and oxides of nitrogen (NOx), are produced by the combustion of fossil fuels and then are converted into fine particles in the atmosphere. Epidemiologic studies have not found consistent evidence of lung cancer risk from occupational exposure to SO2. The IARC has classified strong sulphuric acid aerosol as a known human carcinogen based on increased lung and laryngeal cancer in heavily exposed occupational groups (IARC, 2012b).

Diesel

Diesel exhaust is a ubiquitous component of air pollution worldwide. The IARC recently summarized current estimates of the proportionate contribution of diesel exhaust to ambient PM2.5. These ranged from approximately 2% in remote rural locations to more than 30% in cities (IARC, 2013b). The IARC has classified diesel exhaust as a human carcinogen, based on occupational studies and supporting animal and mechanistic data (IARC, 2013b).

Other

The combustion of fossil fuels contributes known or suspected individual carcinogenic chemicals to urban ambient air. In addition to such known human lung carcinogens as arsenic, chromium, and nickel, several others are of note, and are discussed briefly here.

Radionuclides.  Alpha-​emitting radionuclides can also be mea-

sured in outdoor air. These include isotopes of lead, radium, thorium, and uranium (Natusch, 1978) that are naturally present in fossil fuels and are released by combustion. These contribute a relatively small proportion of the radiation dose received by the general population (National Council on Radiation Protection and Measurements, 2009).

1,3-​butadiene.  1,3-​butadiene is a volatile organic compound used

since the 1930s to produce synthetic rubber. Industrial emissions contribute to its presence in some urban areas. However, the major source in the United States is automotive exhaust. Emissions are substantially greater for vehicles with more than two axles (Sapkota and Buckley, 2003). Levels of 1,3-​butadiene in ambient air generally range from 1 to 10 ppb, about 1000-​fold less than occupational exposure levels (IARC, 1992). 1,3-​butadiene has been classified by the IARC as a human carcinogen (Group 1) based largely on animal experiments, which found increased tumor occurrence at multiple sites, including the lung (IARC, 2012b). Epidemiologic studies of occupationally exposed populations

294

294

PART III:  THE CAUSES OF CANCER

(rubber workers and butadiene monomer production workers) have consistently observed increases in hematopoietic cancers, but not cancers of the respiratory system (IARC, 2012b).

EPIDEMIOLOGIC EVIDENCE ON OUTDOOR AIR POLLUTION AND CANCER

Aldehydes.  Various aldehydes are classified as hazardous air pol-

Research Approaches

lutants by the US EPA. Formaldehyde and acetaldehyde are present in urban ambient air largely because of the combustion of gasoline and diesel fuel (Health Effects Institute, 2010). Exposures to formaldehyde and acetaldehyde in outdoor air tend to be highly correlated. Outdoor concentrations of formaldehyde generally range from 1 to 20 μg/​m3. Levels up to 100 μg/​m3 have been observed in heavy traffic or episodes of air pollution (IARC, 2012b). Combustion of alternative fuels, such as methanol, and oxygenated fuels containing the additive MBTE also yields aldehydes (Health Effects Institute, 1993). Formaldehyde is present in indoor as well as outdoor air. Formaldehyde has been classified as a human carcinogen by the IARC and the National Toxicology Program (IARC, 2006, 2012b; National Toxicology Program, 2014), based on evidence from animal experiments and occupationally exposed workers with an excess of nasal and nasopharyngeal cancer and leukemia.

Point Sources Industrial point sources also contribute to air pollution. Electrical power plants fueled by fossil fuels (e.g., coal, oil, and natural gas) emit known and suspected carcinogens (Natusch, 1978). These include metals and metalloids (chromium, nickel, and arsenic), radionuclides such as radon and uranium, and POM such as benzo(a)pyrene. Non-​ferrous metal smelters emit inorganic arsenic and other metals, and sulphur dioxide (Pershagen et  al., 1977). Municipal solid waste incinerators emit heavy metals (e.g., lead and cadmium), PAHs, organic compounds (such as dioxins), and acidic gases (World Health Organization, 1988). Unfortunately, these sources of air pollution are often located in or near poor, working-​class communities. Reports describing cancer rates around point sources have addressed chemical disasters (e.g., dioxin in Seveso, Italy) (Bertazzi et al., 1993), radiation from the nuclear reactor accident at Three Mile Island, Pennsylvania (Hatch et al., 1990, 1991), and other sources (Pershagen, 1990). Such studies have been useful for examining specific issues but are not essential for addressing the broader issue of air pollution and cancer. Methods have been developed for studying point sources (Elliott, 2006).

Fibers Asbestos fibers contaminate ambient air in rural and urban environments, and apparently have been present in the ambient environment for at least 10,000 years. The literature on levels of asbestos in outdoor air was reviewed as part of a comprehensive report on the health effects of asbestos in public buildings (Health Effects Institute et al., 1991). The median levels of asbestos fibers in rural environments, where there are no known natural sources of asbestos, were on the order of 0.01–​0.001 ng/​m3; few individual measurements exceeded 1  ng/​m3. In urban environments, where there is both a greater prevalence of asbestos-​containing materials and a higher frequency of release of fibers from building materials and vehicular brake linings, the median levels ranged from 0.02 to 10 ng/​m3 and a much larger proportion of individual measurements exceeded 1 ng/​m3 (Health Effects Institute et al., 1991). Epidemiologic studies of occupational groups, such as asbestos miners and asbestos textile production workers, who were exposed to asbestos at concentrations several orders of magnitude greater than those cited in the preceding, have consistently observed increased rates of lung cancer. Based on a large body of experimental and epidemiologic evidence, asbestos is considered to be a known human carcinogen (IARC, 2012a). Lung cancer occurrence was not increased in relation to exposure at levels observed in the ambient urban environment in a study of chrysotile-​asbestos mining regions (Camus et al., 1998). The risks of different fiber types remain contentious.

Epidemiologists have used three approaches to examine the role of indoor and outdoor air pollution, in the etiology of lung and other cancers: cohort, case-​control, and ecologic designs. Historically, some of the earliest studies of outdoor air pollution were based on ecological approaches, comparing lung cancer mortality rates in urban and rural areas (Stocks and Campbell, 1955). Ecological studies have also been used in exploratory assessments of the association of indoor radon with lung cancer. However, the recent and most critical evidence comes largely from cohort and case-​control studies, which use more refined approaches for exposure assessment than earlier studies (IARC, 2015). Cohort studies of outdoor air pollution have usually defined exposure based on residence location. Thus, the information on age, gender, potential confounders such as cigarette smoking, and exposure to air pollution were classified at the individual level, whereas the information on the covariates was ecological. Recent studies have refined exposure estimates based on statistical models of the location of residence (Krewski et  al., 2009; Raaschou-​Nielsen et  al., 2013). The case-​control approach provides investigators with an efficient way to estimate the relative risk of lung cancer in relation to air pollution exposure without requiring access to an existing cohort or the collection of information on an entire population. In case-​control studies, cases of lung cancer that occur in the population are identified and classified according to their exposure to outdoor or indoor air pollution; a sample of the study population—​the controls—​is selected and similarly classified according to their exposures. While most recent data come from cohort studies, nested case-​control studies are useful for assessing biomarkers. In contrast to the cohort and case-​control designs, ecologic, or aggregate-​level, studies do not collect information on individual subjects, but instead compare the incidence or mortality rates of lung or other cancers in relation to the levels of air pollution. This approach uses routinely collected data on both lung cancer rates and regional measurements of air pollution. In these studies, surrogate measures of exposure have been used based on geologic features, housing characteristics, or measurements of indoor concentrations. Relative risks can be estimated from ecologic data, but the interpretation of these estimates is more complicated than for estimates derived from cohort and case-​control studies. In most cases, data are not available to take into account inter-​individual and between-​region differences in other lung cancer risk factors. Ecological studies of air pollution are no longer informative. All three designs potentially suffer from a common problem when applied to the study of air pollution and lung cancer: the difficulty of characterizing accurately the subjects’ exposure to air pollution and mixtures of diverse carcinogens inhaled in indoor and outdoor environments. As described earlier, an individual’s exposure to carcinogens in outdoor and indoor environments may be complex and may occur in multiple microenvironments; therefore it may be difficult to estimate exposure for the purpose of epidemiologic analysis. Some of the exposure assessment approaches used are listed in Table 17–​2 (Di et al., 2016; van Donkelaar et al., 2015); these range from categorical indicators, such as self-​report and residence location, to contemporary geographic information systems and statistical models based on monitoring and satellite data, and land use characteristics. The early lung cancer studies often classified exposure to outdoor air pollution based on urban or rural residence, a crude indicator of current or lifetime exposure. In some subsequent studies, approaches based on residence location were refined by more fully capturing the residential history and incorporating duration of residence into exposure indexes. Exposure estimates have been refined by using available monitoring data in combination with air pollution models that incorporate land-​ use information (e.g., roadways and major point sources and, most

 295



Air Pollution

Table 17–​2. Methods of Assessing Exposure Source of Information Source strength Geographical information Dispersion models Outdoor–​indoor penetration Stationary monitoring Questionnaires and interviews Personal monitoring Human samples Toxicological models

Spatial models

Type of Information Emission rate (mass per time), traffic density Distance of place of residence from the source Spatiotemporal concentration distributions from modeling of emission rates, meteorology, air chemistry, geography Modeling from outdoor concentration, building, and ventilation characteristics Concentration over time and by place Source strength, distance from the source, time activity Continuous or cumulated concentrations over time Concentrations of biomarkers of exposure in biospecimens Concentration and dose of pollutants in target organs modeling from concentration, breathing rate, metabolism Monitoring data, land use regression, satellite data

Source: van Donkelaar et al., 2015; Di et al., 2016.

recently, satellite data) to assign concentrations to the residences where study participants live. For example, the European Study of Cohorts for Air Pollution Effects (ESCAPE) pooled data from 17 cohorts across Europe. Measurements were made in each study site, and then a land-​use regression model was developed to estimate concentrations of particulate matter and nitrogen oxides at the home addresses of participants (Raaschou-​Nielsen et al., 2013). These approaches have been continually refined and are now complemented by the use of satellite data to estimate concentrations of airborne particulate matter and other pollutants (Di et al., 2016; van Donkelaar et al., 2015). There has also been extensive attention given to the development of approaches to estimating exposures to indoor carcinogens, particularly secondhand smoke and radon, and, to a lesser extent, asbestos. While the association of secondhand smoke with lung cancer is not specifically covered in this chapter, the conceptual approaches to assessing exposure to secondhand smoke are covered because of their relevance to the topic of air pollution and cancer. For each of these agents, the validity of exposure assessment approaches and the degree of misclassification and resulting bias in risk estimates have figured prominently in the interpretation of the epidemiological findings. The micro-​environmental model has become the model for exposure assessment. For radon, individual-​and population-​level exposures are largely driven by exposures at home, as mentioned earlier. These reflect both the time spent at home and the home concentrations. Radon concentrations can be measured in homes, using passive and relatively inexpensive but accurate measuring devices that record the average concentration across intervals ranging from days to a year. A technique to measure longer-​term exposures has been developed that uses levels of surface radioactivity in glass (Steck et  al., 2002). Issues related to misclassification based on these approaches are considered in the section of this chapter on indoor radon. For secondhand smoke, exposure measurement has involved qualitative assessments based on questionnaires and quantitative assessments based on biomarkers or air measurements (Apelberg et al., 2013; Avila-​Tang et al., 2013a, 2013b). Study participants can describe the smoking status of their spouse, the number of smokers at home, and the extent of exposure at work or in other settings in the recent past. Cotinine and other biomarkers of tobacco smoke provide objective evidence of exposure within the past 3–​7 days. Self-​reported information on exposures in childhood and at various periods throughout life cannot be validated directly, but can provide qualitative information about past exposures. The issue of misclassification was central to the

295

efforts of the tobacco industry to discredit the extensive evidence that secondhand smoke causes lung cancer (Proctor, 2012). All exposure estimates in studies of air pollutants are subject to some misclassification. Usually this misclassification is non-​differential, in that its effects are similar in subjects with and without lung cancer. This form of misclassification attenuates the association between an air pollutant and cancer (Jurek et  al., 2005; Szpiro et  al., 2011)  and can obscure a monotonic relationship with exposure (Birkett, 1992; Dosemeci et al., 1990; Jurek et al., 2005). Studies may differ in the degree of misclassification, however. This can create the misimpression that their results are in conflict (Lubin et  al., 1995a, 1995b; Rothman and Greenland, 1998). Differential misclassification can spuriously elevate as well as attenuate estimates of association. This form of bias is of less concern than confounding, however. Most studies of air pollution attempt to collect data on other lung cancer risk factors, such as cigarette smoking or occupational exposures. Errors in the measurement of potential confounders can introduce bias, even if air pollution exposures are estimated with relatively little error; the result may be either over-​or underestimation of the air pollution effect (Greenland, 1980).

Epidemiologic Studies of Ambient Air Pollution and Lung Cancer The first studies on lung cancer and air pollution compared lung cancer death rates in urban and rural areas. These studies coincided with the early studies of smoking and lung cancer; consequently, some (but not all) attempted to control for smoking. Most studies found overall excesses of lung cancer mortality of about 30%–​40% in the urban areas. The attribution of these excesses to differences in air quality was strengthened by evidence of urban/​rural differences in ambient levels of carcinogens, such as benzo-​(a)-​pyrene, and by the persistence of the urban excess after adjustment for cigarette smoking status in some studies. Later studies have used increasingly sophisticated methods to estimate exposure and control for potential confounders. Detailed descriptions of these studies are provided in the IARC’s Outdoor Air Pollution (Volume 109; IARC, 2015). Groundbreaking findings came from the Harvard Six Cities Study and the American Cancer Society’s (ACS) Cancer Prevention (CPS) II Study, with initial findings published in 1993 and 1995, respectively. Both found increased risk for lung cancer in association with higher levels of exposure to PM2.5, a finding replicated in many subsequent studies. A meta-​analysis of 18 studies, carried out in support of the IARC’s Volume 109, found that lung cancer risk increased by nearly 10% (meta-​relative risk 1.09; 95% CI: 1.04, 1.14) with each 10 μg/​m3 increase of PM2.5 (Figure  17–1) (Hamra et al., 2014). There was no strong evidence for variation by geographic region or smoking status. In most of the cohort studies, the exposures were assigned based on air pollution data at a single point in time or averaged over extended periods. Consequently, most studies have not examined the relationship between air pollution and lung cancer with respect to the timing of exposure. Analyses of extended follow-​ups of both the Six Cities and ACS cohorts suggest that more recent exposures may have the strongest effects on all-​ cause mortality (Krewski et al., 2000). The current evidence suggests that lung cancer attributable to air pollution may occur among both smokers and non-​smokers (Hamra et al., 2014); therefore, residual confounding and effect modification by air pollution must be considered, especially among the smokers. Most studies report air pollution relative risks adjusted for cigarette smoking, but the adjustment may not have controlled completely for potential confounding. Most cohort studies have information on cigarette smoking only at the beginning of follow-​up. The possibility that changes in tobacco use were correlated with air pollution exposure cannot be excluded, since patterns of smoking cessation have varied geographically in the United States. On the other hand, the association of lung cancer with air pollution was largely unaffected in the Six Cities Study (Dockery et al., 1993), when longitudinal information

296

296

PART III:  THE CAUSES OF CANCER Study by region North America

RR (95% CI)

McDonell et al. 2000 Krewski et al. 2009 Hart et al. 2011 Lipsett et al. 2011 Lepeule et al. 2012 Hystad et al. 2013 Jerrett et al. 2013a Puett et al. 2014 Subtotal (/ 2 = 0.0%, p = 0.490)

Weight

1.39 (0.79, 2.46) 1.09 (1.05, 1.13) 1.18 (0.95, 1.48) 0.95 (0.70,1.28) 1.37 (1.07, 1.75) 1.29 (0.95, 1.76) 1.12 (0.91, 1.37) 1.06 (0.90, 1.24) 1.11 (1.05, 1.16)

0.66 21.19 3.77 2.22 3.20 2.14 — 6.48 39.67

0.81 (0.63, 1.04) 1.11 (0.86, 1.43) 1.05 (1.01, 1.10) 1.39 (0.91, 2.13) 1.03 (0.89, 1.20)

3.11 3.07 20.21 1.17 27.56

Cao et al. 2011 Katanoda et al. 2011 Subtotal (/2 = 91.0%, p = 0.001)

1.03 (1.00, 1.07) 1.24 (1.12, 1.37) 1.13 (0.94, 1.34)

21.25 11.52 32.77

Overall (/2 = 53.0%, p = 0.010)

1.09 (1.04, 1.14)

100.00

Europe Beelen et al. 2008 Carey et al. 2013 Cesaroni et al. 2013 Raaschou-Nielsen et al. 2013 Subtotal (/ 2 = 50.0%, p = 0.112)

Other

0.5

1

2

3

Figure 17–​1.  Estimates of lung cancer relative risk associated with relative risks. Source: Hamra et al. (2014).

on cigarette smoking was used in a reanalysis. Several case-​control studies have adjusted for smoking using time-​varying information and still found increased risk for air pollution exposure (e.g., Nyberg et al., 2000). With one exception (Turner et al., 2011), the studies that show increased lung cancer risks among self-​reported never-​smokers have been small and the effect estimates imprecise. However, Turner and colleagues (2011) estimated that long-​term residential exposure to PM2.5 increased lung cancer mortality by 15% to 27% per 10μg/​m3 increase in exposure in 188,699 lifelong never-​smokers in the American Cancer Society’s CPS II cohort (Turner et al., 2011). A subsequent analysis of this cohort by Turner et  al. (2014) found that the effects of joint exposure to PM2.5 and tobacco smoking were greater than additive, corroborating the results of several earlier studies (IARC, 2013c). Limitations of approaches to exposure estimation also contribute to uncertainty in risk estimates. The ACS and Six Cities studies estimated the exposure of each participant based solely on long-​term average concentrations in the metropolitan area of residence. While this approach may accurately reflect exposure to pollutants that are distributed homogenously over large areas for several decades, it does not capture finer spatial and temporal variation; more recent studies in Europe and North America are now incorporating spatial statistical methods to estimate individual long-​term exposure histories, linking residential histories, measurements of traffic density on nearby streets, and long-​term records of specific air pollutants. With this approach, the variations in exposures can be estimated in relation to time and space, perhaps with less misclassification (Hoek et al., 2002; Nyberg et al., 2000; Reynolds et al., 2001). Hoek and colleagues observed larger relative risk estimates for cardiopulmonary mortality among subjects who lived near major roads than were observed in studies that used larger-​scale urban and regional measurements (Hoek et  al., 2002). The highest relative risk estimates were with exposure 20 or more years before diagnosis (Nyberg et al., 2000). By providing exposure estimates at the individual level, these studies also reduce the possibility of aggregate-​level (ecologic) bias.

Risk Attribution Estimates of the population attributable risk of lung cancer due to outdoor air pollution have been based on markedly different methods and vary by an order of magnitude. The early estimates of burden were

low. Doll and Peto based their estimate on past and current estimates of benzo-​(a)-​pyrene in urban air and extrapolation from occupational studies of PAH-​exposed workers (Doll and Peto, 1981). They estimated that less than 1% of future lung cancer in the United States would be due to air pollution from the burning of fossil fuels. They did note, however, that perhaps 10% of the lung cancers then occurring in large cities might have been due to air pollution. In 1990, the US EPA (US EPA, 1990) estimated that 0.2% of all cancer, and probably less than 1% of lung cancer, could be attributed to air pollution. This estimate was obtained by applying the unit risks for over 20 known or suspected human carcinogens found in outdoor air to estimates of the ambient concentrations and numbers of persons potentially exposed. The Global Burden of Disease project provides the most comprehensive worldwide estimates of the global burden of lung cancer due to ambient and household air pollution (Cohen et  al., 2016; Forouzanfar et al., 2015; Institute for Health Metrics and Evaluation, 2016). Exposure to ambient PM2.5 was estimated to have contributed to 387,000 (UI: 350, 000–​420,000) lung cancer deaths (23.6% of the total), and a resultant loss of 8.3 million (uncertainty interval [UI]: 7.5–​ 9.1 million) disability-​adjusted life years (DALY) worldwide in 2013. The estimated global lung cancer mortality rate attributable to ambient PM2.5 increased from 4.64 to 5.41 per 100,000 from 1990 to 2013. Fifty-​two percent (52%) of global lung cancer mortality due to PM2.5 in 2013 was estimated to have occurred in China. When all causes of death were considered, ambient PM2.5 was the seventh ranking global mortality risk factor in 2013. Exposure to ambient PM2.5 contributed to an estimated 2.93  million (UI:  2.78–​ 3.07 million) premature deaths and a loss of 69.7 million (UI: 65.5–​ 75.5 million) DALY in 2013, or 5.3% of total global deaths and 2.9% of global DALY. Sixty percent of this burden occurs in in low-​and middle-​income countries in East and South Asia. The absolute number of deaths estimated as attributable to ambient PM2.5 increased by 31% from 1990 to 2013. Exposure to ozone was estimated to cause an additional 217,000 (UI: 161,000–​272,000) premature deaths and a loss of 5.1 million (UI: 3.6–​6.6 million) DALY in 2013. Worldwide, ambient PM2.5 was the second leading cause of lung cancer mortality in 2013 after tobacco smoking. Household air pollution (HAP) from burning of solid fuels was the eighth ranking global mortality risk factor in 2013 (Institute for Health

 297



Air Pollution

Metrics and Evaluation, 2013). Exposure to HAP contributed to an estimated 2.89 million (UI: 2.46–​3.30 million) premature deaths and 81.1 million (UI: 70.0–​92.8 million) DALY in 2013, largely in low-​ and middle-​income countries in East and South Asia. The absolute number of deaths attributable to HAP increased by 1.3% from 1990 to 2013. HAP was the fifth ranking cause of lung cancer in 2013, contributing to 127,701 (UI: 57,000–​202,900) lung cancer deaths, and 2.9 million (UI: 1.3–​4.6 million) DALY worldwide in 2013, with the preponderance again in low-​and middle-​income countries in East and South Asia (Institute for Health Metrics and Evaluation, 2013). Between 1990 and 2013 the estimated global lung cancer mortality rate (per 100 persons) attributable to HAP declined by 6.1%.

Diesel Exhaust and Cancer of the Lung and Bladder The epidemiological evidence regarding the carcinogenicity of diesel exhaust has been reviewed extensively and repeatedly, partly because of the large numbers of people exposed and the implications for regulation. Most of the evidence comes from studies of occupational exposure rather than general populations, for whom exposure is difficult to estimate accurately (HEI Diesel Epidemiology Panel, 2015; IARC, 2013b). Epidemiologic evidence indicates that workers exposed to diesel exhaust for prolonged periods in a variety of occupational settings are at increased risk of lung cancer. Workers in jobs that entailed prolonged exposure to diesel exhaust in the trucking and railroad industries and in urban transport were shown to have approximately 20%–​50% higher risk of lung cancer mortality in meta-​analyses of studies dating from the 1950s through the mid-​1990s. This increased risk was independent of tobacco smoking (Cohen and Higgins, 1995; Larkin et  al., 2000; Lipsett and Campleman, 1999). Occupational exposure to diesel exhaust is also linked to increased risk of bladder cancer, based on less consistent evidence. The IARC’s Monograph 105 (“Diesel and gasoline engine exhausts and some nitroarenes”) noted that case-​control studies generally reported an increased risk in occupationally exposed workers but cohort studies did not (Cohen and Higgins, 1995; IARC, 2013b). Until recently, few studies had linked lung cancer risk to quantitative estimates of long-​term exposure to diesel exhaust; the resulting uncertainty limited both causal attribution and quantitative risk assessment (Health Effects Institute, 1999). New studies of underground miners and trucking industry workers have linked long-​term exposure to respirable elemental carbon (EC), a component of diesel particles, to lung cancer in these settings. Underground miners exposed to extremely high levels of diesel particles experienced 2-​to 3-​fold increases in lung cancer mortality in analyses that controlled for tobacco smoking; trucking industry workers experienced a 40% increase (Garshick et al., 2012; Silverman et al., 2012). A recent reanalysis of these studies by the Health Effects Institute corroborated the reported results and concluded that the new studies provided a basis for quantitative risk assessment of lung cancer due to diesel exhaust exposure (HEI Diesel Epidemiology Panel, 2015). IARC’s Monograph 105 provides the most recent and comprehensive summary of the numerous cohort and case-​control studies of diesel exhaust and cancer. This included the newer and more definitive studies, and largely corroborated the conclusions of the earlier meta-​ analyses (IARC, 2013b). Diesel exhaust is now considered by IARC to be “carcinogenic to humans” (Group 1), based on epidemiologic and toxicologic evidence (IARC, 2013b). With the finding that diesel exhaust is carcinogenic, emphasis has shifted to quantitative risk assessment (HEI Diesel Epidemiology Panel, 2015). To this end, Vermeulen et al. pooled the results of three cohort studies (two of truckers and one of miners) (Vermeulen et al., 2014). They estimated that lifetime environmental exposure to 0.8 μg/​m3 EC would cause 21 excess lung cancer deaths per 10,000 persons. They further estimated that “[b]‌ased on broad assumptions regarding past occupational and environmental exposures . . . approximately 6% of annual [US and UK] lung cancer deaths may be due to DEE exposure (p. 175).” This estimate is broadly consistent with earlier estimates by US regulatory agencies. Both the EPA

297

(2002) and the California EPA (1998) based their risk estimates on the occupational epidemiology studies available at the time. Both agencies have concluded that the evidence from occupational studies is consistent with a causal relationship with lung cancer. The California EPA calculated a unit risk value (1998), which ranged from 1 to 24 excess deaths in 10,000 people per 1 μg/​m3 diesel PM lifetime exposure. The US EPA estimated a possible range of lung cancer risk from environmental exposure to diesel exhaust (0.1 to 10 excess deaths in 10,000 people per 1 μg/​m3 diesel PM lifetime exposure). The US EPA, however, acknowledged recent improvements in diesel technology, and uncertainty as to the relevance of these risk estimates to present engines, a concern that was recently underscored by the result of a major toxicologic study of the carcinogenicity of emissions from the latest diesel engines. The ACES study of the Health Effects Institute reported “dramatic improvements in emissions and the absence of any significant health effects (especially cancer)” and concluded that “the overall toxicity of exhaust from modern diesel engines is significantly decreased compared to emissions from traditional-​technology diesel engines (p. 5)” (HEI, 2015; US EPA, 2002).

Childhood Cancer and Exposure to Traffic-​Related Air Pollution Motor vehicle exhaust contains a number of known carcinogens, including benzene, a known leukemogen (Chapter 38). In 1989 Savitz and Feingold, using data from an earlier case-​control study of residential exposure to electric and magnetic fields (Savitz and Feingold, 1989), reported that vehicular traffic density at the residence of cancer cases was associated with an increased risk of leukemia and brain cancer among children in Denver. As interest in the health effects of air pollution on children has increased, and spatial statistical methods, such as the geographic information system (GIS), have become more widely available, this relationship has been studied by others with inconsistent results (IARC, 2015). The IARC’s Outdoor Air Pollution (Volume 109)  reviewed studies of all childhood cancers combined, leukemia and lymphoma, and brain tumors. The evidence was considered to be subject to potential publication bias, but indicative of possible weak association.

INDOOR AIR POLLUTION Carcinogens in indoor air have large potential implications for risk because people spend substantial amounts of time indoors. Indoor air pollution may stem from incoming outdoor air or may originate indoors from tobacco smoking, building materials, soil gas, household products, and combustion from heating and cooking (Spengler, 1991). In more developed countries, two of the most important indoor pollutants that influence lung cancer risk in never-​smokers are passive smoking (US Department of Health and Human Services, 1986, 2006) and radon (National Research Council and Committee on Health Risks of Exposure to Radon, 1999). Asbestos exposure may pose a risk to building occupants, but the resulting risk is estimated to be minimal, particularly if asbestos-​containing materials are appropriately managed in place (Health Effects Institute et al., 1991). Other carcinogens are present in indoor air, but have received less attention in high-​income countries. One major concern in economically developing countries is indoor air pollution from the use of unprocessed solid fuels, notably coal, for cooking and space heating (Martin et al., 2013). The high risk of lung cancer among non-​smoking Chinese women has been mentioned earlier.

Asbestos Asbestos, a well-​established occupational carcinogen, includes several forms of fibrous, naturally occurring silicate minerals that have been widely used in products found in homes and public and commercial buildings (Health Effects Institute et al., 1991). The epidemiologic evidence on asbestos and lung cancer dates to the 1950s, although clinical

298

298

PART III:  THE CAUSES OF CANCER

case series had previously led to the hypothesis that asbestos causes lung cancer (Lynch and Smith, 1939; Wedler, 1944). In a retrospective cohort study published in 1955, Doll observed that asbestos textile workers at a factory in the United Kingdom had a 10-​fold elevation in lung cancer risk and that the risk was most heavily concentrated during the time frame before regulations were implemented to limit asbestos dust in factories (Doll, 1955). A 7-​fold excess of lung cancer was subsequently observed among insulation workers in the United States (Selikoff et al., 1964, 1979). The peak incidence occurred 30–​ 35 years after the initial exposure to asbestos (Selikoff et al., 1980). Subsequently, numerous studies of specific worker groups and of the general population have repeatedly documented the carcinogenicity of asbestos (IARC, 2012a). The risk of lung cancer has been noted to increase with increased exposure to asbestos and to be associated with the principal commercial forms of asbestos. Asbestos and cigarette smoking are both independent causes of lung cancer; in combination, they act synergistically to increase risk in a manner compatible with a multiplicative effect (Hammond et al., 1979; IARC, 2004). With the recognition in the 1970s and 1980s that asbestos-​containing materials were prevalent in indoor environments and a potential source of airborne fibers, concern was raised as to risks to the general population for both mesothelioma and lung cancer. In the United States, programs were initiated to either remove or manage asbestos-​containing materials to limit releases of fibers. Because the concentrations were generally very low, the risks could not be directly addressed using epidemiological approaches. A  risk assessment was carried out by the Health Effects Institute that used all available measurements of indoor asbestos fiber concentrations and a concentration–​risk relationship from studies of workers (Health Effects Institute et  al., 1991). The concentration information considered by the Health Effects Institute showed that exposures in indoor environments were generally quite low and were not elevated in comparison with levels in urban outdoor air. The US EPA estimates that an individual who continuously breathed air containing asbestos at an average of 0.000004 fibers/​cm3 over his or her entire lifetime would theoretically have no more than a 1 in 1 million increased risk of developing cancer from this exposure. Similarly, the EPA estimates that breathing air containing 0.00004 fibers/​cm3 would result in not greater than a 1 in 100,000 increased chance of developing cancer, and air containing 0.0004 fibers/​cm3 would result in not greater than a 1 in 10,000 increased chance of developing cancer (US EPA, 1999). This risk was considered sufficiently low as to not warrant removal of asbestos from buildings. The EPA now promotes in-​place management of asbestos-​containing materials.

Radon Radon is an inert gas produced naturally from radium in the decay series of uranium. It decays into a series of relatively short-​lived particle progeny; two of the short-​lived decay products emit alpha particles that can, by virtue of their high energy and mass, damage DNA in respiratory epithelial cells. Laboratory experiments involving exposure of cells to single alpha particles show that even one “hit” from an alpha particle causes permanent change in a cell. Even though cancer is a multistage phenomenon, this finding implies that radon exposure may result in increased risk at any level (National Research Council and Committee on Health Risks of Exposure to Radon, 1999). Research on radon is now quite consistent regarding carcinogenicity. The various lines of evidence (dosimetric, in vitro, and epidemiological) consistently support a linear no-​threshold risk relationship. Epidemiologic studies of underground miners of uranium and other ores have long established that exposure to radon and its decay products causes lung cancer (Lubin et al., 1995a; National Research Council and Committee on Health Risks of Exposure to Radon, 1999; National Research Council and Committee on the Biological Effects of Ionizing Radiation, 1988). In fact, radon exposure was probably the first occupational respiratory carcinogen to be identified, based on the extremely high rates of lung cancer documented in the underground metal miners of Schneeberg and Joachimsthal and the high levels of radon measured in the mines (National Research Council

and Committee on the Biological Effects of Ionizing Radiation, 1988; Proctor, 2012). While lesser risks have been observed for more recent worker cohorts, the newer epidemiological studies show clear evidence of a continued increase in cancer risk (Hunter et al., 2013; Tirmarche et al., 2012). Cigarette smoking and radon decay products synergistically influence lung cancer risk in a manner that is supra-​ additive but sub-​multiplicative (Lubin et al., 1995a; National Research Council and Committee on Health Risks of Exposure to Radon, 1999). Radon is of broad societal interest because it is a ubiquitous indoor air pollutant, entering buildings in soil gas. On average, indoor exposures to radon for the general population are much less than those received by occupational groups such as uranium miners. For example, even the lowest historical radon concentration in a uranium mine is still roughly an order of magnitude higher than in the average home (Lubin et  al., 1995a). However, measurements of indoor radon concentrations in homes show a log-​normal distribution with all homes having some radon detectable at an average around 1 picocurie per liter (37 becquerels per m3), but with a highly skewed tail of the distribution that corresponds to exposures in uranium mines. As a basis for developing control strategies, risk assessments have estimated the magnitude of the problem and the extent to which exposures must be reduced to protect public health. Because of the prevalence of low and average exposures in the population, these are of greatest concern. After the problem of indoor radon was first widely recognized in the early 1980s, case-​control studies were carried out to estimate the risks directly. These studies complemented the information from cohort studies of underground miners, which generally had much higher levels of exposure. Most studies incorporated measurements of radon concentrations in the homes of lung cancer cases and controls. The interpretation of the findings was limited by exposure misclassification, a particularly difficult problem in estimating lifetime exposure (Lubin et al., 1995a). Measurement of implanted polonium-​210 in the surface of glass (e.g., window glass or glass covering a picture) has been proposed as an indicator of long-​term concentration and has been applied in several epidemiological studies (Brownson and Alavanja, 2000; Field et  al., 2002; Steck et  al., 2002). Nonetheless, the risk estimates from individual case-​control studies have been relatively imprecise. A  meta-​analysis of the findings through the mid-​1990s did show evidence for a significant and positive exposure–​response relationship that was consistent with the extrapolated risks in the miner studies (National Research Council and Committee on Health Risks of Exposure to Radon, 1999). Subsequent pooled analyses of the European and North American case-​control studies provide risk estimates that are also consistent with the risk models based on the underground miner studies (Darby et al., 2005; Krewski et al., 2005). For the purpose of risk assessment and policy formulation, risk models have been developed based on the findings in the underground miners (US EPA, 2003). There are inherent uncertainties in extending such models to the general population, particularly the extrapolation of findings from highly exposed uranium miners to the much lower exposures indoors. The studies of miners lacked information on women and children and had little control over other differences between mines and homes, such as exposure to diesel exhaust. Effect modification by smoking added further uncertainty; the evidence from the miners indicated synergy between smoking and radon, but the combined effect was sub-​multiplicative (National Research Council and Committee on Health Risks of Exposure to Radon, 1999). From the policy perspective, the most critical uncertainty is whether radon causes lung cancer at all exposures—​that is, is there a threshold exposure below which radon does not cause cancer? Biological evidence supports the assumption that a single hit to a cell by an alpha particle causes permanent cellular change, an assumption leading to a non-​threshold dose–​response relationship (Hei et  al., 1994, 1997; National Research Council and Committee on Health Risks of Exposure to Radon, 1999). Additionally, the case-​control study findings to date provide evidence for increased risk down to levels of exposure not substantially greater than that associated with typical indoor concentrations. When combined in a meta-​analysis, studies of the general population show a significant association between indoor radon and lung cancer in the general population that is quantitatively

 29



Air Pollution

compatible with risk models derived from the underground miners. This coherence lends support to using extrapolation of the miner data with a linear, non-​threshold model to estimate the risk of indoor radon. Consequently, the risk model developed by the National Research Council’s Biological Effects of Ionizing Radiation (BEIR) VI Committee (National Research Council and Committee on Health Risks of Exposure to Radon, 1999) and subsequently applied by the US EPA incorporates a linear model without a threshold. The assumptions made by the Environmental Protection Agency (US EPA, 1992b) and the Biological Effects of Ionizing Radiation (BEIR) IV and VI Committees of the National Research Council (National Research Council and Committee on Health Risks of Exposure to Radon, 1999; National Research Council and Committee on the Biological Effects of Ionizing Radiation, 1988)  led to estimates that approximately 21,000 lung cancer deaths per year in the United States are attributable to radon (with an uncertainty range of 8000 to 45,000), an estimate that makes indoor radon the second-​leading cause of lung cancer (US EPA, 2003). The story of research on radiation and lung cancer is relatively complete and illustrates how epidemiological evidence can inform policy decisions.

Secondhand Smoke In 1981 reports were published from Japan (Hirayama, 1981)  and from Greece (Trichopoulos et  al., 1981)  that indicated increased lung cancer risk in non-​smoking women married to cigarette smokers. This once controversial association has subsequently been examined in numerous investigations in the United States and other countries. A causal relationship between secondhand smoke (SHS) and lung cancer is biologically plausibile, given the many carcinogens in sidestream smoke and the lack of a documented threshold dose for respiratory carcinogens in active smokers (IARC, 2004; US Department of Health and Human Services, 2006, 2010). Moreover, genotoxic activity has been demonstrated for many components of SHS (Claxton et  al., 1989; Hillerdal, 1996; IARC, 2004; Lofroth, 1989; Weiss, 1989). Experimental exposure of non-​ smokers to SHS has validated the excretion of 4-​ (methylnitrosamino)-​ 1-​ (3-​ pyridyl)-​1 butanol (NNAL), a tobacco-​specific carcinogen, in urine as a valid biomarker of exposure to tobacco smoke (Anderson et al., 2001; Hecht et al., 1993). Non-​smokers exposed to SHS also have increased concentrations of adducts of tobacco-​related carcinogens (Crawford et al., 1994; Maclure et al., 1989). Through the 1980s and 1990s, the tobacco industry sought to maintain controversy and uncertainty about the carcinogenicity of SHS (Proctor, 2012). In fact, the US courts found the tobacco industry guilty of fraud and deception about the risks of SHS exposure (Kessler, 2006). There are long-​standing authoritative conclusions that SHS causes lung cancer. By 1986, the evidence had mounted, and three consensus reports published in the same year concluded that SHS was a cause of lung cancer. The IARC (1986) concluded that “passive smoking gives rise to some risk of cancer.” In its monograph on tobacco smoking, the agency supported this conclusion on the basis of the characteristics of sidestream and mainstream smoke, the absorption of tobacco smoke materials during involuntary smoking, and the nature of dose–​response relationships for carcinogenesis. In the same year, the National Research Council (1986) and the US Surgeon General (1986) also concluded that involuntary smoking increases the incidence of lung cancer in non-​smokers. In reaching this conclusion, the National Research Council (1986) cited the biological plausibility of the association between exposure to SHS and lung cancer and the supporting epidemiological evidence. Based on a pooled analysis of the epidemiological data adjusted for bias, the report concluded that the best estimate for the excess risk of lung cancer in non-​smokers married to smokers was 25%. The 1986 report of the Surgeon General (1986) characterized involuntary smoking as a cause of lung cancer in non-​smokers. This conclusion was based on the extensive information already available about the carcinogenicity of active smoking, the qualitative chemical similarities between SHS and mainstream smoke, and the epidemiological data on involuntary smoking.

299

In 1992 the US EPA (1992a) published its risk assessment of SHS as a carcinogen. The agency’s evaluation drew on the toxicologic evidence on SHS and the extensive literature on active smoking. A meta-​ analysis of the 31 studies published to that time was central in the decision to classify SHS as a class A carcinogen—​namely, a known human carcinogen. Overall, the analysis found a significantly increased risk of lung cancer in never-​smoking women married to smoking men; for the studies conducted in the United States, the estimated relative risk was 1.19 (90% CI: 1.04, 1.35). Subsequent conclusions reaffirmed those of the earlier reports. A 2004 report by the IARC (2004) reiterated the conclusion with the following statement: “evidence is sufficient to conclude that involuntary smoking is a cause of lung cancer in never-​smokers. The magnitudes of the observed risks are reasonably consistent with predictions based on studies of active smoking in many populations.” The 2006 report of the US Surgeon General (p.  15), which focused on SHS, concluded: “1. The evidence is sufficient to infer a causal relationship between secondhand smoke exposure and lung cancer among lifetime nonsmokers. This conclusion extends to all secondhand smoke exposure, regardless of location and 2. The pooled evidence indicates a 20 to 30 percent increase in the risk of lung cancer from secondhand smoke exposure associated with living with a smoker” (US Department of Health and Human Services, 2006). The burden of lung cancer attributable to passive smoking has been estimated as 31,620 deaths globally (Global Burden of Disease Study 2013 Collaborators, 2015).

EXPOSURE IN LOW-​AND MIDDLE-​INCOME COUNTRIES Current knowledge about ambient air pollution and lung cancer is based largely on the experience of populations of Western industrialized nations. The levels of exposures in low-​and middle-​income countries (LMICs) in Asia, Africa, and the Middle-​East rival and often exceed those commonly observed in high-​income countries. Some LMICs face a double air pollution challenge: the first from increasing levels of ambient air pollution due to urbanization and industrialization, and the second from a continuing reliance on solid fuels for domestic energy needs. The latter is most common among the rural poor, who still lack access to clean energy (Balakrishnan et al., 2014). Although the proportion of the global population that uses solid fuels such as coal or biomass for household cooking or heating has declined over the past three decades, the populations have grown and the number of people who rely on solid fuels has remained largely unchanged at 2.8 billion. These exposures occur mostly in sub-​Saharan Africa and South and East Asia. Forty-​one percent (41%) of global households and 60% percent of households in sub-​Saharan Africa and South Asia were estimated to have burned solid fuels in 2010 (Bonjour et  al., 2013). Indoor burning of solid fuels has also been associated with increased rates of cancer of the upper airways. Exposure to household air pollution is highest in women due to their role in food preparation (Forouzanfar et al., 2015). Exposures to levels of fine particles (PM2.5) and other health-​ damaging air pollutants, such as carbon monoxide, are many-​fold higher in households where solid fuels are burned in unventilated living areas. The emissions from the combustion of solid fuels include known human carcinogens such as polycyclic aromatic hydrocarbons, formaldehyde, and benzene (IARC, 2010a). Household combustion of solid fuels also contributes to ambient air pollution in settings where solid fuels are widely used, and was estimated to be responsible for 12% of global population-​weighted PM2.5, 37% and 26% in sub-​Saharan Africa and South Asia, respectively, in 2010 (Chafe et al., 2014). Indoor burning of coal causes lung cancer in humans and is associated with a 2-​fold increase in lung cancer risk (Hosgood et al., 2011; IARC, 2010a). The risk from indoor combustion of low-​grade “smoky” coal approximates the risks from active smoking. Indoor burning of biomass, such as wood and dung, is also associated with increased lung cancer (Bruce et  al., 2015; Hosgood et  al., 2010). In addition, cooking-​oil emissions from high-​temperature frying may also cause

30

300

PART III:  THE CAUSES OF CANCER

lung cancer (IARC, 2010a). In some parts of China, household coal burning has resulted in markedly elevated lung cancer rates among non-​smoking women, and the venting of emissions and introduction of cleaner burning stoves have been demonstrated to reduce lung cancer rates in exposed women (Hosgood et al., 2011; Lan et al., 2002). The use of cleaner fuels, such as compressed natural gas, is associated with reduced rates of lung cancer and other adverse health outcomes relative to the use of coal and biomass. There has been little research on ambient air pollution and lung cancer among urban residents of the burgeoning cities in LMICs, although mounting levels of urban air pollution from local stationary and, increasingly, mobile sources are recognized as an important environmental problem (World Health Organization [WHO], 2002). In the cities of the poorest developing countries, WHO’s Global Environmental Monitoring System observed average ambient concentrations of total suspended particles of 300 μg/​m3; the exposure in places where coal is used for fuel, such as poor communities in South Africa, may exceed 1 mg/​m3 (WHO, 1990). Estimated levels of PM10 in cities in developing countries also exceed those commonly encountered in Europe and North America (Pandey et  al., 2006). One might predict that the high levels of ambient air pollution found in cities in the developing world would be associated with greater excess lung cancer occurrence than has been observed in Western industrialized settings. Although there are currently few relevant investigations, a case-​control study in Shenyang, China, observed a 200% increase in lung cancer risk after adjustment for age, education, and smoking, among residents in “smoky” areas of the city and a 50% increase among those in “somewhat or slightly smoky” areas (Xu et al., 1989) As a greater proportion of the world’s population moves from rural communities to the rapidly expanding and highly polluted cities of Asia and the Southern Hemisphere, there is a clear need for research on outdoor air pollution in the developing world. Epidemiologic studies in these settings will present even greater challenges than those in the industrialized West. In addition to the generic problem of estimating long-​term exposure to air pollution discussed earlier, the ambient air pollution mixture in urban centers in the developing countries is changing, due in part to the increase in vehicular traffic. Careful planning will be required to characterize these changes over time, including the choice and development of methods to measure indicator pollutants for different sources. In addition, the current dramatic increases in cigarette smoking in the developing world (Peto and Lopez, 2003), and the thoroughly predictable consequences, will complicate the interpretation of studies of air pollution and lung cancer.

FUTURE DIRECTIONS There is now abundant evidence that specific air pollutants increase lung cancer risk in smokers and non-​smokers. Carcinogens can be measured in indoor and outdoor environments; toxicologic data indicate the potential for human carcinogenicity. Epidemiologic research shows evidence of effects of indoor and outdoor air pollution on lung cancer risk, albeit weak for some agents. Various indoor air pollutants are confirmed causes of cancer, as is outdoor air pollution generally. Research is still needed on air pollution and lung cancer to guide policies for protection of public health. Critical remaining questions relate to the magnitude of the risk at the lower pollution levels now present in many high-​income countries and to the specific pollutants leading to increased lung cancer risk. Direct epidemiologic observation of exposed populations can provide the best information for evaluating the magnitude of outdoor air pollution–​related excess lung cancer. In general, large-​scale epidemiologic studies of air pollution and lung cancer are needed if we are to obtain sufficiently informative and precise data on risks. Large numbers of cases will be necessary to measure the effects of air pollution among women and ethnic minorities and to measure the joint effects of air pollution and other factors, such as occupation and smoking. Evidence is particularly needed for low-​and middle-​income countries.

References Anderson KE, Carmella SG, Ye M, et  al. 2001. Metabolites of a tobacco-​ specific lung carcinogen in nonsmoking women exposed to environmental tobacco smoke. J Natl Cancer Inst, 93(5), 378–​381. PMID: 11238699. Apelberg BJ, Hepp LM, Avila-​Tang E, et  al. 2013. Environmental monitoring of secondhand smoke exposure. Tob Control, 22(3), 147–​ 155. PMCID: PMC3639351. Armstrong B, Hutchinson E, Unwin J, and Fletcher T. 2004. Lung cancer risk after exposure to polycyclic aromatic hydrocarbons:  a review and meta-​analysis. Environ Health Perspect, 112(9), 970–​ 978. PMCID: PMC1247189. Autrup H, Daneshvar B, Dragsted LO, et al. 1999. Biomarkers for exposure to ambient air pollution: comparison of carcinogen-​DNA adduct levels with other exposure markers and markers for oxidative stress. Environ Health Perspect, 107(3), 233–​238. Avila-​Tang E, Al-​Delaimy WK, Ashley DL et  al. 2013a. Assessing secondhand smoke using biological markers. Tob Control, 22(3), 164–​ 171. PMCID: PMC3639350. Avila-​Tang E, Elf JL, Cummings KM, et  al. 2013b. Assessing secondhand smoke exposure with reported measures. Tob Control, 22(3), 156–​163. PMCID: PMC3639349. Balakrishnan K, Cohen A, and Smith KR. 2014. Addressing the burden of disease attributable to air pollution in India: the need to integrate across household and ambient air pollution exposures. Environ Health Perspect, 122(1), A6–​A7. PMCID: PMC3888581. Bertazzi A, Pesatori AC, Consonni D, et al. 1993. Cancer incidence in a population accidentally exposed to 2,3,7,8-​ tetrachlorodibenzo-​ para-​ dioxin. Epidemiology, 4(5), 398–​406. PMID: 8399687. Birkett NJ. 1992. Effect of nondifferential misclassification on estimates of odds ratios with multiple levels of exposure. Am J Epidemiol, 136(3), 356–​362. PMID: 1415154. Bonjour S, Adair-​Rohani H, Wolf J, et al. 2013. Solid fuel use for household cooking: country and regional estimates for 1980-​2010. Environ Health Perspect, 121(7), 784–​790. PMCID: PMC3701999. Brauer M, Amann M, Burnett RT, et al. 2012. Exposure assessment for estimation of the global burden of disease attributable to outdoor air pollution. Environ Sci Technol, 46(2), 652–​660. PMCID: PMC4043337. Brauer M, Freedman G, Frostad J, et al. 2016. Ambient air pollution exposure estimation for the Global Burden of Disease 2013. Environ Sci Technol, 50(1), 79–​88. PMID: 26595236. Brownson RC, and Alavanja MC. 2000. Previous lung disease and lung cancer risk among women (United States). Cancer Causes Control, 11(9), 853–​ 858. PMID: 11075875. Bruce N, Dherani M, Liu R, et al. 2015. Does household use of biomass fuel cause lung cancer? A systematic review and evaluation of the evidence for the GBD 2010 study. Thorax, 70(5), 433–​441. PMID: 25758120. California Environmental Protection Agency. 1998. Health Risk Assessment for Diesel Exhaust: Proposed Identification of Diesel Exhaust as a Toxic Air Contaminant. Sacramento:  Office of Environmental Health Hazard Assessment, California Environmental Protection Agency. Camus M, Siemiatycki J, and Meek B. 1998. Nonoccupational exposure to chrysotile asbestos and the risk of lung cancer. N Engl J Med, 338(22), 1565–​1571. PMID: 9603793. Chafe ZA, Brauer M, Klimont Z, et al. 2014. Household cooking with solid fuels contributes to ambient PM2.5 air pollution and the burden of disease. Environ Health Perspect, 122(12), 1314–​1320. PMCID: PMC4256045. Claxton LD, Morin RS, Hughes TJ, and Lewtas J. 1989. A genotoxic assessment of environmental tobacco smoke using bacterial bioassays. Mutat Res, 222(2), 81–​99. PMID: 2645519. Cohen AJ, Brauer M, and Burnett R. 2016. The Global Burden of Disease attributable to ambient air pollution: estimates of current burden and 23-​ year trends from the GBD 2013 Study. Cohen AJ, and Higgins MWP. 1995. Health Effects of Diesel Exhaust: Epidemiology Diesel Exhaust: A Critical Analysis of Emissions, Exposure, and Health Effects. A Special Report of the Institute’s Working Group (pp. 251–​292). Cambridge, MA: Health Effects Institute. Crawford FG, Mayer J, Santella RM, et al. 1994. Biomarkers of environmental tobacco smoke in preschool children and their mothers. J Natl Cancer Inst, 86(18), 1398–​1402. PMID: 8072033. Darby S, Hill D, Auvinen A, et  al. 2005. Radon in homes and risk of lung cancer: collaborative analysis of individual data from 13 European case-​ control studies. BMJ, 330(7485), 223. PMCID: PMC546066. DeMarini DM. 2013. Genotoxicity biomarkers associated with exposure to traffic and near-​road atmospheres:  a review. Mutagenesis, 28(5), 485–​ 505. PMID: 23945473. DeMarini DM, Shelton ML, Abu-​Shakra A, Szakmary A, and Levine JG. 1998. Spectra of spontaneous frameshift mutations at the hisD3052 allele of

 301



Air Pollution

Salmonella typhimurium in four DNA repair backgrounds. Genetics, 149(1), 17–​36. PMCID: PMC1460133. Di Q, Kloog I, Koutrakis P, et  al. 2016. Assessing PM exposures with high spatio-​temporal resolution across the continental United States. Environ Sci Technol, 50(9), 4712–​4721. PMID: 27023334. Dockery DW, Pope CA, 3rd, Xu X, et al. 1993. An association between air pollution and mortality in six U.S. cities. N Engl J Med, 329(24), 1753–​1759. PMID: 8179653. Doll R. 1955. Mortality from lung cancer in asbestos workers. Br J Ind Med, 12(2), 81–​86. PMCID: PMC1037613. Doll R, and Hill AB. 1952. A study of the aetiology of carcinoma of the lung. Br Med J, 2(4797), 1271–​1286. PMCID: PMC2022425. Doll R, and Peto R. 1981. The causes of cancer: quantitative estimates of avoidable risks of cancer in the United States today. J Natl Cancer Inst, 66(6), 1191–​1308. PMID: 7017215. Doll R, Vessey MP, Beasley RW, et  al. 1972. Mortality of gasworkers -​final report of a prospective study. Br J Ind Med, 29(4), 394–​406. PMCID: PMC1009455. Dosemeci M, Wacholder S, and Lubin JH. 1990. Does nondifferential misclassification of exposure always bias a true effect toward the null value? Am J Epidemiol, 132(4), 746–​748. PMID: 2403115. Elliott P. 2006. Point sources of air pollution: investigation of possible health effects using small area methods. In: Ayres J, Maynard R, and Richards R (Eds.), Air Pollution and Health (Vol. 3, pp. 49–​67). London: Imperial College Press. Field RW, Smith BJ, Steck DJ, and Lynch CF. 2002. Residential radon exposure and lung cancer:  variation in risk estimates using alternative exposure scenarios. J Expo Anal Environ Epidemiol, 12(3), 197–​203. PMID: 12032816. Forouzanfar MH, Alexander L, Anderson HR, et al. 2015. Global, regional, and national comparative risk assessment of 79 behavioural, environmental and occupational, and metabolic risks or clusters of risks in 188 countries, 1990–​2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet, 386(10010), 2287–​2323. PMCID: PMC4685753. Garshick E, Laden F, Hart JE, et al. 2012. Lung cancer and elemental carbon exposure in trucking industry workers. Environ Health Perspect, 120(9), 1301–​1306. PMCID: PMC3440130. Georgiadis P, Topinka J, Stoikidou M, et al. 2001. Biomarkers of genotoxicity of air pollution (the AULIS project): bulky DNA adducts in subjects with moderate to low exposures to airborne polycyclic aromatic hydrocarbons and their relationship to environmental tobacco smoke and other parameters. Carcinogenesis, 22(9), 1447–​1457. PMCID: PMC11532867. Global Burden of Disease Study 2013 Collaborators. 2015. Global, regional, and national incidence, prevalence, and years lived with disability for 301 acute and chronic diseases and injuries in 188 countries, 1990–​2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet, 386(9995), 743–​800. PMCID: PMC4561509. Gordon SB, Bruce NG, Grigg J, et al. 2014. Respiratory risks from household air pollution in low and middle income countries. Lancet Respir Med, 2(10), 823–​860. PMCID: PMC5068561. Greenberg A. 1987. Analyses of polycyclic aromatic hydrocarbons. In:  Lioy PJ, and Daisey JM (Eds.), Toxic Air Pollution: A Comprehensive Study of Non-​Criteria Pollutants (pp. 93–​118). Boca Raton, FL: Lewis Publishers. Greenland S. 1980. The effect of misclassification in the presence of covariates. Am J Epidemiol, 112(4), 564–​569. PMID: 7424903. Hammond EC, Selikoff IJ, and Seidman H. 1979. Asbestos exposure, cigarette smoking and death rates. Ann N Y Acad Sci, 330, 473–​490. PMID: 294198. Hamra GB, Guha N, Cohen A et al. 2014. Outdoor particulate matter exposure and lung cancer: a systematic review and meta-​analysis. Environ Health Perspect, 122(9), 906–​911. PMCID: PMC4154221. Hatch MC, Beyea J, Nieves JW, and Susser M. 1990. Cancer near the Three Mile Island nuclear plant: radiation emissions. Am J Epidemiol, 132(3), 397–​412; discussion 413–​397. PMID: 2389745. Hatch MC, Wallenstein S, Beyea J, Nieves JW, and Susser M. 1991. Cancer rates after the Three Mile Island nuclear accident and proximity of residence to the plant. Am J Public Health, 81(6), 719–​724. PMCID: PMC1405170. Health Effects Institute. 1993. Research Priorities for Mobile Air Toxics [HEI Communication Number 2]. Cambridge, MA: Health Effects Institute. Health Effects Institute. 1999. Diesel Emissions and Lung Cancer: Epidemiology and Quantitative Risk Assessment. Cambridge, MA:  Health Effects Institute. Health Effects Institute. 2010. Special Report 17:  Traffic-​ Related Air Pollutions: A Critical Review of the Literature on Emissions, Exposure, and Health Effects. Boston: Health Effects Institute. Health Effects Institute, Asbestos Research Committee, and Literature Review Panel. 1991. Asbestos in Public and Commercial Buildings: A Literature

301

Review and Synthesis of Current Knowledge (p. 227). Cambridge, MA: Health Effects Institute. Hecht SS, Carmella SG, Murphy SE, et al. 1993. A tobacco-​specific lung carcinogen in the urine of men exposed to cigarette smoke. N Engl J Med, 329(21), 1543–​1546. HEI Diesel Epidemiology Panel. 2015. Diesel emissions and lung cancer: an evaluation of recent epidemiological evidence for quantitative risk assessment. Res Rep Health Eff Inst, Spec No. 19(19), 1–​149. PMID: 26846065. Hei TK, Piao CQ, Willey JC, Thomas S, and Hall EJ. 1994. Malignant transformation of human bronchial epithelial cells by radon-​simulated alpha-​ particles. Carcinogenesis, 15(3), 431–​437. PMCID: PMC20515. Hei TK, Wu LJ, Liu SX et al. 1997. Mutagenic effects of a single and an exact number of alpha particles in mammalian cells. Proc Natl Acad Sci U S A, 94(8), 3765–​3770. PMCID: PMC20515. Hillerdal G. 1996. Long-​term survival of patients with lung cancer from a defined geographical area before and after radiological screening. Lung Cancer, 15(1), 21–​30. PMID: 8865120. Hirayama T. 1981. Non-​smoking wives of heavy smokers have a higher risk of lung cancer: a study from Japan. Br Med J (Clin Res Ed), 282(6259), 183–​185. PMCID: PMC1503989. Hoek G, Brunekreef B, Goldbohm S, Fischer P, and van den Brandt PA. 2002. Association between mortality and indicators of traffic-​related air pollution in the Netherlands: a cohort study. Lancet, 360(9341), 1203–​1209. PMCID: PMC12401246. Hoffman FL. 1929. Cancer of the lungs. Am Rev Tuberc, 19, 392–​406. Hoffman FL. 1931. Cancer and smoking habits. Annals Surg, 93(1), 50–​67. PMCID: PMC1398760. Hosgood HD, 3rd, Boffetta P, Greenland S, et  al. 2010. In-​home coal and wood use and lung cancer risk:  a pooled analysis of the International Lung Cancer Consortium. Environ Health Perspect, 118(12), 1743–​1747. PMCID: PMC3002194. Hosgood HD, 3rd, Wei H, Sapkota A, et al. 2011. Household coal use and lung cancer: systematic review and meta-​analysis of case-​control studies, with an emphasis on geographic variation. Int J Epidemiol, 40(3), 719–​728. PMCID: PMC3147068. Hunter N, Muirhead CR, Tomasek L, et  al. 2013. Joint analysis of three European nested case-​ control studies of lung cancer among radon exposed miners:  exposure restricted to below 300 WLM. Health Phys, 104(3), 282–​292. PMID: 23361424. International Agency for Research on Cancer (IARC). 1986. Tobacco smoking. IARC Monogr Eval Carcinog Risk Chem Hum, 38, 35–​ 394. PMID: 3460963. IARC. 1992. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans:  Occupational Exposures to Mists and Vapours from Strong Inorganic Acids; And Other Industrial Chemicals. Lyon, France: International Agency for Research on Cancer. IARC. 2004. Tobacco smoke and involuntary smoking. IARC Monogr Eval Carcinog Risks Hum, 83, 1–​1438. PMCID: PMC4781536. IARC. 2006. Formaldehyde, 2-​ butoxyethanol and 1-​ tert-​ butoxypropan-​ 2-​ol. IARC Monogr Eval Carcinog Risks Hum, 88, 1–​ 478. PMCID: PMC4781641. IARC. 2010a. Household use of solid fuels and high-​temperature frying. IARC Monogr Eval Carcinog Risks Hum, 95, 1–​430. PMCID: PMC5046077. IARC. 2010b. Some non-​heterocyclic polycyclic aromatic hydrocarbons and some related exposures. IARC Monogr Eval Carcinog Risks Hum, 92, 1–​853. PMCID: PMC4781319. IARC. 2012a. Arsenic, metals, fibres, and dusts. IARC Monogr Eval Carcinog Risks Hum, 100(Pt C), 11–​465. PMCID: PMC4781271. IARC. 2012b. Chemical agents and related occupations. IARC Monogr Eval Carcinog Risks Hum, 100(Pt F), 9–​562. PMCID: PMC4781612. IARC. 2013a. Bitumens and bitumen emmissions, and some N-​and S-​ heterocyclic polycyclic aromatic hydrocarbons. IARC Monogr Eval Carcinog Risks Hum, 103, 9–​303. PMCID: PMC4781167. IARC. 2013b. Diesel and gasoline engine exhausts and some nitroarenes. IARC Monogr Eval Carcinog Risks Hum, 105, 9–​699. PMCID: 4781216. IARC. 2013c. IARC Scientific Publication No. 161. Air Pollution and Cancer. Press W Ed. Geneva: World Health Organization. IARC. 2015. Outdoor Air Pollution. Vol. 109. Lyon, France:  International Agency for Research on Cancer. Institute for Health Metrics and Evaluation. 2013. GBD Compare. Available from: https://​vizhub.healthdata.org/​gbd-​compare/​. Accessed August 22, 2016. Institute for Health Metrics and Evaluation. 2016. Global Burden of Disease (GBD). Available from: http://​www.healthdata.org/​gbd. Accessed August 22, 2016. Jurek AM, Greenland S, Maldonado G, and Church TR. 2005. Proper interpretation of non-​differential misclassification effects: expectations vs observations. Int J Epidemiol, 34(3), 680–​687. PMID: 15802377.

302

302

PART III:  THE CAUSES OF CANCER

Katsouyanni K. 2013. Geographical distribution of air pollutants. In: Straif K, Cohen A, and Samet J (Eds.), IARC Scientific Publication No. 161, Air Pollution and Cancer (p. 15). Lyon, France: WHO Press. Kessler G. 2006. Final Opinion Civil Action No. 99–​2496 (GK). Available from:  http://​www.justice.gov/​sites/​default/​files/​civil/​legacy/​2014/​09/​11/​ amended opinion_​0.pdf Accessed on June 8, 2015. Kinney PL, Aggarwal M, Northridge ME, Janssen NA, and Shepard P. 2000. Airborne concentrations of PM(2.5) and diesel exhaust particles on Harlem sidewalks:  a community-​ based pilot study. Environ Health Perspect, 108(3), 213–​218. PMCID: PMC1637978. Kinney PL, Chillrud SN, Ramstrom S, Ross J, and Spengler JD. 2002. Exposures to multiple air toxics in New  York City. Environ Health Perspect, 110 Suppl 4, 539–​546. PMCID: PMC1241202. Krewski D, Burnett RT, Goldberg MS, et al. 2000. Reanalysis of the Harvard Six Cities Study and the American Cancer Society Study of Particulate Air Pollution and Mortality. Investigators’ Reports Parts I and II. Cambridge, MA: Health Effects Institute. Krewski D, Jerrett M, Burnett RT, et  al. 2009. Extended follow-​up and spatial analysis of the American Cancer Society study linking particulate air pollution and mortality. Res Rep Health Eff Inst 140, 5–​114; discussion 115–​136. PMID: 19627030. Krewski D, Lubin JH, Zielinski JM, et al. 2005. Residential radon and risk of lung cancer: a combined analysis of 7 North American case-​control studies. Epidemiology, 16(2), 137–​145. PMID: 15703527. Krzyzanowski M, Apte JS, Bonjour SP, et al. 2014. Air pollution in the mega-​ cities. Curr Environ Health Rep, 1, 185–​191. Kyrtopoulos SA, Georgiadis P, Autrup H, et al. 2001. Biomarkers of genotoxicity of urban air pollution. Overview and descriptive data from a molecular epidemiology study on populations exposed to moderate-​to-​low levels of polycyclic aromatic hydrocarbons: the AULIS project. Mutat Res, 496(1–​ 2), 207–​228. PMCID: PMC11551497. Lan Q, Chapman RS, Schreinemachers DM, Tian L, and He X. 2002. Household stove improvement and risk of lung cancer in Xuanwei, China. J Natl Cancer Inst, 94(11), 826–​835. PMID: 12048270. Larkin EK, Smith TJ, Stayner L, et al. 2000. Diesel exhaust exposure and lung cancer:  adjustment for the effect of smoking in a retrospective cohort study. Am J Ind Med, 38(4), 399–​409. PMID: 10982980. Levy JI, Houseman EA, Ryan L, Richardson D, and Spengler JD. 2000. Particle concentrations in urban microenvironments. Environ Health Perspect, 108(11), 1051–​1057. PMCID: PMC11102296. Lipsett M, and Campleman S. 1999. Occupational exposure to diesel exhaust and lung cancer: a meta-​analysis. Am J Public Health, 89(7), 1009–​1017. PMCID: PMC1508841. Lofroth G. 1989. Environmental tobacco smoke:  overview of chemical composition and genotoxic components. Mutat Res, 222(2), 73–​ 80. PMID: 2645518. Lubin JH, Boice JD, Jr, Edling C, et al. 1995a. Lung cancer in radon-​exposed miners and estimation of risk from indoor exposure. J Natl Cancer Inst, 87(11), 817–​827. PMID: 7791231. Lubin JH, Boice JD, Jr, and Samet JM. 1995b. Errors in exposure assessment, statistical power and the interpretation of residential radon studies. Radiat Res, 144(3), 329–​341. PMID: 7494877. Lynch KM, and Smith WA. 1939. Pulmonary asbestosis. V. A report of bronchial carcinoma and epithelial metaplasia. Am J Cancer, 36, 567–​573. Macklin MT, and Macklin CC. 1940. Does chronic irritation cause primary carcinoma of the human lung? Arch Pathol, 30, 924–​955. Maclure M, Katz RB, Bryant MS, Skipper PL, and Tannenbaum SR. 1989. Elevated blood levels of carcinogens in passive smokers. Am J Public Health, 79(10), 1381–​1384. PMCID: PMC1350179. Martin WJ, 2nd, Glass RI, Araj H, et al. 2013. Household air pollution in low-​ and middle-​income countries: health risks and research priorities. PLoS Med, 10(6), e1001455. PMCID: PMC3672215. Mauderly JL. 1997. Relevance of particle-​induced rat lung tumors for assessing lung carcinogenic hazard and human lung cancer risk. Environ Health Perspect, 105 Suppl 5, 1337–​1346. PMCID: PMC1470153. National Council on Radiation Protection and Measurements. 2009. Ionizing Radiation Exposure of the Population of the United States. Report No. 160. 1st ed. Washington, DC: National Council on Radiation. National Research Council, and Committee on Health Risks of Exposure to Radon. 1999. Health Effects of Exposure to Radon: BEIR VI. Washington, DC: National Academy Press. National Research Council, Committee on Human and Environmental Exposure Science in the 21st Century, Board on Environmental Studies and Toxicology, and Division on Earth and Life Studies. 2012. Exposure Science in the 21st Century:  A  Vision and a Strategy. Washington, DC: National Academies Press.

National Research Council, and Committee on Passive Smoking. 1986. Environmental Tobacco Smoke:  Measuring Exposures and Assessing Health Effects. Washington, DC: National Academy Press. National Research Council, and Committee on the Biological Effects of Ionizing Radiation. 1988. Health Risks of Radon and Other Internally Deposited Alpha-​Emitter: BEIR IV. Washington, D.C. National Research Council, and Panel on Dosimetric Assumptions Affecting the Application of Radon Risk Estimates. 1991. Comparative Dosimetry of Radon in Mines and Homes: Companion to BEIR IV Report. Washington, DC: National Academy Press. National Toxicology Program. 2014. Formaldehyde CAS No. 50-​99-​0 13th Report on Carcinogens (13th ed., p.  11). Triangle Park, NC:  US Department of Health and Human Services, Public Health Service. Natusch DF. 1978. Potentially carcinogenic species emitted to the atmosphere by fossil-​ fueled power plants. Environ Health Perspect, 22, 79–​ 90. PMCID: PMC1637149. Nyberg F, Gustavsson P, Jarup L, et  al. 2000. Urban air pollution and lung cancer in Stockholm. Epidemiology, 11(5), 487–​495. PMID: 10955399. Overholt RH, and Rumel WR. 1940. Clincial studies of primary carcinoma of the lung. An analysis of seventy-​five cases, twenty-​one of which were treated by pneumonectomy or lobectomy. JAMA, 114 (9), 735–​742. Pandey KD, Wheeler D, Ostro B, et  al. 2006. Ambient Particulate Matter Concentrations in Residential and Pollution Hotspot Areas of World Cities: New Estimates Based on the Global Model of Ambient Particulates (GMAPS). Washington, DC:  The World Bank Development Economics Research Group and the Environment Department Working Paper, The World Bank. Perera FP, Hemminki K, Gryzbowska E, et  al. 1992. Molecular and genetic damage in humans from environmental pollution in Poland. Nature, 360(6401), 256–​258. PMID: 1436106. Pershagen G. 1990. Air pollution and cancer. In:  Vainio H, Sorsa M, and McMichael AJ (Eds.), Complex Mixtures and Cancer Risk (pp. 240–​251). Lyon, France: International Agency for Research on Cancer. Pershagen G, Elinder CG, and Bolander AM. 1977. Mortality in a region surrounding an arsenic emitting plant. Environ Health Perspect, 19, 133–​ 137. PMCID: PMC1637390. Peto R, and Lopez AD. 2003. Smoking Data:  Clinical Trial Service Unit. Available from: http://​www.ctsu.ox.ac.uk/​. Accessed January 10, 2016. Proctor R. 2012. Golden Holocaust: Origins of the Cigarette Catastrophe and the Case for Abolition. Berkeley: University of California Press. Raaschou-​Nielsen O, Andersen ZJ, Beelen R, et  al. 2013. Air pollution and lung cancer incidence in 17 European cohorts: prospective analyses from the European Study of Cohorts for Air Pollution Effects (ESCAPE). Lancet Oncol, 14(9), 813–​822. PMID: 23849838. Redmond CK. 1983. Cancer mortality among coke oven workers. Environ Health Perspect, 52, 67–​73. PMCID: PMC1569361. Reynolds P, Elkin E, Scalf R, Von Behren J, and Neutra RR. 2001. A case-​ control pilot study of traffic exposures and early childhood leukemia using a geographic information system. Bioelectromagnetics, Suppl 5, S58–​S68. PMID: 11170118. Rothman KJ, and Greenland S. 1998. Modern Epidemiology. Philadelphia: Lippincott-​Raven. Sapkota A, and Buckley TJ. 2003. The mobile source effect on curbside 1,3-​ butadiene, benzene, and particle-​bound polycyclic aromatic hydrocarbons assessed at a tollbooth. J Air Waste Manag Assoc, 53(6), 740–​748. PMID: 12828334. Savitz DA, and Feingold L. 1989. Association of childhood cancer with residential traffic density. Scand J Work Environ Health, 15(5), 360–​363. PMID: 2477895. Selikoff IJ, Churg J, and Hammond EC. 1964. Asbestos exposure and neoplasia. JAMA, 188, 22–​26. PMID: 14107207. Selikoff IJ, Hammond EC, and Seidman H. 1979. Mortality experience of insulation workers in the United States and Canada, 1943–​1976. Ann N Y Acad Sci, 330, 91–​116. PMID: 294225. Selikoff IJ, Hammond EC, and Seidman H. 1980. Latency of asbestos disease among insulation workers in the United States and Canada. Cancer, 46(12), 2736–​2740. PMID: 7448712. Silverman DT, Samanic CM, Lubin JH, et  al. 2012. The Diesel Exhaust in Miners study:  a nested case-​control study of lung cancer and diesel exhaust. J Natl Cancer Inst, 104(11), 855–​868. PMCID: PMC3369553. Spengler JD. 1991. Sources and concentrations of indoor air pollution. In: Samet JM, and Spengler JD (Eds.), Indoor Air Pollution: A Health Perspective (pp. 33–​67). Baltimore, MD: Johns Hopkins University Press. Steck DJ, Alavanja MC, Field RW, et al. 2002. 210Po implanted in glass surfaces by long term exposure to indoor radon. Health Phys, 83(2), 261–​ 271. PMCID: PMC12132714.

 30



Air Pollution

Stocks P, and Campbell JM. 1955. Lung cancer death rates among non-​smokers and pipe and cigarette smokers:  an evaluation in relation to air pollution by benzpyrene and other substances. Br Med J, 2(4945), 923–​929. PMCID: PMC1981086. Szpiro AA, Paciorek CJ, and Sheppard L. 2011. Does more accurate exposure prediction necessarily improve health effect estimates? Epidemiology, 22(5), 680–​685. PMCID: PMC3195520. Tirmarche M, Harrison J, Laurier D, et al. 2012. Risk of lung cancer from radon exposure: contribution of recently published studies of uranium miners. Ann ICRP, 41(3–​4), 368–​377. PMID: 23089036. Tomatis L. 1990. Air Pollution and Human Cancer. Berlin, Germany: Springer-​Verlag. Trichopoulos D, Kalandidi A, Sparros L, and MacMahon B. 1981. Lung cancer and passive smoking. Int J Cancer, 27(1), 1–​4. PMID: 7251227. Turner MC, Cohen A, Jerrett M, et  al. 2014. Interactions between cigarette smoking and fine particulate matter in the Risk of Lung Cancer Mortality in Cancer Prevention Study II. Am J Epidemiol, 180(12), 1145–​1149. PMID: 25395026. Turner MC, Krewski D, Pope CA, 3rd, et  al. 2011. Long-​term ambient fine particulate matter air pollution and lung cancer in a large cohort of never-​smokers. Am J Respir Crit Care Med, 184(12), 1374–​ 1381. PMID: 21980033. US Department of Health and Human Services. 1986. The Health Consequences of Involuntary Smoking: A Report of the Surgeon General. Washington, DC:  US Department of Health and Human Services, Public Health Service, Office on Smoking and Health. US Department of Health and Human Services. 2006. The Health Consequences of Involuntary Exposure to Tobacco Smoke:  A  Report of the Surgeon General. Atlanta, GA:  US Department of Health and Human Services, Centers for Disease Control and Prevention, Coordinating Center for Health Promotion, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. US Department of Health and Human Services. 2010. How Tobacco Smoke Cuases Disease:  The Biology and Behavioral Basis for Smoking-​ Atrributable Disease: A Report of the Surgeon General. Atlanta, GA: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. US Environmental Protection Agency (EPA). 1992a. Respiratory Health Effects of Passive Smoking:  Lung Cancer and Other Disorders. Washington, DC: US Government Printing Office. US EPA. 1992b. Technical Support Document for the 1992 Citizen’s Guide to Radon. Washington, DC: US Government Printing Office.

303

US EPA. 1999. Integrated Risk Information System (IRIS) on Asbestos. Washington, DC: National Center for Environmental Assessment, Office of Research and Development. US EPA. 2003. EPA Assessment of Risks from Radon in Homes: Air and Radiation (6608J) (p. 98). Washington, DC: Office of Radiation and Indoor Air. US EPA. 2002. Health Assessment Document for Diesel Engine Exhaust (Final 2002). Washington, DC:  US Environmental Protection Agency, Office of Research and Development, National Center for Environmental Assessment. US EPA. 1990. Cancer Risk from Outdoor Exposure to Air Toxics (Vol. 1, Final Report, p.  162). Research Triangle Park, NC:  US Government Printing Office. US EPA. 2016. Our Nation’s Air: Status and Trends Through 2015. Available from: https://​gispub.epa.gov/​air/​trendsreport/​2016/​. Accessed September 9, 2016. van Donkelaar A, Martin RV, Brauer M, and Boys BL. 2015. Use of satellite observations for long-​term exposure assessment of global concentrations of fine particulate matter. Environ Health Perspect, 123(2), 135–​143. PMCID: PMC4314252. Vermeulen R, Silverman DT, Garshick E, et al. 2014. Exposure-​response estimates for diesel engine exhaust and lung cancer mortality based on data from three occupational cohorts. Environ Health Perspect, 122(2), 172–​ 177. PMCID: PMC3915263. Wedler HW. 1944. Asbestosis and pulmonary carcinoma (Asbestose und lungenkrebs). Bull Hygiene, 19(5), 362–​362. Weiss SJ. 1989. Tissue destruction by neutrophils. N Engl J Med, 320(6), 365–​ 376. PMID: 2536474. Winer AM, and Busby WF Jr. 1995. Atmospheric Transport and Transformation of Diesel Emissions (pp. 83–​105). Cambridge, MA: The Health Effects Institute. World Health Organization (WHO). 1988. Emissions of Heavy Metal and PAH Compounds from Municipal Solid Waste Incinerators:  Control Technology and Health Effects:  Report on a WHO Meeting, Florence. Copenhagen: WHO. WHO. 1990. Global Estimates for Health Situation Assessment and Projections. Geneva: WHO. WHO. 2002. The World Health Report:  Reducing Risks, Promoting Healthy Life. Geneva: World Health Organization. Xu ZY, Blot WJ, Xiao HP, et al. 1989. Smoking, air pollution, and the high rates of lung cancer in Shenyang, China. J Natl Cancer Inst, 81(23), 1800–​ 1806. PMID: 2555531. Zielinska B, Samy S, McDonald JD, and Seagrave J. 2010. Atmospheric Transformation of Diesel Emissions. HEI Research Report 147 (p. 86). Boston, MA: Health Effects Institute.

304

 305

18

Water Contaminants KENNETH P. CANTOR, CRAIG M. STEINMAUS, MARY H. WARD, AND LAURA E. BEANE FREEMAN

OVERVIEW Humans have long recognized the hazards of microbial contamination of drinking water. Only since the 1960s, however, have epidemiologic studies systematically examined whether naturally occurring and/​or man-​made pollutants in drinking water affect cancer risk. Ironically, some of the measures taken to reduce microbial hazards have increased exposure to other contaminants. This chapter will begin by discussing three waterborne exposures that affect large numbers of people and have been studied most extensively: inorganic arsenic, disinfection byproducts, and nitrate. Of these, only arsenic and its compounds are currently designated as carcinogenic to humans. We then discuss the evidence concerning two emerging issues: the carcinogenicity of toxins from cyanobacteria, an ancient and ubiquitous family of prokaryotic organisms formerly known as blue-​green algae, now affected by climate change, and the methods of studying cancer in local communities where the water supply has been contaminated by industrial chemicals. Methodologic challenges complicate studies of these issues. The long induction period following exposure, especially when the exposure affects early stages of carcinogenesis, requires that researchers reconstruct historical levels that existed many decades in the past. The exposed populations must be sufficiently large, stable, and well defined that epidemiologic studies can detect associations with specific cancer sites. The presence of multiple exposures complicates efforts to evaluate individual exposures in relation to a ­priori hypotheses. Despite these obstacles, there have been important advances in studying the potential carcinogenicity of waterborne pollutants, as well as new opportunities to integrate genetic and metabolomic approaches into future studies.

severe symptoms and even death may follow shortly after ingestion of even a small inoculum of bacteria. The short incubation period virtually eliminates the opportunity for those exposed to cholera to move or change their water supply before the onset of disease. Snow’s map depicting the cross-​sectional clustering of cases along the distribution pipes of the contaminated water system strongly implicated drinking water as the vector for disease transmission, and argued against ‘miasma’ as a plausible alternative explanation. In contrast to the study of infectious agents, epidemiologic studies of cancer typically confront exposures for which the induction period may last 20, 50, or more years, depending on the stage(s) of carcinogenesis that are affected. Retrospective studies (case control or historical cohort) and ecological analyses are usually more feasible than prospective studies for estimating risk, although these studies require that researchers reconstruct historical exposures that occurred many decades in the past. Mobility creates logistical problems in identifying and tracking the exposed population (Meliker et  al., 2007; Nuckols et al., 2011), and in characterizing exposure at different periods of life. Studies often rely on a single, contemporaneous questionnaire seeking information about historical levels of water consumption that are assumed to accurately represent lifetime patterns. Exposed populations may not be sufficiently large and/​or well defined that an epidemiologic study can detect associated levels of risk in site-​specific cancers. Despite these obstacles, there has been substantial progress in studying the potential carcinogenicity of waterborne pollutants and in developing opportunities to integrate genetic and metabolomic analyses into this area of research (Antonelli et al., 2014; Cantor et al., 2010). The present chapter will discuss developments in the field since the third edition of this text (Cantor et al., 2006).

INTRODUCTION

INORGANIC ARSENIC

Water is essential to life. However, drinking water may be contaminated by biological, chemical, and other pollutants that lead to ill health and death. Access to clean water has been a prerequisite for successful human settlements since ancient times. The engineering feats of the Romans more than 2000 years ago (Gagarin, 2010) and the Incas prior to the Spanish Conquest (Wright et al., 1997) attest to the central importance of uncontaminated water in human cultures. People have long taken steps to reduce microbial contamination of drinking water. Efforts to assess the relationship between waterborne contaminants and cancer are relatively recent, however. Not until the 1960s did epidemiologic studies begin to evaluate systematically whether naturally occurring and/​or man-​made chemicals and other substances in drinking water affect cancer risk (International Agency for Research on Cancer [IARC], 2012a). Epidemiologic studies of water pollution and cancer confront special challenges that are not generally problematic when studying microbial contaminants in drinking water. These can be illustrated by contrasting the seminal study of waterborne cholera in mid-​nineteenth-​century London by Dr. John Snow (1855) with an evaluation of cancer risk in relation to chemical contamination of water. In the case of cholera, the incubation period is measured in hours, or at most a few days. The consequences of infection are rapid and unambiguous;

Exposure Human exposure to inorganic arsenic occurs primarily through water ingestion, although exposure can also occur via air, food, and tobacco smoking (IARC, 2012a). The most common source of high exposure is from drinking water contaminated with arsenic from naturally occurring geologic sources. Industrial exposure occurs through the inhalation of arsenic-​containing particulates, especially during the mining and smelting of copper, tin, and lead; some agricultural workers continue to be exposed to arsenical pesticides. Medications containing arsenic were a staple of the pharmacopeia until the 1940s; patients were treated for a wide range of ailments, particularly dermatologic conditions (Cantor et al., 2006; Cuzick et al., 1992). With the exception of arsenic trioxide, used to treat promyelocytic leukemia (Coombs et  al., 2015), medicinal arsenic is no longer a significant source of exposure. In natural waters, inorganic arsenic occurs primarily as pentavalent arsenate [As(V)] or trivalent arsenite [As(III)] (Cantor et  al., 2006). Arsenic can also occur in organic forms such as arsenobetaine or other arsenosugars. These occur more commonly in food (e.g., seafood) and are considerably less toxic than inorganic arsenic. Tens of millions of people worldwide are exposed to naturally occurring inorganic arsenic

305

306

306

PART III:  THE CAUSES OF CANCER

in drinking water, including an estimated 50  million in Bangladesh, 30 million in India (mostly in West Bengal), 15 million in China, and millions more in Europe and South and Central America (Ravenscroft, 2007). Exposures in Bangladesh and West Bengal are primarily the result of widespread introduction of tube wells, beginning in the 1970s, to reduce exposure to microbial pathogens in surface water. By 1993, widespread arsenic contamination of tube wells was confirmed, leading to enormous public health concerns (Smith et al., 2000). A 1998–​ 1999 survey of 3534 tube wells throughout Bangladesh found that 25%–​27% of wells had arsenic concentrations > 50 µg/​l, five times the current World Health Organization recommended maximum of 10 µg/​l, and 9% had concentrations > 200 µg/​l. It was estimated that 57 million people in Bangladesh were exposed to water with arsenic concentrations > 10 µg/​l, and 35 million were exposed to concentrations > 50 µg/​l (British Geological Survey, Government of the People’s Republic of Bangladesh, 2001). Mitigation efforts have reduced the proportion of wells with arsenic concentrations > 50 µg/​l from 25% to 13% (Flanagan et al., 2012). Despite this, an estimated 22 million people in Bangladesh still rely on the highly contaminated wells. In most other areas with naturally contaminated water, arsenic water concentrations are generally lower than those in Bangladesh. According to the Environmental Protection Agency (EPA) in the United States, an estimated 12% of all public water systems (5252 of 43,443) have arsenic concentrations between 5–​10 µg/​l (US EPA, 2000a). Several million people in the United States are thought to be exposed to higher levels from drinking water drawn from unregulated private wells. Arsenic concentrations > 50 µg/​l have been reported in many US states, but levels > 200 µg/​l are rare (Chappells et al., 2014). A case-​control study of skin, bladder, and kidney cancer conducted in parts of Hungary, Romania, and Slovakia reported that 25% of the population was exposed to an average concentration of arsenic in drinking water of > 10 µg/​l, and 8% were exposed to > 50 µg/​l. Levels above 200 µg/​l were reported in only a few wells (Hough et al., 2010). Another source of arsenic exposure is food, which, on average, accounts for about 8–​10 μg of ingested arsenic per day (Dabeka et al., 1993; Tao and Bolger, 1999). Arsenic can be taken up by plants grown in soil with elevated arsenic levels. Some arsenic in soil can result from previous use of arsenic-​based pesticides, or irrigation with arsenic-​ contaminated water (Rahman and Hasegawa, 2011). Exposure to arsenic in rice has received considerable attention. Rice actively extracts arsenic from soil, and detectable levels have been seen in many rice products. In 2012, the US Food and Drug Administration (FDA) found arsenic in almost all of the nearly 200 brands of rice, rice cakes, rice baby foods, and rice cereals it tested (US FDA, 2012). A 2011 study of 229 pregnant women in the United States found that women who ate rice had significantly higher urinary arsenic levels than those who did not. Eating 0.56 cups of cooked rice per day was associated with an arsenic intake equivalent to that from drinking one liter per day of water with an arsenic concentration of 10 µg/​l (the current standard for drinking water specified by the United States and WHO; Gilbert-​Diamond et al., 2011). Rice also absorbs arsenic from cooking water as well as during growth. Thus, intake from rice is likely higher in areas where arsenic-​ contaminated water is used for cooking (Rahman and Hasegawa, 2011).

Exposure Measurement in Epidemiologic Studies Residential Exposure

Epidemiologic studies of arsenic and cancer often use residential histories in combination with historical records of water concentration to estimate exposure in regions with highly contaminated drinking water. For example, Marshall et al. investigated cancer mortality following a 13-​year period of very high arsenic exposure in a desert area (Region II) in northern Chile (Marshall et al., 2007). The period of high exposure began in 1958 when two rivers with arsenic water concentrations near 860 µg/​l were piped to the city for drinking, and ended in 1971 when an arsenic treatment plant was installed. The epidemiologists measured cancer mortality rates for the years before, during, and after this high exposure period, and compared these with death rates in a socio-​demographically similar region in Chile (Region V). Figure 18–​1

shows the temporal trend of arsenic concentration in drinking water in relation to the rate ratio (RR) estimates for lung and bladder cancer mortality. The RR estimates for both cancer sites started to increase about 10 years after the large initial increase in arsenic exposure and continued to rise until the late 1980s. Subsequent analyses have shown that cancer mortality rates in this area remained high up to 2000, the last year assessed, approximately 40  years after the high exposures began and approximately 30 years after the maximum exposure was ended (Smith et al., 2012).

Biologic Metrics of Arsenic Exposure

Humans excrete arsenic primarily through the urine, and urinary levels of inorganic arsenic and its metabolites are commonly used to estimate recent internal absorption (IARC, 2012a; NRC, 1999). The biological half-​life of arsenic in urine following ingestion is approximately 39–​ 59 hours (Buchet et al., 1981); thus, urinary concentrations are most valuable for assessing recent exposures or exposures among people whose intake levels have not changed much over time. Arsenic can also be measured in blood, hair, and nails (Orloff et  al., 2009). The half-​life of arsenic in blood is fairly short, so blood levels are only useful for detecting recent exposures. Arsenic levels in the hair and nails have a longer half-​life, so these can be used to assess exposures over several months or more. Hair and nails can be contaminated by external exposures to arsenic, so specialized cleaning procedures are employed when using hair and nails to assess internally absorbed arsenic (Orloff et al., 2009).

Statistical Modeling of Exposure

Historical exposure records are rarely available, especially in regions where private wells are common. Several studies have used statistical modeling to estimate arsenic exposures from sources that lack direct measurements. For example, in a bladder cancer case-​control study in New England, investigators measured arsenic concentrations in water samples from 2611 current and 448 past residences of study subjects (Nuckols et al., 2011). In this region, almost half the population obtains drinking water from private wells. The available measurements covered residences that accounted for only 27% of the total lifetime person-​years of participants in the study. In order to estimate exposures during the remaining residential history of the subjects, the researchers developed statistical regression models that combined empirical measurements from the sampled wells with geological data for two major strata and geochemical data for public water supplies drawn from different sources. Model results were applied to estimate the arsenic level at each historic residence where direct data were not available. In a validation survey done as part of this investigation, measures of sensitivity, specificity, and overall agreement varied. For example, using a dichotomous cutoff of 5 µg/​l, overall agreement between measured and predicted level of drinking water arsenic from bedrock wells (from one of the models) varied between 57% and 70%, depending on the buffer radius that was used for the estimate. Sensitivity varied from 58% to 78% and specificity from 52% to 73%. While these results are adequate for some purposes, the level of exposure misclassification in these estimates is high enough to compromise findings in epidemiologic analyses, highlighting the difficulties that can occur using modeling approaches to estimate past exposure. Other modeling or statistical approaches have been used, including geological averages, kriging, and using the arsenic concentration in nearest neighboring wells. Reported correlation coefficients between the arsenic concentrations predicted by these methods and concentrations in known wells in validation surveys have ranged from fair to good (i.e., correlation coefficients between 0.4 and 0.8) (Dauphine et  al., 2013; James et al., 2014; Meliker et al., 2008).

Cancer Outcomes Experimental Studies in Animals

Until recently, arsenic has been one of the few human carcinogens without an animal model. More recent experiments using arsenical

 307

307

Water Contaminants

4

600

3

450

3

450

2

300

2

300

1

150

1

150

0

Rate Ratio

600

Arsenic µg/L

Rate Ratio

4

0

0 1950

1960

1970 1980 Years

1990

0 1950

2000

Bladder Cancer Mortality Men aged 30 and over 600

6

300

4

150

2 0

0 2000

Rate Ratio

8

1990

1990

2000

600

10

450

1970 1980 Years

1970 1980 Years

12

Arsenic µg/L

Rate Ratio

10

1960

1960

Bladder Cancer Mortality Women aged 30 and over

12

1950

Arsenic µg/L

Lung Cancer Mortality Women aged 30 and over

Lung Cancer Mortality Men aged 30 and over

450

8 300

6 4

150

2 0

Arsenic µg/L



0 1950

1960

1970 1980 Years

1990

2000

Figure 18–​1.  Lung and bladder cancer mortality rate ratios comparing Region II with Region V, Chile, for men and women aged 30 and above, separately, as estimated by Poisson regression with smoothing. The circles are the mortality rate ratios plotted at the midpoint of each successive 3-​year period. The shading represents the 95% confidence intervals of these rate ratios. Histograms (gray lines) of the population-​weighted average arsenic water concentrations for Region II, from 1950 to 1994 in 5-​year increments, are also presented (vertical axes at right). The dramatic rise and decline in arsenic water concentrations were in the years 1958 and 1970, respectively. They were incorrectly drawn in the original published graphs shown here.

methylation metabolites and early-​life exposures have shown increases in cancer in rodents (IARC, 2012a; Tokar et al., 2010a). There are important interspecies differences in the metabolism of arsenic between laboratory animal species and humans. These may account for differences in the toxicity and dose–​response relationships in different species (Vahter, 1999b). For example, mice are very effective at metabolizing arsenic. In rats, methylated inorganic arsenic accumulates in red blood cells; storage inside red blood cells may limit toxicity to other organs. This does not occur in humans. Consequently, long-​term feeding studies of laboratory animals may not reliably predict the carcinogenic effects in humans. The designation of inorganic arsenic as a human carcinogen by the IARC is based on data from humans rather than animals (IARC, 2012a), and refers primarily to cancers of the skin, bladder, and lung. Associations have been observed with cancers of the kidney, liver, and prostate, but these are not currently regarded as definitive (IARC, 2012a). Among the first human evidence linking arsenic to cancer in humans were case reports describing skin and internal cancers following ingestion of medicinal arsenic, arsenic in drinking water, or wine contaminated with arsenical pesticides. The earliest quantitative epidemiologic studies of ingested arsenic and cancer were ecological in design, conducted primarily in areas with chronic arsenicism and highly contaminated drinking water, such as Taiwan, Argentina, and Chile. The arsenic exposure in Taiwan began in the 1940s, long before the tragic exposures in Bangladesh, but also resulted from a shift from surface water to wells to avoid microbial contaminants in

surface water. Ecological studies in the 1960s through 1990s reported elevated mortality from cancers of the skin, bladder, lung, liver, and kidney (Cantor et al., 2006). More recently, cohort and case-​control studies that collected individual data on arsenic drinking water levels and potential confounders have confirmed these associations. For example, in a prospective cohort study of 8086 residents of northwestern Taiwan, relative risks of urothelial cancer associated with residential arsenic water concentrations of < 10, 10–​49.9, 50–​99.9, 100–​299.9, and ≥ 300 µg/​l at the time of study recruitment were 1.00 (reference), 1.85 (95% confidence interval [CI]: 0.45, 7.61), 2.19 (CI: 0.43, 11.1), 5.50 (CI: 1.39, 21.8), and 10.8 (CI: 2.90, 40.3), respectively (Chen et al., 2010b). Relative risks for lung cancer were 1.00 (reference), 1.10 (CI: 0.74, 1.63), 0.99 (CI: 0.59, 1.68), 1.54 (CI: 0.97, 2.46), and 2.25 (CI: 1.43, 3.55) (Chen et al., 2010a). In northern Chile, arsenic concentrations in drinking water have been measured longitudinally in almost all sources for decades, allowing researchers to estimate lifetime exposures (Ferreccio et al., 2000). A 2007–​2010 case-​control study in Chile reported odds ratios for lifetime average arsenic water concentrations of < 26, 26–​79, 80–​197, and > 197 µg/​l of 1.00 (reference), 0.92 (95% CI: 0.52, 1.61), 2.62 (CI: 1.53, 4.50), and 6.00 (CI: 3.38, 10.64), respectively, for bladder cancer, and 1.00 (reference), 0.98 (CI: 0.62, 1.53), 1.70 (CI: 1.05, 2.75), and 3.18 (CI: 1.90, 5.30) for lung cancer (Steinmaus et al., 2013). Elevated odds ratios were also reported for kidney cancers, helping to confirm previous ecologic mortality observations for kidney cancer (Chen et al., 1988; Ferreccio et al., 2013a). Extensive summaries of other epidemiologic studies on arsenic and cancer in humans and experimental

308

308

PART III:  THE CAUSES OF CANCER

animal studies are provided elsewhere (Agency for Toxic Substances and Disease Registry [ATSDR], 2007; IARC, 2012a; NRC, 1999, 2014). Interestingly, not all of the effects of arsenic on cancer are detrimental. As mentioned previously, arsenic trioxide is an effective treatment for promyelocytic leukemia (Coombs et  al., 2015), once considered among the most fatal forms of leukemia. Also provocative is that a study from northern Chile found a 49% decrease in breast cancer mortality among women over age 30 years during the 13-​year period of high arsenic water concentrations (near 860 µg/​l) in the area (Smith et al., 2014). In fact, during the latter part of the high-​exposure period (1965–​1970), breast cancer mortality among exposed women under age 60 years was 70% lower than that among unexposed women. Regional differences in diagnosis, treatment, and confounding variables were found to be unlikely causes for the lower risk. Although this finding has not yet been replicated elsewhere, its biologic plausibility is supported by in vivo evidence of increased apoptosis of breast cancer cells at exposure levels similar to those in the Chile study area (Sun et al., 2011; Xia et al., 2012).

Metabolism

Factors That May Affect Risk Several factors have been reported to influence susceptibility to arsenic-​ related cancers. These include demographic traits (age, gender), metabolic capability, genetic susceptibility, nutrition, and smoking.

Demographic Factors

Age at exposure appears to modify susceptibility to certain arsenic-​ related cancers. Exposure in utero or during childhood is more strongly associated with lung and bladder cancer than the same level of exposure in adulthood. In the Chile case-​control study (noted earlier), odds ratios of 5.24 (95% CI:  3.05, 9.00) for lung cancer and 8.11 (95% CI: 4.31, 15.25) for bladder cancer were seen among adults exposed to high arsenic water concentrations in utero or as children. Odds ratios were lower among those whose exposure began later (Steinmaus et al., 2014). Ecologic studies in Chile also found evidence of increases in kidney, laryngeal, and liver cancer mortality among adults who had been highly exposed in early life (Smith et al., 2012). These findings are supported by studies in mice showing that prenatal arsenic exposure increases the occurrence of lung, liver, and other tumors later in life, whereas the same level of arsenic exposures in adulthood does not affect risk (Tokar et al., 2010a). The mechanisms by which in utero or long-​term/​early-​life exposure affect risk are unknown, although the in utero period is a time of major epigenetic reorganization, and arsenic has been linked to epigenetic changes in laboratory animals and humans (Kile et al., 2014). Epigenetic changes could affect gene expression and regulation later in life. Other proposed mechanisms relate to the effects of arsenic on the developing immune system or other rapidly developing organs, to exposure before the development of detoxification enzymes or to the higher dose relative to body mass index (BMI) in fetuses and children compared to adults (Miller et al., 2001).

As (V)

As(III) GSH GSSG

OH I O = Asv – OH I O–

In humans, ingested inorganic arsenic is metabolized through a complex series of reduction and methylation steps (Figure 18–​2). Ingested inorganic pentavalent arsenate [As(V)] is rapidly reduced to arsenic trioxide [As(III)], which then accepts a methyl group from S-​adenosyl methionine (SAM) to produce the methylated pentavalent species monomethylarsonic acid, MMA(V). MMA(V) is then reduced to trivalent monomethylarsonous acid, MMA(III). An additional methyl group can then be added to yield dimethylarsinic acid, DMA(V), which can be reduced to small amounts of DMA(III) (NRC, 2014). As mentioned, urine is the primary route of arsenic excretion in humans. A potentially useful marker of an individual’s ability to metabolize inorganic arsenic is the percentage that different metabolites contribute to the total amount in urine. Although there is substantial inter-​ individual variation, on average, As(III) and As(V) together account for 10%–​30% of the total, MMA(III) and MMA(V) contribute 10%–​ 20%, and DMA(III) and DMA(V) combined account for the remaining 60%–​70% (Hopenhayn-​Rich et al., 1993; Vahter, 1999a). The methylation of arsenic was previously thought to contribute primarily to detoxification, since the pentavalent metabolites, MMA(V) and DMA(V), are acutely less toxic in vitro and more rapidly excreted than arsenic in its inorganic form (Vahter and Concha, 2001). However, more recent in vitro studies have shown that MMA(III) may be more toxic, at least acutely, than inorganic arsenic (Styblo et al., 2000). This suggests that MMA(III) could be the principal toxicant and that people who are less able to metabolize MMA(III) to DMA(V) may be more susceptible to arsenic toxicity. MMA(III) is difficult to measure in epidemiologic studies because it is rapidly oxidized to its pentavalent form in urine (Kalman et al., 2014). Epidemiologic studies have found that people with high urinary proportions of total MMA or low proportions of total DMA have 2–​5 times higher risk of arsenic-​related cancers and other diseases than individuals who do not (Smith and Steinmaus, 2009). For example, in the highly exposed southwest coast of Taiwan, odds ratios for urothelial carcinoma were calculated for different strata of lifetime exposure to arsenic, further stratified according to high or low proportions of urinary inorganic arsenic excreted as MMA (%MMA). Subjects in the lower category of both cumulative exposure (< 20 mg/​l-​years) and %MMA (< 11.40%) served as the reference group. The odds ratios for urothelial cancer in subjects with cumulative exposure ≥ 20 mg/​l-​year were 1.5 (95% CI: 0.4, 5.9) in those with %MMA < 11.40%, versus 3.7 (95% CI: 1.2, 11.6) in those with higher %MMA values (Huang et al., 2008). Similar patterns have been seen in Chile for lung cancer, where odds ratios for arsenic water concentrations >200 ug/​l were 3.16 (95% CI: 1.59, 6.32) in subjects with low %MMA values (< 12.5%) but 6.81 (95% CI: 3.24, 14.31) in subjects with higher %MMA values (Melak et al., 2014).

Genetic Susceptibility

Polymorphisms in genes that code for the enzymes involved in arsenic methylation have been linked to poor arsenic metabolism. The most consistent evidence for this involves the arsenic (3+) methyltransferase (AS3MT) genes (Figure 18–​2), for which

MMA (V) SAM

SAH OH I O = AsV – CH3

OH I OH – AsIII – OH AS3MT

MMA (III) OH I AsIII – CH3 I OH

DMA (V) SAM

SAH

AS3MT

CH3 I O = AsV – CH3 I OH

Figure 18–​2.  Arsenic metabolism: in humans, ingested inorganic arsenic is metabolized through a series of reduction and methylation steps. In the past, this was thought to be primarily a detoxification process since DMA(V) is more readily excreted and is less toxic than inorganic arsenic (As(V) or As(III)). However, more recent evidence suggests that the intermediary species, MMA(III), may be more toxic than As(V) or As(III). Abbreviations: As(V) = arsenate; GSH = glutathione; GSSG = glutathione disulfide; As(III) = arsenite; SAM = S-​adenosyl methionine; SAH = S-​adenosyl homocysteine; AS3MT = arsenic-​3-​methyltransferase; MMA(V) = monomethylarsonic acid; MMA(III) = monomethylarsonous acid; DMA(V) = dimethylarsinic acid.

 309



Water Contaminants

polymorphisms have been statistically significantly associated with urinary methylation patterns (Antonelli et al., 2014). For example, in an arsenic-​exposed region in Mexico, subjects with the C allele of Met287Thr in the AS3MT gene had higher mean urinary %MMA levels (15 vs. 10%, p < 0.05) and a higher odds of having arsenic-​ related premalignant skin lesions (odds ratio [OR]  =  4.28; 95% CI:  1.0, 18.5) (Valenzuela et  al., 2009). Polymorphisms in other genes have also been linked to arsenic metabolism, including several glutathione-​s-​transferase genes.

Nutrition

The availability of certain micronutrients may also affect a person’s ability to methylate arsenic. For example, low intakes of methionine, protein, and choline have been shown to impair arsenic methylation in experimental animals (Vahter and Marafante, 1987). Low body mass or weight has been associated with increased risks of arsenic-​ related skin lesions in India and Bangladesh (Ahsan et al., 2006; Guha Mazumder et  al., 1998). Folate affects one carbon metabolism, and therefore may modify the transfer of methyl groups from S-​adenosyl methionine that occur in arsenic methylation (Hall and Gamble, 2012; Kile and Ronnenberg, 2008). In arsenic-​exposed areas of Bangladesh, folate supplementation was associated with decreased blood arsenic levels and an increased percentage of inorganic arsenic excreted as DMA (Gamble et al., 2006; Peters et al., 2015). Folate deficiency has been linked to arsenic-​caused skin lesions (Melkonian et  al., 2012); most of these studies have been cross-​sectional, and clear prospective associations between folate intake and cancer have not been seen to date. Selenium increases the non-​enzymatic methylation of inorganic arsenic, and has been associated with reduced arsenic toxicity and increased arsenic excretion in animal and laboratory tests (Zwolak and Zaporowska, 2012). Both selenium and arsenic affect DNA methylation in vivo and in vitro, suggesting that in cytosine DNA methyltransferase, selenium, and arsenic compete for methyl donation from SAM. Higher levels of blood selenium have been associated with lower percentages of MMA in blood or urine (Basu et  al., 2011; Pilsner et  al., 2011), and with lower risks of premalignant arsenic-​related skin lesions (Chen et al., 2007), but little information is available on whether selenium modifies cancer risks related to arsenic.

Other Factors

Synergistic associations have been seen between arsenic exposure in either air or water, and tobacco smoking (Chen et  al., 2004; Hertz-​ Picciotto et al., 1992). For example, in Chile, arsenic water concentrations > 200 µg/​l were associated with lung cancer odds ratios of 8.0 (95% CI:  1.7, 52.3) among never-​smokers, but 32.0 (95% CI:  7.22, 198.0) in ever-​smokers (Ferreccio et al., 2000). Some evidence of synergy between arsenic in water and certain occupational exposures has been reported in relation to premalignant or malignant skin lesions or other cancers for pesticides, silica, asbestos, and wood dust (Ferreccio et al., 2013b; Melkonian et al., 2011). Recently, evidence of synergy between arsenic and excess BMI were reported in a cancer case-​ control study in Chile (Steinmaus et al., 2015).

Cancer Risks at Lower Exposures The most convincing epidemiologic evidence of the carcinogenicity of ingested arsenic comes from studies involving high exposure levels, that is, drinking water arsenic concentrations > 100 µg/​l. Findings at exposure levels below this are mixed (Gibb et  al., 2011)  (Table 18–​1). This issue is important because millions of people worldwide are exposed to these lower levels, but the associated cancer risks are unknown. Linear extrapolation from higher dose studies suggests that the cancer risks at (lifetime) exposures near the US and World Health Organization limits for arsenic in water (10 µg/​l) may be on the order of one extra cancer case for every 300 people (NRC, 2001). Some have suggested that the dose–​response relationship between ingested arsenic and cancer is not linear and that relatively low arsenic exposure levels may be associated with little or no excess risk (Cohen et al., 2013).

309

Much of the controversy about the effects of arsenic at low exposure levels reflects the difficulty of measuring long-​term exposures. Because of the long latency of arsenic-​associated cancer, retrospective studies must collect or estimate exposure data from the distant past (which are not commonly available), and prospective studies must follow subjects over a period of decades (which can be prohibitively expensive). Assessing past or long-​term exposure patterns can be difficult if subjects change water sources throughout their lives, or use different water sources with different arsenic concentrations throughout the day (i.e., at home, work, and school). Single measurements of arsenic in urine or hair will likely only represent a small portion of lifetime exposure for many people. In addition, among people with relatively low drinking water arsenic concentrations (e.g., < 10 µg/​ l), food-​borne arsenic can become a more important exposure source (Meliker et al., 2006). Because food consumption patterns and arsenic concentrations in foods may change over time, assessing past or long-​ term exposure to arsenic from food can also be challenging. Inaccurate or missing data can increase misclassification of exposure, can lead to errors in estimating study subjects’ true arsenic exposure, and therefore can limit the ability of a study to identify true associations (Cantor and Lubin, 2007). Other issues relate to statistical power and the limits at which relatively small increases in risk can be detected in epidemiologic studies. In other words, if arsenic concentrations that are currently allowed in drinking water were truly associated with increased cancer risks, these increases would expected to be fairly small (e.g., relative risks of < 1.15 at the current US and World Health Organization limits of 10 µg/​l) (Gibb et al., 2011). Such small increases in risk, while important from a public health standpoint, are difficult to detect in epidemiologic studies without very large sample sizes, detailed exposure assessment, and rigorous control of confounding. Even small increases in risk can be important if the exposure and endpoints are common. Lung and bladder cancer are among the most common cancer types worldwide. Given the potential importance of studies of low exposure, such studies must be carefully designed and executed to provide meaningful estimates of risk (Cantor and Lubin, 2007).

Mechanism(s) of Carcinogenicity Currently, the primary mechanisms by which arsenic exerts its carcinogenic effects in humans are unknown. Numerous pathways have been proposed. Evidence suggests that arsenic does not interact directly with DNA (Nesnow et al., 2002), but may be genotoxic through other mechanisms. One possibility is that the metabolism of arsenic generates reactive oxygen species and/​or decreases antioxidant defenses, leading to DNA damage and chromosomal changes (Kitchin and Conolly, 2010). Chromosomal alterations from arsenic exposure are evidenced by the induction of micronuclei in human lymphocytes and in exfoliated bladder cells (Cantor et al., 2006). Arsenic may also alter enzymes involved in DNA repair (Andrew et al., 2009; Hsu et al., 2008) or DNA methylation. Both hypo-​and hypermethylation of DNA have been associated with arsenic exposure in exposed populations (Bailey and Fry, 2014; Ren et al., 2011). The role of these changes in human cancer remains unclear. Cytotoxicity following sulfhydral protein binding of reactive metabolite species and subsequent regenerative cell proliferation or hyperplasia might also lead to the development of cancer (Cohen et al., 2013). Modulations in the immune system have been linked to cancer risks (Finn, 2012), and arsenic has been associated with a variety of inflammatory and infectious diseases, as well as to biomarkers of altered immune response that may be related to immune suppression and altered immune surveillance of cancerous cells (Dangleben et al., 2013). Increasing evidence regarding the role of stem cells in cancer development, as well as the links between early-​ life arsenic exposure and adult disease, has led to theories regarding the role of stem cells in arsenic carcinogenicity (Tokar et al., 2011). In a rat kidney stem cell or partially differentiated progenitor cell line, low-​level chronic arsenic exposure induced phenotypic changes of cancer, including increases in colony formation (indicative of the loss of contact inhibition), increased proliferation, increases in matrix

310

Table 18–​1. Cancer Studies in Populations with Arsenic Water Concentrations < 100 µg/​L Study

Area/​Outcome

Results

Bates et al., 1995

Utah Bladder cancer

Cumulative arsenic ≥ 53 mg OR = 1.41 (90% CI: 0.7, 2.9)

Lewis et al., 1999

Utah Multiple cancers

Cumulative arsenic > 5000 μg/​L-​yr Lung cancer: SMR = 0.44 (p < 0.05) in males and 0.22 (p > 0.05) in females

Karagas et al., 2001

New Hampshire Skin cancer

Toenail arsenic > 97th percentile OR = 2.07 (0.92, 4.66)

Steinmaus et al., 2003

California/​Nevada Bladder cancer

Arsenic intake > 80 μg/​day OR = 1.78 (0.89, 3.56) for exposures > 40 years before cancer diagnosis

Karagas et al., 2004

New Hampshire Bladder cancer

Toenail arsenic > 97th percentile OR = 2.17 (0.92, 5.11) among smokers

Lamm et al., 2004

United States Bladder cancer

County median arsenic concentration ≥ 3 μg/​L SMR = 0.94 (0.90, 0.98)

Heck et al., 2009

New Hampshire and Vermont Lung cancer

Toenail arsenic ≥ 0.1137 μg/​g OR = 0.89 (0.46, 1.75) All types OR = 2.75 (1.00, 7.57) Squamous/​ small cell

Gilbert-​ Diamond et al., 2013

New Hampshire Skin cancer

Urine arsenic ≥ 5.31 μg/​L OR = 1.43 (0.91, 2.27)

Meliker et al., 2007

Michigan Multiple cancers

Meliker et al., 2010

Michigan Bladder cancer

Garcia-​ Esquinas et al., 2013 (Strong Heart Study) Dauphine et al., 2013

Arizona, Oklahoma, and North/​South Dakota Multiple cancers

Exposed counties Lung cancer: SMR = 1.02 (99% CI: 0.98, 1.06) in males and 1.02 (99% CI: 0.96, 1.07) in females Bladder cancer: SMR = 0.94 (99% CI: 0.82, 1.08) in males and 0.98 (99% CI: 0.80, 1.19) in females Time weighted lifetime average water arsenic concentration > 10 μg/​L OR = 1.10 (0.65, 1.86) 80th vs. 20th percentile urinary arsenic HR = 3.30 (1.28, 8.48) Prostate cancer HR = 2.46 (1.09, 5.58) Pancreas cancer HR = 1.56 (1.02, 2.39) Lung cancer Bladder cancer: insufficient number of cases Arsenic water concentration ≥ 85 μg/​L OR = 1.39 (0.55, 3.53) for exposures > 40 years before cancer diagnosis

Baastrup et al., 2008

Denmark Multiple cancers

California/​Nevada Lung cancer

Cumulative exposure 5 mg IRR = 1.00 (0.98, 1.03) Lung cancer IRR = 1.01 (0.98, 1.04) Bladder cancer IRR = 0.94 (0.81, 1.09) Kidney cancer IRR = 0.99 (0.97, 1.01) Skin cancer

Description 1. Case-​control study 2. No past arsenic water concentration measurements 3. No data on private wells 4. OR > 1.0 but not statistically significant in all subjects. Corresponding OR in smokers = 3.32 (90% CI: 1.1, 10.3) 1. Retrospective cohort mortality study 2. Rare population: rural Mormon population 1900+. 3. Lifestyle factors (smoking, diet, other) not assessed 4. Unusual results: low SMR for lung cancer potentially an indication of major confounding 1. Case-​control study 2. Toenails measure only a few months of exposure: past exposure unknown 3. OR > 1.0 but not statistically significant 4. No clear dose–​response relationship 1. Case-​control study 2. Low power to assess past exposures 3. < 9% of subjects lived in highly exposed areas > 40 years before diagnosis 4. OR > 1.0 but not statistically significant 1. Case-​control study 2. Toenails measure only a few months of exposure: past exposure unknown 3. OR > 1.0 but not statistically significant 4. OR not increased in never smokers 5. No clear dose–​response relationship 1. Ecologic mortality study 2. County median used despite wide variability within counties 3. Only a fraction of all wells within exposed counties were measured 4. No data on past exposure 5. No data on smoking, diet, or other potential confounders 1. Case-​control study 2. ORs much lower without multiple adjustments; the confounder causing this is unknown 3. Toenails measure only a few months of exposure: past exposures unknown 4. No clear dose–​response relationship 1. Cross-​sectional: urine samples measured at the time of cancer diagnosis 2. Urine measures only a few weeks of exposure 3. No direct data on past exposure 4. OR > 1.0 but not statistically significant 5. OR not elevated in long-​term residents 1. Ecologic mortality study 2. Ecologic exposure data based on county 3. Wide variability in exposure within counties 4. Past exposure not assessed 5. No data on smoking, diet, or other potential confounders 1. Case-​control study 2. Exposure in past private wells based on statistical modeling 1. Prospective cohort mortality study 2. Unusual findings: highest hazard ratios are for outcomes not commonly linked to arsenic (prostate and pancreas cancer) 3. Urine measures only a few weeks of exposure 1. Case-​control study 2. Low power to assess past exposures 3. < 15% of subjects lived in highly exposed areas > 40 years before diagnosis 4. OR > 1.0 but not statistically significant 1. Retrospective cohort study 2. Very low exposures: median exposures of 0.7–​2.1 ug/​L and 95th percentiles of 2.0–​2.5 ug/​L 3. Exposures from food not assessed

 31



311

Water Contaminants

Table 18–​1.  Continued Study

Area/​Outcome

Chen et al., 2010a

Northeastern Taiwan Lung cancer

Chen et al., 2010b

Northeastern Taiwan Urothelial cancer

Chen et al., 2004 Han et al., 2009

Northeastern and southwest Taiwan Lung cancer Idaho, US Multiple cancers

Buchet and Lison, 1998

Belgium Multiple cancers

Ferreccio et al., 2000

Chile Lung cancer

Kurttio et al., 1999

Finland Bladder and kidney cancer

Steinmaus et al., 2014

Chile Lung cancer

Results

Description

For arsenic water concentration < 100 µg/​L: RR = 1.10 (0.74, 1.63) for 10–​49.9 µg/L​ RR = 0.99 (0.59, 1.68) for 50–​99.9 ug/​L For arsenic water concentration < 100 µg/​L: RR = 1.85 (0.45, 7.61) for 10–​49.9 µg/L​ RR = 2.19 (0.43, 11.1) for 50–​99.9 ug/​L For arsenic water concentration 10–​99 µg/​L: RR = 1.09 (0.63, 1.91) County average arsenic concentrations No statistically significant correlations with lung, bladder, or kidney cancer incidence in adjusted analyses Arsenic water concentrations 20–​50 µg/​L Males SMR = 1.05 (0.94, 1.18) Lung cancer SMR = 0.92 (0.63, 1.35) Bladder cancer SMR = 1.13 (0.67, 1.91) Kidney cancer Females SMR = 1.24 (0.83, 1.87) Lung cancer SMR = 1.20 (0.63, 2.31) Bladder cancer SMR = 0.91 (0.45, 1.81) Kidney cancer Arsenic water concentrations < 89 µg/​L OR = 0.3 (0.1, 1.2) for 10–​29 µg/​L OR = 1.8 (0.5, 6.9) for 30–​59 µg/​L OR = 4.1 (1.8, 9.6) for 60–​89 µg/​L Arsenic water concentrations ≥ 0.5 µg/​L RR = 1.51 (0.67, 3.38) Bladder cancer RR = 1.07 (0.46, 2.52) Kidney cancer For exposures > 10 years before cancer diagnosis As conc (µg/​L) OR   (90% CI)   6.5 1.00 (ref) 23.0 1.43 (0.82, 2.52) 58.5 2.01 (1.14, 3.52)

1. Cohort study 2. Arsenic water concentrations measured at the residence at the time of study recruitment 1. Cohort study 2. Arsenic water concentrations measured at the residence at the time of study recruitment 1. Cohort study 2. Includes some subjects in Chen et al., 2010 3. Less than lifetime exposure in many subjects 1. Ecologic study 2. Used county average arsenic well concentrations 3. Mostly low exposures 4. Did not account for latency 1. Ecologic mortality study 2. Limited exposure data 3. Exposures from water and air, subjects living near zinc smelters

1. Case-​control study 2. Exposure based on average concentrations for the years 1958–​1970 1. Retrospective cohort study 2. Very low exposures 3. Elevated risk ratios seen in smokers: RR = 10.3 (1.16, 92.6) for bladder cancer 1. Case-​control study 2. Median arsenic water concentrations 3. Some data obtained in interviews with next-​of-​kin

As: arsenic; CI: confidence interval; HR: hazard ratio; IRR: incidence rate ratio; OR: odds ratio; RR: relative risk; SMR: standardized mortality ratio. 95% confidence intervals in parentheses unless otherwise noted.

metalloproteinases, and dysregulated signaling pathways (Tokar et al., 2013). Other studies have shown that arsenic may affect stem cells in other cell types, including skin and prostate (Sun et al., 2012; Tokar et al., 2010b). Although a number of different mechanisms have been proposed, the mechanism or combination of mechanisms most likely responsible for the human health effects of arsenic is unknown.

Opportunities for Prevention The most effective and practical way to reduce cancer risks from arsenic is to reduce exposure. Several countries have promulgated regulatory limits for arsenic concentrations in drinking water. In the United States, these apply only to public water sources, not household wells. Large municipal water suppliers have reduced arsenic concentrations through various methods, including the development of new water sources (e.g., tapping new aquifers, desalinization of ocean water), mixing water from wells with higher arsenic concentrations with water with lower arsenic concentrations, and coagulation with iron salts and filtration (Albuquerque Bernalillo County Water Utility Authority,

2010; Sancha et al., 2000; Su et al., 2011; US EPA, 2012). Other treatments include reverse osmosis, activated alumina, microfiltration, and ion exchange (US EPA, 2000b). Smaller reverse osmosis filters are commonly used to reduce arsenic concentrations in private domestic wells, where most of the other approaches are prohibitively expensive (Slotnick et al., 2006). The effectiveness of these filters may vary considerably due to inconsistent maintenance and other factors (George et al., 2006; Thomson et al., 2000; Walker et al., 2008). A variety of approaches have been used to reduce arsenic exposures in Bangladesh. A mass survey of over 5 million wells measured arsenic concentrations and used colored paint to indicate the level of contamination. Wells with arsenic concentrations < 50 µg/​l were painted green; wells over this level (approximately 1.4 million) were painted red (Johnston and Sarker, 2007). Contamination in this area primarily affected tube wells that accessed shallower aquifers (i.e., < 150 m). Mitigation efforts have closed highly contaminated wells, expanded the use of surface water, and created deep tube wells, shallow dug wells, and sand pond filters. Educational programs were implemented in highly exposed areas in Araihazar, Bangladesh. Participation in

312

312

PART III:  THE CAUSES OF CANCER

a school-​ based educational and intervention program resulted in a five-​fold increase in switching to a lower arsenic well, and significant declines in urinary arsenic levels among participants (Khan et al., 2015). Chelation therapy with British anti-​lewisite (BAL; dimercaprol) or its analogs unithiol (DMPS) and succimer (DMSA) is used to treat very high-​dose acute poisonings (e.g., occupational accidents, homicide, or suicide attempts). Chelation has not been shown to be appropriate for prevention or treatment of arsenic exposures from food or drinking water (Kosnett, 2013). Biomonitoring and early detection have been proposed in areas where arsenic exposures have already been reduced (Adonis et al., 2014). To date, the effectiveness of such programs has not been established.

DISINFECTION BYPRODUCTS Exposure The disinfection of drinking water is one of the triumphs of public health, responsible for significant reductions in mortality and morbidity due to infectious disease (Cutler and Miller, 2005). An unintended consequence of this approach, however, has been the formation of disinfection byproducts (DBPs) that may affect cancer risk. DBPs are formed by the reaction of chlorine or other disinfectants with organic matter; they include a complex mixture of compounds that depends on both the treatment regimen and characteristics of the source water. The formation of specific compounds is influenced by the level and type of organic matter present, the existence of bromide and/​or iodide, and the pH and temperature of the water (Richardson et al., 2007). Trihalomethanes (THMs) comprise one category of DBP. They were discovered in 1974 (Rook, 1974)  and are composed of chloroform (typically the dominant THM), dibromochloromethane, bromodichloromethane, and bromoform. More than 700 DBPs have now been identified. Major classes that are regulated in the United States include the THMs, haloacetic acids (HAAs), and oxyhalides (bromate and chlorite). There are also multiple unregulated classes, such as the halonitromethanes, iodo-​ acids and other halo-​ acids, chlorinated hydroxy furanones (e.g., MX), haloamides and haloacetonitriles, nitrosamines, and aldehydes (Richardson et  al., 2007). Many of these compounds have been shown to be mutagenic and/​or genotoxic. As indicated earlier, only three classes of compounds are regulated in the United States; it is assumed that reduced levels of these compounds will reduce overall exposure to DBPs (US EPA, 2003). The regulatory maximum contaminant level (MCL) in the United States is 80 ppb total for total THM, 60 ppb for the HAAs, 10 μg/​l for bromate, and 1000 μg/​l for chlorite (US EPA, 2006). Concern about bladder cancer was a key consideration when the rules and cost–​benefit analysis were updated in 2006 (US EPA, 2005). In 1991, the IARC determined that there was inadequate evidence to classify the use of chlorinated drinking water as carcinogenic to humans (IARC, 1991). Since then, IARC has evaluated individual DBPs, none of which is listed as a Class I carcinogen (IARC, 2004, 2013, 2014). Chloral and chloral hydrate were designated probable human carcinogens (Group 2A), whereas dichloroacetic acid, trichloroacetic acid, dibromoacetic acid, bromochloroacetic acid, and MX were listed as possible human carcinogens (Group 2B) (IARC, 2004, 2013, 2014). People can be exposed to DBPs through ingestion, dermal absorption, and inhalation. While ingestion has received the most attention in epidemiologic studies, recent work has evaluated exposure through inhalation or dermal absorption of low molecular weight, volatile, non-​polar constituents (Richardson et  al., 2007). Studies have demonstrated increased THM levels in blood following exposure, with higher THM blood levels associated with showering than with ingestion (Backer et al., 2000). Additionally, studies of swimmers in chlorinated pools show increased THM concentrations in exhaled air (Aggazzotti et  al., 1993; Caro and Gallego, 2007; Kogevinas et  al.,

2010; Lourencetti et  al., 2012), blood (Aggazzotti et  al., 1990), and urine (Caro and Gallego, 2007, 2008).

Types of Cancer Soon after THMs were identified, ecologic studies conducted in the United States, Norway, and Finland showed elevated cancer rates associated with chlorination byproducts or surrogate measures (Cantor et al., 2006). The exposures in these studies were broadly defined, with some studies comparing the use of chlorinated to unchlorinated water, and others comparing surface water (which tends to have higher levels of DBPs) to groundwater. Levels of THM and mutagenicity were estimated. The most commonly identified cancer sites with elevated rates were bladder, colon, and rectum. Subsequently, a number of death-​certificate-​based case-​control studies were conducted, focusing on these and other sites. Information on cause of death and current residence was obtained from death certificates (Cantor et  al., 2006). Although these studies had methodologic limitations, including a lack of historical exposure information and information on potential confounders, they generally supported an association between chlorinated byproducts and risk of cancers of the bladder, colon, and rectum (Cantor et al., 2006). Following the ecological studies of cancer mortality, a series of case-​ control studies of incident cancers collected more information on longer-​term exposures, and variables that might confound or interact with DBPs. The site with the strongest evidence of a causal relationship was bladder, although risks associated with DBP were generally modest, with ORs around 1.5 associated with estimated total THM levels of 50 μg/​L (Costet et al., 2011). Three cohort and eight case-​control studies evaluated this association, in addition to several reviews and meta-​analyses (Costet et  al., 2011; Villanueva et al., 2003). The cohort studies and all but two of the case-​control studies were reviewed in the previous edition of this text and will not be discussed here (Cantor et  al., 2006). Many of the earlier studies focused on surrogates of exposure to DBPs, such as use of a chlorinated water source for drinking water, or duration of use of surface versus groundwater. A meta-​analysis published in 2003 reported an odds ratio of 1.2 (95% CI: 1.1, 1.4) associated with ever consuming chlorinated water, and 1.4 (95% CI: 1.2, 1.7) for long-​term exposure (Villanueva et al., 2003). Although results were generally consistent among studies, variation existed particularly with respect to apparent sex differences. Some studies showed stronger effects among men. However, none of the individual studies had adequate power to detect associations among women. Since the previous edition of this text, two case-​control studies of bladder cancer have been conducted, one from six regions in Spain (Villanueva et al., 2007) and another in three northern New England states in the United States (Beane Freeman et al., 2017). Both expanded upon previous exposure assessment efforts by considering routes of exposure other than ingestion, such as showering, bathing, and use of swimming pools (Costet et  al., 2011; Villanueva et  al., 2004). In Spain, THM levels were estimated by using historical information from individual municipalities (Villanueva et al., 2006). Compared to participants in the lowest quartile of long-​term average THM concentration in water (< 8 µg/​l), those in the highest quartile (> 49 μg/​l) had a two-​fold increase of bladder cancer risk (OR = 2.10; 95% CI: 1.09, 4.02). In a metric that incorporated both the THM concentration and the amount of water ingested, there was a non-​significant increase in bladder cancer risk among those with the highest intake of THM (> 35 μg/​d) compared to those with no exposure (OR = 1.35; CI: 0.92, 1.99) (Villanueva et  al., 2007). Information about the duration of showering and bathing was combined with the average THM concentration to evaluate inhalation and dermal exposure. Those with the highest exposure levels had significantly elevated risk (OR = 1.83; CI: 1.17, 2.87). Bladder cancer risk was also elevated among those who reported ever use of swimming pools, another potentially large source of DBP exposure (OR = 1.57; CI: 1.18, 2.09), although risk did not increase with increasing total lifetime hours of use. The results for ingestion and concentration were stronger among men than women, whereas the

 31



Water Contaminants

associations with showering/​bathing and swimming pool exposures were similar in both sexes. In the population-​based case control study in New England (Beane Freeman et al., 2017), historical information from public utilities and residential histories were used to estimate total THM exposures, as well as levels of chlorinated and brominated compounds. At the 95th percentile of average daily intake (> 103.9 µg/​day), bladder cancer risk was significantly associated with total THM (OR = 1.53; 95% CI: 1.01, 2.32) and brominated THMs (OR = 1.98; 95% CI: 1.19, 3.29). For cumulative intake (mg) of chlorinated and brominated THMs from age 10 through diagnosis or interview, bladder cancer risk was elevated among those with > 95th percentile of total THM (OR = 1.45; 95% CI: 0.95, 2.2), and chlorinated and brominated THMs (OR = 1.77; CI: 1.05, 2.99 and OR = 1.78; CI: 1.05, 3.00, respectively). There was no evidence of bladder cancer risk with swimming pool use (OR = 0.94; CI: 0.55, 1.59 for > 12,174 lifetime hours compared to those who never used swimming pools). Although both studies showed increased risk of bladder cancer associated with increased THM levels, the risk estimates in the New England study were lower than in Spain. It should be noted that the estimated exposure to THMs was also much lower in New England (Beane Freeman et  al., 2017; Salas et  al., 2013). Differences in the association with swimming pool use may be due to chance or to differences in swimming pool disinfection that are not well understood. The DBPs in swimming pools are affected by the same characteristics as drinking water. The specific DBPs formed in swimming pools vary by the disinfection processes used (Lee et  al., 2010), the number of swimmers in the pool, and the presence of urine, other body fluids, and personal care products (Chowdhury et al., 2014). These issues should be studied further. Much less epidemiologic research has been conducted on the association between DBP exposure and cancers other than bladder. Colon and rectum cancer are the sites with the most evidence for an association. A case-​control study in Spain and Italy reported no association between total THMs and colorectal cancer, although there was a suggestive increase in risk among men with the highest quartile of brominated THM exposure (OR = 1.43; 95% CI: 0.83, 2.46, p trend = 0.04) (Villanueva et al., 2017). There was no association with showering and bathing in water with higher levels of total THMs, chloroform, or brominated compounds, nor were risks elevated in women. There was a suggestion of an interaction between brominated THMs and four genetic polymorphisms in the CYP2E1 gene. An interaction between this gene and total THMs was observed in the Spanish study of bladder cancer. Results from this study were similar for both colon and rectal cancers. Other studies, however, have suggested that the risks may differ for colon and rectum cancer. For example, a case-​control study of colon and rectum cancer in Ontario, Canada, reported increased risk of colon cancer in men, with increasing cumulative exposure to THMs, duration of exposure to chlorinated surface water, and duration of exposure to THM levels > 50 µg/​l. No associations were observed with rectum cancer or in women (King et al., 2000). A case-​control study in Iowa reported increased risk of rectal cancer (Hildesheim et al., 1998). Recently, two studies have suggested an increased risk of rectal cancer at higher levels of bromoform exposure. A case control study of rectal cancer in New York showed a statistically significantly increased risk at the highest levels of bromoform exposure (1.69–​15.43 µg/​day), but not with other THMs (Bove et al., 2007). An ecologic study in Australia evaluated colon and rectal cancers in relation to total THMs and individual species of THMs (Rahman et  al., 2014). The authors reported a statistically significantly increased incidence rate ratio for colon cancer with an increasing interquartile range of bromoform (IQR = 2 µg/​l) in men, but not women. No other individual species were associated with risk, nor was total THM level. There were no significant associations with rectal cancer. Although the study was ecological, a strength was that it evaluated all THMs, separately and in combination. A recent analysis of kidney cancer in a cohort of postmenopausal women in Iowa found no association between kidney cancer and long-​term average total THM (HRQ4vsQ1 = 0.70; 95% CI: 0.41=, 1.20), HAA5 (HRQ4vsQ1 = 0.65; 95% CI; 0.39, 1.08;

313

Table 18–​3), nor with individual THMs or HAAs in models adjusted for water nitrate level (Jones et  al., 2016). More recently, a small study conducted within the National Health and Nutrition Survey demonstrated an association between blood levels of brominated, but not chlorinated, THMs and cancer mortality overall (Min and Min, 2016). Other sites that have been evaluated include brain, esophagus, pancreas, and childhood leukemia (Cantor et al., 2006). Each of these sites has at least one epidemiologic study that shows some evidence of association with increasing exposure to DBPs. As indicated, for both colon and bladder cancer there is some suggestion that risk may differ by sex, with many studies showing associations with THMs only in men (Costet et al., 2011; King et al., 2000). However, it should be noted that most studies are underpowered to detect associations in women. The two largest studies of bladder cancer, with the most women, did not demonstrate sex differences (Beane Freeman et al., 2017; Cantor et al., 1987).

Mechanisms of Carcinogenesis Given the number of compounds identified as DBPs, it is not surprising that several mechanisms for carcinogenicity have been proposed. Some DBPs are genotoxic or mutagenic, either alone or as part of a mixture; chloroform is considered to be mutagenic only at very high levels of exposure (Richardson et al., 2007). Water samples from five European countries demonstrated cytotoxicity in mammalian cells that increased with the number of DBP compounds identified in each sample (Jeong et al., 2012). Although there were differences in genotoxicity among the samples, it did not appear that the genomic DNA damage in these samples correlated with the chemical analyses. Therefore, it may be that the observed associations were due to unidentified DBPs, or perhaps to combinations of DBPs. There has been interest in the effect of genetic susceptibility in modifying risk. In an analysis from the Spanish bladder cancer study, Cantor and colleagues (Cantor et al., 2010) demonstrated significant interaction between genetic variation in key metabolizing pathways and DBP levels on bladder cancer risk. Exposed individuals with high-​risk variants of GSTT1, GSTZ1, and CYP2E1 were at increased risk compared with those without the high-​ risk forms. Other studies have not reported attempts to replicate these findings. It has been suggested that epigenetic mechanisms may also be important, particularly with regard to long-​term, lower-​level exposures. In controls from the bladder cancer case-​control study from Spain, the levels of DNA methylation at some transposons (LINE-​1 5-​ methylcytosine levels) in granulocytes were associated with increased lifetime exposure to THM (Salas et al., 2014). There was also some evidence that levels of methylation modified the association between THM levels and bladder cancer risk (p-​interaction = 0.03). In another epigenome-​wide study, genes that have been associated with cancer at several sites, including colorectal and bladder, were found to have methylation levels that differed by THM level (Salas et al., 2015).

NITRATE Exposure Since the mid-​1920s, human activities have doubled the natural rate at which nitrogen is deposited onto land. Most important has been the production and application of nitrogen fertilizers, the combustion of fossil fuels, and replacement of natural vegetation with nitrogen-​fixing crops such as soybeans (Davidson EA, 2012; Vitousek et al., 1997). In 1909 the Haber-​Bosch process was developed to produce ammonia by reacting nitrogen gas with hydrogen. This allowed a dramatic increase in the production of synthetic fertilizers and expansion of global agriculture. Fertilizer production has increased exponentially since 1960; half of all synthetic fertilizers have been produced since 1985 (Howarth, 2008). As a result, nitrogen inputs to the land have steadily increased. Approximately half of all applied nitrogen drains from agricultural fields to contaminate surface and groundwater (Davidson EA,

314

314

PART III:  THE CAUSES OF CANCER

2012). Nitrate levels in groundwater under agricultural land can be several to 100 times higher than levels under natural vegetation (Nolan BT, 2000). Approximately 15% of the US population relies on privately owned household wells for drinking water (Hutson, 2004). As mentioned, these are not regulated by the EPA. The US Geological Survey assessed available private well measurements in US aquifers sampled in 1991–​2004 and found that 8% exceeded the 10 mg/​l maximum contaminant level (MCL) for nitrate-​nitrogen (NO3-​N) (Focazio et al., 2006). Other private well surveys show that a significant proportion of wells, particularly those less than 30 meters in depth in agricultural areas, exceed the MCL for nitrate (Centers for Disease Control and Prevention, 1998; Johnson and Kross, 1990). Almost all public water supplies have nitrate levels below the MCL. However, in the past few decades, nitrate levels have risen to levels approaching the MCL in some public supplies located in agricultural areas (Ward et al., 2010).

Mechanisms of Carcinogenesis The MCL for nitrate in drinking water was established at 10 mg/​l NO3-​ N to protect against infant methemoglobinemia, not cancer or other health outcomes (US EPA, 2016). Nevertheless, nitrate is a precursor compound of N-​nitroso compounds (NOC), many of which are potent animal carcinogens. NOC cause cancer in every animal species tested and at multiple organ sites (IARC, 2010; Lijinsky, 1986). Approximately 5% of nitrate ingested in drinking water or in the diet is reduced to nitrite by bacteria. This occurs mostly because of the oral microbiome, but continues in the stomach, small intestine, colon, and bladder when nitrate-​reducing bacteria are present (IARC, 2010). Nitrite can then react with amines and amides to form nitrosamines and nitrosamides, the two main types of NOC. Nitrate is found in many foods, with the highest levels occurring in certain green leafy and root vegetables (NRC, 1981). Average daily intakes are estimated to be in the range of 30–​130 mg/​day as nitrate (IARC, 2010). Endogenous NOC formation is blocked by ascorbic acid, alpha-​tocopherol, and other polyphenols. Because these inhibitors are present at high levels in most vegetables, dietary nitrate intake from vegetables may not result in substantial NOC formation (Mirvish, 1995). Drinking water contributes the majority of nitrate intake when levels are near the MCL (Chilvers et al., 1984; IARC, 2010). Sources of exposure to preformed NOC include processed meats and fish, beer, certain occupations, cosmetics, and some drugs. It is estimated that 45%–​75% of human exposure to NOC comes from in vivo formation, however (Tricker, 1997). An IARC Working Group reviewed the epidemiologic studies of ingested nitrate and nitrite through mid-​2006. The evidence for dietary nitrite ingestion and cancer is considered limited (IARC, 2010), based on epidemiologic studies of stomach and esophageal cancer. However, another recent IARC review concluded that processed meats, which usually contain added nitrite and/​ or nitrate, cause cancer in humans, based mostly on the evidence for colorectal cancer (Bouvard et al., 2015). Endogenous NOC formation due to ingestion of water with elevated nitrate has been demonstrated in human volunteers (Mirvish et  al., 1992; Møller et al., 1989; Vermeer et al., 1998), generally under conditions of low antioxidant intake (e.g., low vitamin C). NOC also have been measured in human feces after ingestion of nitrate via drinking water (Rowland et al., 1991). In each of these four studies, increased nitrosation was observed at nitrate concentrations that exceeded the MCL. Some of the NOCs that were formed were known carcinogens. In the hundreds of animal studies conducted since the 1970s, NOCs cause tumors of the upper and lower gastrointestinal tract, urinary tract, lung, thyroid, breast, and ovary. Transplacental exposure induces brain tumors in offspring (Rice, 1989). Long-​term exposure to lower NOC concentrations has a stronger carcinogenic effect than short-​term higher exposures (Lijinsky, 1986). The IARC Working Group classified ingested nitrate or nitrite as probably carcinogenic to humans (2A), when ingested under conditions that result in endogenous nitrosation (IARC, 2010). This designation was based on human mechanistic studies, sufficient evidence in animals for the carcinogenicity of nitrite in combination with amines or amides, and limited epidemiologic evidence for dietary nitrite. The

epidemiologic evidence for drinking water nitrate was considered inadequate. In human studies, it is important to assess effect modification by exposure to modifiers of endogenous nitrosation. To date, there are few studies of any cancer site that have assessed long-​term exposure from drinking water and potential interactions with dietary factors that affect NOC formation.

Epidemiologic Studies Most early epidemiologic studies were ecologic studies of mortality from stomach cancer that used exposure estimates based on measurements of nitrate in drinking water concurrent with death from cancer. Results were mixed, with some studies showing positive associations, many showing no association, and a few showing inverse associations. The results of ecologic studies through 1995 were reviewed in Cantor (1997). Since that review, an ecologic study of cancer incidence in Slovakia (Gulis et al., 2002) evaluated 20-​year average nitrate levels in public water supplies (highest quartile: 6–​10 mg/​l NO3-​N) and found a positive correlation with stomach cancer incidence among women but not men. Other ecologic studies of cancer incidence were largely null: endpoints in these studies included the brain (Barrett et al., 1998; van Leeuwen et al., 1999), bladder and colon (Gulis et al., 2002; van Leeuwen et al., 1999), esophagus and stomach (Barrett et al., 1998), kidney (Gulis et al., 2002), ovary (van Leeuwen et al., 1999), and non-​ Hodgkin lymphoma (NHL) (Cocco et al., 2003; Law et al., 1999; van Leeuwen et  al., 1999). Among the limitations of ecologic studies is their inability to assess individual-​level exposure or dietary factors that might interact with nitrate metabolism and affect the endogenous formation of NOC. They also do not identify potentially susceptible subgroups or other interactions. Since the previous version of this chapter was published, 8 case control and 2 cohort studies have evaluated historical nitrate levels in public water supplies in relation to several cancers. The levels were largely below 10 mg/​l NO3-​N. Most of these studies evaluated potential confounders and factors affecting nitrosation. Details of 9 case-​ control and 2 cohort studies published through 2005 were presented in the previous version of this chapter (Cantor et al., 2006). Table 18–​2 shows the study designs and results of more recent studies, including the results of periodic follow-​ups of a cohort study of postmenopausal women in Iowa (Inoue-​Choi et  al., 2012, 2015; Jones et  al., 2016, 2017; Ward et al., 2010). In the first analysis of drinking water nitrate in the Iowa cohort with follow-​up through 1998, Weyer and colleagues reported that ovarian and bladder cancer were significantly associated with the long-​term average residential nitrate levels at the time of enrollment (Weyer et  al., 2001). They also observed significant inverse associations for uterine and rectal cancer, but no associations with cancers of the breast, colon, rectum, pancreas, kidney, lung, melanoma, NHL, or leukemia. Analyses of public water supply nitrate concentrations and cancers of the thyroid, breast, ovary, bladder, and kidney were published after additional follow-​up of the cohort. The exposure assessment was improved by (a) the computation of average nitrate levels and years of exposure at or above 5 mg/​l NO3-​N, based on time in residence (vs. using one public water supply average nitrate estimate used by Weyer and colleagues); and (b) by estimation of THM levels and dietary nitrite intake. Thyroid cancer was evaluated for the first time after follow-​up of the cohort through 2004. A total of 40 cases was identified (Ward et al., 2010). Among women with > 10 years of public water supply whose levels exceeded 5 mg/​l NO3-​N for 5 years or more, thyroid cancer risk was 2.6 times higher than that of women whose supplies never exceeded 5 mg/​l. Higher dietary nitrate intake was also associated with increased thyroid cancer risk. With follow-​up through 2010, the risk of ovarian cancer remained significantly increased among women in the highest quartile of average nitrate in public supplies (Inoue-​Choi et al., 2015). Ovarian cancer risk among private well users was also elevated compared to the lowest nitrate quartile for public supplies. Associations were stronger when vitamin C intake was below median levels with a significant interaction for users of private wells. Higher intake of processed meat sources of dietary nitrite was also associated with increased risk. Overall, breast

 315

Table 18–​2. Case-​Control and Cohort Studies of Drinking Water Nitrate Levels and Cancer (2004–​2017) First Author (Year) Country

Study Design, Years Regional Description

Exposure Description

Cancer Sites Included

Summary of Findingsa,b

Comments

> 11 mg/​L NO3-​N vs. ND OR = 1.0 (CI: 0.4, 2.2); exclude bottled water users OR = 1.5 (CI: 0.6, 3.8) > 1.5 mg/​L NO2-​N vs. ND OR = 2.1 (CI: 0.6, 7.4); exclude bottled water users OR = 5.2 (CI: 1.2, 23.3) Well as water source for entire pregnancy including periconceptual period (vs. public supply) associated with increased risk in Canada and Seattle but not other centers Private wells: > 5.0 mg/​L NO3-​N vs. ND OR = 0.8 (CI: 0.2, 2.5) Public supply average: > 2.9 mg/​ L NO3-​N vs. < 0.63 OR = 1.2 (CI: 0.6, 2.2) Years > 5mg/​L NO3-​N: 10+ vs. 0 OR = 1.4 (CI: 0.7, 2.9)

Strongest associations for astroglial tumors with dipstick nitrate and nitrite measurements at pregnancy residence

Nitrate intake from water (highest quintile > 1.7 mg/​day NO3-​N [median in quintile = 2.4 mg/​day] vs. lowest RR = 1.11 (CI: 0.87, 1.41; p-​trend = 0.14) Public supply average: > 2.8 mg/​ L NO3-​N vs. < 0.62 OR = 0.89 (CI: 0.57, 1.39); Years > 5mg/​ L NO3-​N 11+ vs. 0 OR = 1.03 (CI: 0.66, 1.60) Joint effect for 11+ years > 5 mg/​ L NO3-​N and red meat (> 1.2 servings/​day) OR = 1.91 (CI: 1.04, 3.51); p-​interaction = 0.01 Private wells > 10.0 mg/​L NO3-​N vs. < 0.5: Colorectal cancer OR = 1.52 (CI: 0.95, 2.44); proximal colon cancer: OR = 2.91 (CI: 1.52, 5.56); no association for distal colon cancer or rectal cancer

No effect modification by vitamin C, E, smoking

Average nitrate quartiles (> 4.32 vs. < 2.45 mg/​L NO3-​N): stomach OR = 1.2 (CI: 0.5, 2.7); esophagus OR = 1.3 (CI: 0.6, 3.1); Years > 10 mg/​L NO3-​N (9+ vs. 0): stomach OR = 1.1 (CI: 0.5, 2.3); esophagus OR = 1.2 (CI: 0.6, 2.7) Private well users (> 4.5mg/​L NO3-​ N vs. < 0.5) stomach OR = 5.1 (CI: 0.5, 52; 4 cases, 13 controls); esophagus OR = 0.5 (CI: 0.1, 2.9; 8 cases; 13 controls)

> 2.7mg/​L NO3-​N & > 30g/​ day processed meat vs. low intakes both: stomach OR = 2.0 (CI: 0.8, 4.8); p-​interaction = 0.21; no interaction for esophagus; no interaction with vitamin C, red meat for either cancer

Mueller (2004) 6 countries

Population-​based case-​control Incidence, 1976–​1994 Los Angeles County, San Francisco Bay Area, Seattle–​ Puget Sound region of the United States; Paris, France; Milan, Italy; Valencia, Spain; Winnipeg, Canada

Dipstick measurements of nitrate and nitrite in the pregnancy water supply among women who had not moved (185 cases, 341 controls); population excluding bottled water users (131 cases, 241 controls)

Childhood (< 15 years) malignant brain tumors

Ward (2006) United States

Population-​based case-​control Incidence, 1998–​2000 Iowa

Non-​Hodgkin lymphoma

Zeegers (2006) Netherlands

Cohort Incidence, 1986–​1995 204 municipal registries across the Netherlands

Nitrate levels in public water supplies among those with nitrate estimates for > 70% of person-​ years > 1960 (181 cases, 142 controls); nitrate measurements for private well users at time of interviews (1998–​2000; 54 cases, 44 controls). 1986 nitrate level in 364 pumping stations, exposure data available for 871 cases, 4359 members of the subcohort

Ward (2007) United States

Population-​based case control Incidence, 1986–​1989 Iowa

Nitrate levels in public water supplies among those with nitrate estimates for > 70% of person-​ years > 1960 (201 cases, 1244 controls)

Kidney (renal cell carcinoma)

McElroy (2008) United States

Population-​based case-​control, women Incidence, 1990–​1992 and 1999–​2001 Wisconsin

Colorectum

Ward (2008) United States

Population-​based case control Incidence, 1988–​1993 Nebraska

Limited to women in rural areas with no public water system (475 cases, 1447 controls); nitrate levels at residence (presumed to be private wells) estimated by kriging using data from a 1994 representative sample of 289 private wells Controls from prior study of lymphohematopoetic cases and controls interviewed in 1992–​ 1994; proxy interviews for 80%, 76%, 61% of stomach, esophagus, controls, respectively. Nitrate levels (1965–​1985) in public water supplies for > 70% of person-​years (79 distal stomach, 84 esophagus, 321 controls); private wells sampled at interview (15 stomach, 22 esophagus, 44 controls)

Bladder

Stomach and esophagus (adenocarcinomas only)

No effect modification by vitamin C, smoking

Joint effects of water nitrate and vitamin C showed similar pattern to red meat (p-​interaction = 0.13)

Interactions with dietary intakes not evaluated; no interaction with smoking

(continued)

316

Table 18–​2  Continued First Author (Year) Country

Study Design, Years Regional Description

Exposure Description

Cancer Sites Included

Ward (2010) United States

Cohort of women ages 55–​69 Incidence, 1986–​2004 Iowa

Nitrate levels in public water supplies (1955–​1988) and private well use among women > 10 years at residence (21,977 women; 40 thyroid cases); no measurements for private wells

Thyroid

Inoue-​Choi (2012) United States

Cohort of women ages 55–​69 Incidence, 1986–​2008 Iowa

Nitrate levels in public water supplies (1955–​1988) and private well use among women > 10 years at residence (20,147 women; 1751 breast cases); no measurements for private wells

Breast

Inoue-​Choi (2015) United States

Cohort of women ages 55–​69 Incidence, 1986–​2010 Iowa

Ovary

Espejo-​Herrera (2015) Spain

Hospital-​based case-​control Incidence, 1998–​2001 Asturias, Alicante, Barcelona, Valles-​Bages, Tenerife provinces

Jones (2016) United States

Cohort of women ages 55–​69 Incidence, 1986–​2010 Iowa

Nitrate levels in public water supplies (1955–​1988) and private well use among women > 10 years at residence with nitrate and trihalomethane estimates (17,216 women; 190 ovarian cases); no measurements for private wells Adjusted for total trihalomethanes (TTHM) (1955–​1988), measured TTHM levels in 1980s, prior years estimated by expert) Nitrate levels in public supplies (1979–​2010) and bottled water (measurements of brands with highest consumption based on a Spanish survey); analyses limited to those with > 70% of residential history with nitrate estimate (531 cases, 556 controls) Nitrate levels in public water supplies (1955–​1988) and private well use among women > 10 years at residence with nitrate and trihalomethane estimates (20,945 women; 170 bladder cases); no measurements for private wells Adjusted for total trihalomethanes (TTHM) (1955–​1988), measured TTHM levels in 1980s, prior years estimated by expert)

Summary of Findingsa,b

Comments

Average nitrate quartiles (> 2.46 vs < 0.36 mg/​L NO3-​N) HR = 2.18 (CI: 0.83, 5.76; p-​trend = 0.02) Years > 5 mg/​L (> 5 years vs. 0) HR = 2.59 (CI: 1.09, 6.19; p-​trend = 0.04); private well (vs. < 0.36 mg/​L NO3-​N) HR = 1.13 (CI: 0.83, 3.66) Average nitrate quintiles (> 3.8 vs. < 0.32 mg/​L NO3-​N) HR = 1.14 (CI: 0.95, 1.36; p-​trend = 0.11); private well (vs. < 0.32 mg/​L NO3-​N) HR = 1.14 (CI: 0.97, 1.34) Subgroup with folate > 400 µg/​ d: (> 3.8 vs. < 0.32 mg/​L NO3-​ N) HR = 1.40 (CI: 1.05, 1.87; p-​trend = 0.04); private well (vs. < 0.32 mg/​L NO3-​N) HR = 1.38 (CI: 1.05, 1.82) No association among those with low folate < 400 µg/​d Highest vs. lowest quartile average (> 2.98 mg/​L vs. < 0.47 mg/​L NO3-​ N) HR = 2.03 (CI: 1.22, 3.38; p-​ trend = 0.003), adjusted for TTHM; years > 5 mg/​L (> 4 years vs. 0) HR = 1.52 (CI: 1.00, 2.31; p-​ trend = 0.05), adjusted for TTHM Private well users (vs. < 0.47 mg/​L NO3-​N) HR = 1.53 (CI: 0.93, 2.54)

Dietary nitrate intake quartiles positively associated with risk (p-​trend = 0.046)

Bladder

Average level (age 18-​interview) > 2.26 vs. 1.13 mg/​L NO3-​N OR = 1.04 (CI: 0.60, 1.81) Years > 2.15 mg/​L NO3-​N (75th percentile): > 20 vs. 0 years OR = 1.41 (CI: 0.89, 2.24)

No interaction with vitamin C, E, red meat, processed meat, average THM level

Bladder

Highest vs. lowest quartile average (> 2.98 vs. < 0.47 mg/​L NO3-​N) HR = 1.47 (CI: 0.91, 2.38; p-​ trend = 0.11), adjusted for average TTHM levels Years > 5 mg/​L (> 4 years vs. 0) HR = 1.61 (CI: 1.05, 2.47; p-​ trend = 0.03), adjusted for average TTHM levels Private well users (vs. < 0.47 mg/​L NO3-​N) HR = 1.53 (CI: 0.93, 2.54)

Significant interaction with smoking: Current smokers with > 2.98 mg/​L NO3-​N vs. non-​smokers < 0.47 mg/​L NO3-​N HR = 3.67 (CI: 1.43, 9.38); p-​interaction = 0.03; no significant interaction with vitamin C, TTHM levels

Interaction with vitamin C and smoking not evaluated

Associations with average water nitrate and private well use were stronger among women with < median vitamin C intake (average nitrate p-​trend = 0.005; p-​interaction = 0.33; private well p-​interaction = 0.01) No interaction with red meat intake

 317

Espejo-​Herrera (2016) Spain, Italy

Pooled case-​control studies Incidence, 2008–​2013 Spain (9 provinces) and population-​based controls; Italy (two provinces) and hospital-​ based controls

Espejo-​Herrera (2016) Spain

Hospital-​based case-​control Incidence, 2008–​2013 Spain (8 provinces)

Jones (2017) United States

Cohort of women ages 55–​69 Incidence, 1986–​2010 Iowa

Nitrate levels in public supplies (2004–​2010) for 349 water supply zones, bottled water (measurements of brands with highest consumption based on surveys in Spain and Italy), and private wells and springs (measurements in 2013 in 21 municipalities of León, Spain, the area with highest non-​public water use). Analyses include those with nitrate estimates for > 70% of period 30 years before interview (1869 cases, 3530 controls) Nitrate levels in public supplies (2004–​2010), bottled water (measurements of brands with highest consumption based on a Spanish survey), and private wells and springs (measurements in 2013 in 21 municipalities of León, Spain, the area with highest non-​ public water use). Analyses include women with > 70% of period from age 18 to 2 years before interview (1245 cases, 1520 controls) Nitrate levels in public water supplies (1955–​1988) and private well use among women > 10 years at residence with nitrate and trihalomethane estimates (20,945 women; 163 kidney cases); no measurements for private wells Adjusted for total trihalomethanes (TTHM) (1955–​1988), measured levels in 1980s, prior year levels estimated by expert)

Colorectal cancer

Water nitrate intake based on average nitrate levels (estimated 30 to 2 years prior to interview) and water intake (L/​day) > 2.3 vs. < 1.1 mg /​day NO3-​N OR = 1.49 (CI: 1.24, 1.78), adjusted for diet/​other colorectal cancer risk factors; colon OR = 1.52 (1.24, 1.86), rectum OR = 1.62 (1.23, 2.14)

Stronger associations for men and among those with high red meat intake; no significant interaction with red meat, vitamin C, E, fiber

Breast

Water nitrate intake based on average nitrate levels (age 18 to 2 years prior to interview) and water intake (L/​day). Post-​menopausal women: > 2.0 vs. 0.5 mg/​day NO3-​N OR = 1.32 (0.93, 1.86); Premenopausal women: > 1.4 vs. 0.4 mg/​day NO3-​N OR = 1.14 (0.67, 1.94)

Water nitrate intake estimated from age 18 to 30 and from 15 to 2 years before interview showed similar results. Stronger associations among postmenopausal women with high red or processed meat intake; no significant interaction with red meat, vitamin C, E, smoking

Kidney

95th percentile vs. lowest quartile of average nitrate level (> 5.00 vs. < 0.47 mg/​L NO3-​N) HR = 2.23 (CI: 1.19, 4.17; p-​trend = 0.35), adjusted for TTHM Years >5 mg/​L (> 4 years vs. 0) HR = 1.54 (CI; 0.97, 2.44; p-​ trend = 0.09), adjusted for THM Private well users (vs. < 0.47 mg/​L NO3-​N) HR = 0.96 (CI: 0.59, 1.58)

No significant interaction with smoking, vitamins C or E

ND: not detected a Nitrate or nitrite levels presented in the publications as mg/​L of the ion were converted to mg/​L as NO3-​N or NO2-​N b Odds ratios (OR) for case-​control studies, incidence rate ratios (RR), and hazard ratios (HR) for cohort studies, and 95% confidence intervals (CI).

318

318

PART III:  THE CAUSES OF CANCER

cancer risk was not associated with water nitrate levels with follow-​up through 2008 (Inoue-​Choi et  al., 2012). Among women with folate intake ≥ 400 μg/​day, risk was significantly increased for those in the highest average nitrate quintile (HR = 1.40; 95% CI: 1.05, 1.87) and among private well users (HR = 1.38; 95% CI: 1.05, 1.82), compared to those with the lowest average nitrate quintile. There was no association with nitrate exposure among women with lower folate intake. Analyses of effect modification by other dietary factors were not presented. With follow-​up through 2010, there were 130 bladder cancer cases among women who had used publicly supplied water > 10 years. Risk remained elevated among women with the highest average nitrate levels and was significantly increased among women whose drinking water concentration exceeded 5 mg/​l NO3-​N for at least 4 years (Jones et al., 2016). Risk estimates were not changed by adjustment for THM levels. Smoking, but not vitamin C intake, modified the association with nitrate in water; increased risk was apparent only in current smokers. Dietary nitrite intake was not associated with risk. With follow-​ up through 2010, there were 125 kidney cancer cases among women in the public water supply analysis; risk was significantly increased among those in the 95th percentile of average nitrate (> 5.0 mg/​L NO3-​N) compared with the lowest quartile (HR = 2.2; 95% CI: 1.2, 4.2). There was no positive trend with the average nitrate level and no increased risk for women using private wells, compared to those with low average nitrate in their public water supply (Jones et al., 2016). Higher risk was evident in the 95th percentile of dietary nitrite intake from processed meats but with no significant trend. In contrast to the positive findings for bladder cancer among the cohort of Iowa women, a cohort study of men and women aged 55–​69 in the Netherlands with lower nitrate levels in public supplies found no association between water nitrate ingestion (median in top quintile = 2.4 mg/​day NO3-​N) and bladder cancer risk (Zeegers et al., 2006). Dietary intake of vitamins C and E and history of cigarette smoking did not modify this observation. A hospital-​based case-​control study of bladder cancer in multiple areas of Spain (Espejo-​Herrera et  al., 2015) assessed lifetime water sources and usual intake of tap water. Nitrate levels in public water supplies were low, with almost all average levels below 2 mg/​l NO3-​N. Risk of bladder cancer was not associated with the nitrate level in drinking water or with estimated nitrate ingestion, and there was no evidence of interaction with factors affecting endogenous nitrosation. Several case-​control studies conducted in the Midwestern United States obtained lifetime histories of drinking water sources and estimated exposure for users of public water supplies. In contrast to findings of an increased risk of NHL associated with higher average nitrate levels in Nebraska public water supplies in an earlier study (Ward et al., 1996), there was no association with similar concentrations in public water sources in a case-​control study of NHL in Iowa (Ward et  al., 2006). A  study of renal cell carcinoma in Iowa (Ward et  al., 2007)  found no association with the level of nitrate in public water supplies, including the number of years that the supply exceeded 5 or 10 mg/​l NO3-​N. Higher nitrate levels in public water supplies increased risk among subgroups that reported above the median intake of red meat intake or below the median intake of vitamin C, however. These interactions were statistically significant. A small case-​control study of adenocarcinoma of the stomach and esophagus among men and women in Nebraska (Ward et  al., 1997)  estimated nitrate levels among long-​term users of public water supplies and found no association between average nitrate levels and risk. A case-​control study among rural women in Wisconsin estimated nitrate levels in private wells using spatial interpolation of nitrate concentrations from a 1994 water quality survey and found a significant increased risk of proximal colon cancer among women estimated to have nitrate levels > 10 mg/​l NO3-​N compared to levels < 0.5 mg/​l. Risks of distal colon cancer and rectal cancer were not associated with the nitrate level (McElroy et al., 2008). Water nitrate ingestion from public supplies, bottled water, and private wells and springs over the adult lifetime was estimated in analyses that pooled case-​control studies of colorectal cancer in Spain and Italy (Espejo-​Herrera et  al., 2016a). Risk of colorectal cancer was significantly increased among those with > 2.3 mg/​day NO3-​N (vs. < 1.1 mg/​day); risks were stronger among men, but there were

no significant interactions with red meat, vitamins C and E, and fiber intakes. Postmenopausal and premenopausal breast cancers were not associated with water nitrate ingestion in a hospital-​based case-​control study in Spain (Espejo-​Herrera et al., 2016b). Animal studies demonstrate that in utero exposure to nitrosamides can cause brain tumors in the exposed offspring. Water intake during pregnancy was estimated in a multicenter case-​control study of childhood brain tumors in five countries based on the maternal residential water source (Mueller et al., 2004). Nitrate/​nitrite levels in water supplies were measured using a dipstick method for a subset of the women; however, most of these measurements occurred many years after the pregnancy. Drinking water from private wells versus public water supplies was not consistently associated with risk of childhood brain tumors. However, higher nitrite levels (> 1.5 mg/​l nitrite-​N) in the drinking water were associated with significantly increased risk of childhood brain tumors, especially astroglial tumors.

Biomarkers of Genetic Damage Two cross-​sectional studies of the genotoxic effects of nitrate in drinking water included individuals drinking well water with nitrate concentrations ranging from 11 to 65 mg/​l as NO3-​N. The first study found no increase in the frequency of peripheral lymphocyte sister chromatid exchanges with increasing levels of nitrate (Kleinjans et  al., 1991). A  subsequent study employed the hypoxanthine-​guanine phosphoribosyltransferase (HPRT) variant frequency test in peripheral lymphocytes (van Maanen et  al., 1996). An increased prevalence of HPRT variants in subjects drinking medium and high levels of nitrate was observed. An inverse correlation between the labeling index in lymphocytes and nitrate exposure was suggestive of an exposure-​related immunosuppressive effect. Future studies examining the genotoxic potential of nitrate in drinking water can make valuable contributions to understanding the adverse effects of nitrate exposure and warrant further exploration.

CYANOBACTERIAL TOXINS Cyanobacteria (formerly known as blue-​green algae) are an ancient and ubiquitous family of prokaryotic organisms, with fossil remains dating back ~3.5 billion years. Photosynthesis by these bacteria is likely responsible for earth’s oxygen-​enriched atmosphere, and subsequent evolution of higher organisms (Olson, 2006; Paerl et al., 2011; Schopf, 2000). Cyanobacteria from more than 40 genera produce a wide variety of toxins (cyanotoxins). Among these are hepato-​, nephro-​, neuro-​, and dermatotoxins belonging to several chemical classes (Chorus and Bartram, 1999). Adverse human health effects range from minor skin irritation to death from severe liver damage and possibly amyotrophic lateral sclerosis (Cox and Sacks, 2002). Evidence for carcinogenicity is strongest for microcystins (MCs), nodularin, and cylindrospermopsin (Zegura et al., 2011). Microcystins, usually the most common toxins found in fresh water, are cyclic peptides containing seven amino acids, with two variable positions. There are at least 80 analogs. Microcystin-​LR, the most thoroughly studied, contains lysine (L) and arginine (R) in the variable positions (Butler, 2009). Epidemiologic evidence implicating cyanotoxins (particularly microcystin-​LR) as human carcinogens comes largely from southeastern coastal China. Here, elevated rates of hepatocellular carcinoma (HCC) have long been recognized. The main causes are known to be endemic infection with hepatitis B (and more recently C) viruses, in combination with consumption of grain contaminated by aflatoxin. An additional cofactor, however, may be consumption of surface water contaminated by cyanotoxins. Many studies in this region have reported much higher risk of HCC among populations that regularly used surface waters (largely pond, ditch, or river water) as their primary water source, compared to those who consumed water from shallow or deep wells (Shen et al., 1985; Su, 1979; Yu, 1989; Zhang, 1993). For example, in the period 1972–​1981 in Qidong County, Shen

 319



Water Contaminants

et al. (1985) observed annual incidence rates of HCC of 141.4, 72.3, 43.5, 22.3, and 11.7 per 100,000 for water consumers of house ditch, field ditch, river, shallow wells, and deep wells, respectively. Among 99 HCC cases and 99 matched controls in Guangxi Autonomous Region, Zhang et al. (1993) found an odds ratio of 3.7 (95% CI: 1.3, 11.0) for consuming pond/​ditch water, compared with well water. In a review, Yu (1989) reported several studies of incidence and mortality from HCC conducted in Qidong, Nanhue, and Haimen counties in 1973–​1983. These showed a consistent pattern of higher incidence and mortality rates among those who drank water from ponds and ditches than among those who only used wells. For example, the standardized incidence ratios (SIR) in Qidong were 2.6 and 1.4 per 100,000 for pond and ditch water users, and 0.34 for shallow well water consumers. Variability in hepatitis virus infection or aflatoxin (Yu, 1989) in the affected populations did not explain the risk differences. In a public health intervention (1973–​1978), the population was encouraged to shift drinking water source from surface to wells, resulting in substantially decreased HCC risks (Su, 1979). Prior to 1990, researchers suspected that chemical contaminants in surface waters in the endemic region probably accounted for the association between liver cancer and drinking surface water in the endemic areas. Pesticides and other agricultural chemicals were considered strong candidates. Empirical evidence for their involvement was lacking, however. Extracts from cyanobacteria have long been known to be highly toxic to humans (Schwimmer and Schwimmer, 1955). In particular, gastroenteritis, hepatotoxicity, and occasionally death from these toxins have been observed (Chorus and Bartram, 1999; Griffiths and Saker, 2003; Jochimsen et  al., 1998). Experimental demonstration of skin tumor–​promoting activity in mice that ingest extracts of Microcystis aeruginosa (common in freshwater) was reported in 1989 (Falconer and Buckley, 1989), followed by experimental results showing that microcystin-​LR specifically promotes liver tumors in rodents (Fujiki and Suganuma, 2011; Lian et al., 2006; Nishiwaki-​Matsushima et al., 1992). Field results from the endemic HCC region of China have demonstrated the presence of microcystin in surface water. A series of 989 water samples, tested with an enzyme-​linked immunosorbent assay (ELISA) (detection limit of 50 pg/​ml) revealed that 17% of pond/​ditch water samples (average = 101 pg/​ml), 32% of river water samples (160 pg/​ml), and 4% of shallow well water samples (68 pg/​ml) were positive for microcystin. No microcystin was detected in deep well water (Ueno et al., 1996). Environmental surveys and epidemiologic studies of HCC incidence or mortality from this era further implicated microcystin as a risk factor for HCC (Chen and Kensler, 2014; Harada et al., 1996; Ling, 2000; Yu et al., 2001; Yu and Chen, 1994). Adding to this evidence is statistical modeling of HCC case-​control data that showed a multiplicative interaction between consumption of water from ponds or ditches and infection with hepatitis B virus and aflatoxin ingestion (Zhao et al., 1994), consistent with the promotional effects of MC-​LR seen experimentally. Most of the risk from consumption of microcystin results from its joint effect with hepatitis B infection. Compared to groundwater users negative for hepatitis B antibody (HBab) (OR [ref] = 1.00), the OR estimate among subjects who were negative for HBAb but consumed pond/​ditch water was 1.74; among subjects positive for HBAb who consumed groundwater, the OR was 6.57, and among subjects who were both HBab positive and consumed ditch water, the OR was 11.42. The finding that high-​risk populations consumed water contaminated with microcystin, and that this interacted with hepatitis virus infection to cause HCC, suggests that microcystin is an important risk factor for HCC. Individual-​level data on dose–​response relationships are lacking, however. Collecting such information would be difficult, given the poor predictability and temporal variability of microcystin blooms and the long induction period for liver cancer. Further research on whether microcystin interacts with other HCC risk factors would be valuable. Studies in China have also found an association between likely exposure to microcystin in drinking water and elevated risk of colorectal cancer (Chen et  al., 2003; Zhou et  al., 2002). In 2006, an IARC

319

Working Group classified microcystin-​LR as “possibly carcinogenic to humans,” microcystis extracts as “not classifiable as to their carcinogenicity to humans,” and nodularin also as “not classifiable” (IARC, 2010). A limited number of studies have evaluated the association between cyanobacteria and liver cancer in other populations. In central Serbia, more than 80% of drinking water reservoirs have experienced cyanobacterial blooms over the past 80 years, many with measureable levels of microcystin-​LR (Svircev et al., 2013). The mean incidence of HCC was elevated in three Serbian districts where cyanobacterial blooms occurred annually for the past 25 years. The incidence rate of HCC in these districts averaged 27/​100,000 in the period 1999–​ 2008, nearly four times the rate (7.2/​100,000) in another region of the country that uses groundwater. The geographic distribution of other risk factors (hepatitis B virus, hepatitis C virus, and liver cirrhosis, as measured by mortality), did not explain the localized increases in risk. An ecologic study in Florida, which evaluated HCC incidence among people consuming publically-​supplied surface waters with a history of contamination by microcystins, relative to those who used uncontaminated groundwater sources, was equivocal (Fleming et al., 2002). Microcystin-​LR and nodularin are potent inhibitors of protein phosphatases 1 and 2A; they belong to the okadaic acid class of tumor promotors (Fujiki and Suganuma, 1994, 2011). Microcystin-​LR upregulates production of tumor necrosis factor α gene (TNF-​α), which has tumor-​promoting activity, as well as affecting early-​response genes such as c-​jun, jun B, jun D, c-​fos, fos B, and fra-​1 (Sueoka et al., 1997). The World Health Organization in 1998 first recommended a provisional limit for microcystin-​LR in drinking water of 1 µg/​l (WHO, 2011). Many countries have accepted this level and have issued recommendations for its implementation (Burch, 2008). A Health Advisory issued by the US EPA calls for a limit of 0.3 µg/​l total microcystin for infants and preschool children and 1.6 µg/​l for older children and adults (US EPA, 2015). Few countries have issued legally binding limits (Burch, 2008). Cyanobacteria proliferate rapidly in fresh and marine waters under suitable environmental conditions. So-​ called “water blooms” are exacerbated by phosphorus and nitrogen runoff from agricultural and municipal sources, and by warming of lakes and other drinking water sources driven by the changing global climate (Elliott, 2012; Heisler et al., 2008; Newcombe et al., 2012; O’Reilly et al., 2015; Paerl and Paul, 2012). The public health implications of these combined factors deserves further study.

COMMUNITIES EXPOSED TO INDUSTRIAL CHEMICALS IN DRINKING WATER Here, we describe three settings in which elevated cancer risks have been observed in communities exposed to water contaminated by industrial chemicals. Such studies are intrinsically opportunistic. In one instance (Toms River, NJ), elevated rates of childhood cancers were observed. The exposure involved poorly defined mixtures and did not allow the attribution of risk to specific chemical(s). In the other two scenarios, the predominant chemical exposures were known and observations of elevated risk of specific types of cancer were consistent with occupational exposures and/​or experimental findings.

Camp Lejeune At Camp Lejeune, a US Marine Corps Base Camp in coastal North Carolina, two of eight drinking water systems that served most of the base’s population (including family members of Marines) were contaminated by solvents from the 1950s through 1985, as reported by the US ATSDR (2016). One supply was contaminated by an off-​ base dry cleaner, primarily with perchloroethylene (PCE) (tetrachloroethylene), reaching a maximum measured level of 215 ug/​l; other contaminants included trichloroethylene (TCE), trans-​1,2-​dichloroethylene, and VC, a degradation product of PCE. Another affected water supply was contaminated by sources located on the base

320

320

PART III:  THE CAUSES OF CANCER

involving industrial spills, waste disposal sites, and leaking underground storage tanks. The primary contaminant was TCE (maximum detected level of 1400 ug/​l); levels of PCE were lower (maximum of 100 ug/​l), and benzene was detectable. A third water supply that served 2100 family housing units was not contaminated. The US ATSDR conducted a detailed assessment of exposure and estimated historical average monthly levels of TCE, PCE, VC, and benzene exposure in water provided to work areas and housing units on the base (Maslia et al., 2007, 2013). Two retrospective cohort studies evaluated the risk of total and site-​specific cancer mortality. One cohort was comprised of Marines and Navy personnel who began service in 1975–​1985 (N = 154,932); another included civilian personnel employed at the base during 1973–​ 1985 (N = 4,647). Follow-​up of both cohorts began in 1979 and ended in 2008 (Bove et al., 2014a, 2014b) The primary comparison was with uniformed personnel (N = 154,969) and civilian employees (N = 4690), respectively, at Camp Pendleton, a Marine base of similar size and demographic composition located in California, with no known history of comparable drinking water contamination. Among Marines and Navy personnel, mortality due to all cancers combined was elevated (hazard ratio [HR] = 1.10; 95% CI: 1.00, 1.20). Hazard ratio estimates were also increased for cancer sites of “primary interest” (based on occupational studies or experimental evidence), although these results were not statistically significant. These included kidney cancer (HR = 1.35; CI: 0.84, 2.16), liver cancer (HR = 1.42; CI: 0.92, 2.20), esophageal cancer (HR = 1.43; CI: 0.85, 2.38), cervical cancer (HR = 1.33; CI: 0.24, 7.32), Hodgkin lymphoma (HR = 1.47; CI: 0.71, 3.06), and multiple myeloma (HR = 1.68; CI: 0.76, 3.72). There was an increasing trend of risk of kidney cancer mortality with total contaminant level, and of Hodgkin lymphoma with TCE and with benzene. Increases in risk (not statistically significant) were also found among the civilian employees for mortality from all cancer combined (HR = 1.12; CI: 0.92, 1.36), kidney cancer (HR = 1.92; CI: 0.58, 6.34) and multiple myeloma (HR = 1.84; CI: 0.45, 7.58). A study of incident cancer is underway (2017). A separate case-​control study measured mortality from 13 hematopoietic cancers among children born in the interval 1968–​1985 to mothers with household exposure to drinking water at Camp Lejeune. Odds ratio estimates of 1.6 (CI: 0.5, 4.8) and 1.6 (CI: 0.5, 4.7) were observed for first trimester exposure to any PCE or any VC, respectively (Ruckart et al., 2013). Another small case-​control study, of 71 male breast cancer cases identified from the Department of Veterans Affairs cancer registry, found an adjusted OR of 1.14 (CI: 0.65, 1.97) for ever having been stationed at Camp Lejeune (Ruckart et al., 2015). As mentioned, the 95% CI associated with most cancer sites in the Camp Lejeune studies did not exclude the null hypothesis. When viewed in isolation, these investigations do not provide evidence of causal associations for the contaminants involved. However, the observed risks are consistent with positive epidemiologic evidence from more highly exposed occupational populations, where the exposure levels are better documented, and there are mechanistic and/​or experimental data that support a causal relationship. Based on the overall evidence, primarily from occupational settings, an IARC working group classified TCE, the contaminant found at highest levels at Camp Lejeune, as “carcinogenic to humans” (IARC, 2014). The US EPA has come to a similar conclusion (Chiu et  al., 2013), as has a draft report from the US National Toxicology Program (National Toxicology Program, 2015). Vinyl chloride and benzene have been classified as human carcinogens (IARC, 2012b), and PCE was evaluated by an IARC working group that designated it a “probable” human carcinogen (IARC, 2014).

Community Exposed to Perfluorooctanoic Acid (PFOA) and Related Compounds Perfluorooctanoic acid (PFOA or C8) has been used historically in the production of Teflon, Gore-​ Tex, non-​ stick cookware surfaces, and other products. Manufacturing using PFOA began at the DuPont Washington Works on the Ohio-​West Virginia border in 1951. Nearby populations were exposed, primarily through contaminated drinking

water from private wells and public water supply wells in six nearby districts. Environmental fate and transport modeling revealed that extensive contamination of groundwater occurred via deposition of airborne PFOA from plant stacks and subsequent transport to the aquifer (Shin et al., 2011b). There were also direct releases into the Ohio River. A  class-​action suit brought by local residents alleging health damages was resolved by a pretrial settlement agreement calling for DuPont to fund epidemiologic studies in the surrounding communities and among plant workers (Frisbee et al., 2009). This was implemented by the “C8 Science Panel,” composed of three senior chronic disease epidemiologists who designed and oversaw the studies, reviewed data, and reported to the court whether selected adverse health outcomes were “more probably than not” caused by exposure to PFOA. This criterion is a legal rather than a scientific concept. Any positive results reported would require DuPont to support continuing medical monitoring for the relevant condition(s). Blood serum measurements of PFOA concentrations by the C8 Health Project in 2005–​2006 revealed an overall median concentration of 28.2 ng/​ml in this population, compared with 4 ng/​ml in the United States overall, as measured in the National Health and Nutrition Examination Survey (Steenland et al., 2009). A number of studies were conducted to evaluate the risk of 55 health outcomes potentially associated with PFOA. These studies included individual exposure assessment in a community (N  =  43,449) (Shin et  al., 2011a, 2013), a survey (including personal medical information and background) of PFOA concentrations in blood (serum) and drinking water in the six major contaminated water districts (Winquist et  al., 2013), a geographic analysis of cancer incidence in the region (Vieira et  al., 2013), a cohort mortality study of exposed workers (Steenland and Woskie, 2012), and a study of incident cancers among adults (Barry et  al., 2013). The cancer incidence study included 32,254 adult community residents or plant workers, most of whom had participated in a 2005–​ 2006 baseline survey with serum PFOA measurements. Yearly serum concentrations were estimated for each person from 1952 to 2011 (Shin et  al., 2011a). Subsequent interviews were held in 2008–​2011 to assess health problems (later confirmed by medical records and/​or cancer registry information). Estimated cumulative serum PFOA concentrations were positively associated with kidney and testicular cancer. The HR estimates for kidney cancer with increasing exposure quartiles were 1.0, 1.23, 1.48, and 1.58 (linear trend test p  =  0.18); those for testicular cancer were 1.0, 1.04, 1.91, and 3.17 (trend p = 0.04). Based on findings from this study and the literature, the C8 Science Panel concluded that “there is a probable link between PFOA and both testicular and kidney cancer” (C8 Science Panel, 2012). The Panel also concluded, “there is no probable link between PFOA and either thyroid cancer or melanoma, for which limited but insufficient evidence was found.” A working group of the IARC classified PFOA as “possibly carcinogenic to humans,” based primarily on these epidemiologic findings (IARC, 2016). The epidemiologic evaluation of the public health impact of widespread exposure to PFOA played a central role in responding to the health concerns of the community and in establishing possible human carcinogenicity, consistent with laboratory observations that it causes cancer of the testicles, liver, and pancreas in rodents (Kennedy, 2015).

Toms River, New Jersey Responding to residents of Toms River, New Jersey, who reported an unusual number of childhood cancers, the New Jersey Department of Health and Senior Services (NJDHSS), in collaboration with the U.S. Agency for Toxic Substances and Disease Registry (ATSDR) and community organizations, undertook several studies (Maslia et  al., 2005; NJ Department of Health, 2016). A 1997 analysis of cancer incidence in Dover Township, the location of Toms River, found a significant elevation in the incidence of all childhood cancers combined (90 observed, 67 expected) for the years 1979–​1995, based on the New Jersey State cancer registry (Berry, 1997). None

 321



Water Contaminants

321

of the other 32 townships in Ocean County had a significantly elevated rate. The risk of pediatric cancers was especially elevated in the Toms River census tracts. Overall, 24 cases were observed, with 14.1 expected. In females under age 5 years, 10 cases were observed, with 1.6 expected. The NJDHSS subsequently conducted two case-​control studies of leukemia and neurological cancers, one based on an extensive interview of parents and another based on birth certificates. These used 4 and 10 controls, respectively, per case, matched on age and gender. An important element of these studies involved detailed modeling by ATSDR of the water distribution system in order to generate month-​by-​month estimates of potential individual residential exposures during the time period 1962–​1996 for each of several public well water sources. Two sources were contaminated by chemicals from separate Superfund sites (Maslia et al., 2001). A former Ciba Chemical Corporation (later, Ciba-​Geigy) plant opened in 1952 and produced organic dyes and intermediate products, epoxy resins, and specialty chemicals. Until 1966, the plant discharged treated wastewater directly into the Toms River. After 1966, the wastewater was pumped via a 10-​ mile pipeline into the Atlantic Ocean. The pipeline experienced four major breaks over time, causing local chemical contamination. Tens of thousands of drums with solid and liquid chemical wastes were disposed on site, resulting in extensive soil and groundwater contamination. A down-​gradient well field of the local water supply (Holly Street) was contaminated with dyes, nitrobenzene, and other compounds during the mid-​1960s. The other Superfund site is the Reich Farm, where, for 4 months in 1971, over 4500 drums of toxic waste from a Union Carbide plant were dumped illegally. These leaked into the sandy subsoil, forming a plume that contaminated another public supply well field (Parkway) (Fagin, 2013). A modeling study of public drinking water reconstructed the monthly well-​field origins of water provided to each residence during the relevant time period (Maslia et al., 2000, 2001). However, the residential exposures were estimated after extensive remediation, when most chemical contaminants were no longer detectable, obviating the possibility of linking specific chemicals or mixtures with modeled exposures from individual well fields. This shortcoming in exposure assessment, and the relatively small numbers of cases, limited the utility of the case control studies. Air pollution from the Ciba-​Geigy plant was also modeled and used in case-​control data analyses. Parents of 40 children diagnosed with leukemia or nervous system cancer while residing in Dover Township in 1979–​1996 participated in the interview study, as did 159 matched controls. Birth mothers and fathers were interviewed by telephone, providing information on residence location, volume of water consumed, parents’ occupations, and several other factors. The focus of most statistical analyses was the relative level of consumption of water originating in the contaminated Holly Street or Parkway well fields. No consistent associations were observed with postnatal exposures (Fagliano et  al., 2003). Prenatal exposure to water from the Parkway well field in the period 1982–​ 1996 was associated with leukemia among female children of all ages (OR = 3.4; CI:1.1, 10.4). Findings among male children were essentially null. The study, albeit small, was well designed and was supported by an exposure assessment that was extensive and detailed, but that lacked important information on specific chemical contaminants. Cancer risk was not measured in adults in Toms River who consumed contaminated water. The complex story of the origin and findings of these studies conducted by NJDHSS and ATSDR and the historical, social, legal, and political context of water contamination in Toms River are described by Fagin (2013).

particles, metals and metalloids other than arsenic, and water hardness, the epidemiologic evidence at this time is limited and equivocal.

OTHER WATERBORNE EXPOSURES

FUTURE DIRECTIONS

Classes of waterborne carcinogens not described here are addressed elsewhere in the current or previous edition of this text (Cantor et al., 2006). Schistosomiasis is an established infectious cause of bladder cancer (Chapters 24 and 52); radionuclides such as radon and radium 226 and 228 are naturally occurring contaminants of drinking water (Chapter 13). For some other waterborne factors, such as asbestiform

Inorganic Arsenic

CLIMATE CHANGE The capacity of the global atmosphere to hold water increases by about 7% per 1oC warming, leading to increasing water vapor in the air and more intense precipitation events (Trenberth, 2011). By the year 2015, average annual global land surface temperature had increased by 0.90oC (1.62oF) above the average for the twentieth century (National Oceanic & Atmospheric Administration [NOAA], 2015). Climate change is associated with major perturbations in the hydrologic cycle, and therefore in the quantity and quality of water available for human consumption. With respect to its possible effects on carcinogenic water contaminants, research is currently focused in two areas: (1) increases in cyanobacterial growth patterns, and (2) elevated levels of disinfection byproducts. The first is occurring in response to elevated temperatures of freshwater bodies and the release of plant nutrients, especially nitrate and phosphate, from agricultural regions (Elliott, 2012; Heisler et al., 2008; Newcombe et al., 2012; O’Reilly et al., 2015; Paerl and Paul, 2012). Increasing levels of disinfection byproducts may result from the mobilization of precursor humic and fulvic acids from soils during more intense rainfall events (Delpla et al., 2016). The design of most existing water treatment plants predates concerns about climate change. Increasing attention is now directed on enhancing these facilities to cope with changes in water quantity and quality due to both high rainfall and, conversely, water shortages (Levine et al., 2016).

EFFECTS OF EXPOSURE MISCLASSIFICATION IN EPIDEMIOLOGIC STUDIES The degree to which risk estimates from epidemiologic studies of cancer reflect “true” risks depends on the accuracy of the exposure estimates during the relevant exposure period, commonly measured in decades prior to appearance of a tumor. Relatively small errors in estimating historical exposure can have profound effects on the observed risk. When a true risk is present and the misclassification of exposure in case-​control studies is non-​differential (similar among cases and non-​cases), the risk estimate is typically biased toward the null (Cantor and Lubin, 2007). For example, given a “true” relative risk of 2.0 (an actual 100% risk increase), the observed relative risk would be decreased to 1.5 (an observed 50% increase) if the correlation coefficient between the actual exposure and the estimated exposure from the study were 0.6 (Vineis, 2004). This level of exposure misclassification is common in studies of environmental factors and leads to special concern in low-​exposure settings, where expected excess risks are relatively small and the error in exposure estimates can lead to erroneously missing important risks. Careful consideration of this is warranted when evaluating the findings of epidemiologic studies of lower levels of arsenic in drinking water (< 100 ug/​l) and for all studies of disinfection byproducts and of nitrate in drinking water. A corollary is that when an association between cancer risk and a waterborne exposure is consistently observed among different populations and varying circumstances (e.g., bladder cancer and DBP), quantitative estimates of risk almost certainly underestimate the true risk. When planning future studies, careful assessment of statistical power issues, in the face of unavoidable errors in historical exposure assessment, must be conducted.

Inorganic arsenic is an established human carcinogen, with most evidence regarding ingested exposure from populations with prolonged consumption of drinking water at concentrations well over 100 µg/​l. In the observed range, the dose–​response curve appears to be linear. An

32

322

PART III:  THE CAUSES OF CANCER

unanswered question concerns the level of risk and shape of the dose–​ response curve when exposures are below 100 µg/​l. Addressing this issue is fraught with many methodologic challenges. The most feasible approach would be to evaluate risk in populations with exposure in the 50–​100 µg/​l range, if such groups could be identified. Other research questions concern the mechanism(s) of action and the role of genetic polymorphisms in arsenic metabolism and cancer risk. Better understanding of the mechanism(s) will likely shed light on the effects of low exposure levels. Elevated MMA in the urine of exposed individuals places them at elevated cancer risk and risk of other adverse health outcomes. The reasons for this observation are not well understood.

Disinfection Byproducts Considerable evidence implicates exposure to disinfection byproducts in the risk of bladder cancer, although the specific agents associated with this risk have yet to be determined. Currently, the data for other cancer sites are limited and sometimes inconsistent. Well-​designed studies of colon and rectal cancers, renal cell carcinoma, and brain cancer would be informative. It may not be possible to determine which of the compounds in the complex and varying mixtures that constitute DBPs are most strongly linked to risk. However, it is feasible to refine exposure assessment in future studies to specific types or classes of DBPs in order to guide mitigation efforts. To that end, epidemiologic studies should include lifetime residential histories that can then be linked to public water supply databases in order to estimate long-​term exposures. Of particular importance is the question of whether risk differs for brominated and chlorinated compounds. Future work should also investigate the degree of correlation that other chemical classes have with THMs or HAAs. Direct measurement and estimation of other classes would be valuable where possible. Another major knowledge gap is the extent to which inhalation and dermal routes of exposure are important. Exposure studies suggest that these routes could result in internal doses of the volatile DBPs much higher than ingestion. The two epidemiologic studies of bladder cancer that evaluated these routes of exposure are inconsistent. Further evaluation is warranted.

Nitrate Although the number of analytic epidemiologic studies with historical exposure data has increased since the early 1990s, there remain few studies of any single cancer site, making interpretation difficult. The recent analytic studies have generally included historical data of nitrate levels from public water supplies and have evaluated potential confounders and factors affecting nitrosation. However, users of private wells, which often have higher nitrate levels than public sources of drinking water, are usually excluded due to a lack of long-​term measurements of nitrate. Recent efforts to model nitrate levels in private wells for epidemiologic studies using geographic information systems (Aschebrook-​Kilfoy et al., 2012; Wheeler et al., 2015) will be important for understanding the relationship with nitrate ingestion over a broader range of intake. With the increase in use of nitrogen-​based fertilizers, nitrate levels in water supplies have been increasing in many areas worldwide; therefore, additional studies, preferably of populations with well-​characterized and higher exposures, are warranted.

Cyanotoxins Microcystin is a likely cofactor for elevated rates of HCC in southeastern coastal China. The expanding worldwide contamination of freshwater used for drinking with nitrate and phosphorous, combined with increased water temperatures, is leading to more blooms of cyanobacteria and consequent contamination with microcystin and other cyanotoxins. Further studies in China and other regions of the world are warranted and should include other possibly carcinogenic cyanotoxins such as nodularin and cylindrospermopsin.

References Adonis MI, Diaz J, Miranda VR, et al. 2014. Biomarkers for screening of lung cancer and pre-​neoplastic lesions in a high risk Chilean population. Biol Res, 47(1), 62. PMCID: PMC4335787. Aggazzotti G, Fantuzzi G, Tartoni PL, and Predieri G. 1990. Plasma chloroform concentrations in swimmers using indoor swimming pools. Arch Environ Health, 45(3), 175–​179. PMID: 2386423. Aggazzotti G, Fantuzzi G, Righi E, et  al. 1993. Chloroform in alveolar air of individuals attending indoor swimming pools. Arch Environ Health, 48(4), 250–​254. PMID: 8357275. Ahsan H, Chen Y, Parvez F et al. 2006. Arsenic exposure from drinking water and risk of premalignant skin lesions in Bangladesh:  baseline results from the Health Effects of Arsenic Longitudinal Study. Am J Epidemiol, 163(12), 1138–​1148. PMID: 16624965. Albuquerque Bernalillo County Water Utility Authority. 2010. Water Resources Education. Arsenic Removal Facility. Available from:  http://​www.abcwua.org/​education/​31_​Arsenic.html. Accessed June 12, 2015. Andrew AS, Mason RA, Kelsey KT, et al. 2009. DNA repair genotype interacts with arsenic exposure to increase bladder cancer risk. Toxicol Lett, 187(1), 10–​14. PMCID: PMC2680739. Antonelli R, Shao K, Thomas DJ, Sams R, 2nd, and Cowden J. 2014. AS3MT, GSTO, and PNP polymorphisms:  impact on arsenic methylation and implications for disease susceptibility. Environ Res, 132, 156–​ 167. PMID: 24792412. Aschebrook-​Kilfoy B, Heltshe SL, Nuckols JR, et  al. 2012. Modeled nitrate levels in well water supplies and prevalence of abnormal thyroid conditions among the Old Order Amish in Pennsylvania. Environ Health, 11, 6. PMCID: PMC3305600. Agency for Toxic Substances and Disease Registry (ATSDR). 2007. Toxicological Profile for Arsenic. Available from:  http://​www.atsdr.cdc. gov/​toxprofiles/​tp.asp?id=22&tid=3. Accessed February 11, 2016. ATSDR. 2016. Camp Lejeune Drinking Water Public Health Assessment, U.S. Marine Corps Base Camp Lejeune, North Carolina. Atlanta, GA: (ATSDR) Agency for Toxic Substances and Disease Registry. Baastrup R, Sorensen M, Balstrom T, et al. 2008. Arsenic in drinking-​water and risk for cancer in Denmark. Environ Health Perspect, 116(2), 231–​237. PMID: 18288323. Backer LC, Ashley DL, Bonin MA, et al. 2000. Household exposures to drinking water disinfection by-​products: whole blood trihalomethane levels. J Expo Anal Environ Epidemiol, 10(4), 321–​326. PMID: 10981726. Bailey K and Fry RC. 2014. Long-​term health consequences of prenatal arsenic exposure: links to the genome and the epigenome. Rev Environ Health, 29(1–​2), 9–​12. PMCID: PMC4049250. Barrett JH, Parslow RC, McKinney PA, Law GR, and Forman D. 1998. Nitrate in drinking water and the incidence of gastric, esophageal, and brain cancer in Yorkshire, England. Cancer Causes Control, 9(2), 153–​159. PMID: 9578292. Barry V, Winquist A, and Steenland K. 2013. Perfluorooctanoic acid (PFOA) exposures and incident cancers among adults living near a chemical plant. Environ Health Perspect, 121(11–​12), 1313–​1318. PMCID: PMC3855514. Basu A, Mitra S, Chung J, et  al. 2011. Creatinine, diet, micronutrients, and arsenic methylation in West Bengal, India. Environ Health Perspect, 119(9), 1308–​1313. PMCID: PMC3230402. Bates M, Smith A, and Cantor K. 1995. Case-​control study of bladder cancer and arsenic in drinking water. Am J Epidemiol, 141(6), 523–​530. PMID: 7900719. Beane Freeman L, Cantor KP, Karagas M, et al. 2017. Exposure to disinfection by-​products and bladder cancer in New England. Environ Health Perspect. 125(6):067010. PMID: 28636529. Berry M. 1997. Childhood Cancer Incidence Health Consultation: A Review and Analysis of Cancer Registry Data, 1979–​1995 for Dover Township, Ocean County, New Jersey. Trenton, New Jersey: New Jersey Department of Health and Senior Services. Bouvard V, Loomis D, Guyton KZ, et al. 2015. Carcinogenicity of consumption of red and processed meat. Lancet Oncol, 16(16), 1599–​ 1600. PMID: 26514947. Bove FJ, Ruckart PZ, Maslia M, and Larson TC. 2014a. Evaluation of mortality among marines and navy personnel exposed to contaminated drinking water at USMC base Camp Lejeune: a retrospective cohort study. Environ Health, 13(1), 10. PMCID: PMC3943370. Bove FJ, Ruckart PZ, Maslia M, and Larson TC. 2014b. Mortality study of civilian employees exposed to contaminated drinking water at USMC Base Camp Lejeune: a retrospective cohort study. Environ Health, 13, 68. PMCID: PMC4237831.

 32



Water Contaminants

Bove GE, Jr, Rogerson PA, and Vena JE. 2007. Case control study of the geographic variability of exposure to disinfectant byproducts and risk for rectal cancer. Int J Health Geogr, 6, 18. PMCID: PMC1890278. British Geological Survey. Government of the People’s Republic of Bangladesh MoLG, Rural Development and Co-​operatives, Department of Public Health Engineering. 2001. Arsenic Contamination of Groundwater in Bangladesh. BGS Technical Report WC/​00/​19, Volume 1, Kinniburg D, and Smedley P (Eds.). Buchet JP and Lison D. 1998. Mortality by cancer in groups of the Belgian population with a moderately increased intake of arsenic. Int Arch Occup Environ Health, 71(2), 125–​130. PMID: 9553788. Buchet JP, Lauwerys R, and Roels H. 1981. Comparison of the urinary excretion of arsenic metabolites after a single oral dose of sodium arsenite, monomethylarsonate, or dimethylarsinate in man. Int Arch Occup Environ Health, 48(1), 71–​79. PMID: 6894292. Burch MD. 2008. Effective doses, guidelines and regulations. Adv Exp Med Biol, 619, 831–​853. PMID: 18461792. Butler N, Carlisle JC, Linville R, and Washburn B. 2009. Microcystins: A Brief Overview of Their Toxicity and Effects, with Special Reference to Fish, Wildlife, and Livestock. Sacramento, CA, USA: California EPA. C8 Science Panel. 2012. Probable Link Evalution of Cancer. Available from: http:// ​ w ww.c8sciencepanel.org/ ​ p dfs/ ​ P robable_ ​ L ink_ ​ C 8_ ​ C ancer_​ 16April2012_​v2.pdf. Cantor KP Hoover R, Hartge P, et  al. 1987. Bladder cancer, drinking water source, and tap water consumption: a case-​control study. J Natl Cancer Inst, 79(6), 1269–​1279. PMID: 3480378. Cantor KP. 1997. Drinking water and cancer. Cancer Causes Control, 8(3), 292–​308. PMID: 9498894. Cantor KP, Ward MH, Moore LE, and Lubin JH. 2006. Water contaminants. In:  Schottenfeld D, and Fraumeni JF, Jr. (Eds.), Cancer Epidemiology and Prevention (3rd ed., pp. 382–​404). New  York:  Oxford University Press. Cantor KP, and Lubin JH. 2007. Arsenic, internal cancers, and issues in inference from studies of low-​level exposures in human populations. Toxicol Appl Pharmacol, 222(3), 252–​257. PMCID: PMC2692340. Cantor KP, Villanueva CM, Silverman DT, et  al. 2010. Polymorphisms in GSTT1, GSTZ1, and CYP2E1, disinfection by-​ products, and risk of bladder cancer in Spain. Environ Health Perspect, 118(11), 1545–​1550. PMCID: PMC2974691. Caro J and Gallego M. 2007. Assessment of exposure of workers and swimmers to trihalomethanes in an indoor swimming pool. Environ Sci Technol, 41(13), 4793–​4798. PMID: 17695931. Caro J and Gallego M. 2008. Alveolar air and urine analyses as biomarkers of exposure to trihalomethanes in an indoor swimming pool. Environ Sci Technol, 42(13), 5002–​5007. PMID: 18678040. Centers for Disease Control and Prevention (CDC). 1998. A Survey of the Quality of Water Drawn from Domestic Wells in Nine Midwest States (p. 25). Atlanta, GA: CDC. Chappells H, Parker L, Fernandez CV, et al. 2014. Arsenic in private drinking water wells:  an assessment of jurisdictional regulations and guidelines for risk remediation in North America. J Water Health, 12(3), 372–​392. PMID: 25252340. Chen CJ, Kuo TL, and Wu MM. 1988. Arsenic and cancer [letter]. Lancet, 1, 414–​415. Chen CL, Chiou HY, Hsu LI, et  al. 2010a. Ingested arsenic, characteristics of well water consumption and risk of different histological types of lung cancer in northeastern Taiwan. Environ Res, 110(5), 455–​462. PMID: 19735913. Chen CL, Chiou HY, Hsu LI, et  al. 2010b. Arsenic in drinking water and risk of urinary tract cancer:  a follow-​ up study from northeastern Taiwan. Cancer Epidemiol Biomarkers Prev, 19(1), 101–​110. PMID: 20056628. Chen CL, Hsu LI, Chiou HY, et al. 2004. Ingested arsenic, cigarette smoking, and lung cancer risk:  a follow-​up study in arseniasis-​endemic areas in Taiwan. JAMA, 292(24), 2984–​2990. PMID: 15613666. Chen JG and Kensler TW. 2014. Changing rates for liver and lung cancers in Qidong, China. Chem Res Toxicol, 27(1), 3–​6. PMCID: PMC3946948. Chen K, Cai J, Liu XY, et  al. 2003. Nested case-​control study on the risk factors of colorectal cancer. World J Gastroenterol, 9(1), 99–​ 103. PMCID: PMC4728259. Chen Y, Hall M, Graziano JH, et  al. 2007. A prospective study of blood selenium levels and the risk of arsenic-​ related premalignant skin lesions. Cancer Epidemiol Biomarkers Prev, 16(2), 207–​213. PMCID: PMC1924921. Chilvers C, Inskip H, Caygill C, et al. 1984. A survey of dietary nitrate in well-​ water users. Int J Epidemiol, 13(3), 324–​331. PMID: 6490303.

323

Chiu WA, Jinot J, Scott CS, et al. 2013. Human health effects of trichloroethylene: key findings and scientific issues. Environ Health Perspect, 121(3), 303–​311. PMCID: PMC3621199. Chorus I and Bartram J (Eds.). 1999. Toxic Cyanobacteria in Water: A Guide to Their Public Health Consequences, Monitoring and Management. London: WHO. Chowdhury S, Alhooshani K, and Karanfil T. 2014. Disinfection byproducts in swimming pool: occurrences, implications and future needs. Water Res, 53, 68–​109. PMID: 24509344. Cocco P, Broccia G, Aru G, et al. 2003. Nitrate in community water supplies and incidence of non-​Hodgkin’s lymphoma in Sardinia, Italy. J Epidemiol Community Health, 57(7), 510–​511. PMCID: PMC1732520. Cohen SM, Arnold LL, Beck BD, Lewis AS, and Eldan M. 2013. Evaluation of the carcinogenicity of inorganic arsenic. Crit Rev Toxicol, 43(9), 711–​ 752. PMID: 24040994. Coombs CC, Tavakkoli M, and Tallman MS. 2015. Acute promyelocytic leukemia: where did we start, where are we now, and the future. Blood Cancer J, 5, e304. PMCID: PMC4450325. Costet N, Villanueva CM, Jaakkola JJ, et  al. 2011. Water disinfection by-​ products and bladder cancer:  is there a European specificity? A  pooled and meta-​analysis of European case-​control studies. Occup Environ Med, 68(5), 379–​385. PMID: 21389011. Cox PA and Sacks OW. 2002. Cycad neurotoxins, consumption of flying foxes, and ALS-​PDC disease in Guam. Neurology, 58(6), 956–​959. PMID: 11914415. Cutler D and Miller G. 2005. The role of public health improvements in health advances: the twentieth-​century United States. Demography, 42(1), 1–​22. PMID: 15782893. Cuzick J, Sasieni P, and Evans S. 1992. Ingested arsenic, keratoses, and bladder cancer. Am J Epidemiol, 136(4), 417–​421. PMID: 1415161. Dabeka RW, McKenzie AD, Lacroix GM, et al. 1993. Survey of arsenic in total diet food composites and estimation of the dietary intake of arsenic by Canadian adults and children. J AOAC Int, 76(1), 14–​25. PMID: 8448438. Dangleben NL, Skibola CF, and Smith MT. 2013. Arsenic immunotoxicity: a review. Environ Health, 12(1), 73. PMCID: PMC3848751. Dauphine DC, Smith AH, Yuan Y, et al. 2013. Case-​control study of arsenic in drinking water and lung cancer in California and Nevada. Int J Environ Res Public Health, 10(8), 3310–​3324. PMCID: PMC3774440. Davidson EA, David MB, Galloway JN, et al. 2012. Excess nitrogen in the U.S. environment: trends, risks, solutions. Issues Ecol, 15, 1–​16. Delpla I, Scheili A, Guilherme S, Cool G, and Rodriguez MJ. 2016. Variations of disinfection by-​product levels in small drinking water utilities according to climate change scenarios: a first assessment. J Water Clim Change, 7(1), 1–​15. Elliott JA. 2012. Is the future blue-​green? A review of the current model predictions of how climate change could affect pelagic freshwater cyanobacteria. Water Res, 46(5), 1364–​1371. PMID: 22244968. Espejo-​ Herrera N, Cantor KP, Malats N, et  al. 2015. Nitrate in drinking water and bladder cancer risk in Spain. Environ Res, 137, 299–​307. PMID: 25601732. Espejo-​Herrera N, Gracia-​Lavedan E, Boldo E, et al. 2016. Colorectal cancer risk and nitrate exposure through drinking water and diet. Int J Cancer, 139(2), 334–​346. PMID: 26954527. Espejo-​ Herrera N, Gracia-​ Lavedan E, Pollan M, et  al. 2016b. Ingested nitrate and breast cancer in the Spanish Multicase-​Control Study on Cancer (MCC-​Spain). Environ Health Perspect, 124(7), 1042–​ 1049. PMCID: PMC4937871. Fagin D. 2013. Toms River:  A  Story of Science and Salvation. New York: Bantam Books. Fagliano JA, Berry M, Kohler BA, Klotz JB, and Imtiaz R. 2003. Case-​Control Study of Childhood Cancers in Dover Township, Ocean County, New Jersey, Vol I–​Vol V. Trenton, New Jersey, USA: New Jersey Department of Health and Senior Services in cooperation with Agency for Toxic Substances and Disease Registry. Falconer IR and Buckley TH. 1989. Tumor promotion by microcystis sp a blue-​green-​alga occurring in water-​supplies. Med J Australia, 150(6), 351–​351. Ferreccio C, Gonzalez C, Milosavjlevic V, et al. 2000. Lung cancer and arsenic concentrations in drinking water in Chile. Epidemiology, 11(6), 673–​679. PMID: 11055628. Ferreccio C, Smith AH, Duran V, et al. 2013a. Case-​control study of arsenic in drinking water and kidney cancer in uniquely exposed Northern Chile. Am J Epidemiol, 178(5), 813–​818. PMCID: PMC3817454. Ferreccio C, Yuan Y, Calle J, et al. 2013b. Arsenic, tobacco smoke, and occupation:  associations of multiple agents with lung and bladder cancer. Epidemiology, 24(6), 898–​905. PMID: 24036609.

324

324

PART III:  THE CAUSES OF CANCER

Finn OJ. 2012. Immuno-​oncology:  understanding the function and dysfunction of the immune system in cancer. Ann Oncol, 23 Suppl 8, viii6–​9. PMCID: PMC4085883. Flanagan SV, Johnston RB, and Zheng Y. 2012. Arsenic in tube well water in Bangladesh:  health and economic impacts and implications for arsenic mitigation. Bull World Health Organ, 90(11), 839–​ 846. PMCID: PMC3506399. Fleming LE, Rivero C, Burns J, et al. 2002. Blue green algal (cyanobacterial) toxins, surface drinking water, and liver cancer in Florida. Harmful Algae, 1(2), 157–​168. Focazio MJ, Tipton D, Shapiro SD, and Geiger LH. 2006. The chemical quality of self-​supplied domestic well water in the United States. Ground Water Monit Remediat, 26(3), 92–​104. Frisbee SJ, Brooks AP, Jr., Maher A, et al. 2009. The C8 health project: design, methods, and participants. Environ Health Perspect, 117(12), 1873–​1882. PMCID: PMC2799461. Fujiki H and Suganuma M. 1994. Tumor necrosis factor-​alpha, a new tumor promoter, engendered by biochemical studies of okadaic acid. J Biochem, 115(1), 1–​5. PMID: 8188614. Fujiki H and Suganuma M. 2011. Tumor promoters: microcystin-​LR, nodularin and TNF-​alpha and human cancer development. Anticancer Agents Med Chem, 11(1), 4–​18. PMID: 21269254. Gagarin MF (Ed.). 2010. The Oxford Encyclopedia of Ancient Greece and Rome, Volume 1. Oxford: Oxford University Press. Gamble MV, Liu X, Ahsan H, et  al. 2006. Folate and arsenic metabolism: a double-​ blind, placebo-​ controlled folic acid-​ supplementation trial in Bangladesh. Am J Clin Nutr, 84(5), 1093–​1101. PMCID: PMC2046214. Garcia-​Esquinas E, Pollan M, Umans JG, et  al. 2013. Arsenic exposure and cancer mortality in a US-​based prospective cohort: the Strong Heart Study. Cancer Epidemiol Biomarkers Prev, 22(11), 1944–​1953. PMID: 23800676. George CM, Smith AH, Kalman DA, and Steinmaus CM. 2006. Reverse osmosis filter use and high arsenic levels in private well water. Arch Environ Occup Health, 61(4), 171–​175. PMCID: PMC3849398. Gibb H, Haver C, Gaylor D, et  al. 2011. Utility of recent studies to assess the National Research Council 2001 estimates of cancer risk from ingested arsenic. Environ Health Perspect, 119(3), 284–​290. PMCID: PMC3059988. Gilbert-​Diamond D, Cottingham KL, Gruber JF, et al. 2011. Rice consumption contributes to arsenic exposure in US women. Proc Natl Acad Sci U S A, 108(51), 20656–​20660. PMCID: PMC3251121. Gilbert-​Diamond D, Li Z, Perry AE, Spencer SK, Gandolfi AJ, and Karagas MR. 2013. A population-​based case-​control study of urinary arsenic species and squamous cell carcinoma in New Hampshire, USA. Environ Health Perspect, 121(10), 1154–​1160. PMID: 23872349. Griffiths DJ and Saker ML. 2003. The Palm Island mystery disease 20 years on: a review of research on the cyanotoxin cylindrospermopsin. Environ Toxicol, 18(2), 78–​93. PMID: 12635096. Guha Mazumder DN, Haque R, Ghosh N, et al. 1998. Arsenic levels in drinking water and the prevalence of skin lesions in West Bengal, India. Int J Epidemiol, 27(5), 871–​877. PMID: 9839746. Gulis G, Czompolyova M, and Cerhan JR. 2002. An ecologic study of nitrate in municipal drinking water and cancer incidence in Trnava District, Slovakia. Environ Res, 88(3), 182–​187. PMID: 12051796. Hall MN and Gamble MV. 2012. Nutritional manipulation of one-​ carbon metabolism: effects on arsenic methylation and toxicity. J Toxicol, 2012, 595307. PMCID: PMC3317163. Han YY, Weissfeld JL, Davis DL, and Talbott EO. 2009. Arsenic levels in ground water and cancer incidence in Idaho: an ecologic study. Int Arch Occup Environ Health, 82(7), 843–​849. PMID: 18941772. Harada K, Oshikata M, Uchida H, et al. 1996. Detection and identification of microcystins in the drinking water of Haimen City, China. Nat Toxins, 4(6), 277–​283. PMID: 9029553. Heck JE, Andrew AS, Onega T, et al. 2009. Lung cancer in a U.S. population with low to moderate arsenic exposure. Environ Health Perspect, 117(11), 1718–​1723. PMID: 20049123. Heisler J, Glibert PM, Burkholder JM, et al. 2008. Eutrophication and harmful algal blooms: a scientific consensus. Harmful Algae, 8(1), 3–​13. Hertz-​Picciotto I, Smith AH, Holtzman D, Lispett M, and Alexeef G. 1992. Synergism between occupational arsenic exposure and smoking in the induction of lung cancer. Epidemiology, 3(1), 23–​31. Hildesheim ME, Cantor KP, Lynch CF, et  al. 1998. Drinking water source and chlorination byproducts. II. Risk of colon and rectal cancers. Epidemiology, 9(1), 29–​35. PMID: 9430265. Hopenhayn-​Rich C, Smith AH, and Goeden HM. 1993. Human studies do not support the methylation threshold hypothesis for the toxicity of inorganic arsenic. Environ Res, 60(2), 161–​177.

Hough RL, Fletcher T, Leonardi GS, et al. 2010. Lifetime exposure to arsenic in residential drinking water in Central Europe. Int Arch Occup Environ Health, 83(5), 471–​481. PMID: 20401490. Howarth RW. 2008. Coastal nitrogen pollution: a review of sources and trends globally and regionally. Harmful Algae, 8(1), 14–​20. Hsu LI, Chiu AW, Huan SK, et  al. 2008. SNPs of GSTM1, T1, P1, epoxide hydrolase and DNA repair enzyme XRCC1 and risk of urinary transitional cell carcinoma in southwestern Taiwan. Toxicol Appl Pharmacol, 228(2), 144–​155. PMID: 18199464. Huang YK, Huang YL, Hsueh YM, et  al. 2008. Arsenic exposure, urinary arsenic speciation, and the incidence of urothelial carcinoma:  a twelve-​year follow-​up study. Cancer Causes Control, 19(8), 829–​839. PMID: 18351295. Hutson SS, Barber NL, Kenney JF, Linsey KS, Lumia DS, and Maupin MA. 2004. Estimated Use of Water in the United States in 2000. Washington, DC: US Geological Survey. International Agency for Research on Cancer (IARC). 1991. Chlorinated drinking-​ water; chlorination by-​products; some other halogenated compounds; cobalt and cobalt compounds. IARC Monogr Eval Carcinog Risks Hum, 52. IARC. 2004. Some drinking water disinfectants and contaminants, including arsenic. IARC Monogr Eval Carcinog Risks Hum, 84. IARC. 2010. Ingested nitrate and nitrite, and cyanobacterial peptide toxins. IARC Monogr Eval Carcinog Risks Hum, 94, 329–​412. IARC. 2012a. A review of human carcinogens:  arsenic, metals, fibres, and dusts. IARC Monogr Eval Carcinog Risks Hum, 100(Pt C). IARC. 2012b. Chemical agents and related occupations. IARC Monogr Eval Carcinog Risks Hum, 100(Pt F), 9–​562. IARC. 2013. Some chemicals in industrial and consumer products, some food contaminants and flavourings, and water chlorination by-​products. IARC Monogr Eval Carcinog Risks Hum, 101. IARC. 2014. Trichloroethylene, tetrachloroethylene, and some other chlorinated agents. IARC Monogr Eval Carcinog Risks Hum, 106. IARC. 2016. Perfluorooctanoic acid, tetrafluoroethylene, dichloromethane, 1,2-​ dichloropropane, and 1,3-​propane sultone. IARC Monogr Eval Carcinog Risks Hum, 110. Inoue-​Choi M, Jones RR, Anderson KE, et al. 2015. Nitrate and nitrite ingestion and risk of ovarian cancer among postmenopausal women in Iowa. Int J Cancer, 137(1), 173–​182. PMCID: PMC4405451. Inoue-​Choi M, Ward MH, Cerhan JR, et  al. 2012. Interaction of nitrate and folate on the risk of breast cancer among postmenopausal women. Nutr Cancer, 64(5), 685–​694. PMCID: PMC3403733. James KA, Meliker JR, Buttenfield BE, et al. 2014. Predicting arsenic concentrations in groundwater of San Luis Valley, Colorado:  implications for individual-​level lifetime exposure assessment. Environ Geochem Health, 36(4), 773–​782. PMID: 24429726. Jeong CH, Wagner ED, Siebert VR, et  al. 2012. Occurrence and toxicity of disinfection byproducts in European drinking waters in relation with the HIWATE epidemiology study. Environ Sci Technol, 46(21), 12120–​ 12128. PMCID: PMC4790091. Jochimsen EM, Carmichael WW, An JS, et al. 1998. Liver failure and death after exposure to microcystins at a hemodialysis center in Brazil. N Engl J Med, 338(13), 873–​878. PMID: 9516222. Johnson CJ and Kross BC. 1990. Continuing importance of nitrate contamination of groundwater and wells in rural areas. Am J Ind Med, 18(4), 449–​456. PMID: 2248249. Johnston RB and Sarker MH. 2007. Arsenic mitigation in Bangladesh: national screening data and case studies in three upazilas. J Environ Sci Health A Tox Hazard Subst Environ Eng, 42(12), 1889–​1896. PMID: 17952790. Jones RR, Weyer PJ, Dellavalle CT, et al. 2017. Ingested nitrate, disinfection by-​ products, and kidney cancer risk in older women. Epidemiology, [epub ahead of print]. PMID: 28252454. Jones RR, Weyer PJ, Dellavalle CT, et al. 2016. Nitrate from drinking water and diet and bladder cancer among postmenopausal women in Iowa. Environ Health Perspect, 124(11), 1751–​1758. PMID: 27258851. Kalman DA, Dills RL, Steinmaus C, et al. 2014. Occurrence of trivalent monomethyl arsenic and other urinary arsenic species in a highly exposed juvenile population in Bangladesh. J Expo Sci Environ Epidemiol, 24(2), 113–​120. PMID: 23549402. Karagas MR, Stukel TA, Morris JS, et al. 2001. Skin cancer risk in relation to toenail arsenic concentrations in a US population-​based case-​control study. Am J Epidemiol, 153(6), 559–​565. PMID: 11257063. Karagas MR, Tosteson TD, Morris JS, et al. 2004. Incidence of transitional cell carcinoma of the bladder and arsenic exposure in New Hampshire. Cancer Causes & Control, 15, 265–​272. PMID: 15286466. Kennedy GL and Symons JM, 2015. Carcinogenicity of perfluoroalkyl compounds. In: DeWitt J (Ed.), Toxicological Effects of Perfluoroalkyl and Polyfluoroalkyl Substances (pp. 265–​304). London: Springer-​Verlag.

 325



Water Contaminants

Khan K, Ahmed E, Factor-​Litvak P, et  al. 2015. Evaluation of an elementary school-​based educational intervention for reducing arsenic exposure in Bangladesh. Environ Health Perspect, 123(12), 1331–​ 1336. PMCID: PMC4671245. Kile ML and Ronnenberg AG. 2008. Can folate intake reduce arsenic toxicity? Nutr Rev, 66(6), 349–​353. PMCID: PMC4013823. Kile ML, Houseman EA, Baccarelli AA, et al. 2014. Effect of prenatal arsenic exposure on DNA methylation and leukocyte subpopulations in cord blood. Epigenetics, 9(5), 774–​782. PMCID: PMC4063836. King WD, Marrett LD, and Woolcott CG. 2000. Case-​control study of colon and rectal cancers and chlorination by-​products in treated water. Cancer Epidemiol Biomarkers Prev, 9(8), 813–​818. PMID: 10952098. Kitchin KT and Conolly R. 2010. Arsenic-​induced carcinogenesis-​-​oxidative stress as a possible mode of action and future research needs for more biologically based risk assessment. Chem Res Toxicol, 23(2), 327–​335. PMID: 20035570. Kleinjans JC, Albering HJ, Marx A, et al. 1991. Nitrate contamination of drinking water:  evaluation of genotoxic risk in human populations. Environ Health Perspect, 94, 189–​193. PMCID: PMC1567968. Kogevinas M, Villanueva CM, Font-​Ribera L, et al. 2010. Genotoxic effects in swimmers exposed to disinfection by-​products in indoor swimming pools. Environ Health Perspect, 118(11), 1531–​1537. PMCID: PMC2974689. Kosnett MJ. 2013. The role of chelation in the treatment of arsenic and mercury poisoning. J Med Toxicol, 9(4), 347–​354. PMCID: PMC3846971. Kurttio P, Pukkala E, Kahelin H, Auvinen A, and Pekkanen J. 1999. Arsenic concentrations in well water and risk of bladder and kidney cancer in Finland. Environ Health Perspect, 107(9), 705–​710. PMID: 10464069. Lamm S, Engel A, Kruse M, Feinleib M, Byrd D, Lai S, and Wilson R. 2004. Arsenic in drinking water and bladder cancer mortality in the US:  an analysis based on 133 U.S. counties and thirty years of observation. J Occ Environ Med, 48, 298–​306. PMID: 15091293. Law G, Parslow R, McKinney P, and Cartwright R. 1999. Non-​Hodgkin’s lymphoma and nitrate in drinking water:  a study in Yorkshire, United Kingdom. J Epidemiol Community Health, 53(6), 383–​384. PMCID: PMC1756892. Lee J, Jun MJ, Lee MH, et al. 2010. Production of various disinfection byproducts in indoor swimming pool waters treated with different disinfection methods. Int J Hyg Environ Health, 213(6), 465–​474. PMID: 20961810. Levine A, Yang Y, and Goodrich J. 2016. Enhancing climate adaptation capacity for drinking water treatement facilities. J Water Clim Change, 7, 485–​497. Lewis DR, Southwick JW, Ouellet-​Hellstrom R, Rench J, and Calderon RL. 1999. Drinking Water Arsenic in Utah: A Cohort Mortality Study. Environ Health Perspect, 107(5), 359–​365. PMID: 10210691. Lian M, Liu Y, Yu SZ, et al. 2006. Hepatitis B virus x gene and cyanobacterial toxins promote aflatoxin B-​1-​induced hepatotumorigenesis in mice. World J Gastroenterol, 12(19), 3065–​3072. Lijinsky W. 1986. The significance of N-​nitroso compounds as environmental carcinogens. Environ Carcinog Rev, 4(1), 1–​45. Ling B. 2000. Health impairments arising from drinking water polluted with domestic sewage and excreta in China. Schriftenr Ver Wasser Boden Lufthyg, 105, 43–​46. PMID: 10842792. Lourencetti C, Grimalt JO, Marco E, et al. 2012. Trihalomethanes in chlorine and bromine disinfected swimming pools:  air-​water distributions and human exposure. Environ Int, 45, 59–​67. PMID: 22572118. Marshall G, Ferreccio C, Yuan Y, et al. 2007. Fifty-​year study of lung and bladder cancer mortality in Chile related to arsenic in drinking water. J Natl Cancer Inst, 99(12), 920–​928. PMID: 17565158. Maslia ML, Reyes JJ, Gillig RE, et al. 2005. Public health partnerships addressing childhood cancer investigations:  case study of Toms River, Dover Township, New Jersey, USA. Int J Hyg Environ Health, 208(1–​2), 45–​54. PMID: 15881978. Maslia ML, Sautner JB, Aral MM, et al. 2000. Using water-​distribution system modeling to assist epidemiologic investigations. J Water Res Pl-​ASCE, 126(4), 180–​198. Maslia ML, Sautner JB, Faye RE, et al. 2007. Analysis of Groundwater Flow, Contaminant Fate and Transport, and Distribution of Drinking Water at Tarawa Terrace and Vicinity, U.S. Marine Corps Base Camp Lejeune, North Carolina: Historical Reconstruction and Present-​Day Conditions—​ Executive Summary. Available from:  http://​www.atsdr.cdc.gov/​sites/​ lejeune/​tarawaterrace.html. Maslia ML, Sautner JB, and Mustafa MA. 2001. Summary of Findings: Historical Reconstruction of the Water-​ Distribution System Serving the Dover Township Area, New Jersey. Atlanta, GA: Agency for Toxic Substances and Disease Registry (ATSDR). Maslia ML, Suarez-​Soto RJ, Sautner JB, et al. 2013. Analyses and Historical Reconstruction of Groundwater Flow, Contaminant Fate and Transport,

325

and Distribution of Drinking Water Within the Service Areas of the Hadnot Point and Holcomb Boulevard Water Treatment Plants and Vicinities, U.S. Marine Corps Base Camp Lejeune, North Carolina, Chapter A: Summary and Findings. Available from: http://​www.atsdr.cdc. gov/​sites/​lejeune/​hadnotpoint.html. McElroy JA, Trentham-​Dietz A, Gangnon RE, et  al. 2008. Nitrogen-​nitrate exposure from drinking water and colorectal cancer risk for rural women in Wisconsin, USA. J Water Health, 6(3), 399–​409. PMID: 19108561. Melak D, Ferreccio C, Kalman D, et al. 2014. Arsenic methylation and lung and bladder cancer in a case-​control study in northern Chile. Toxicol Appl Pharmacol, 274(2), 225–​231. PMCID: PMC4344188. Meliker JR, Avruskin GA, Slotnick MJ, et al. 2008. Validity of spatial models of arsenic concentrations in private well water. Environ Res, 106(1), 42–​ 50. PMCID: PMC2271042. Meliker JR, Franzblau A, Slotnick MJ, and Nriagu JO. 2006. Major contributors to inorganic arsenic intake in southeastern Michigan. Int J Hyg Environ Health, 209(5), 399–​411. PMID: 16731038. Meliker JR, Slotnick MJ, Avruskin GA, et al. 2007. Individual lifetime exposure to inorganic arsenic using a space-​time information system. Int Arch Occup Environ Health, 80(3), 184–​197. PMID: 16897097. Meliker JR, Slotnick MJ, AvRuskin GA, et al. 2010. Lifetime exposure to arsenic in drinking water and bladder cancer: a population-​based case-​ control study in Michigan, USA. Cancer Causes Control, 21(5), 745–​ 757. PMID: 20084543. Melkonian S, Argos M, Pierce BL, et al. 2011. A prospective study of the synergistic effects of arsenic exposure and smoking, sun exposure, fertilizer use, and pesticide use on risk of premalignant skin lesions in Bangladeshi men. Am J Epidemiol, 173(2), 183–​191. PMCID: PMC3011951. Melkonian S, Argos M, Chen Y, et  al. 2012. Intakes of several nutrients are associated with incidence of arsenic-​ related keratotic skin lesions in Bangladesh. J Nutr, 142(12), 2128–​2134. PMCID: PMC3497963. Miller MD, Marty AM, Arcus A, et al. 2001. Differences between children and adults: implications in risk assessment at California EPA. Int J Toxicol, 21, 403–​418. Min JY and Min KB. 2016. Blood trihalomethane levels and the risk of total cancer mortality in US adults. Environ Pollut, 212, 90–​96. PMID: 26840521. Mirvish SS. 1995. Role of N-​nitroso compounds (NOC) and N-​nitrosation in etiology of gastric, esophageal, nasopharyngeal and bladder cancer and contribution to cancer of known exposures to NOC. Cancer Lett, 93(1), 17–​48. PMID: 7600541. Mirvish SS, Grandjean AC, Møller H, et al. 1992. N-​nitrosoproline excretion by rural Nebraskans drinking water of varied nitrate content. Cancer Epidemiol Biomarkers Prev, 1(6), 455–​461. PMID: 1302557. Møller H, Landt J, Pedersen E, et al. 1989. Endogenous nitrosation in relation to nitrate exposure from drinking water and diet in a Danish rural population. Cancer Res, 49(11), 3117–​3121. PMID: 2720669. Mueller BA, Searles Nielsen S, Preston-​Martin S, et al. 2004. Household water source and the risk of childhood brain tumours: results of the SEARCH International Brain Tumor Study. Int J Epidemiol, 33(6), 1209–​1216. PMID: 15567873. National Oceanic & Atmospheric Administration (NOAA). 2015. Global Highlights:  Calendar Year 2015. Available from:  http://​www.ncdc.noaa. gov/​sotc/​summary-​info/​global/​201512. Accessed September 23, 2016. National Toxicology Program. 2015. Report on Carcinogens: Monograph on Trichloroethylene. Research Triangle Park, NC: US Department of Health and Human Services. Nesnow S, Roop BC, Lambert G, et al. 2002. DNA damage induced by methylated trivalent arsenicals is mediated by reactive oxygen species. Chem Res Toxicol, 15(12), 1627–​1634. PMID: 12482246. Newcombe G, Chorus I, Falconer I, and Lin TF. 2012. Cyanobacteria: impacts of climate change on occurrence, toxicity and water quality management. Water Res, 46(5), 1347–​1348. PMID: 22265251. New Jersey Department of Health. 2016. Toms River Township Childhood Cancer Investigation. Available from:  http://​www.nj.gov/​health/​eohs/​ dovertwp.shtml -​ Case-​control%20studies. Accessed April 18, 2016. Nishiwaki-​Matsushima R, Ohta T, Nishiwaki S, et al. 1992. Liver tumor promotion by the cyanobacterial cyclic peptide toxin microcystin-​LR. J Cancer Res Clin Oncol, 118(6), 420–​424. PMID: 1618889. Nolan BT and Stoner JD, 2000. Nutrients in groundwaters of the conterminous United States, 1992–​1995. Environ Sci Technol, 34, 1156–​1165. National Research Council (NRC). 1981. The Health Effects of Nitrate, Nitrite, and N-​Nitroso Compounds. Washington DC: Committee on Nitrite and Alternative Curing Agents in Foods, National Research Council. NRC. 1999. Arsenic in Drinking Water (pp. 229–​ 250). Washington, DC:  Subcommittee on Arsenic in Drinking Water, National Research Council.

326

326

PART III:  THE CAUSES OF CANCER

NRC. 2001. Arsenic in Drinking Water 2001 Update. Washington, DC: Subcommittee to Update the 1999 Arsenic in Drinking Water Report, National Research Council. NRC. 2014. Critical Aspects of EPA’s IRIS Assessment of Inorganic Arsenic:  Interim Report. National Research Council. Available from: http://​www.nap.edu/​catalog.php?record_​id=18594. Accessed June 2, 2014. Nuckols JR, Freeman LE, Lubin JH, et al. 2011. Estimating water supply arsenic levels in the New England Bladder Cancer Study. Environ Health Perspect, 119(9), 1279–​1285. PMCID: PMC3230387. O’Reilly CM, Sharma S, Gray DK, et  al. 2015. Rapid and highly variable warming of lake surface waters around the globe. Geophysical Research Letters, 42(24), 10,773–​10,781. Olson JM. 2006. Photosynthesis in the Archean era. Photosynth Res, 88(2), 109–​117. PMID: 16453059. Orloff K, Mistry K, and Metcalf S. 2009. Biomonitoring for environmental exposures to arsenic. J Toxicol Environ Health B Crit Rev, 12(7), 509–​ 524. PMID: 20183531. Paerl HW, Hall NS, and Calandrino ES. 2011. Controlling harmful cyanobacterial blooms in a world experiencing anthropogenic and climatic-​induced change. Sci Total Environ, 409(10), 1739–​1745. PMID: 21345482. Paerl HW and Paul VJ. 2012. Climate change: links to global expansion of harmful cyanobacteria. Water Res, 46(5), 1349–​1363. PMID: 21893330. Peters BA, Hall MN, Liu X, et al. 2015. Folic acid and creatine as therapeutic approaches to lower blood arsenic: a randomized controlled trial. Environ Health Perspect, 123(12), 1294–​1301. PMCID: PMC4671237. Pilsner JR, Hall MN, Liu X, et  al. 2011. Associations of plasma selenium with arsenic and genomic methylation of leukocyte DNA in Bangladesh. Environ Health Perspect, 119(1), 113–​118. PMCID: PMC3018489. Rahman MA and Hasegawa H. 2011. High levels of inorganic arsenic in rice in areas where arsenic-​contaminated water is used for irrigation and cooking. Sci Total Environ, 409(22), 4645–​4655. PMID: 21899878. Rahman MB, Cowie C, Driscoll T, et al. 2014. Colon and rectal cancer incidence and water trihalomethane concentrations in New South Wales, Australia. BMC Cancer, 14, 445. PMCID: PMC4088985. Ravenscroft P. 2007. Predicting the Global Distribution of Natural Arsenic Contamination of Groundwater. Symposium on Arsenic: The Geography of a Global Problem. Royal Geographical Society, London. Available from:  http://​www.geog.cam.ac.uk/​research/​projects/​arsenic/​symposium/​ S1.2_​P_​Ravenscroft.pdf. Accessed March 16, 2012. Ren X, McHale CM, Skibola CF, et  al. 2011. An emerging role for epigenetic dysregulation in arsenic toxicity and carcinogenesis. Environ Health Perspect, 119(1), 11–​19. PMCID: PMC3018488. Rice JM, Rehm S, Donovan PJ, and Perantoni AO. 1989. Comparative transplacental carcinogenesis by directly acting and metabolism-​dependent alkylating agents in rodents and nonhuman primates. IARC Sci Publ, 96, 17–​34. Richardson SD, Plewa MJ, Wagner ED, Schoeny R, and Demarini DM. 2007. Occurrence, genotoxicity, and carcinogenicity of regulated and emerging disinfection by-​products in drinking water: a review and roadmap for research. Mutat Res, 636(1–​3), 178–​242. PMID: 17980649. Rook JJ. 1974. Formation of haloforms during chlorination of natural waters. Water Treat Exam, 23, 234–​243. Rowland IR, Granli T, Bockman OC, Key PE, and Massey RC. 1991. Endogenous N-​nitrosation in man assessed by measurement of apparent total N-​nitroso compounds in faeces. Carcinogenesis, 12(8), 1395–​1401. PMID: 1860160. Ruckart PZ, Bove FJ, and Maslia M. 2013. Evaluation of exposure to contaminated drinking water and specific birth defects and childhood cancers at Marine Corps Base Camp Lejeune, North Carolina: a case-​control study. Environ Health, 12, 104. PMCID: PMC3880212. Ruckart PZ, Bove FJ, Shanley E, 3rd, and Maslia M. 2015. Evaluation of contaminated drinking water and male breast cancer at Marine Corps Base Camp Lejeune, North Carolina: a case control study. Environ Health, 14, 74. PMCID: PMC4571057. Salas LA, Bustamante M, Gonzalez JR, et al. 2015. DNA methylation levels and long-​term trihalomethane exposure in drinking water: an epigenome-​wide association study. Epigenetics, 10(7), 650–​661. PMCID: PMC4622514. Salas LA, Cantor KP, Tardon A, et al. 2013. Biological and statistical approaches for modeling exposure to specific trihalomethanes and bladder cancer risk. Am J Epidemiol, 178(4), 652–​660. PMCID: PMC3736753. Salas LA, Villanueva CM, Tajuddin SM, et  al. 2014. LINE-​1 methylation in granulocyte DNA and trihalomethane exposure is associated with bladder cancer risk. Epigenetics, 9(11), 1532–​1539. PMCID: PMC4622716. Sancha A, O’Ryan R, and Perez O. 2000. The removal of arsenic from drinking water and associated costs:  the Chilean case. Interdisciplinary

Perspectives on Drinking Water Risk Assessment and Management. Proceedings of the Santiago (Chile) Symposium, September 1998. IAHS Pub. no. 260. Schopf JW. 2000. Solution to Darwin’s dilemma:  discovery of the missing Precambrian record of life. Proc Natl Acad Sci U S A, 97(13), 6947–​6953. PMCID: PMC34368. Schwimmer M and Schwimmer D. 1955. The Role of Algae and Plankton in Medicine. New York: Grune & Stratton. Shen ZC, Li WG, and Chen JG. 1985. An epidemiological study of drinking water and PLC in Qidong (in Chinese). Jiangsu Med J, 11, 28–​30. Shin HM, Vieira V, Ryan PB, et  al. 2011a. Retrospective exposure estimation for perfluorooctanoic acid (PFOA) for participants in the C8 Health Project. Epidemiology, 22(1), S83–​S83. Shin HM, Vieira VM, Ryan PB, et  al. 2011b. Environmental fate and transport modeling for perfluorooctanoic acid emitted from the Washington Works Facility in West Virginia. Environ Sci Technol, 45(4), 1435–​1442. PMID: 21226527. Shin HM, Vieira VM, Ryan PB, Steenland K, and Bartell SM. 2013. Retrospective exposure estimation and predicted versus observed serum perfluorooctanoic acid concentrations for participants in the C8 Health Project. Environ Health Perspect, 199(4), 1760. Slotnick MJ, Meliker JR, and Nriagu JO. 2006. Effects of time and point-​of-​use devices on arsenic levels in Southeastern Michigan drinking water, USA. Sci Total Environ, 369(1–​3), 42–​50. PMID: 16750243. Smith AH, Lingas EO, and Rahman M. 2000. Contamination of drinking-​water by arsenic in Bangladesh: a public health emergency. Bull World Health Organ, 78(9), 1093–​1103. PMCID: PMC2560840. Smith AH, Marshall G, Liaw J, et al. 2012. Mortality in young adults following in utero and childhood exposure to arsenic in drinking water. Environ Health Perspect, 120(11), 1527–​1531. PMCID: PMC3556614. Smith AH, Marshall G, Yuan Y, et al. 2014. Rapid reduction in breast cancer mortality with inorganic arsenic in drinking water. EBioMedicine, 1(1), 58–​63. PMCID: PMC4286879. Smith AH and Steinmaus CM. 2009. Health effects of arsenic and chromium in drinking water: recent human findings. Annu Rev Public Health, 30, 107–​122. PMCID: PMC2762382. Snow J. 1855. On the Mode of Communication of Cholera, 2nd edition. London: Churchill. Steenland K, Jin C, MacNeil J, et al. 2009. Predictors of PFOA levels in a community surrounding a chemical plant. Environ Health Perspect, 117(7), 1083–​1088. PMCID: PMC2717134. Steenland K and Woskie S. 2012. Cohort mortality study of workers exposed to perfluorooctanoic acid. Am J Epidemiol, 176(10), 909–​917. PMID: 23079607. Steinmaus C, Castriota F, Ferreccio C, et al. 2015. Obesity and excess weight in early adulthood and high risks of arsenic-​related cancer in later life. Environ Res, 142, 594–​601. PMCID: PMC4664040. Steinmaus C, Ferreccio C, Acevedo J, et  al. 2014. Increased lung and bladder cancer incidence in adults after in utero and early-​ life arsenic exposure. Cancer Epidemiol Biomarkers Prev, 23(8), 1529–​ 1538. PMCID: PMC4344186. Steinmaus CM, Ferreccio C, Romo JA, et al. 2013. Drinking water arsenic in northern chile: high cancer risks 40 years after exposure cessation. Cancer Epidemiol Biomarkers Prev, 22(4), 623–​630. PMCID: PMC3848421. Styblo M, Del Razo LM, Vega L, et al. 2000. Comparative toxicity of trivalent and pentavalent inorganic and methylated arsenicals in rat and human cells. Arch Toxicol, 74(6), 289–​299. PMID: 11005674. Su CC, Lu JL, Tsai KY, and Lian Ie B. 2011. Reduction in arsenic intake from water has different impacts on lung cancer and bladder cancer in an arseniasis endemic area in Taiwan. Cancer Causes Control, 22(1), 101–​108. PMID: 21052815. Su D. 1979. Drinking water and liver cell cancer: an epidemiologic approach to the etiology of this disease in China. Chin Med J, 92, 748–​756. Sueoka E, Sueoka N, Okabe S, et al. 1997. Expression of the tumor necrosis factor alpha gene and early response genes by nodularin, a liver tumor promoter, in primary cultured rat hepatocytes. J Cancer Res Clin Oncol, 123(8), 413–​419. PMID: 9292703. Sun RC, Board PG, and Blackburn AC. 2011. Targeting metabolism with arsenic trioxide and dichloroacetate in breast cancer cells. Mol Cancer, 10, 142. PMCID: PMC3240126. Sun Y, Tokar EJ, and Waalkes MP. 2012. Overabundance of putative cancer stem cells in human skin keratinocyte cells malignantly transformed by arsenic. Toxicol Sci, 125(1), 20–​29. PMCID: PMC3243751. Svircev Z, Drobac D, Tokodi N, et  al. 2013. Epidemiology of primary liver cancer in Serbia and possible connection with cyanobacterial blooms. J Environ Sci Health C Environ Carcinog Ecotoxicol Rev, 31(3), 181–​200. PMID: 24024518.

 327



Water Contaminants

Tao SS and Bolger PM. 1999. Dietary arsenic intakes in the United States: FDA Total Diet Study, September 1991–​December 1996. Food Addit Contam, 16(11), 465–​472. PMID: 10755138. Thomson B, Fox K, and O’Grady M. 2000. Arsenic removal by point-​of-​use treatment systems at San Ysidro, NM. Paper presented at the American Water Works Annual Conference, Denver, CO. Tokar EJ, Benbrahim-​Tallaa L, Ward JM, et al. 2010a. Cancer in experimental animals exposed to arsenic and arsenic compounds. Crit Rev Toxicol, 40(10), 912–​927. PMCID: PMC3076186. Tokar EJ, Diwan BA, and Waalkes MP. 2010b. Arsenic exposure transforms human epithelial stem/​progenitor cells into a cancer stem-​like phenotype. Environ Health Perspect, 118(1), 108–​115. PMCID: PMC2831952. Tokar EJ, Person RJ, Sun Y, Perantoni AO, and Waalkes MP. 2013. Chronic exposure of renal stem cells to inorganic arsenic induces a cancer phenotype. Chem Res Toxicol, 26(1), 96–​105. PMCID: PMC3921970. Tokar EJ, Qu W, and Waalkes MP. 2011. Arsenic, stem cells, and the developmental basis of adult cancer. Toxicol Sci, 120 Suppl 1, S192–​S203. PMCID: PMC3043086. Trenberth KE. 2011. Changes in precipitation with climate change. Climate Res, 47(1-​2), 123–​138. Tricker AR. 1997. N-​nitroso compounds and man: sources of exposure, endogenous formation and occurrence in body fluids. Eur J Cancer Prev, 6(3), 226–​268. PMID: 9306073. US Environmental Protection Agency (EPA). 2000a. Arsenic Occurrence in Public Drinking Water Supplies. Washington, DC: US Environmental Protection Agency. EPA-​815-​R-​00-​23. US EPA. 2000b. Technologies and Costs for Removal of Arsenic from Drinking Water. Washington, DC:  Office of Water, US Environmental Protection Agency. Report No. 815R-​00-​028. US EPA. 2003. National Primary Drinking Water Regulations:  Stage 2 Disinfectants and Disinfection Byproducts Rule; National Primary and Secondary Drinking Water Regulations:  Approval of Analytical Methods for Chemical Contaminants (Vol. 168). Washington, DC:  US Environmental Protection Agency. US EPA. 2005. Economic Analysis for the Final Stage 2 Disinfectant and Disinfectant Byproducts Rule. Washington, DC:  US Environmental Protection Agency. US EPA. 2006. National Primary Drinking Water Regulations:  Stage 2 Disinfectants and Disinfection Byproducts Rule, The Federal Register (Vol. 71, pp. 388). Washington, DC: US Environmental Protection Agency. US EPA. 2012. Fallon, NV: Pooling Resources to Construct Arsenic Treatment Facility. Available from:  http://​water.epa.gov/​lawsregs/​rulesregs/​sdwa/​ arsenic/​nv_​fallon.cfm. Accessed May 3, 2015. US EPA. 2015. Drinking Water Health Advisory for the Cyanobacterial Microcystin Toxins. Washington, DC:  US Environmental Protection Agency. EPA-​820R15100. US EPA. 2016. Regulated Drinking Water Contaminants:  Inorganic. Available from: https://​www.epa.gov/​ground-​water-​and-​drinking-​water/​table-​regulated-​ drinking-​water-​contaminants -​ Inorganic. Accessed September 23, 2016. US Food and Drug Administration (FDA). 2012. Arsenic in Rice: Full Analytical Results from Rice/​Rice Product Sampling, September 2012. Available from:  http://​www.fda.gov/​Food/​FoodborneIllnessContaminants/​Metals/​ ucm319916.htm. Accessed November 29, 2012. Ueno Y, Nagata S, Tsutsumi T, et al. 1996. Detection of microcystins, a blue-​ green algal hepatotoxin, in drinking water sampled in Haimen and Fusui, endemic areas of primary liver cancer in China, by highly sensitive immunoassay. Carcinogenesis, 17(6), 1317–​1321. PMID: 8681449. Vahter M and Marafante E. 1987. Effects of low dietary intake of methionine, choline or proteins on the biotransformation of arsenite in the rabbit. Toxicol Lett, 37(1), 41–​46. PMID: 3590229. Vahter M. 1999a. Variation in human metabolism of arsenic. In: Chappell WR, Abernathy CO, and Calderon RL (Eds.), Arsenic: Exposure and Health Effects (pp. 267–​279). Amsterdam: Elsevier. Vahter M. 1999b. Methylation of inorganic arsenic in different mammalian species and population groups. Sci Prog, 82 (Pt 1)(Pt 1), 69–​88. PMID: 10445007. Vahter M and Concha G. 2001. Role of metabolism in arsenic toxicity. Pharmacol Toxicol, 89(1), 1–​5. PMID: 11484904. Valenzuela OL, Drobna Z, Hernandez-​Castellanos E, et al. 2009. Association of AS3MT polymorphisms and the risk of premalignant arsenic skin lesions. Toxicol Appl Pharmacol, 239(2), 200–​207. PMCID: PMC2910909. van Leeuwen JA, Waltner-Toews D, Abernathy T, Smit B, Shoukri M, 1999. Associations between stomach cancer incidence and drinking water contamination with atrazine and nitrate in Ontario (Canada) agroecosystems, 1987–​1991. Int J Epidemiol 28, 836–​840.

327

van Maanen JM, Welle IJ, Hageman G, et al. 1996. Nitrate contamination of drinking water: relationship with HPRT variant frequency in lymphocyte DNA and urinary excretion of N-​nitrosamines. Environ Health Perspect, 104(5), 522–​528. PMCID: PMC1469364. Vermeer IT, Pachen DM, Dallinga JW, Kleinjans JC, and van Maanen JM. 1998. Volatile N-​nitrosamine formation after intake of nitrate at the ADI level in combination with an amine-​rich diet. Environ Health Perspect, 106(8), 459–​463. PMCID: PMC1533225. Vieira V, Hoffman K, and Fletcher T. 2013. Assessing the spatial distribution of perfluorooctanoic acid exposure via public drinking water pipes using geographic information systems. Environ Health Toxicol, 28, e2013009. PMCID: PMC3761072. Villanueva CM, Cantor KP, Cordier S, et  al. 2004. Disinfection byproducts and bladder cancer:  a pooled analysis. Epidemiology, 15(3), 357–​367. PMID: 15097021. Villanueva CM, Cantor KP, Grimalt JO, et  al. 2006. Assessment of lifetime exposure to trihalomethanes through different routes. Occup Environ Med, 63(4), 273–​277. PMCID: PMC2078091. Villanueva CM, Cantor KP, Grimalt JO, et  al. 2007. Bladder cancer and exposure to water disinfection by-​products through ingestion, bathing, showering, and swimming in pools. Am J Epidemiol, 165(2), 148–​156. PMID: 17079692. Villanueva CM, Fernandez F, Malats N, Grimalt JO, and Kogevinas M. 2003. Meta-​analysis of studies on individual consumption of chlorinated drinking water and bladder cancer. J Epidemiol Community Health, 57(3), 166–​173. PMCID: PMC1732410. Villanueva CM, Gracia-​Lavedan E, Bosetti C, et  al. 2017. Colorectal cancer and long-​term exposure to trihalomethanes in drinking water:  a multicenter case-​control study in Spain and Italy. Environ Health Perspect, 125(1), 56–​65. PMID: 27383820. Vineis P. 2004. A self-​fulfilling prophecy: are we underestimating the role of the environment in gene-​environment interaction research? Int J Epidemiol, 33(5), 945–​946. PMID: 15319401. Vitousek PM, Aber JD, Howarth RW, et al. 1997. Human alteration of the global nitrogen cycle: sources and consequences. Ecol Appl, 7(3), 737–​750. Walker M, Seiler RL, and Meinert M. 2008. Effectiveness of household reverse-​osmosis systems in a Western U.S. region with high arsenic in groundwater. Sci Total Environ, 389(2–​3), 245–​252. PMID: 17919687. Ward MH, Cerhan JR, Colt JS, and Hartge P. 2006. Risk of non-​Hodgkin lymphoma and nitrate and nitrite from drinking water and diet. Epidemiology, 17(4), 375–​382. PMID: 16699473. Ward MH, Heineman EF, Markin RS, and Weisenburger DD. 2008. Adenocarcinoma of the stomach and esophagus and drinking water and dietary sources of nitrate and nitrite. Int J Occup Environ Health, 14(3), 193–​197. PMID: 18686719. Ward MH, Kilfoy BA, Weyer PJ, et  al. 2010. Nitrate intake and the risk of thyroid cancer and thyroid disease. Epidemiology, 21(3), 389–​ 395. PMCID: PMC2879161. Ward MH, Mark SD, Cantor KP, et  al. 1996. Drinking water nitrate and the risk of non-​ Hodgkin’s lymphoma. Epidemiology, 7(5), 465–​ 471. PMID: 8862975. Ward MH, Rusiecki JA, Lynch CF, and Cantor KP. 2007. Nitrate in public water supplies and the risk of renal cell carcinoma. Cancer Causes Control, 18(10), 1141–​1151. PMID: 17717631. Ward MH, Sinha R, Heineman EF, et al. 1997. Risk of adenocarcinoma of the stomach and esophagus with meat cooking method and doneness preference. Int J Cancer, 71(1), 14–​19. PMID: 9096659. Weyer PJ, Cerhan JR, Kross BC, et al. 2001. Municipal drinking water nitrate level and cancer risk in older women: the Iowa Women’s Health Study. Epidemiology, 12(3), 327–​338. PMID: 11338313. Wheeler DC, Nolan BT, Flory AR, DellaValle CT, and Ward MH. 2015. Modeling groundwater nitrate concentrations in private wells in Iowa. Sci Total Environ, 536, 481–​488. PMID: 26232757. Winquist A, Lally C, Shin HM, and Steenland K. 2013. Design, methods, and population for a study of PFOA health effects among highly exposed mid-​ Ohio Valley community residents and workers. Environ Health Perspect, 121(8), 893–​899. PMCID: PMC3734501. World Health Organization (WHO). 2011. WHO Guidelines for Drinking-​ Water Quality, 4th edition. Available from:  http://​www.who.int/​water_​ sanitation_​health/​publications/​2011/​dwq_​guidelines/​en. Wright K, Kelly J, and Zegarra, AV. 1997. Macchu Picchu: ancient hydraulic engineering. J Hydraul Eng, 123, 838–​843. Xia J, Li Y, Yang Q, et  al. 2012. Arsenic trioxide inhibits cell growth and induces apoptosis through inactivation of notch signaling pathway in breast cancer. Int J Mol Sci, 13(8), 9627–​9641. PMCID: PMC3431819.

328

328

PART III:  THE CAUSES OF CANCER

Yu S, Zhao N, and Zi X. 2001. The relationship between cyanotoxin (microcystin, MC) in pond-​ditch water and primary liver cancer in China. Zhonghua Zhongliu Zazhi, 23(2), 96–​99. PMID: 11783030. Yu SZ. 1989. Drinking water and primary liver cancer. In: Tang ZY, Wu MC, and Xia SS (Eds.), Primary Liver Cancer (pp. 30–​37). Beijing:  China Academic Publishers. Yu SZ and Chen G. 1994. Blue-​green algae toxins and liver cancer. Chinese J Cancer Res, 6(1), 9–​17. Zeegers MP, Selen RF, Kleinjans JC, Goldbohm RA, and van den Brandt PA. 2006. Nitrate intake does not influence bladder cancer risk:  the Netherlands cohort study. Environ Health Perspect, 114(10), 1527–​1531. PMCID: PMC1626405. Zegura B, Straser A, and Filipic M. 2011. Genotoxicity and potential carcinogenicity of cyanobacterial toxins: a review. Mutat Res, 727(1–​2), 16–​41. PMID: 21277993.

Zhang MD. 1993. [A population-​based case control study of primary liver cancer in Fusui]. Zhonghua Liu Xing Bing Xue Za Zhi, 14(1), 14–​18. PMID: 8504447. Zhao N, Yu S, and Sun WM. 1994. [Interaction among the relative risk factors of primary liver cancer in a case-​control study]. Zhonghua Liu Xing Bing Xue Za Zhi, 15(2), 90–​93. PMID: 7923340. Zhou L, Yu H, and Chen K. 2002. Relationship between microcystin in drinking water and colorectal cancer. Biomed Environ Sci, 15(2), 166–​171. PMID: 12244757. Zwolak I and Zaporowska H. 2012. Selenium interactions and toxicity: a review. Selenium interactions and toxicity. Cell Biol Toxicol, 28(1), 31–​ 46. PMID: 21913064.

 329

19

Diet and Nutrition MARJORIE L. MCCULLOUGH AND WALTER C. WILLETT

OVERVIEW The formal study of diet, nutrition, and cancer is still relatively young, with most epidemiological studies and randomized controlled trials (RCTs) having occurred in the past 20–​30 years. Despite the methodological challenges of studying diet and cancer in free-​living populations, there is scientific consensus that overweight and obesity increase the risk of certain cancers, as well as growing evidence that dietary patterns rich in vegetables, fruits, and whole grains and low in red and processed meat are associated with lower risk of colorectal cancer and total cancer mortality. Although it is more difficult to isolate the specific components of diet that affect risk, several key factors appear to play a role. Dietary composition appears to operate by affecting energy intake and also independent of energy intake. Despite extensive research, evidence does not support an important impact of the macronutrient composition of diet on cancer risk. Individual nutrients and phytochemicals, including folate, vitamin D, calcium, and carotenoids (such as lycopene), appear to be associated with lower risk of certain cancers and are under active investigation. However, thus far, RCTs do not support chemoprevention with high-​dose antioxidant supplements, and in some cases these may cause harm. The assessment of lifelong diet and timing of exposures continues to be challenging. Future directions in diet and cancer research include the examination of early life exposures in relation to carcinogenicity, the role of the microbiome in cancer etiology, and risk factors for cancer subtypes defined by the molecular characteristics of tumors.

INTRODUCTION The relationship between nutritional factors and human cancer has been debated for decades, but it is only in the past 20–​30 years that large epidemiologic studies and RCTs have begun to examine the issue systematically. Dietary factors are believed to contribute to the large geographic and temporal variations in cancer rates observed among human populations and to changes in rates among migrants who move to countries with higher or lower rates than the country of origin. A series of ecological reports in the 1970s showed high correlations between specific dietary factors and cancer rates (e.g., fat and breast cancer, meat and colon cancer) across countries, and stimulated hypotheses on diet and cancer (Armstrong and Doll, 1975). Initially, case-​control studies supported a role of these and other dietary factors in cancer etiology (WCRF/​AICR, 1997). Over time, studies of diet and cancer became progressively larger, and prospective investigations, described herein, refuted some but not all of these associations (WCRF/​AICR, 2007). Meta-​analyses and pooling of individual-​level data have also improved the precision and reproducibility of risk estimates. Current and future prospective studies using repeated measures of diet, biological markers, and tumor molecular phenotype will continue to advance our understanding of the link between diet and cancer etiology. This chapter focuses on research regarding dietary composition in relation to the most common forms of cancer, the most thoroughly studied exposures, and issues of current interest. Because of the broad nature of the topic, we will only briefly mention related topics such as energy balance, obesity and body composition (Chapter 20), physical inactivity and sedentary behavior (Chapter 21), and alcohol

consumption (Chapter 12). Associations with diet are also covered in cancer-​specific chapters, where applicable. The goal of this chapter is to provide a “wide-​angle” perspective on these interrelated factors that affect cancer.

Other Authoritative Reviews The World Cancer Research Fund (WCRF) and its American affiliate, the American Institute for Cancer Research (AICR), have a valuable series of comprehensive systematic literature reviews of diet, physical activity, obesity, and alcohol and cancer. These have broader coverage of nutritional issues than the reviews by the International Agency for Research on Cancer (IARC) and are updated more frequently. The most recent complete WCRF/​AICR report, published in 2007 (WCRF/​ AICR, 2007), synthesized the evidence for 20 cancers and based its conclusions largely on meta-​analyses. The WCRF/​AICR “Continuous Update Project (CUP)” follows a staggered schedule for updating the evidence for each cancer site and more recently for cancer survivors. An expert panel reviews and judges the evidence according to criteria for inferring causality, classifying relationships as “convincing,” “probable,” “limited-​suggestive,” “limited-​no conclusion,” and “substantial effect on risk unlikely.” Throughout this chapter, we will refer to the WCRF/​AICR reports, in addition to studies of nutritional biomarkers and dietary patterns (not covered in the reports), and to other high quality meta-​analyses and individual studies as appropriate.

Estimates of Attributable Fraction Estimates of the proportion of cancers that can be attributed to dietary factors varies depending on the definition of “diet,” population analyzed (e.g., national, global, individual studies), methodology employed, and period of study. Diet is a complex entity that can be defined narrowly to include only the composition and volume of food consumed, or more broadly to encompass related factors such as energy balance, body weight, distribution of body fat, and metabolic byproducts of digestion. In 1981, Doll and Peto estimated that 35% of cancer deaths in the United States in the late 1970s could be attributed to diet, with a range of acceptable estimates of 10%–​70% (Doll and Peto, 1981). This figure was calculated by summing the estimated fraction of deaths from specific cancer sites that might be avoided by practical dietary means (for example, stomach 90%, female breast 50%, bladder 20%, other types 10%) (Doll and Peto, 1981). The Doll and Peto analysis did not provide separate estimates of the attributable fractions for body weight and physical inactivity, although the text stated that “[o]‌vernutrition should perhaps come first rather than last on a list of aspects of diet which may affect the incidence of cancer, even though the relevant mechanisms remain obscure.” In 2015, the WCRF/​AICR proposed new estimates, based on systematic reviews of the updated literature, and regional data on the prevalence of known risk factors for various types of cancer. The latest WCRF/​AICR update estimated that 20 to 22% of all incident cancers in the United States and United Kingdom were due to the combination of diet, physical inactivity, and overweight/​obesity; this figure rose to 29% when the analysis was restricted to the 13 most common cancers (WCRF International, 2015). The estimated fractions of all cancers preventable by diet in China and Brazil were about 15%, where obesity

329

30

330

PART III:  The Causes of Cancer

rates are lower (estimates rose to 19 and 22% for the 13 most common cancers, respectively). The estimates for the United States were lower than those of Doll and Peto (1981), probably because WCRF/​AICR considered only specific dietary factors for which the evidence for causality was judged to be “convincing” or “probable.” Other estimates have attempted to separate the effects of dietary composition from related factors such as overweight/​obesity, physical inactivity, and alcohol consumption. Colditz and Wei attributed 5.0% of incident cancers in the United States specifically to dietary composition, largely because of high consumption of red and processed meat and low consumption of folate (Colditz and Wei, 2012). Parkin proposed similar attributable fraction estimates for the United Kingdom, although consideration was also given to low consumption of fruit, vegetables, and fiber and high intake of salt as well as meat (Parkin, 2011). In 2003, the IARC estimated attributable fractions for low fruit and vegetable intake alone at 5%–​12% (IARC, 2003). Given that many dietary hypotheses remain unresolved and that the long-​ term impact of early life exposures is largely unexplored, estimates of the percentage of cancers attributable to dietary composition alone may continue to change beyond the current estimates of 5%–​12%.

Measures of Dietary Intake Measurement of diet in free-​living populations is a comprehensive field of study with specific methodological considerations, and readers are referred elsewhere for a detailed description (Willett WC, 2013). Most epidemiologic studies of cancer risk in humans use Food Frequency Questionnaires (FFQs) to estimate diet, because these are designed to assess usual intake efficiently for time periods of up to 1  year, thereby avoiding the day-​to-​day fluctuations that affect 24-​hour recalls. FFQs consist of a food list and questions on frequency of consumption, and some include portion sizes (Willett WC, 2013). FFQs can be simple or complex, short or long, and are population specific, reflecting commonly consumed foods and constituents of interest in a study population. Their use has been shown to be sufficiently valid for assessing intake of foods and nutrients, and identifying important diet–​disease relationships in large population studies (Block et al., 1990; Hu et al., 1999; Rimm et al., 1992; Willett et al., 1985). FFQs compare well with biochemical indicators of intake and more detailed questionnaire assessments (Willett WC, 2013). Weighed food records or 24-​hour dietary recalls reflect only very recent intake. However, if collected repeatedly over a year, they could approximate usual intake, and have the advantage of measuring exactly what participants eat and drink (assuming an adequately detailed nutrient database). Whereas serial 24-​hour recalls were prohibitively expensive in the past due to interviewer and training costs, these methods may prove feasible with technological advancements (Kirkpatrick et  al., 2014), including applications to smartphones and tablets. Whether large populations will provide data on a sufficient number of days to provide meaningful assessment of usual diet remains to be demonstrated. Combining these methods with an FFQ may lead to improved precision in dietary exposure assessment.

Biomarkers Biochemical indicators of diet can be useful in some situations, but for many dietary factors of interest, such as total fat, fiber, and sucrose, no useful indicators exist. Recently, studies of the food metabolome have identified potential biomarkers of foods such as red meat (3-​ methylhistidine) (Cross et al., 2011) and whole grains (alkylresorcinols) (Kyro et al., 2014a, 2014b; Scalbert et al., 2014). To be useful in epidemiological studies, the within-​person variation of biomarkers will need to be relatively low (Van Dam and Hunter, 2013). Other biomarkers of interest include DNA specimens that permit the testing of nutritional hypotheses using Mendelian randomization. This approach examines variation in genes of known function to study the causal effect of a related exposure on cancer risk in non-​experimental studies (Gray and Wheatley, 1991). Studies of gene–​diet interactions also hold promise, but these require very large numbers, and few clear examples exist at present.

Inherent Challenges in Measuring Dietary Intake Diet is a complex exposure composed of hundreds of nutrients and non-​nutrient components, some known, some yet to be fully characterized. Dietary behaviors tend to cluster, complicating efforts to isolate the effects of individual foods. Nutrients and foods are consumed in combinations that may enhance or block their absorption, metabolism, or physiologic effects. For example, tomatoes consumed as part of a highly processed diet with large amounts of refined carbohydrate may have a different effect on cancer risk than tomatoes consumed as part of a Mediterranean diet alongside fish, vegetables, nuts, and olive oil. Food processing (e.g., milling, fermentation), storage (e.g., salting, drying, freezing, bottling) and preparation methods (e.g., steaming, grilling) can affect nutrient concentration in a food, and can add carcinogens or unknown substances. The timing of a nutritional exposure may modify its effects on cancer risk, although this is seldom recognized in advance. Vitamin supplements are used by over half the US population (Dickinson et  al., 2014), and some vitamin formulations provide concentrated, high doses that may have different physiologic effects than when derived from food. In addition, individuals eat varied diets formed by individual preference, cultural models, intolerance, and influenced by lifestyle and availability. The complexity of diet and the challenge in quantifying intake in free-​living populations are aptly illustrated in a photojournalistic series on the typical diet in the week of a family in various parts of the world (Menzel, 2005).

Dietary Patterns One method of capturing multiple correlated aspects of diet is to characterize intake in terms of dietary patterns instead of single foods or nutrients. This is illustrated by contrasts between the “Mediterranean diet” and the “Western diet,” as discussed in Chapter  20, and is expanded on later in this chapter.

Study Designs The study designs most commonly used to examine the relationship of diet and cancer risk in humans are summarized in Figure 19–​1 (Harris et  al., 2009). Each design contributes valuable information, but all have strengths and weaknesses. Ecological studies and migrant studies play an important role in generating hypotheses about diet and cancer, including interest in the 1970s in the fat–​breast cancer and meat– colon cancer hypotheses (Armstrong and Doll, 1975). Major changes in disease rates within a population provide key evidence that exposures acquired during life have an important effect on cancer risk in populations. Through the 1990s, most information on diet and cancer was obtained from case-​control studies. Subsequently, concern about recall bias (the diagnosis influencing self-​reported diet) and selection bias (overly health-​conscious controls being more likely to participate) has limited their use in studies of diet and cancer. Indeed, many of the findings from early case-​control studies have not been confirmed in subsequent research. Public health guidelines for diet and nutrition tend to place the most weight on RCTs and prospective cohort studies, which have complementary strengths and weaknesses, and for which the exposure precedes cancer diagnosis. Although RCTs can reduce or eliminate confounding, they are expensive and are not always feasible or ethical. For example, an RCT of processed meat in relation to incident colorectal cancer is both infeasible and unethical, analogous to prohibitions against RCTs of smoking or excess weight gain and cancer risk. Systematic literature reviews and meta-​analyses can also contribute importantly to understanding diet–​cancer relationships, provided that the inclusion or exclusion of individual studies is adequately explained and justified and the definitions of dietary and other exposures are appropriately harmonized across studies. Publication bias is a serious concern in meta-​analyses of published studies. Large pooled analyses of individual-​level data from consortia of prospective studies have advantages over meta-​analyses, because publication bias is minimized, original data are harmonized, and a uniform approach

 31



Diet and Nutrition Strength of Evidence Randomized trials of disease outcomes

Prospective cohorts of disease outcomes

Randomized trials of physiologic measures

Retrospective case-control studies of disease outcomes

Ecologic studies

Prevalence studies

Animal studies

Case series/Case reports

Figure  19–​1.  Hierarchy of quality and strength of evidence of different study designs for evaluating causation of chronic disease. Studies of physiological endpoints may not fully account for all effects of an intervention; thus evidence is strongest when derived from well-​conducted studies of disease endpoints. Well-​controlled prospective observational studies and well-​performed randomized clinical trials have different, and highly complementary, strengths and limitations. The major strength of randomized trials is to minimize residual confounding; limitations include potential lower generalizability to external populations, the inability to evaluate the effects of many risk factors of interest (e.g., smoking, biomarker levels) due to ethical or practical limitations, inadequacy of duration and/​or dose tested, challenges in testing multiple doses, noncompliance, unblinding, differential loss to follow-​up, and treatment crossover. The major limitation of prospective observational studies is potential residual confounding, whereas major strengths generally include the converse of each of the major limitations of randomized trials. Thus, evidence is most robust when studies of both designs provide concordant results. Source:  Harris WS, et al. (2009).

can be used to define exposure, cut-​points, covariates, and statistical modeling (Smith-​Warner et al., 2006). Recognizing these methodologic considerations, the remainder of this chapter discusses elements of diet that have received particular attention with regard to cancer risk.

ENERGY BALANCE There is strong evidence that energy balance has important effects on the incidence of some cancers, as described in detail in Chapter 20. Experimental studies in animals during the first half of the twentieth century indicated that energy restriction profoundly reduced the development of mammary tumors (Tannenbaum, 1942; Tannenbaum and Silverstone, 1953). This finding has consistently been replicated in a wide variety of mammary and other tumor models (Birt et al., 1992; Nair et al., 1995; Ross and Bras, 1971; Weindruch and Walford, 1982). For example, 30% restriction in energy intake reduces mammary tumors by as much as 90% (Boissonneault et  al., 1986). In humans, energy balance is best assessed by determining body size and composition, or changes in weight or body composition, rather than by comparing energy intake to expenditure, since both parameters are measured with poor precision in free-​living humans (Willett WC, 2013). The IARC found sufficient evidence that excess body fat (starting at ≥25 kg/​m2) is associated with increased risk of cancers of the gastric cardia, colon and rectum, liver, gall bladder, pancreas, breast (post-​menopause), corpus uteri, ovary, kidney (renal cell), thyroid and multiple myeloma, meningioma, and esophageal

331

adenocarcinoma (IARC; Lauby-​Secretan, Scoccianti, Loomis, et al., 2016). The strongest empirical support for mechanisms explaining these associations involves the endocrine and metabolic effects of obesity. Hormonal effects primarily involve alterations of peptide (e.g., IGF, insulin) and steroid (estradiol, testosterone) hormones and their binding proteins (Calle and Kaaks, 2004). For certain cancers, such as esophageal adenocarcinoma and gall bladder tumors, chronic local inflammation from esophageal reflux or recurrent gallstones plays an important role. Adipose tissue also generates multiple pro-​ inflammatory mediators. The effects of dietary composition on long-​term weight change and adiposity are of great scientific interest and public health importance. In 2007, WCRF/​AICR characterized the evidence as “probable” that energy-​dense foods, sugary drinks, and fast foods increase risk (WCRF/​AICR, 2007). A meta-​analysis of 53 randomized clinical trials found that weight loss was slightly greater with low-​carbohydrate than with low-​fat diets (Tobias et al., 2015). Many aspects of diet consistent with a Mediterranean dietary pattern appear to contribute to long-​term weight maintenance (Mozaffarian et al., 2011). In a 6-​year trial of diet and weight loss with only 2 years of active intervention, those assigned to a Mediterranean diet had greater sustained weight loss compared with those on low-​fat or high-​fat diets (Shai et al., 2008).

MACRO-​AND MICRONUTRIENTS The relationship of diet to cancer risk can be evaluated in terms of either individual dietary constituents or broad dietary patterns. This section of the chapter discusses recent epidemiologic research on macro-​and micronutrients and their relationship (or lack of relationship) to cancer etiology, biology, and prevention.

Dietary Fat In the 1980s, dietary fat was postulated to be the most important aspect of diet with respect to cancer risk. This hypothesis was largely fueled by the striking international correlations between per capita fat intake and incidence or death rates from breast, colon, and prostate cancer (Armstrong and Doll, 1975; Prentice et al., 1988). For breast cancer, the positive associations seen in some case-​control studies (Howe et al., 1990; WCRF/​AICR, 1997) have not generally been replicated in prospective studies (Key et  al., 2011; Smith-​Warner et  al., 2001; WCRF/​AICR, 2010), although a recent large European study found a weak positive association of saturated fatty acids with ER+PR+ and HER2-​tumor subtypes (Sieri et  al., 2014). The lack of agreement between the case-​control and cohort studies reinforces concerns about recall and selection biases in case-​control studies of diet. The Women’s Health Initiative (WHI) study, designed and implemented at the end of the previous century (The Women’s Health Initiative Study Group, 1998), included a randomized clinical trial to test the effect of a low-​fat diet intervention (by instruction) on breast cancer incidence. The initial goal was to reduce fat intake among women in the intervention arm to 20% of calories, and compare this to women consuming approximately 37% of calories from fat. The actual contrast was smaller, however, because women in the intervention arm reported consuming 27% of calories from fat and the control arm averaged 33%. No differences were observed in plasma lipid fractions known to be influenced by dietary fat between the intervention and control groups, indicating that little or no difference in fat intake was achieved between the two groups (Willett, 2010). At trial completion, there was a non-​significantly lower risk of breast cancer in the intervention arm (Prentice RL et al., 2006). In an extended follow-​up after active intervention, there was no difference in rates of breast or other cancers between the two randomized groups (Thomson et al., 2014b). This inability to test the original hypothesis illustrates at least two serious limitations in using randomized trials to study dietary factors and cancer etiology. The first is the difficulty of achieving and sustaining the desired contrast in the exposure of interest. The second is that even a small change in the exposure (in this case dietary fat intake)

32

332

PART III:  The Causes of Cancer

can secondarily change other factors (in this case, weight loss of 1–​2 kg among women on the lower fat diet) (Prentice et al., 2006). Had there been a statistically significant reduction in the incidence rate of breast or other cancers, the intervention could not have distinguished weight loss from reductions in fat intake. In a second large randomized trial of dietary fat reduction, conducted among women with high mammographic density, a non-​significant increase in risk of breast cancer was seen among those randomized to a low-​fat diet (Martin et al., 2011). The debate over dietary fat intake and breast cancer incidence continues with respect to the timing of exposure, type of fat, and disease subtypes. Limited data suggest that animal fat intake during young adulthood may influence risk of premenopausal breast cancer (Farvid et  al., 2014b). High intake of polyunsaturated fatty acids increases mammary cancer incidence in rodents, but has not been associated with increased risk of breast cancer in humans (Smith-​Warner et al., 2001). The findings regarding total dietary fat intake and prostate cancer have been inconsistent. Although earlier studies suggested a positive relationship with animal fat (WCRF/​AICR, 1997), this association has weakened over time (WCRF/​AICR, 2014b). A  large international collaboration including over 5000 prostate cancer cases and > 6600 controls found no strong evidence that circulating specific fatty acids were important predictors of prostate cancer risk (Crowe et al., 2014). An unexpected modest positive association was observed with two omega-​3 fatty acids, eicosapentaenoic and docosapentaenoic acid. The significance of this finding is unclear, since it conflicts with meta-​analyses of studies of fish consumption, the major source of these fatty acids (Szymanski et  al., 2010), which is inversely associated with prostate cancer incidence (Lovegrove et  al., 2015; Szymanski et al., 2010). A limitation of this study is that it included all incident cases of prostate cancer, instead of restricting the endpoint to advanced or fatal cases. Since many cases are diagnosed by prostate specific antigen (PSA) screening, dietary factors associated with screening tend to be positively associated with overall incidence of prostate cancer.

Carbohydrates Carbohydrates comprise a diverse group of macronutrients that may differ in their effects on chronic disease risk. For example, carbohydrates with a high glycemic index (Wolever and Jenkins, 1986) cause larger spikes in postprandial blood glucose and insulin concentrations than those with a lower glycemic index (Miller, 1994). High glycemic carbohydrates also cause higher fasting insulin levels in insulin-​resistant states (Pereira et  al., 2002). Fasting plasma insulin concentrations are inversely correlated with insulin-​like growth factor-​binding protein 1 (IGFBP-​1) and thus increase bioactive IGF-​ 1 concentrations (Giovannucci, 2001). The glycemic index (GI) of a carbohydrate reflects its potential to raise blood sugar, whereas the glycemic load (GL) considers both the GI and the amount of carbohydrate consumed. Neither parameter was associated with the risk of colorectal or pancreatic cancer in a meta-​analysis (Mulholland et al., 2009). In contrast, the risk of endometrial cancer is consistently associated with increased GL. An increase of 50 units of GL is associated with a 15% higher risk (WCRF/​AICR, 2013). Endometrial cancer is also the cancer site most strongly associated with obesity and insulin resistance. It has not yet been determined whether subsets of people are at higher cancer risk from dietary GL. Grain consumption in the United States has increased by 41% over the past several decades (Wells and Buzby, 2008), concurrent with national dietary guidelines to reduce total dietary fat intake. Approximately 15% of total energy intake in the United States is from added sugars, with soda and energy drinks contributing 36% of added sugars (Welsh et  al., 2011). Consumption of sugar-​sweetened beverages correlates with increasing energy intake (Vartanian et  al., 2007), and contributes to weight gain and increasing prevalence of overweight and obesity (WCRF/​AICR, 2007). Therefore, excess sugar consumption, especially as beverages, likely increases risk of cancer indirectly through weight gain.

Protein Epidemiologic studies have not found a clear association between an overall high intake of protein in adulthood and risk of cancer. Most studies find no evidence of a harmful association with some foods that are major sources of protein, such as fish, poultry, and plant foods. Red and processed meat and dairy products are other major protein sources, discussed later in the chapter.

MICRONUTRIENTS Vitamin D Vitamin D is best known for its role in maintaining calcium homeostasis and bone health. It is found naturally in a limited number of foods (e.g., fatty fish and eggs), and is added to milk and other products in the United States and many other countries. It is also available from dietary supplements at a wide range of doses. The most important source of vitamin D, however, is cutaneous production due to UVB radiation. The recommended dietary allowance for vitamin D for adults is 600 IU/​day, with a tolerable upper limit of 4000 IU (Institute of Medicine, 2011). A cup of milk contains about 100 IU of vitamin D. In comparison, approximately 10,000–​20,000 IU is produced by a fair-​skinned adult in a bathing suit who develops a light pink sunburn from UVB exposure (Holick, 2004). Higher levels of sun exposure convert the skin precursors of vitamin D to inert metabolites, so that vitamin D toxicity from sun exposure is highly unlikely. The potential benefits of sun exposure for vitamin D production need to be balanced against the harmful effects of UV radiation for skin cancer, however (see Chapter 62.4). Seasonal variation occurs in the circulating storage form of vitamin D [25(OH)D], with a peak during summer months and a nadir in mid-​winter. This variation is accentuated in populations who live further from the equator (McCullough et al., 2010). In 1941, Frank Apperly observed lower cancer mortality rates in North American regions that have higher levels of solar radiation (Apperly, 1941), an observation later hypothesized to reflect variations in vitamin D exposure (Garland and Garland, 1980). In vitro and in vivo animal studies provide strong biological support for a role for vitamin D in cancer prevention. The active form of the vitamin, 1,25 dihydroxyvitamin D (1,25(OH)2D), also called calcitriol, is the natural ligand for the vitamin D receptor. This receptor is a nuclear transcription factor that modulates expression of over 200 genes, including some that regulate cell growth, limit inflammation, and reduce levels of VEGF (Feldman et al., 2014). The main storage form of vitamin D [25(OH)D] has a half-​life of 3–​4 weeks. Serum concentrations of 25(OH)D reflect input from diet, supplements, and skin synthesis and are considered the preferred biomarker of exposure. Prospective cohort studies with pre-​diagnostic measures of 25(OH)D provide the strongest support for an inverse relationship between vitamin D and cancer risk in humans. The most consistent evidence is for colorectal cancer, where the incidence rate is approximately one-​third lower in those with high versus low circulating levels of 25(OH)D (Lee et al., 2011a; Ma et al., 2011). The evidence for other cancers is mixed (Helzlsouer, 2010; Kim and Je, 2014; Stolzenberg-​Solomon et al., 2009; Wolpin et al., 2012). For prostate cancer the evidence is highly inconsistent (Ahn et al., 2008; Xu et al., 2014), although some studies suggest a “U-​shaped” association, with higher risks at either end of the 25(OH)D distribution (Kristal et al., 2014). Vitamin D metabolism in the prostate may vary by stage and grade of the tumor (Tannour-​Louet et  al., 2014), complicating the interpretation of the role of vitamin D in prostate cancer. Case-​control studies of breast cancer suggest lower risk with higher 25(OH)D concentrations measured after diagnosis (Yin et al., 2010), but the findings are generally weak or null in prospective studies (Kim and Je, 2014), suggesting possible disease-​related effects on vitamin D status. Determining optimal 25(OH)D levels for disease prevention is challenging because circulating concentrations vary considerably across populations, and assays differ in their quantification of 25(OH)D (Binkley et al., 2014). A large, ongoing, international consortium of 21

 3



Diet and Nutrition

prospective cohorts (Circulating Biomarkers of Breast and Colorectal Cancer Consortium) is addressing this issue by calibrating serum 25(OH)D levels to a single laboratory and using the same method for standardizing serum 25(OH)D concentrations to account for seasonal variation in analyses of prediagnostic vitamin D and breast and colorectal cancer (Ziegler, et al., 2015). Preliminary results show statistically significant inverse associations between circulating 25(OH)D (up to 100 nmol/​L) and colorectal (unpublished), but not breast cancer risk (Visvanathan et al., 2015). Results of RCTs of vitamin D supplementation and cancer outcomes have not definitively confirmed or negated a protective effect of vitamin D and cancer, due to questions about dose and crossover. The large Women’s Health Initiative found no reduction in the risk of colorectal (Wactawski-​Wende et  al., 2006)  or breast (Chlebowski et  al., 2008)  cancer among women randomized to 400 IU vitamin D plus 1000 mg calcium compared to the placebo group. However, 400 IU is now considered a low dose of vitamin D, and most women (including those on placebo) were taking vitamin D by the end of the study. Secondary analyses of osteoporosis trials reported some lowering of cancer risk, but the number of cases was generally too small to be conclusive (Avenell et al., 2012; Lappe et al., 2007). No association with recurrent polyps was observed in a recent trial of 1000 IU vitamin D and 1200 IU calcium (Baron et al., 2015). Ongoing trials of vitamin D supplementation and chronic disease risk in the United States (Manson et al., 2012) and elsewhere (Manson and Bassuk, 2015) are testing higher doses.

Calcium Higher calcium intake has been shown to reduce the incidence of bowel tumors in animals (Lamprecht and Lipkin, 2001), possibly by precipitating bile acids. Calcium supplementation reduced the recurrence of adenomatous polyps by about 20% in one randomized trial (Baron et al., 1999), but not in a recent trial of similar design (Baron et al., 2015). In observational studies, a pooled analysis of cohort studies found a statistically significant 20% lower risk of colorectal cancer in those with the highest versus lowest total calcium intake (Cho et al., 2004). Higher calcium intake has been associated with lower breast cancer risk in some prospective studies (Cui and Rohan, 2006), but not others (Larsson et al., 2009). The aforementioned randomized trial of 1000 mg calcium and 400 IU vitamin D found no reduction in either colorectal (Wactawski-​Wende et al., 2006) or postmenopausal breast cancer incidence (Chlebowski et  al., 2008)  over 7  years of the trial. It is possible that the generally high calcium intake in both trial arms limited the ability to detect a modest reduction in risk. Higher calcium intake (> 2000 mg/​day) from diet and nutritional supplements was associated with increased risk of total, lethal, and high-​grade prostate cancer in 24-​years of follow-​up of men in the Health Professionals Follow-​ up Study (Wilson et  al., 2015). Phosphorous intake was correlated with calcium and was also associated with increased risk. In addition, a meta-​analysis of 32 prospective studies found that dietary calcium, and its major source, dairy product intake, was significantly associated with higher total prostate cancer risk, but supplemental calcium intake was the only form significantly associated with increased risk of fatal prostate cancer (Aune et  al., 2015). Measured levels of total and ionized serum calcium were associated with higher risk of fatal prostate cancer, but only during the early years of follow-​up, suggesting that serum calcium might be a marker of extant prostate cancer (Schwartz and Skinner, 2012). Thus, while evidence supports a modest benefit of higher calcium intake in colorectal cancer prevention, the practical implications are presently unclear because high calcium intake or dairy product consumption has been associated with higher risk of prostate cancer (Kushi et al., 2012).

Folate Folate deficiency has long been known to cause tumors in animals, possibly by influencing gene expression through DNA methylation

333

or by increasing the misincorporation of uracil in DNA (Blount and Ames, 1994). Considerable evidence from both case-​ control and cohort studies supports a lower risk of colon cancer with higher folate intake (Kennedy et al., 2011), and some evidence suggests that higher folate intake may attenuate the risk of colorectal cancer associated with alcohol (Nan et al., 2013). An association between a functional polymorphism in the folic acid metabolizing gene, methylene tetrahydrofolate reductase (MTHFR), and incidence of colorectal cancer adds support for a causal relationship (Zhao et al., 2013). Results of studies of circulating folate and risk of colorectal cancer have been inconsistent, but blood levels of folate are influenced by genetic determinants of metabolism as well as intake. For example, in an analysis that included folate intake and MTHFR genotype, higher blood levels due to diet were associated with lower risk of colorectal cancer, but higher levels due to genotype were associated with higher risk, potentially due to blockage of a metabolic pathway that reduces risk (Lee et al., 2012). This illustrates some of the complexity of using biomarkers of dietary intake. In a randomized controlled trial among patients with a recent history of colorectal adenomas, folic acid supplementation increased risk of more advanced adenomas (Cole et  al., 2007), raising concerns that very high intakes of this vitamin may actually be carcinogenic. These conflicting observations have been postulated to reflect different stages of carcinogenesis, whereby high folate intake might inhibit the initiation of neoplasia but promote the growth and/​ or progression of existing lesions. Reassuringly, recent studies of folate intake continue to find an inverse association with colorectal cancer risk, even among older individuals with very high intakes of folate (Gibson et al., 2011; Stevens et al., 2011). In a detailed analysis of the timing of folate intake in relation to colorectal adenoma and cancer, greater folate intake was inversely associated with the risk of colorectal cancer when exposure occurred before the development of adenoma (Lee et al., 2011b). Lower risk of colorectal cancer was observed only after a 12-​year delay. Also reassuringly, mortality due to colorectal cancer continued to decline steadily after implementation of folic acid fortification in 1998 in the United States (http://​seer.cancer. gov/​statfacts/​html/​colorect.html). The role of folate in numerous other cancers has been studied, but the results are inconsistent and provide no clear evidence of an association.

Fiber Interest in the relation between fiber intake and colon cancer was stimulated by Burkitt’s observation that colon cancer was rare in areas of Africa where fiber consumption and stool bulk were high (Burkitt, 1971). Fiber has been hypothesized to dilute potential carcinogens and accelerate transit through the colon. The anti-​carcinogenic mechanisms of a diet high in fiber and low in fat were explored in a recent study of the microbiome. In a 2-​week food exchange experiment, O’Keefe et al. (2015) gave African American men a high-​fiber, low-​fat African-​ style diet and rural Africans a high-​fat, low-​fiber Western-​style diet. In comparison with their usual diets, the interventions resulted in reciprocal alterations in mucosal biomarkers of cancer risk and in aspects of the microbiota and metabolome related to cancer risk. In observational studies, dietary fiber was inversely associated with risk of colorectal cancer in early case-control studies (Howe et  al., 1992; Trock et  al., 1990), but not consistently in subsequent prospective studies. In a prospective US cohort of almost a half million retired men and women, no association was found for dietary fiber intake and colorectal cancer risk (Schatzkin et al., 2007). Similarly, in a large pooled analysis of over 700,000 men and women from multiple prospective cohort studies, no association was seen except a possible small increase in risk in those with very low fiber intake (Park et al., 2005). In contrast, a 2012 analysis from the large European Prospective Investigation into Cancer and Nutrition (EPIC) cohort found statistically significantly lower colorectal cancer risk with higher intake of dietary fiber, including fiber from fruit, vegetables, and cereals (Murphy et al., 2012). The most recent WCRF/​AICR update of the epidemiologic evidence for colon cancer ranks “foods containing dietary fibre” (see http://​www.wcrf. org/​sites/​default/​files/​Colorectal-​Cancer-​2011-​Report.pdf) to be

34

334

PART III:  The Causes of Cancer

convincingly related to lower risk (WCRF/​AICR, 2011). However, the inconsistency among large prospective studies remains unexplained, and is one of the few examples of important heterogeneity in findings among such studies. RCTs of supplemental wheat bran fiber (Alberts et  al., 2000) and isphaghula husk (psyllium fiber) (Bonithon-​Kopp et al., 2000) failed to reduce the risk of recurrent adenomatous polyps. A trial that combined instructions for both high fiber intake and a reduction in total fat also failed to reduce recurrent adenomatous polyps (Schatzkin et al., 2000). Higher intake of fiber has also been hypothesized to lower breast cancer risk by interfering with enterohepatic metabolism of estrogens (Gorbach and Goldin, 1987). Although most individual prospective cohort studies have not found a statistically significant association of breast cancer risk with greater dietary fiber intake, two meta-​analyses of 10 (Dong et  al., 2011)  and 15 (Aune et  al., 2012a) prospective cohort studies found a statistically significant 5%–​10% lower risk of breast cancer associated with each 10 g of increase in dietary fiber intake. In the large EPIC cohort, a weak inverse relation was seen with fiber from vegetable sources, but not with fiber from grains or fruits; this inverse association was primarily with estrogen receptor–​negative (ER–​) breast cancer (Ferrari et  al., 2013). Earlier life exposure to fiber may also impact breast cancer risk. In a recent study, higher fiber intake during high school was associated with a 16% lower risk of breast cancer overall, and a 24% lower risk of premenopausal breast cancer (Farvid et  al., 2016). Whether fiber or other components of high-​fiber foods, such as carotenoids, are responsible for this beneficial association is unclear.

Carotenoids Carotenoids, ubiquitous in fruits and vegetables, are fat-​soluble pigments that absorb light energy for plant photosynthesis and protect chlorophyll from damage. Over 700 naturally occurring carotenoids have been identified, and six (α-​carotene, β-​carotene, β-​cryptoxanthin, lycopene, and lutein+zeaxanthin) comprise greater than 95% of total carotenoids in human blood (Maiani et al., 2009). These compounds are hypothesized to reduce cancer risk by lowering oxidative stress and inflammation. Pro-​vitamin A carotenoids (α-​carotene, β-​carotene, and β-​cryptoxanthin) are precursors to vitamin A, which may influence carcinogenesis by regulating cell differentiation (Maiani et al., 2009). Preclinical and biomarker-​ based studies in the 1980s (Menkes et al., 1986; Stahelin et al., 1984) provided evidence for a protective relationship between carotenoid intake and risk of lung cancer after controlling for cigarette smoking. Inconsistencies were noted, however, in that these studies did not show a relationship with preformed vitamin A (Willett and Colditz, 2013), and a pooled analysis of seven prospective cohort studies provided little support for a protective role of β-​carotene intake and lung cancer (Mannisto, Smith-​Warner, Spiegelman, et  al., 2004). Counter to the hypothesis that β-​carotene might protect against lung cancer, two large trials in smokers and/​or asbestos workers, the Alpa-​Tocopherol, Beta-​Carotene (ATBC) (The Alpha-​ Tocopherol Beta-​ Carotene Cancer Prevention Study Group, 1994) and the CARET (Omenn et al., 1996) trials reported statistically significant increases in lung cancer occurrence among those randomized to 20–​30 mg β-​carotene per day. This is several times the amount typically present in the diet. However, in mostly non-​smokers, the Physicians’ Health Study (Hennekens et al., 1996) and the Women’s Health Study (Lee et al., 1999) observed no effect of β-​carotene supplements on lung or overall cancer incidence. Many explanations have been offered for the lack of benefit, or even harm, in the randomized trials, including interference with the metabolism of other beneficial carotenoids by the relatively high dose of β-​carotene in the supplements, or by induction of carcinogen-​activating enzymes in the highly oxidized lungs of smokers (Liu et al., 2003; Wang and Russell, 1999). The lack of benefit from β-​carotene supplements does not exclude the possibility that other nutrients or biologically active constituents in fruits and vegetables might prove to be protective in the future. However, epidemiologic evidence and randomized trials both indicate

no benefit of high β-​carotene intake for lung cancer, and current cancer prevention guidelines (Kushi et al., 2012) and the US Preventive Services Task Force (Moyer and Force, 2014) specifically recommend against taking β-​carotene supplements. Neither preformed vitamin A nor carotenoid intake has been consistently associated with risk of colon cancer, although these relationships have been examined in several studies (WCRF/​AICR, 2011). Similarly, β-​carotene—​as well as vitamins C and E—​did not affect the recurrence of adenomatous colon polyps (Greenberg et al., 1994). Recent pooled analyses of prospectively collected blood suggest that several carotenoids may reduce breast cancer risk (Bakker et al., 2016; Eliassen et  al., 2015). Both β-​carotene and α-​carotene serum levels were inversely associated with ER–​breast tumors. These findings are consistent with a pooled analysis in which vegetable consumption was associated inversely with ER–​but not ER+ breast cancer (Jung et al., 2013). The question of whether lycopene, or tomato products, its major food source, protect against prostate cancer risk has been studied extensively in epidemiologic and clinical studies. Lycopene is the most abundant carotenoid measured in prostate tissue (Arab et  al., 2001). Although the totality of the evidence for prostate cancer has not been convincing (WCRF/​AICR, 2014b), a recent large pooled analysis reported statistically significantly lower risk of aggressive prostate cancer with higher prediagnostic circulating lycopene levels (Key et al., 2015), adding important support to the possibility of benefit.

Vitamins C and E Extensive damage to DNA, protein, and lipids can occur from oxidant byproducts of smoking and normal metabolism. Although DNA repair mechanisms and antioxidant defenses exist, they are imperfect (Ames et  al., 1995). Antioxidants may reduce the risk of cancer by neutralizing free radicals and reactive oxygen species that can damage DNA. In humans, vitamin E is the major lipid-​soluble, membrane-​localized antioxidant, and vitamin C is the major water-​soluble antioxidant. Vitamin C can interfere with formation of nitrosamines—​carcinogens formed endogenously from nitrogenous precursors in diet and tobacco smoke—​in the stomach. However, chemoprevention trials of high-​risk populations have not provided strong support for a benefit from antioxidants against cancer. Chemoprevention trials of stomach cancer in high-​risk populations have shown regression of gastric dysplasia from several antioxidant nutrients (Correa et  al., 2000), but not a specific benefit from vitamin C supplements (Blot et  al., 1993). A  secondary analysis of the aforementioned ATBC trial found a 34% lower incidence of prostate cancer in heavy smokers who received supplemental α-​tocopherol (50 mg/​day), despite no effect on lung cancer (The Alpha-​ Tocopherol Beta-​Carotene Cancer Prevention Study Group, 1994). The Selenium and Vitamin E Cancer Prevention Trial (SELECT), initiated partly because of the ATBC trial results for vitamin E, found no benefit from selenium for prostate cancer, and an unexpected borderline adverse effect among men randomized to vitamin E alone (400 IU/​ day vitamin E of all rac α-​tocopheryl acetate) or in combination with selenium (200 µg/​d of L-​selenomethionine). This adverse effect was sustained and became statistically significant during post-​trial follow-​ up (Klein et al., 2011). The SELECT trial tested a higher dose of vitamin E than ATBC. However, an RCT using a similar vitamin E dose in mostly non-​smokers found no effect of the vitamin on prostate cancer (Gaziano et al., 2009). Overall, epidemiologic studies and RCTs have not supported a role of vitamin C or vitamin E in cancer prevention (Fortmann et al., 2013; WCRF/​AICR, 2007). Despite the lack of demonstrated benefit, any aspect of diet that acted by blocking the initial steps in carcinogenesis would need to be present at that time, presumably two or three decades earlier in the case of smoking and lung cancer. This is far longer than the duration of any of the intervention trials.

Selenium Selenium defends against oxidative stress through selenoproteins, including selenium-​dependent glutathione peroxidases. The amount

 35



Diet and Nutrition

of selenium in food may vary up to 10-​fold, depending on the selenium content of soil where the plant was grown or where the animal was raised. This variability makes food composition databases for selenium unreliable (Panel on Dietary Antioxidants and Related Compounds, 2000). Therefore, most epidemiologic evidence on selenium and cancer risk is from randomized trials and biomarkers of selenium exposure. For example, several prospective studies have reported inverse associations between selenium levels and prostate cancer incidence and progression using measures of selenium in toenails (Vogt et al., 2003; Yoshizawa et al., 1998), as a marker of selenium intake over the past year, or in the plasma or serum (Brooks et  al., 2001; Li et  al., 2004; Nomura et  al., 2000). Selenium was strongly associated with reduced prostate cancer risk as a secondary endpoint in one small trial of selenium supplementation and skin cancer (Clark et al., 1996). However, the SELECT trial (Klein et al., 2001), the aforementioned large randomized placebo controlled trial with selenium (200 µg) and vitamin E (400 IU) was ended 4  years early, with the study showing no protective effect of selenium on total prostate cancer incidence (Lippman et  al., 2009). This interpretation of this trial was limited because it included only one fatal case of prostate cancer. Recent research suggests that polymorphisms in genes that code for selenoproteins should be considered when examining the relationship between selenium biomarker concentrations and prostate cancer risk (Xie et al., 2016).

Vitamin and Mineral Combinations Few trials have tested combinations of multivitamins and/​or minerals in relation to cancer incidence. In a nutritionally depleted population in Linxian, China, a trial of over 25,000 men and women found that a cocktail of 50 μg selenium, 30 mg vitamin E, and 15 mg β-​carotene reduced total mortality, cancer-​specific mortality, and gastric cancer, compared to placebo (Blot et  al., 1993). Ten-​year post-​trial follow-​ up showed sustained benefits (Qiao et al., 2009). In contrast, as noted earlier, most vitamin and/​or mineral supplement trials in nutritionally replete populations have not shown benefit (Fortmann et  al., 2013), although benefits among subgroups with less than optimal diet cannot be excluded. Recently, two RCTs conducted in populations presumed to be nutritionally replete found reduced incidence of all cancers combined in individuals randomized to relatively low-​dose, multi-​nutrient supplements. In the Physicians’ Health Study (Gaziano et al., 2012), randomization of adult men to a Centrum Silver multivitamin, which contained 26 nutrients at the US RDA level (recommended level from diet for most US adults), was associated with a significant, 8% lower incidence of all cancers combined. A  similar study in US women is ongoing. The French SUVIMAX study of men and women (Hercberg et  al., 2004)  reported a statistically significant 31% lower risk of all cancers combined in men, but not women, who were randomized to an antioxidant mix (120 mg of ascorbic acid, 30 mg of vitamin E, 6 mg of β-​carotene, 100 μg of selenium, and 20 mg of zinc) versus placebo. Women in that trial had higher baseline plasma antioxidant status. Based on the preponderance of data on RCTs of high-​dose, individual supplements, several organizations recommend against taking antioxidant or other supplements for cancer prevention (Kushi et  al., 2012; WCRF/​ AICR, 2007), or prevention of chronic diseases (Fortmann et al., 2013). However, a possible modest benefit of lower dose supplements in cancer prevention deserves further study, especially in groups with suboptimal diets.

Sodium Foods preserved in salt or other sources of sodium are thought to contribute to a higher risk of gastric cancer in Asia and South America, where salting fish and/​or brining vegetables is common practice. High concentrations of salt irritate the gastric epithelium, especially in the context of Helicobacter pylori (H. pylori) infection (see Chapter 31). In the United States and other Western countries, dietary salt or sodium has not been associated with increased risk of stomach or other cancers. Because salt preservation is an issue in certain populations,

335

however, global cancer-​ prevention guidelines recommend limiting sodium intake (WCRF/​AICR, 2007).

Plant Polyphenols Polyphenols are secondary metabolites widely distributed in plant foods. There are four main classes:  flavonoids, lignans, phenolic acids and stilbenes. Polyphenols are of considerable interest in anticancer research because of their antioxidant, anti-​inflammatory, anti-​ estrogenic, and anticarcinogenic properties (Adolphe et  al., 2010). Advances in measurement and reporting of the polyphenol content of foods (Harnly et al., 2006; Rothwell et al., 2013) facilitate population studies of these compounds. Herein, we focus on evidence for lignans and flavonoids.

Lignans

Lignans, found primarily in flaxseed and other seeds, berries, tea, and wine, comprise one of three main groups of plant compounds classified as phytoestrogens, the other two being isoflavonoids and coumestans (Webb and McCullough, 2005). Studies in humans, animals, and cell culture provide support for a chemopreventive action of lignans, potentially through anti-​ estrogenic, anti-​ angiogenic, proapoptotic, and antioxidant mechanisms. The two primary lignans, matairesinol (MAT) and secoisolariciresinol (SEC), are converted by intestinal microflora to the biologically relevant mammalian lignans, enterodiol (EDL) and enterolactone (ENL). These latter two compounds, measured in blood and urine, reflect only very recent intake. Some studies support an inverse association of these biomarkers with risk of postmenopausal breast (Velentzis et al., 2009) and bladder (Zamora-​Ros et al., 2014) cancer. While evidence is inconsistent for dietary lignan intake and prostate cancer risk (Saarinen et al., 2010), a recent meta-​ analysis supports an inverse association with circulating enterolactone (He et al., 2015). The short half-​life of these circulating compounds and the large between-​individual variability in metabolism present challenges to research in this area.

Flavonoids

Flavonoids are bioactive, polyphenolic non-​nutrient constituents in plants. Common classes of flavonoids include anthocyanidins, flavonols, flavanones, flavones, flavanols, proanthocyanidins, and isoflavones, the latter being the most thoroughly studied. Isoflavones, found primarily in soy foods, have weak estrogenic activity and are therefore of interest in relation to hormone-​dependent cancers. Because isoflavones were shown to increase estrogenic markers in MCF-​7 cells, an ER+ breast cancer cell line, and to produce estrogenic effects in rodent reproductive tissues (Messina and Wood, 2008), there has been concern among medical professionals and the general public that eating soy food could fuel breast cancer growth. However, soy is metabolized differently in humans than it is in mice and rats, so findings in rodents may not apply to people (Setchell et al., 2011). A meta-​analysis of soy feeding studies in humans does not support effects of soy consumption on circulating estrogen (Fritz et al., 2013). Another meta-​analysis of 14 prospective studies showed that in Asian women, those who ate the most (compared to the least) soy isoflavones had a 24% lower risk of incident breast cancer, while there was no association in Western countries such as the United States (Dong and Qin, 2011). However, consumption levels in US women would typically correspond to levels in the reference group from studies in Asian countries. A systematic review concluded that soy intake equivalent to 2–​3 servings daily (25–​50 mg/​d, similar to the traditional Japanese diet) may reduce breast cancer risk (Fritz et al., 2013). It is also important to consider differences in lifelong soy exposure in Asian and Western countries when comparing associations between soy and breast cancer across these populations. In Chinese women, consistently higher soy consumption during adolescence and adulthood was associated with 60% lower risk of pre-​menopausal breast cancer, but not with postmenopausal breast cancer (Lee SA et al., 2009). Isoflavone exposure in utero may likewise be important (Nagata, 2010). For prostate cancer, dietary intake and serum biomarkers of daidzein and

36

336

PART III:  The Causes of Cancer

genistein (major isoflavone types), but not total isoflavones, were associated with a significantly lower risk of total prostate cancer in a recent meta-​analysis (He et al., 2015). Inter-​individual variability in the ability of gut microbiota to convert daidzein to equol may explain some of the inconsistencies in the literature on soy and cancer (Lampe, 2009; Vastag, 2007). Of note, few studies have comprehensively evaluated whether soy supplements or isolated soy proteins (found in energy bars and other processed foods) influence cancer risk. Flavonoid classes other than isoflavones are more commonly consumed in Western populations. Despite considerable interest in these other flavonoids with respect to human cancer, the evidence remains limited, in part because flavonoid databases have only recently permitted comprehensive analyses of these compounds in relation to cancer risk (Peterson et al., 2015). Also, flavonoid content varies by plant species and can be affected by factors such as processing method. Food sources of these compounds include green tea (flavanols), black tea (theaflavins and thearubigins), vegetables (flavonols), berries (anthocyanidins), celery and garlic (flavones), citrus fruits (flavanones), and cocoa (proanthocyanidins), among others. Emerging studies from European and US cohorts support a cancer protective role of flavanols in relation to hepatocellular cancer risk (Zamora-​Ros et al., 2013), flavonols and urothelial cell carcinoma (Zamora-​Ros et  al., 2014), anthocyanidins, flavonols, flavones, flavanols, and total flavonoids and gastric cancer (Zamora-​Ros et  al., 2012), and flavonols and flavanones and ovarian cancer (Cassidy et  al., 2014). However, not all studies are consistent (Wang et al., 2009), and some even suggest harm (Romagnolo and Selmin, 2012).

Food Contaminants A comprehensive review of potentially carcinogenic food and beverage contaminants is beyond the scope of this chapter. Herein we focus on some key contaminants that are established or suspected cancer risk factors. Aflatoxins are mycotoxins produced primarily by the common fungus Aspergillus flavus and the closely related species A. parasiticus, and found primarily in corn, peanuts, cottonseed, and tree nuts, in tropical or semitropical areas (IARC, 2012b). They are classified as Group 1 carcinogens by the IARC (IARC, 2012b), and are an established cause of liver cancer. Presence of the hepatitis B virus sensitizes liver cells to the mutagenic effects of aflatoxin, and recent studies document decreases in liver cancer incidence in high-​risk areas in China coinciding with lower aflatoxin biomarkers in urine and blood (Chen et  al., 2013; Sun et  al., 2013). Reductions in aflatoxin exposure are currently the main reason for the decrease in liver cancer at older ages in these high-​risk areas. Neonatal vaccination against hepatitis B will become more important as vaccinated cohorts age (Sun et al., 2013). An industrial chemical, acrylamide, has been classified as “probably carcinogenic” in humans (Group 2A) by IARC since 1994, based primarily on genotoxicity experiments in animals (IARC, 1994). Discovery in the first decade of the 2000s that acrylamide could be formed in (mostly) carbohydrate-​containing foods during high-​heat cooking stimulated a surge of research into the association of estimated acrylamide intake and cancer risk. Dietary acrylamide assessment using an FFQ was shown to correlate moderately (r  =  0.34) with hemoglobin adducts of acrylamide and its metabolite, glycidamide (Wilson et  al., 2009). Two large meta-​analyses (Pelucchi et  al., 2011; Virk-​Baker et al., 2014) found no association between dietary acrylamide and risk of several cancers. Potential for confounding by tobacco (an important source of acrylamide) and difficulty in assessing exposure in a rapidly changing marketplace remain challenges in studying the health effects of this compound. Bisphenyl A  (BPA) is an industrial chemical used in materials intended to come into contact with food (e.g., reusable plastic bottles, feeding-​bottles, microwave ovenware, storage containers, etc.), whereas the epoxy resins are used for internal coating of food and beverage cans (European Food Safety Authority, 2006). First produced in 1891, its estrogenic potential was hypothesized in the 1930s (Dodds and Lawson, 1936). Because of its ability to activate

the human estrogen receptor, it is classified as an endocrine disruptor (Geens et al., 2012). Although BPA is not thought to be a direct carcinogen, studies in rodents show that it sensitizes mammary tissue to effects of chemical carcinogens (Rochester, 2013). BPA exposure may have significant implications for human health and fertility, especially exposure at developmental ages, and in sensitive populations. In 2012, the FDA banned its use in baby bottles and sippy cups (FDA, 2012a, 2012b). Pesticides and genetic modification of crops to resist pests or improve crop yield are frequently the focus of public concern. Studies of the effects of herbicides and pesticides in humans are complicated, however, because the amounts in foods are highly variable and not known to the consumer. For many of these compounds, no practical biomarker of exposure has been identified. The IARC (Guyton et al., 2015) and the state of California (State of California, 2015) have concluded that glyphosate (Roundup), an extensively used herbicide, is a class II carcinogen. This is troublesome because Roundup use is currently widespread for commercial and non-​commercial purposes. Cadmium, used in nickel-​cadmium batteries and for stabilizing plastics, is a heavy metal that enters the food supply from water contaminated by landfills. It bioacumulates in the liver and kidney, and has a half-​life of 7–​16 years (Kjellstrom and Nordberg, 1978). Since 1993, the IARC has classified cadmium as a Group 1 carcinogen, citing sufficient evidence in humans for the carcinogenicity of cadmium and cadmium compounds (IARC, 2012a). Cadmium is thought to influence cancer risk through increasing oxidative stress and reducing DNA damage repair. It is also considered an endocrine disrupter since it has the potential to mimic estrogens and androgens in the body. Studies generally find no association of dietary estimates of cadmium with breast cancer risk (Wu et al., 2015), potentially because of difficulty estimating exposure. A  meta-​analysis of mostly case-​control studies reported a significant increased risk of breast cancer with higher urinary cadmium concentrations (Larsson et al., 2015).

SPECIFIC FOODS AND BEVERAGES Fruits and Vegetables Fruits and vegetables contain numerous constituents thought to influence carcinogenesis, including vitamins, phytochemicals, and dietary fiber (IARC, 2003). Based largely on case-​control studies, the 1997 WCRF/​AICR report designated the evidence “convincing” that fruits and vegetables lower the risk of several cancers (WCRF/​AICR, 1997). Ten years later, the 2007 WCRF/​AICR report concluded that the evidence for fruits, vegetables, and cancer risk was “probable,” primarily for gastrointestinal cancers (WCRF/​AICR, 2007). This shift was chiefly due to weaker findings from prospective cohort studies, which are not subject to the recall and selection biases of case-​control studies. Nevertheless, fruits and vegetables may influence cancer risk in more nuanced ways. A  higher vegetable intake was associated with a significantly lower risk of ER–​breast cancer in two large prospective analyses (Emaus et al., 2016; Jung et al., 2013), including a pooling project of 20 cohort studies (Jung et al., 2013) (Figure 19–​2). In another pooled analysis of 13 prospective cohort studies, intake of fruits and vegetables was inversely associated with risk of renal cell cancer (Lee JE et al., 2009), although the WCRF CUP for kidney cancer considered the totality of the evidence to be limited (WCRF/​AICR, 2015a). In a meta-​analysis of published studies, intake of fruits and vegetables was associated with a nonlinear lower risk of colorectal cancer, with higher risk reported among those with the lowest intakes, and a flattened association at higher intake levels (Aune et al., 2011). Fruits and vegetables are consistently inversely associated with cancers of the esophagus and other aerodigestive tract sites, although this is based largely on case-​control studies (WCRF/​AICR, 2007). As described later in this chapter, overall diet patterns higher in fruits and vegetables are associated with lower risk of certain cancers, cancer mortality, cardiovascular diseases (CVD), and death from all causes. It is worth noting that fruits and vegetables are highly heterogeneous in composition, and some types of fruits and vegetables may

 37



Diet and Nutrition

337

1.2 P-trend, 0.06

Relative Risk

1

P-trend, < 0.001

0.8 0.6

ER-positive ER-negative

0.4 0.2 0 Q1

Q2

Q3

Q4

Q5

Quintile of Vegetable Intake

Figure  19–​2. Association of vegetable intake and RR of breast cancer by hormone receptor status. The study included 993,466 women from 20 prospective cohort studies, with 19,869 estrogen receptor positive (ER+) and 4821 ER− breast cancers diagnosed during 11–​20 years of follow-​up. Source: Jung S, et al. (2013).

formation of nitrosamines in the gut is catalyzed by heme iron, resulting in oxidative DNA damage (Joosen et  al., 2009). Polycyclic aromatic hydrocarbons (PAHs) and heterocyclic aromatic amines (HCA or HAA) are carcinogens formed during high-​heat cooking of meat (Sinha et al., 1998a, 1998b). In a study from the NIH-​AARP Cohort including 300,948 US men and women (Cross et al., 2010) in which dietary intakes were linked with a meat carcinogen database, a positive association was observed for heme iron, nitrate from processed meats, and heterocyclic amine intake, specifically 2-​amino-​3,8-​dimethylimidazo[4,5-​f]quinoxaline (MeIQx) and 2-​amino-​3,4,8-​trimethylimidazo[4,5-​f]quinoxaline (DiMeIQx). Not all studies of specific meat carcinogens are consistent (Ollberding et  al., 2012). The 2015 US Dietary Guidelines Advisory Committee report recommended that individuals reduce their intake of red and processed meat (regardless of fat content) because of its associations with cancer and other diseases and the environmental impact on agriculture and greenhouse gasses (US Dietary Guidelines Advisory Committee, 2015). However, this recommendation did not survive opposition from industry lobbyists and Congress (US Department of Health and Human Services and US Department of Agriculture, 2015).

Grains have potential deleterious effects. Potatoes and some fruit juices, for example, have a high glycemic load and increase insulin secretion. In the United States, potatoes and orange juice are the most commonly consumed vegetables and fruit, respectively, and broccoli and legumes are among the least frequently consumed vegetables (Produce for Better Health Foundation, 2015). Therefore, careful attention should be paid to the types of foods that contribute to risk estimates.

Red and Processed Meat The IARC Monographs Programme recently reviewed the evidence on red and processed meat in relation to cancer, and classified processed meat (e.g., hot dogs, bacon, sausage, deli meats, etc.) as a Group 1 carcinogen, based on sufficient evidence in humans for colorectal cancer (Bouvard et  al., 2015). Consumption of unprocessed red meat (e.g., beef, pork, lamb) was classified as probably carcinogenic to humans (Group 2A), based on limited evidence in humans for colorectal cancer, and strong mechanistic evidence supporting a carcinogenic effect (Bouvard et al., 2015). These conclusions are consistent with those of the WCRF/​AICR Continuous Update Report (WCRF/​AICR, 2011), which considered the association between red and processed meats and colorectal cancer to be “convincing.” A meta-​analysis of prospective cohort studies reported 17%–​18% higher risk of colorectal cancer for each 100 g of red or 50 g of processed meat consumed per day (Chan et al., 2011). The IARC noted that consumption of processed meat is also positively associated with stomach cancer; consumption of red meat is positively associated with pancreatic and prostate cancer (Bouvard et  al., 2015). Because the relationships for colorectal cancer appear linear up to 140 g/​day (Chan et  al., 2011), the evidence does not allow for a determination of a safe level of consumption. The American Cancer Society’s cancer prevention guidelines recommend reducing consumption in general, rather than indicating a specific amount (Kushi et al., 2012). The timing when meat is consumed may be important. For example, a large pooled analysis of eight cohort studies observed no association between breast cancer and consumption of red or total meat in adulthood (Missmer et al., 2002). However, the consumption of red meat during high school was associated with higher risk of breast cancer in young women (Farvid et al., 2014a). It was considered plausible that breast tissue would be most susceptible to carcinogens during adolescence. Furthermore, modeling substitution of red meat with poultry, fish, or legumes and nuts was associated with 14%–​24% lower risks of overall or postmenopausal breast cancer. There are several mechanisms by which the consumption of red and processed meats could influence colorectal carcinogenesis. The

Whole-​grain foods, made from the entire grain seed, are much higher in fiber and many vitamins, minerals, and other phytochemicals than processed (refined) flour products. While research on whole grains in relation to cancer risk has generally been inconsistent, some studies support an inverse association with gastrointestinal cancers. In the NIH-​AARP Diet and Health Study of 291,988 men and 197,623 women, whole-​grain foods were associated with a significant 20% lower risk of colorectal cancer (Schatzkin et al., 2007), and a statistically borderline inverse association with rare small intestinal tumors (Schatzkin et al., 2008). The evidence for other cancers, including pancreatic (Chan et al., 2007) and breast (Egeberg et al., 2009), is inconclusive. Whole-​grain foods are strongly protective against CVD (Cho et al., 2013) and diabetes (de Munter et al., 2007). As described later in this chapter, dietary patterns rich in whole grains, nuts, vegetables, fruits, and low in red and processed meat are associated with lower risk of certain cancers.

Dairy Products In the United States, dairy products are the major source of dietary calcium and vitamin D.  Consistent with the evidence for calcium and vitamin D, epidemiologic evidence supports an inverse association between dairy products and colorectal cancer (Cho et al., 2004). A  recent meta-​ analysis further suggests that milk and total dairy products (but not cheese) lower the risk of colon but not rectal cancer (Aune et al., 2012b). High consumption of milk during childhood and adolescence is associated with increased height (Berkey et al., 2009), which in turn is associated with multiple forms of cancer (WCRF/​AICR, 2007). Milk consumption also increases blood levels of IGF-​1 (Cadogan et  al., 1997; Rich-​Edwards et  al., 2007), which have been associated with higher risks of breast and prostate cancer (Key et al., 2010; Neuhouser et al., 2013; Rowlands et al., 2009). Prospective studies of breast cancer risk in relation to dairy intake have been inconsistent and mostly null (Missmer et al., 2002), although some studies suggest that breast cancer incidence may be lower with higher intakes of low-​fat dairy products during mid-​and later life (Cui and Rohan, 2006; Shin et al., 2002). In contrast, high intake of dairy products has been associated with an increased risk of other hormonally related cancers, including prostate (Aune et al., 2015), ovary (Genkinger et al., 2006), and endometrium, especially among women not using hormone therapy (Ganmaa et al., 2012). Evidence on milk and dairy consumption during childhood and cancer risk in adulthood remains limited; in one report, adolescent milk consumption was not significantly associated with future risk of breast cancer (Linos et al., 2010).

38

338

PART III:  The Causes of Cancer

Soy Soy foods (such as tofu, tempeh, edamame, miso, many veggie burgers, and other products made with soy flour) contain isoflavones which, as described earlier, have weak estrogenic activity and can act as anti-​estrogens by competitive inhibition for the estrogen receptor (Nagata, 2010). Because of this, soy foods have been hypothesized to reduce the risks of breast and other hormonally related cancers. In a recent meta-​analysis including 14 epidemiologic studies, Asian women who ate the most (compared to the least) soy isoflavones had a 24% lower risk of developing breast cancer, while there was no association in Western countries such as the United States (Dong and Qin, 2011). The benefit of soy has been observed primarily with consumption during childhood or early adult life (Lee SA et al., 2009).

relative risk for drinking 4 or more cups per day compared with less than 2 cups per day was 0.25 (95% CI: 0.11, 0.62); this was weakened after control for biomarkers of hepatocellular injury and inflammation, suggesting that these mechanisms may underlie the association (Aleksandrova et al., 2015). Whether coffee is related to a lower risk of cancers of the kidney (Lee et al., 2007), breast (Bhoo-​Pathy et al., 2015), colon and rectum (Li et  al., 2013), or prostate (particularly lethal or advanced forms) (Wilson et al., 2011) is the subject of continued research. The 2016 IARC Expert Review Committee also reviewed the evidence for maté (a traditional South American caffeine-​rich infused drink), and beverage temperature. The IARC concluded that drinking very hot beverages at above 65 degrees Centigrade (149 degrees Fahrenheit), including maté, was “probably carcinogenic to humans” (Group 2A) (Loomis et al., 2016). As most of the evidence in humans is from case-​control studies, it will be important to confirm these findings in additional prospective studies.

Coffee In addition to caffeine, coffee contains multiple biologically active compounds such as chlorogenic acid, kahweol, and N-​methylpyridinium, which in animals and in vitro induce apoptosis, inhibit inflammation, angiogenesis, and metastasis, and regulate genes involved in DNA repair and detoxification processes (Bohn et al., 2014). In 1991, an expert Working Group organized by the International Agency for Research on Cancer (IARC) Monographs Program classified coffee as a 2B carcinogen, concluding that “[c]‌offee is possibly carcinogenic to the human urinary bladder” (IARC Working Group on the Evaluation of Carcinogenic Risks to Humans, 1991); however, a 2016 review by IARC concluded that coffee drinking was considered unclassifiable as to its carcinogenicity to humans (Group 3) (Loomis et al., 2016). Most of the studies considered in the 1991 IARC review were case-​control studies, which are susceptible to recall and selection biases. A 2012 meta-​analysis of coffee consumption and bladder cancer risk continued to show positive associations for case-​control studies, but estimates from five prospective cohort studies were null (Zhou et al., 2012). Further, coffee consumption is generally associated with smoking, which is strongly related to bladder cancer and difficult to control for. The potential for residual confounding due to smoking is a concern for the relationship of coffee consumption with other cancers as well. In the AARP cohort, there was a two-​fold increased risk of mortality due to cancer in age-​adjusted models for both men and women; after controlling for smoking and other risk factors, risk estimates were substantially attenuated and were no longer statistically significant in any category of consumption, although the trend remained statistically significant in men only (Freedman et  al., 2012). Significant interactions by smoking status were observed (p < 0.01), with weak inverse trends in never smokers and either a positive or no trend in current and former smokers. The WCRF/​ AICR CUP considers the protective association between coffee consumption and endometrial cancer to be “probable” (WCRF/​AICR, 2013) although a recent meta-​analysis concluded that there is no association (Yang TO et al., 2015). Of 12 cancers examined in relation to coffee consumption in the Prostate, Lung, Colorectal, and Ovarian (PLCO) screening cohort, a statistically significant inverse association was observed only for endometrial cancer; the association was non-​statistically significantly inverse when limiting the analysis to never smokers (Hashibe et al., 2015). Despite methodologic limitations, coffee consumption has been strongly and consistently related to lower risk of hepatocellular cancer (HCC) (Bravi et al., 2013), and the 2014 WCRF/​AICR Liver Cancer CUP considers it “probable” that coffee consumption protects against liver cancer (WCRF/​AICR, 2015b). In a recent pooling project of nine US cohorts including over 1 million persons, higher coffee consumption was associated with lower risk of HCC (relative risk for > 3 cups/​ day vs. non-​drinker = 0.73; 95% CI: 0.53, 0.99, P, trend cups/​day = < 0.0001) (Petrick et al., 2015). The relative risk was stronger for caffeinated coffee than for decaffeinated coffee, although the confidence interval for the latter was wide. In the European EPIC cohort, the

Tea Tea is rich in polyphenols and other compounds with antioxidant and anti-​cancer properties. Green tea, more commonly consumed in Asian countries, is particularly rich in catechins, a type of flavonoid. In contrast, black tea, more commonly consumed in Western countries, is higher in theaflavins and thearubigens (Yuan et al., 2011). Case-​control studies of tea consumption and cancer risk generally find stronger inverse associations than prospective studies. A  meta-​analysis of 57 prospective studies (including 20 from the US and 22 from Asia) that included over 8 million individuals and almost 50,000 incident cancer cases found tea consumption to be inversely associated with oral cancer (RR 0.72; 95% CI: 0.54, 0.95, high vs. low tea consumption), but with none of the other cancers examined (gastric, colon, rectum, lung, liver, pancreas, breast, prostate, ovarian, bladder, or glioma). Thus, the evidence to date suggests that the relationship between tea and cancer is likely to be minimal, but additional large studies of oral cancer would be valuable to confirm this finding.

Alcohol As described in Chapter  12, alcohol is considered a Group  1 carcinogen by the IARC (IARC, 2010), and an established cause of cancers of the upper gastrointestinal tract, liver, colorectum, and breast (female). In contrast, there is evidence for lack of carcinogenicity for kidney cancer and non-​Hodgkin lymphoma. For pancreatic cancer, risk begins to increase at greater than 3 servings/​day (Gapstur et al., 2011), whereas the increase for breast cancer begins at as few as 3 servings/​week (Chen et al., 2011; Smith-​Warner et al., 1998). On the other hand, alcohol is associated with lower CVD risk and mortality (Thun et  al., 1997), and is an integral part of certain healthful dietary patterns, such as the Mediterranean diet (Trichopoulou et  al., 2003). Dietary recommendations for cancer prevention advise limiting alcohol consumption among those who drink to no more than one drink/​day for women and two drinks/​day for men (WCRF/​AICR, 2007).

DIETARY PATTERNS Characterizing “diet” as an overall diet pattern provides a more complete picture of dietary intake than studies of single foods or nutrients. Dietary patterns also potentially reflect additive and synergistic effects of the components of diet on disease risk. This approach may be particularly useful in the study of cancer etiology, where associations with certain dietary risk factors may be small and difficult to detect in isolation. The two most common approaches to characterize dietary patterns in observational studies involve either statistically driven empirical methods, such as factor analysis, or creation of a priori dietary indexes or scores (Hu, 2002), as described later in this chapter.

 39



Diet and Nutrition

Vegetarians Levels of vegetarianism range from complete exclusion of all animal products (including fish and dairy; “vegan”) to lacto-​ovo vegetarian (includes eggs and dairy), to pescovegetarian (includes fish). In the Adventist Health Study II, compared to non-​ vegetarians, vegetarians of any type—​defined according to protein sources reported on a FFQ—​had a statistically significant 28% lower risk of colorectal cancer (Orlich et al., 2015). Likewise, in a recent British study (Key et al., 2014), vegans, vegetarians, and pescovegetarians had a significantly lower risk of several cancers, compared to meat eaters.

Mediterranean Diet Probably the most studied dietary pattern has been the Mediterranean diet, characterized by high consumption of olive oil (or unsaturated fats in some scoring systems), fruits, vegetables, nuts, legumes, and unprocessed cereals, low consumption of meat and meat products, low consumption of dairy (with the exception of long preservable cheeses), and moderate intake of alcohol, usually wine consumed with meals (Trichopoulou et  al., 2014). A  meta-​analysis of 11 cohort studies found that higher (vs. lower) scores reflecting a Mediterranean-​like diet were associated with a lower risk of all cancers combined (relative risk = 0.87; 95% CI: 0.81, 0.93), and specific cancer sites (colorectum, breast, prostate, stomach, liver, pancreas, head and neck, and respiratory cancers) (Schwingshackl and Hoffmann, 2015). This conclusion was further supported by a recent analysis of three cohort studies in which adherence to a Mediterranean diet, calculated using identical scoring methods, was associated with lower all-​cancer mortality (Liese et  al., 2015). The Spanish PREDIMED randomized trial randomized participants to three diet pattern interventions: a Mediterranean diet (via instruction) with supplementary nuts, a Mediterranean diet with supplementary extra virgin olive oil, or a control (low-​fat) diet (Estruch et al., 2013). Both Mediterranean diet groups had unequivocal health benefits for CVD; although based on a small number of cases, the group randomized to extra virgin olive oil had a substantially lower risk of breast cancer compared to the control group (RR = 0.32; 95% CI: 0.13, 0.79) (Toledo et al., 2015).

Other Diet Patterns Empirically derived patterns, which use statistical approaches to extract unique correlated eating behaviors within a population, typically identify two or three such patterns, each explaining up to 5%–​ 10% of the variation in diet. As expected, patterns vary across study populations, but two have been commonly identified across international borders:  the first, a so-​called “Western” pattern, is higher in red and processed meat, potatoes, sugar, and lower in vegetables and whole fruit; the second, a “Prudent” pattern, is higher in fruits, vegetables, poultry, fish, and nuts, and low in high-​fat dairy, meat, and sugar. The “Western” pattern is positively associated with colon cancer incidence (Magalhaes et al., 2011) and with cancer recurrence and mortality among colorectal cancer patients (Meyerhardt et al., 2007). Although the Prudent pattern is associated with lower colorectal cancer risk, it appears that this is not as strong as avoiding a Western-​type diet (Fung et al., 2003). A priori indices, such as those used to study the Mediterranean diet, quantify the degree of dietary concordance with hypothesized healthy diet patterns using a set of criteria, with low scores generally reflecting poor concordance, and high scores reflecting high concordance to multiple dietary components. Healthy diet scores have been based on dietary guidelines, hypothesized general or disease-​specific diets, culturally defined healthful models of eating, and specific mechanisms mediated by diet, such as its antioxidant or anti-​inflammatory potential. Recently, associations of four hypothesized healthful diet patterns were examined in relation to cause-​specific mortality using a uniform approach to diet score calculations and analysis across three prospective cohort studies (Liese et al., 2015). Scores reflecting dietary concordance with the Mediterranean diet, the Dietary Approaches to Stop

339

Hypertension (DASH) diet, the Healthy Eating Index (reflecting US Dietary Guidance), and an Alternate Healthy Eating Index (reflecting the Harvard Healthy Eating Plate & Food Guide Pyramid) were each associated with an approximate 20% statistically significant reduction in cancer mortality, comparing the highest to lowest quintile (Liese et al., 2015). These diet scores share several commonalities, including being rich in a variety of plant foods and generally higher in healthy protein sources, and lower in sugar and saturated and trans fat, but vary in the specific fatty acids, protein sources, and alcohol recommendations. Of individual cancers, diet scores have tended to be most consistently inversely associated with gastrointestinal cancers. Despite the many advantages of studying diet patterns, such exposures are more complex to interpret, as they may mask subcomponents that are minimally or maximally predictive of disease outcome.

Cancer Prevention Guidelines Dietary guidelines focused on cancer prevention emphasize not only healthy diet, but also behaviors that affect body weight, physical activity, and alcohol consumption (Kushi et al., 2012; WCRF/​AICR, 2007). These recommendations encourage diets that are mostly plant-​ based, high in non-​starchy vegetables, whole fruit, and whole (vs. refined) grains, and low in red and processed meat. Scores reflecting the diet, physical activity, body weight, and alcohol recommendations of cancer prevention guidelines have been developed and examined in relation to cancer incidence (Kabat et al., 2015; Romaguera et al., 2012; Thomson et al., 2014a) and mortality (McCullough et al., 2011; Vergnaud et  al., 2013). Lifelong non-​ smokers whose dietary patterns, weight, and physical activity more closely correspond with the American Cancer Society (ACS) guidelines have a 24%–​30% lower death rate from cancer, and a 42% lower all-​cause mortality rate, compared to those whose patterns are least consistent (McCullough et al., 2011). Another large study of the AARP cohort reported that, in analyses controlling for smoking, the incidence rates for 14 out of 25 cancer sites were significantly lower in study participants whose behaviors were consistent with the ACS guidelines than in those who were not (Kabat et al., 2015). In the Women’s Health Initiative cohort, significant inverse associations with cancer incidence and death rates were reported among white, black, and Hispanic postmenopausal women, with stronger associations appearing in the last two racial/​ethnic groups (Thomson et al., 2014a). In models with mutual adjustment, the individual score components, including diet, generally remained independently associated with lower cancer risk. Results using the global WCRF/​AICR score, with different weighting for lifestyle factors, found qualitatively similar inverse associations with cancer risk and mortality (Romaguera et al., 2012; Vergnaud et al., 2013). Overall, these studies underscore the significant potential for cancer prevention through combined health behaviors including diet, physical activity, and healthy body weight.

Effects of Diet on Cancer Survival Approximately 14.5 million children and adult cancer survivors were alive on January 1, 2014 (American Cancer Society, 2014). Compared to the evidence base on diet and cancer prevention, the study of diet in relation to cancer recurrence, cancer-​specific survival, and overall survival is in its infancy. Therefore guidelines on nutrition and physical activity for cancer survivors also recommend following cancer prevention guidelines (Rock et al., 2012). Timing is a central issue in studies of diet and cancer survival. For a cancer patient, the relevant time is after the diagnosis, and many of the existing studies assessed diet before diagnosis. Optimally an epidemiologic study should have dietary assessments both before and after diagnosis because these tend to be correlated, and we would want to know their independent contribution to survival. Only a few randomized trials of dietary interventions and cancer survival have been conducted, as described in the following paragraph. Breast cancer survivors comprise over 3.1 million survivors in the United States, larger than any other group (American Cancer Society,

340

340

PART III:  The Causes of Cancer

2014). A report based on a systematic and quantitative literature review found no convincing or probable dietary factors associated with recurrence, cause-​specific, or total mortality for breast cancer survivors (WCRF/​AICR, 2014a), although limited support exists for soy foods, high dietary fiber, and low saturated fat. A limitation is that many studies contributing to this report only examined diet prior to diagnosis. In Western diets where soy consumption is less common, lignans are the predominant phytoestrogen. A  recent population-​based case-​control study showed that pre-​diagnostic dietary lignan intake was associated with statistically significantly lower risk of breast cancer mortality (McCann et al., 2010). The Women’s Intervention Nutrition Study (WINS) trial, a low-​fat dietary intervention trial, found a 24% marginally significant lower risk of recurrence (Chlebowski et al., 2006); however, because women on the low-​fat arm also lost weight, it is possible that the effect was due to weight loss and not because of a reduction in dietary fat per se (Gapstur and Khan, 2007). The Women’s Healthy Eating and Living (WHEL) study found no difference in breast cancer outcomes between women counseled on a high fruit, vegetable, fiber, low-​fat diet versus those counseled on eating “five a day” (fruits and vegetables) (Pierce et al., 2007). As breast cancer patients typically have a relatively good prognosis, they are subject to the same diseases as the general population and actually tend to be at higher risk of other chronic diseases, including CVD, relative to non-​ cancer survivors (Rock et al., 2012). Thus, understanding the potential role of diet on both cancer and non-​cancer outcomes in this patient group has strong clinical and population health relevance. As would be expected, a growing literature in breast cancer survivors from prospective cohort studies suggests that women who consume healthy diet patterns (assessed via diet scores) rich in a variety of plant foods and whole grains and low in red and processed meat after diagnosis have lower risk of non-​breast-​cancer causes of death (George et al., 2014; Inoue-​Choi et al., 2013; Izano et al., 2013; Kim et al., 2011); however, none of the diet scores examined was associated with breast cancer–​ specific mortality. For colorectal cancer survivors, post-​diagnostic dietary intake from two cohorts suggests a deleterious effect of a Western dietary pattern, high red and processed meat intake, and a high glycemic load diet on colorectal cancer recurrence or survival (Van Blarigan and Meyerhardt, 2015). In another type of analysis, post-​diagnostic data were used to predict plasma vitamin D status. In turn, higher predicted vitamin D was associated with lower colorectal cancer-specific and overall mortality (Ng et al., 2009). Like breast cancer, clinical recurrence and death often occur many years after a prostate cancer diagnosis, thus providing a wide time window for potential dietary and lifestyle influences on the outcome (Chan et  al., 2014). Although the data on post-​diagnostic diet and survival are limited, a higher prostate-​specific mortality with higher intake of saturated fat and a lower total mortality with higher intake of vegetable fats were observed in a recent report from the Physician’s Health Study (Van Blarigan et  al., 2015). Similar findings were observed in another population of men with prostate cancer (Richman et al., 2013). Adherence to a Western dietary pattern has been associated with higher prostate-​specific and total mortality, and adherence to a Prudent dietary pattern was associated with lower total mortality (Yang M et al., 2015). Numerous short-​term interventions have been conducted with changes in PSA or other surrogate variables as outcomes; possible benefits have been seen with several combinations of nutritional supplements, but clinical outcomes are lacking (Hackshaw-​ McGeagh et al., 2015). The available evidence supports the same recommendations for cancer survival as for cancer prevention (Rock et al., 2012). Fortunately, these recommendations also apply to the prevention of CVD, and thus provide broad benefits for the prevention of chronic disease.

GENE–​DIET INTERACTIONS Nutrigenomics seeks to identify genetic polymorphisms and epigenetic alterations that modify individual response to nutrients (Teegarden et al., 2012; Zeisel, 2007). The underlying goal is to identify individuals

who might benefit most from increased or decreased exposure to specific dietary factors. The same concept underlies “personalized medicine” or “personalized prevention.” The potential for practical applications of this concept has weakened, especially for prevention, with the realization that the effects of common genetic variants on disease risk are considerably smaller than anticipated 15  years ago. Apart from this application, the identification of gene–​environment interactions can be of value in etiologic studies if Mendelian randomization supports a particular mechanism for a diet and disease relationship that cannot be tested in a randomized clinical trial (Willett W, 2013). Furthermore, the identification of a subset of individuals who are uniquely responsive to the effects of a particular dietary factor can increase the statistical power of observational studies of this relationship. The study of gene–​diet interactions in relation to cancer risk has evolved from studies of single nucleotide polymorphisms (SNPs) in candidate genes to genome-​wide association studies (GWAS) using a more agnostic approach. A  recent systematic literature review of gene–​diet interactions and colorectal cancer risk in candidate gene studies found suggestive evidence for interactions between meat, cruciferous vegetables, dietary fiber, calcium, vitamins, and alcohol and SNPs in the ABCB1, NFKB1, GSTM1, GSTT1, CCND1, VDR, MGTM, IL10, and PPARG genes (Andersen et  al., 2013). More recently, a comprehensive GWAS approach examined diet–​ gene interactions with colorectal cancer risk in over 9000 cases and 9000 controls from 10 studies (Figueiredo et al., 2014). The study uncovered strong evidence for a gene–​diet interaction and colorectal cancer risk between a genetic variant (rs4143094) on chromosome 10p14 near the gene GATA3 and processed meat consumption (p = 8.7E-​09), although the functional significance of this interaction is still unknown (Figueiredo et al., 2014). The impact of epigenetic modifications can vary from full gene silencing to complete expression, and the effect may be permanent or transient (Teegarden et al., 2012). Multiple bioactive food components, including folate, selenium, polyphenols, isothiocyanates, and others, have been shown to affect epigenetic processes in vitro; however, more evidence is needed from animal and human studies (Ong et al., 2011). At present, no gene–​environment interactions involving nutritional factors and cancer have been identified with sufficient certainty and impact to have clinical application.

SUMMARY Methodologic Considerations Existing methodologic approaches have identified important relationships between nutrition and cancer, especially regarding dietary patterns, anthropometric measures, and alcohol consumption. Diet composition can be reasonably well characterized by food frequency questionnaires, repeated measures of short-​term intake (e.g., multiple food records and recalls), and biological markers of dietary intake. The methods of assessing diet composition continue to improve as nutrient databases are expanded, technologies are adapted and/​or developed to measure dietary intake more precisely, new biomarkers are discovered, and metabolites that mediate between food intake and biological effects on cancer risk are identified. Further improvements in these methodologies will likely increase the precision of estimates of absolute or relative risk. New biomarkers are needed to assess aspects of diet that cannot be evaluated from self-​report; markers of food processing and contaminants would be particularly useful. It is inherently difficult to separate the effects of highly correlated nutrients and foods in observational studies; distinguishing among them may be facilitated by Mendelian randomization studies and short-​ term randomized studies with intermediate endpoints. These issues become unimportant in studies of overall dietary patterns. No single type of study is likely to resolve most hypothesized diet and cancer relationships. Large randomized trials of diet and cancer incidence may be definitive if “positive,” but are not a gold standard, as they may be unable to detect important effects due to poor adherence

 341

Table 19–1.  Relationship of Dietary Factors with Risk of Selected Individual Cancer Sitesa Dietary Factor

Colorectum

Breast

Prostate

Obesity

↑↑

↑↑˅

↑˅

Abdominal fatness

↑↑

↑

Lung

Stomach

Esophagus

Pancreas

Liver

Ovary

Endometrium

All Cancers

↑↑˅

↑↑˅

↑↑

↑↑

↑

↑↑

↑↑

Macronutrients/​energy balance

↑§

Carbohydrates/​sugars

↑

Glycemic Load ↑↑

↑↑

↑

↑

Vitamin D [25(OH)D]

↓

φ

Non-​linear

Conflicting

Calcium

↓

Folate

↓

Fiber

↓↓

Height

↑↑

↑

↑

Nutrients (from food unless otherwise specified)

↑

§

§

↓˅

Lycopene↓§

↓

ß-​carotenesupplements

φ

↑↑˅

Vitamin E supplements

φ

Selenium supplements

φ

Carotenoids

Antioxidant (combin​ation) supplements

§

ϕ

φ ↑↑

Salt preservation ↑↑arsenic in drinking water

Contaminants

↑↑Aflatoxin

Foods ↓**

Fruits ↓˅

Vegetables Red meat

↑↑

Processed meat

↑↑

Other protein sources, fish, poultry, nuts

↓

Whole grains

↓

Dairy or milk

↓

Soy Coffee

↑

§

↓

↓^

↓

↓^

**

↑†

↑† ↑↑

†

↑^ ↓

§

↑˅§ ↓˅ φ

↓↓§

↓ (continued)

342

Table 19–1. Continued Dietary Factor

Colorectum

Breast

Prostate

Lung

Stomach

Esophagus

Pancreas

Liver

Ovary

Endometrium

φ

Tea Alcohol

All Cancers

↑↑˅

↑↑

↑↑

↑↑

↑§

↑↑

↑↑

↓

↓

Diet patterns* Empirical diet patterns “Western” diet

↑↑

“Prudent” diet

↓

↓˅

Mediterranean diet

↓

↓§

“DASH” diet

↓

USDA Healthy Eating Index

↓

↓

Harvard “Alternate” Healthy Eating Index

↓

↓

Dietary Inflammatory Index

↑

Dietary indices ↓˅

↓ ↓

↓

↓

↓ ↓

Cancer lifestyle guidelines‡ ACS guidelines

↓

↓

↓˅

↓

↓

WCRF/AICR Guidelines

↓

↓

↓

↓

↓

↓˅

↓˅

↓

↓↓

↓

↓

↓↓

Two arrows denotes convincing or consistent evidence; One arrow denotes probable association; Grey arrow denotes only one or limited studies; ϕ mark=unlikely to be an association; ˅ denotes lower risk in subgroups, e.g. obesity and postmenopausal (not premenopausal) breast cancer, aggressive prostate, gastric cardia and esophageal adenocarcinoma; vegetables, carotenoids and ER-​breast cancer; beta-​carotene supplements and lung cancer in smokers; processed meat and gastric cardia cancer; fruit, vegetables, processed meat and squamous cell esophageal cancer; dairy and aggressive prostate cancer; soy and breast cancer primarily in Asian countries; alcohol and colorectal cancer risk convincing in men, probable in women. For diet patterns, prudent diet related to lower risk of ER-​breast tumors, Mediterranean diet associated with esophageal squamous cell cancer; ACS guidelines related to lower liver, lung and pancreatic cancer in men only. § denotes author conclusions (note: breast cancer and red meat includes early life exposure); † also noted in IARC 2015 conclusions *Evidence on diet patterns is based on a limited number of studies. ‡ In addition to food-based recommendations, the ACS and WCRF/​AICR scores also include body mass index, physical activity and alcohol consumption. Sources: WCRF/​AICR 2007 report or CUP reports if available; **This association may be confounded by smoking. a

 34



Diet and Nutrition

(especially with complex dietary patterns), limited duration, and difficulty blinding the intervention. For most hypotheses, the best evidence will probably come from replicated cohort studies in combination with short-​term randomized trials with intermediate biomarkers.

Estimates of Attributable Fraction The population attributable fraction for diet and all cancers, based on more precise estimates for relationships that are established with reasonable certainty, is about 20% overall, which is weaker than previously estimated, and much of this is related to overweight and inactivity. This also does not include the contribution of dietary composition to adiposity, which has become better established in recent years. The proportion contributed by dietary composition independent of adiposity is unclear because some dietary factors have yet to be identified or established with sufficient certainty, and associations are likely underestimated because of measurement error or misspecification of temporal relationships. Estimates based on reasonably documented relationships suggest an etiologic contribution of 5%–​12%, but this could be appreciably higher.

Specific Diet and Cancer Relationships Progress in diet and cancer prevention over the past 30+ years has been slow compared to accrual of knowledge relating diet to CVD and diabetes, but several findings come into focus: 1. Individual nutritional supplements in high doses are unlikely to prevent cancer and may even be harmful, but evidence suggests a modest benefit of RDA-​level multi-​nutrient supplements that may be most important in subsets with suboptimal diets. 2. Certain constituents (nutrients, phytochemicals) may significantly reduce the risk of various cancers. Low folate intake likely contributes to colon and possibly other cancers, and considerable evidence supports a role of plant foods, lycopene, vitamin D, and constituents of coffee in human cancer. 3. Processed and red meat consumption is likely to increase risk of colorectal and possibly other cancers, and dairy foods are likely to influence risk of colorectal and prostate cancers in opposite directions. 4. Alcohol consumption increases the risk of breast, colorectal, esophageal, and other cancers but is unlikely to increase risk of kidney cancer or non-​Hodgkin lymphoma. 5. Several culturally or a priori defined diet patterns, including the Mediterranean diet, are associated with lower cancer risk and mortality in prospective epidemiological studies. These diet patterns are higher in plant foods (vegetables, fruits, cereals, vegetable protein) and fish, and lower in processed meat and red meat and generally moderate in dairy and alcohol (Table 19–​1).

FUTURE DIRECTIONS The precision of dietary estimates can be improved by combining several different approaches to exposure measurement, including the use of biomarkers, food frequency questionnaires, and multiple 24-​hour recalls or food records. Pooled analyses of dietary intakes or biomarkers in relation to cancer outcomes across multiple prospective studies can minimize variation due to chance and biases from selective recall and publication. This is optimal if measurements of absolute intake and biomarker concentration are standardized and/​or calibrated across studies. Probably the greatest limitation of currently available evidence is that it derives mainly from diets assessed in mid-​or later life with follow-​up of less than 15 years. Dietary assessments earlier in life, including in utero and during puberty, combined with long follow-​ up using repeated dietary assessments, may uncover exposures that are relevant. Technological advancements in biomarker analysis permit comprehensive examination of an individual’s metabolome, microbiome, genome, and epigenome. Use of these methods to assess internal exposures and identify subsets of individuals who are particularly

343

responsive to the dietary factor being investigated could increase our ability to detect diet and cancer relationships and to provide targeted interventions. Cohort studies with prospectively collected archived stool, blood, and urine samples will be positioned to evaluate proposed biomarkers from “omics” studies in relation to cancer risk. Molecular characterization is increasingly used in oncology and cancer epidemiology to identify tumor subtypes that differ in their clinical characteristics and etiology. As with the identification of gene–​ environment interactions, identifying biologically relevant exposures and pathways related to specific cancer subtypes can increase the ability to detect relationships and to develop support for causality. This work will benefit from collaborative efforts to provide the large numbers of study participants and cases required. References Adolphe JL, Whiting SJ, Juurlink BH, Thorpe LU, and Alcorn J. 2010. Health effects with consumption of the flax lignan secoisolariciresinol diglucoside. Br J Nutr, 103(7), 929–​938. PMID: 20003621. Ahn J, Peters U, Albanes D, et al. 2008. Serum vitamin D concentration and prostate cancer risk:  a nested case-​control study. J Natl Cancer Inst, 100(11), 796–​804. Alberts DS, Martinez ME, Roe DJ, et al. 2000. Lack of effect of a high-​fiber cereal supplement on the recurrence of colorectal adenomas. N Engl J Med, 342, 1156–​1162. Aleksandrova K, Bamia C, Drogan D, et  al. 2015. The association of coffee intake with liver cancer risk is mediated by biomarkers of inflammation and hepatocellular injury:  data from the European Prospective Investigation into Cancer and Nutrition. Am J Clin Nutr, 102(6), 1498–​ 1508. PMCID: PMC4658462. American Cancer Society. 2013. Breast cancer facts & fi­ gures  2013–​2014. Atlanta: American Cancer Society. American Cancer Society. 2014. Cancer treatment and survivorship facts & ­figures 2014–​2015. Atlanta: American Cancer Society. Ames BN, Gold LS, and Willett WC. 1995. The causes and prevention of cancer. Proc Natl Acad Sci USA, 92, 5258–​5265. Andersen V, Holst R, and Vogel U. 2013. Systematic review: diet-​gene interactions and the risk of colorectal cancer. Aliment Pharmacol Ther, 37(4), 383–​391. PMCID: PMC3565452. Apperly. 1941. The relation of solar radiation to cancer mortality in North America. Cancer Res, 1, 191–​195. Arab L, Steck-​Scott S, and Bowen P. 2001. Participation of lycopene and beta-​ carotene in carcinogenesis: defenders, aggressors, or passive bystanders? Epidemiologic Rev, 23(2), 211–​230. Armstrong B, and Doll R. 1975. Environmental factors and cancer incidence and mortality in different countries, with special reference to dietary practices. Int J Cancer, 15, 617–​631. Aune D, Chan DS, Greenwood DC, et al. 2012a. Dietary fiber and breast cancer risk: a systematic review and meta-​analysis of prospective studies. Ann Oncol, 23(6), 1394–​1402. PMID: 22234738. Aune D, Lau R, Chan DS et al. 2011. Nonlinear reduction in risk for colorectal cancer by fruit and vegetable intake based on meta-​analysis of prospective studies. Gastroenterology, 141(1), 106–​118. PMID: 21600207. Aune D, Lau R, Chan DS, et al. 2012b. Dairy products and colorectal cancer risk: a systematic review and meta-​analysis of cohort studies. Ann Oncol, 23(1), 37–​45. PMID: 21617020. Aune D, Navarro Rosenblatt DA, Chan DS, et al. 2015. Dairy products, calcium, and prostate cancer risk: a systematic review and meta-​analysis of cohort studies. Am J Clin Nutr, 101(1), 87–​117. PMID: 25527754. Avenell A, MacLennan GS, Jenkinson DJ, et al. 2012. Long-​term follow-​up for mortality and cancer in a randomized placebo-​controlled trial of vitamin D(3) and/​or calcium (RECORD trial). J Clin Endocrinol Metab, 97(2), 614–​622. PMID: 22112804. Bakker MF, Peeters PH, Klaasen VM, et al. 2016. Plasma carotenoids, vitamin C, tocopherols, and retinol and the risk of breast cancer in the European Prospective Investigation into Cancer and Nutrition cohort. Am J Clin Nutr, 103(2), 454–​464. PMID: 26791185. Baron JA, Beach M, Mandel JS, et al. Calcium supplements for the prevention of colorectal adenomas. N Engl J Med 1999;340:101–​7. Baron JA, Barry EL, Mott LA et al. 2015. A trial of calcium and vitamin D for the prevention of colorectal adenomas. N Engl J Med, 373(16), 1519–​ 1530. PMCID: PMC4643064. Berkey CS, Colditz GA, Rockett HR, Frazier AL, and Willett WC. 2009. Dairy consumption and female height growth: prospective cohort study. Cancer Epidemiol Biomarkers Prev, 18(6), 1881–​1887. PMCID: PMC3740511. Bhoo-​Pathy N, Peeters PH, Uiterwaal CS, et al. 2015. Coffee and tea consumption and risk of pre-​and postmenopausal breast cancer in the European

34

344

PART III:  The Causes of Cancer

Prospective Investigation into Cancer and Nutrition (EPIC) cohort study. Breast Cancer Res, 17, 15. PMCID: PMC4349221. Binkley N and Sempos CT for the Vitamin D Standardization Program. 2014. Standardizing vitamin D assays:  the way forward. J Bone Miner Res, 29(8), 1709–​1714. PMID: 24737265. Birt DF, Kris ES, Choe M, and Pelling JC. 1992. Dietary energy and fat effects on tumor promotion. Cancer Res, 52(7 Suppl), 2035s–​2039s. PMID: 1544138. Block G, Hartman AM, and Naughton D. 1990. A reduced dietary questionnaire: development and validation. Epidemiol, 1, 58–​64. Blot WJ, Li JY, Taylor PR, et al. 1993. Nutrition intervention trials in Linxian, China: supplemenation with specific vitamin/​mineral combinations, cancer incidence, and disease-​specific mortality in the general population. J Natl Cancer Inst, 85(18), 1483–​1492. Blount BC, and Ames BN. 1994. Analysis of uracil in DNA by gas chromatography-​mass spectrometry. Analytical Biochemistry 219, 195–​200. Bohn SK, Blomhoff R, and Paur I. 2014. Coffee and cancer risk, epidemiological evidence, and molecular mechanisms. Mol Nutr Food Res, 58(5), 915–​930. PMID: 24668519. Boissonneault GA, Elson CE, and Pariza MW. 1986. Net energy effects of dietary fat on chemically induced mammary carcinogenesis in F344 rats. J Natl Cancer Inst, 76(2), 335–​338. PMID: 3080637. Bonithon-​Kopp C, Kronborg O, Giocosa A, Rath U, and Faivre J. 2000. Calcium and fibre supplementation in the prevention of colorectal adenoma recurrence: a randomised intervention trial. Lancet, 356, 1300–​1306. Bouvard V, Loomis D, Guyton KZ, et  al. 2015. Carcinogenicity of consumption of red and processed meat. Lancet, 16, 1599–​1600 PMID: 26514947 Bravi F, Bosetti C, Tavani A, Gallus S, and La Vecchia C. 2013. Coffee reduces risk for hepatocellular carcinoma: an updated meta-​analysis. Clin Gastroenterol Hepatol, 11(11), 1413–​1421 e1411. PMID: 23660416. Brooks JD, Metter EJ, Chan DW, et  al. 2001. Plasma selenium level before diagnosis and the risk of prostate cancer development. J Urol, 166(6), 2034–​2038. PMID: 11696701. Burkitt DP. 1971. Epidemiology of cancer of the colon and rectum. Cancer, 28, 3–​13. Cadogan J, Eastell R, Jones N, and Barker ME. 1997. Milk intake and bone mineral acquisition in adolescent girls: randomised, controlled intervention trial. BMJ, 315(7118), 1255–​1260. PMCID: PMC2127785. Calle EE, and Kaaks R. 2004. Overweight, obesity and cancer: epidemiological evidence and proposed mechanisms. Nature Rev, 4, 579–​591. Cassidy A, Huang T, Rice MS, Rimm EB, and Tworoger SS. 2014. Intake of dietary flavonoids and risk of epithelial ovarian cancer. Am J Clin Nutr, 100(5), 1344–​1351. PMCID: PMC4196485. Chan DS, Lau R, Aune D et al. 2011. Red and processed meat and colorectal cancer incidence: meta-​analysis of prospective studies. PLoS One, 6(6), e20456. PMCID: PMC3108955. Chan JM, Van Blarigan EL, and Kenfield SA. 2014. What should we tell prostate cancer patients about (secondary) prevention? Curr Opin Urol, 24(3), 318–​323. PMCID: PMC4084902. Chan JM, Wang F, and Holly EA. 2007. Whole grains and risk of pancreatic cancer in a large population-​based case-​control study in the San Francisco Bay Area, California. Am J Epidemiol, 166(10), 1174–​1185. PMID: 17881383. Chen JG, Egner PA, Ng D, et al. 2013. Reduced aflatoxin exposure presages decline in liver cancer mortality in an endemic region of China. Cancer Prev Res (Phila), 6(10), 1038–​1045. PMCID: PMC3800239. Chen WY, Rosner B, Hankinson SE, Colditz GA, and Willett WC. 2011. Moderate alcohol consumption during adult life, drinking patterns, and breast cancer risk. JAMA, 306(17), 1884–​ 1890. PMCID: PMC3292347. Chlebowski RT, Blackburn GL, Thomson CA, et al. 2006. Dietary fat reduction and breast cancer outcome:  interim efficacy results from the Women’s Intervention Nutrition Study. J Natl Cancer Inst, 98(24), 1767–​1776. PMID: 17179478. Chlebowski RT, Johnson KC, Kooperberg C, et  al. 2008. Calcium plus vitamin D supplementation and the risk of breast cancer. J Natl Cancer Inst, 100(22), 1581–​1591. Cho E, Smith-​Warner SA, Spiegelman D et al. 2004. Dairy foods, calcium, and colorectal cancer: a pooled analysis of 10 cohort studies. J Natl Cancer Inst, 96, 1015–​1022. Cho SS, Qi L, Fahey GC, Jr., and Klurfeld DM. 2013. Consumption of cereal fiber, mixtures of whole grains and bran, and whole grains and risk reduction in type 2 diabetes, obesity, and cardiovascular disease. Am J Clin Nutr, 98(2), 594–​619. PMID: 23803885.

Clark LC, Combs GF Jr, Turnbull BW, et al. 1996. Effects of selenium supplementation for cancer prevention in patients with carcinoma of the skin: a randomized controlled trial. Nutritional Prevention of Cancer Study Group. JAMA, 276(24), 1957–​1963. Colditz GA, and Wei EK. 2012. Preventability of cancer:  the relative contributions of biologic and social and physical environmental determinants of cancer mortality. Annu Rev Public Health, 33, 137–​ 156. PMCID: PMC3631776. Cole BF, Baron JA, Sandler RS, et al. 2007. Folic acid for the prevention of colorectal adenomas: a randomized clinical trial. JAMA, 297(21), 2351–​ 2359. PMID: 17551129. Correa P, Fontham ETH, Bravo JC, et al. 2000. Chemoprevention of gastric dysplasia: randomized trial of antioxidant supplements and anti-​Helicobacter pylori therapy. J Natl Cancer Inst, 92(23), 1881–​1888. Cross AJ, Ferrucci LM, Risch A, et al. 2010. A large prospective study of meat consumption and colorectal cancer risk:  an investigation of potential mechanisms underlying this association. Cancer Res, 70(6), 2406–​2414. PMCID: PMC2840051. Cross AJ, Major JM, and Sinha R. 2011. Urinary biomarkers of meat consumption. Cancer Epidemiol Biomarkers Prev, 20, 1107–​1111. Crowe FL, Appleby PN, Travis RC, et  al. 2014. Circulating fatty acids and prostate cancer risk: individual participant meta-​analysis of prospective studies. J Natl Cancer Inst, 106(9). PMCID: PMC4188122. Cui Y, and Rohan TE. 2006. Vitamin D, calcium, and breast cancer risk:  a review. Cancer Epidemiol Biomarkers Prev, 15(8), 1427–​1447. de Munter JS, Hu FB, Spiegelman D, Franz M, and van Dam RM. 2007. Whole grain, bran, and germ intake and risk of type 2 diabetes:  a prospective cohort study and systematic review. PLoS Med, 4(8), e261. PMCID: PMC1952203. Dickinson A, Blatman J, El-​Dash N, and Franco JC. 2014. Consumer usage and reasons for using dietary supplements: report of a series of surveys. J Am Coll Nutr, 33(2), 176–​182. PMID: 24724775. Dodds EC, and Lawson W. 1936. Synthetic œstrogenic agents without the phenanthrene nucleus. Nature, 137, 996. Doll R, and Peto R. 1981. The causes of cancer: quantitative estimates of avoidable risks of cancer in the United States today. J Natl Cancer Inst, 66, 1191–​1308. Dong JY, He K, Wang P, and Qin LQ. 2011. Dietary fiber intake and risk of breast cancer: a meta-​analysis of prospective cohort studies. Am J Clin Nutr, 94(3), 900–​905. PMID: 21775566. Dong JY, and Qin LQ. 2011. Soy isoflavones consumption and risk of breast cancer incidence or recurrence:  a meta-​analysis of prospective studies. Breast Cancer Res Treat, 125(2), 315–​323. PMID: 21113655. Egeberg R, Olsen A, Loft S, et al. 2009. Intake of whole grain products and risk of breast cancer by hormone receptor status and histology among postmenopausal women. Int J Cancer, 124(3), 745–​750. PMID: 19004010. Eliassen AH, Liao X, Rosner B, et  al. 2015. Plasma carotenoids and risk of breast cancer over 20 y of follow-​up. Am J Clin Nutr, 101(6), 1197–​1205. PMCID: PMC4441811. Emaus MJ, Peeters PH, Bakker MF, et al. 2016. Vegetable and fruit consumption and the risk of hormone receptor-​defined breast cancer in the EPIC cohort. Am J Clin Nutr, 103(1), 168–​177. PMID: 26607934. Estruch R, Ros E, Salas-​Salvado J, et al. 2013. Primary prevention of cardiovascular disease with a Mediterranean diet. N Engl J Med, 368(14), 1279–​ 1290. PMID: 23432189. European Food Safety Authority. 2006. Opinion of the Scientific Panel on Food Additives, Flavourings, Processing Aids and Materials in Contact with Food on a request from the Commission related to 2,2-​ BIS(4-​ HYDROXYPHENYL)PROPANE (Bisphenol A). EFSA, 428, 1–​75. Farvid MS, Cho E, Chen WY, Eliassen AH, and Willett WC. 2014a. Dietary protein sources in early adulthood and breast cancer incidence: prospective cohort study. BMJ, 348, g3437. PMCID: PMC4051890. Farvid MS, Cho E, Chen WY, Eliassen AH, and Willett WC. 2014b. Premenopausal dietary fat in relation to pre-​and post-​ menopausal breast cancer. Breast Cancer Res Treat, 145(1), 255–​ 265. PMCID: PMC4126228. Farvid MS, Eliassen AH, Cho E, et  al. 2016. Dietary fiber intake in young adults and breast cancer risk. Pediatrics, 137(3), 1–​11. PMID: 26908709. FDA. 2012a. Food additive petitions: American Chemistry Council. Available at:  https://​plastics.americanchemistry.com/​Product-​Groups-​and-​Stats/​ PolycarbonateBPA-​Global-​Group/​FDA-​Petition-​Letter-​ACC.pdf FDA. 2012b. Indirect food additives: polymers. Federal Register (Vol. 77, pp. 41899–​41902): FDA, Washington, D.C. Feldman D, Krishnan AV, Swami S, Giovannucci E, and Feldman BJ. 2014. The role of vitamin D in reducing cancer risk and progression. Nat Rev Cancer, 14(5), 342–​357. PMID: 24705652.

 345



Diet and Nutrition

Ferrari P, Rinaldi S, Jenab M, et  al. 2013. Dietary fiber intake and risk of hormonal receptor-​defined breast cancer in the European Prospective Investigation into Cancer and Nutrition study. Am J Clin Nutr, 97(2), 344–​353. PMID: 23269820. Figueiredo JC, Hsu L, Hutter CM, et al. 2014. Genome-​wide diet-​gene interaction analyses for risk of colorectal cancer. PLoS Genet, 10(4), e1004228. PMCID: PMC3990510. Fortmann SP, Burda BU, Senger CA, Lin JS, and Whitlock EP. 2013. Vitamin and mineral supplements in the primary prevention of cardiovascular disease and cancer: An updated systematic evidence review for the U.S. Preventive Services Task Force. Ann Intern Med, 159(12), 824–​ 834. PMID: 24217421. Freedman ND, Park Y, Abnet CC, Hollenbeck AR, and Sinha R. 2012. Association of coffee drinking with total and cause-​specific mortality. N Engl J Med, 366(20), 1891–​1904. PMCID: PMC3439152. Fritz H, Seely D, Flower G, et  al. 2013. Soy, red clover, and isoflavones and breast cancer:  a systematic review. PLoS One, 8(11), e81968. PMCID: PMC3842968. Fung T, Hu FB, Fuchs C, et  al. 2003. Major dietary patterns and the risk of colorectal cancer in women. Arch Intern Med, 163, 309–​314. Ganmaa D, Cui X, Feskanich D, Hankinson SE, and Willett WC. 2012. Milk, dairy intake and risk of endometrial cancer:  a 26-​year follow-​up. Int J Cancer, 130(11), 2664–​2671. PMCID: PMC3359127. Gapstur SM, Jacobs EJ, Deka A, et al. 2011. Association of alcohol intake with pancreatic cancer mortality in never smokers. Arch Intern Med, 171(5), 444–​451. PMID: 21403041. Gapstur SM, and Khan S. 2007. Fat, fruits, vegetables, and breast cancer survivorship. JAMA, 298(335–​336). PMID: 17635896 Garland CF, and Garland FC. 1980. Do sunlight and vitamin D reduce the likelihood of colon cancer? Int J Epidemiol, 9(3), 227–​231. Gaziano JM, Glynn RJ, Ghristen WG, et  al. 2009. Vitamins E and C in the prevention of prostate and total cancer in men. The Physician’s Health Study II Randomized Controlled Trial. JAMA, 301, 52–​ 62. PMCID: PMC2774210 Gaziano JM, Sesso HD, Christen WG, et al. 2012. Multivitamins in the prevention of cancer in men: the Physicians’ Health Study II randomized controlled trial. JAMA, 308(18), 1871–​1880. PMCID: PMC3517179. Geens T, Aerts D, Berthot C, et al. 2012. A review of dietary and non-​dietary exposure to bisphenol-​ A. Food Chem Toxicol, 50(10), 3725–​ 3740. PMID: 22889897. Genkinger JM, Hunter DJ, Spiegelman D, et al. 2006. Dairy products and ovarian cancer:  a pooled analysis of 12 cohort studies. Cancer Epidemiol Biomarkers Prev, 15(2), 364–​372. PMID: 16492930. George SM, Ballard-​Barbash R, Manson JE, et  al. 2014. Comparing indices of diet quality with chronic disease mortality risk in postmenopausal women in the Women’s Health Initiative Observational Study: evidence to inform national dietary guidance. Am J Epidemiol, 180(6), 616–​625. PMCID: PMC4157698. Gibson TM, Weinstein SJ, Pfeiffer RM, et  al. 2011. Pre-​and postfortification intake of folate and risk of colorectal cancer in a large prospective cohort study in the United States. Am J Clin Nutr, 94(4), 1053–​1062. PMCID: PMC3173023. Giovannucci E. 2001. Insulin, insulin-​like growth factors and colon cancer: a review of the evidence. J Nutr, 131, 3109S–​3120S. Gorbach SL, and Goldin BR. 1987. Diet and the excretion and enterohepatic cycling of estrogens. Prev Med, 16(4), 525–​531. PMID: 3628202. Gray R, and Wheatley K. 1991. How to avoid bias when comparing bone marrow transplantation with chemotherapy. Bone Marrow Transplant, 7(Suppl 3), 9–​12. Greenberg ER, Baron JA, Tosteson TD, et  al. 1994. A clinical trial of antioxidant vitamins to prevent colorectal adenoma. Polyp Prevention Study Group. N Engl J Med, 331(3), 141–​147. PMID: 8008027. Guyton KZ, Loomis D, Grosse Y, et  al. 2015. Carcinogenicity of tetrachlorvinphos, parathion, malathion, diazinon, and glyphosate. Lancet Oncol, 16(5), 490–​491. PMID: 25801782. Hackshaw-​McGeagh LE, Perry RE, Leach VA, et al. 2015. A systematic review of dietary, nutritional, and physical activity interventions for the prevention of prostate cancer progression and mortality. Cancer Causes Control, 26(11), 1521–​1550. PMCID: PMC4596907. Harnly JM, Doherty RF, Beecher GR et  al. 2006. Flavonoid content of U.S. fruits, vegetables, and nuts. J Agric Food Chem, 54(26), 9966–​9977. PMID: 17177529. Harris WS, Mozaffarian D, Rimm E, et al. 2009. Omega-​6 fatty acids and risk for cardiovascular disease:  a science advisory from the American Heart Association Nutrition Subcommittee of the Council on Nutrition, Physical Activity, and Metabolism; Council on Cardiovascular Nursing; and

345

Council on Epidemiology and Prevention. Circulation, 119(6), 902–​907. PMID: 19171857. Hashibe M, Galeone C, Buys SS, et al. 2015. Coffee, tea, caffeine intake, and the risk of cancer in the PLCO cohort. Br J Cancer, 113(5), 809–​816. PMCID: PMC4559834. He J, Wang S, Zhou M, et al. 2015. Phytoestrogens and risk of prostate cancer:  a meta-​analysis of observational studies. World J Surg Oncol, 13, 231. PMCID: PMC4521376. Helzlsouer K. 2010. Overview of the Cohort Consortium Vitamin D Pooling Project of Rarer Cancers. Am J Epidemiol, 172(1), 4–​ 9. PMID: 20562193 Hennekens CH, Buring JE, Manson JE, et  al. 1996. Lack of effect of long-​ term supplementation with beta carotene on the incidence of malignant neoplasms and cardiovascular disease. N Engl J Med, 334, 1145–​1149. Hercberg S, Galan P, Preziosi P, et al. 2004. The SU.VI.MAX study. A randomized, placebo-​controlled trial of the health effects of antioxidant vitamins and minerals. Arch Intern Med, 164, 2335–​2342. PMID: 15557412 Holick MF. 2004. Sunlight and vitamin D for bone health and prevention of autoimmune diseases, cancers, and cardiovascular disease. Am J Clin Nutr, 80(Suppl), 1678S–​1688S. PMID: 15585788 Howe GR, Benito E, Castelleto R, et  al. 1992. Dietary intake of fiber and decreased risk of cancers of the colon and rectum:  evidence from the combined analysis of 13 case-​control studies. J Natl Cancer Inst, 84(24), 1887–​1896. Howe GR, Hirohata T, Hislop TG, et al. 1990. Dietary factors and risk of breast cancer: combined analysis of 12 case-​control studies. J Natl Cancer Inst, 82(7), 561–​569. PMID: 2156081. Hu FB. 2002. Dietary pattern analysis: a new direction in nutritional epidemiology. Curr Opin Lipidol, 13(1), 3–​9. Hu FB, Rimm E, Smith-​Warner SA, et al. 1999. Reproducibility and validity of dietary patterns assessed with a food-​frequency questionnaire. Am J Clin Nutr, 69(2), 243–​249. PMID: 9989687. IARC. 1994. Acrylamide. In:  International Agency for Research on Cancer (Ed.),: Some industrial chemicals. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans (Vol. 60). Lyon, France: IARC Press. IARC. 2003. Fruit and vegetables. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans (Vol. 8). Lyon: IARC Press. IARC. 2010. Alcohol consumption and ethyl carbamate. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans (Vol. 96). Lyon, France: IARC Press. IARC. 2012a. Arsenic, metals, fibres, and dust. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans (Vol. 100C). Lyon, France: IARC Press. IARC. 2012b. Chemical agents and related occupations. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans (Vol. 100F). Lyon, France: IARC Press. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. 1991. Coffee, tea, mate, methylxanthines and methylglyoxal. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans (Vol. 51, pp. 1–​513). Lyon, France: IARC Press. Inoue-​Choi M, Robien K, and Lazovich D. 2013. Adherence to the WCRF/​ AICR guidelines for cancer prevention is associated with lower mortality among older female cancer survivors. Cancer Epidemiol Biomarkers Prev, 22(5), 792–​802. PMCID: PMC3650116. Institute of Medicine. 2011. Dietary reference intakes for calcium and vitamin D. Washington, DC: Institute of Medicine (IOM). Izano MA, Fung TT, Chiuve SS, Hu FB, and Holmes MD. 2013. Are diet quality scores after breast cancer diagnosis associated with improved breast cancer survival? Nutr Cancer, 65(6), 820–​826. PMCID: PMC3776311. Joosen AM, Kuhnle GG, Aspinall SM, et  al. 2009. Effect of processed and red meat on endogenous nitrosation and DNA damage. Carcinogenesis, 30(8), 1402–​1407. PMCID: PMC2718076. Jung S, Spiegelman D, Baglietto L, et al. 2013. Fruit and vegetable intake and risk of breast cancer by hormone receptor status. J Natl Cancer Inst, 105(3), 219–​236. PMCID: PMC3593764. Kabat GC, Matthews CE, Kamensky V, Hollenbeck AR, and Rohan TE. 2015. Adherence to cancer prevention guidelines and cancer incidence, cancer mortality, and total mortality: a prospective cohort study. Am J Clin Nutr, 101(3), 558–​569. PMCID: PMC4340061. Kennedy DA, Stern SJ, Moretti M, et  al. 2011. Folate intake and the risk of colorectal cancer:  a systematic review and meta-​ analysis. Cancer Epidemiol, 35(1), 2–​10. PMID: 21177150. Key TJ, Appleby PN, Cairns BJ, et al. 2011. Dietary fat and breast cancer: comparison of results from food diaries and food-​frequency questionnaires in the UK Dietary Cohort Consortium. Am J Clin Nutr, 94(4), 1043–​1052. PMID: 21865329.

346

346

PART III:  The Causes of Cancer

Key TJ, Appleby PN, Crowe FL, et  al. 2014. Cancer in British vegetarians: updated analyses of 4998 incident cancers in a cohort of 32,491 meat eaters, 8612 fish eaters, 18,298 vegetarians, and 2246 vegans. Am J Clin Nutr, 100 (Suppl 1), 378S–​385S. PMCID: PMC4144109. Key TJ, Appleby PN, Reeves GK, Roddam AW, and Endogenous Hormones Breast Cancer Collaborative Group. 2010. Insulin-​like growth factor 1 (IGF1), IGF binding protein 3 (IGFBP3), and breast cancer risk: pooled individual data analysis of 17 prospective studies. Lancet Oncol, 11(6), 530–​542. PMCID: PMC3113287. Key TJ, Appleby PN, Travis RC, et al. 2015. Carotenoids, retinol, tocopherols, and prostate cancer risk: pooled analysis of 15 studies. Am J Clin Nutr, 102(5), 1142–​1157. PMCID: PMC4625592. Kim EH, Willett WC, Fung T, Rosner B, and Holmes MD. 2011. Diet quality indices and postmenopausal breast cancer survival. Nutr Cancer, 63(3), 381–​388. PMCID: PMC3113520. Kim Y, and Je Y. 2014. Vitamin D intake, blood 25(OH)D levels, and breast cancer risk or mortality: a meta-​analysis. Br J Cancer, 110(11), 2772–​2784. PMCID: PMC4037823. Kirkpatrick SI, Subar AF, Douglass D, et al. 2014. Performance of the automated self-​ administered 24-​ hour recall relative to a measure of true intakes and to an interviewer-​administered 24-​h recall. Am J Clin Nutr, 100(1), 233–​240. PMCID: PMC4144101. Kjellstrom T, and Nordberg GF. 1978. A kinetic model of cadmium metabolism in the human being. Environ Res, 16(1–​3), 248–​269. PMID: 679914. Klein EA, Thompson IM, Jr., Tangen CM, et al. 2011. Vitamin E and the risk of prostate cancer: the Selenium and Vitamin E Cancer Prevention Trial (SELECT). JAMA, 306(14), 1549–​1556. Klein EA, Thompson IM, Lippman SM, et al. 2001. SELECT: the next prostate cancer prevention trial. Selenum and Vitamin E Cancer Prevention Trial. J Urol, 166(4), 1311–​1315. PMID: 11547064. Kristal AR, Till C, Song X, et  al. 2014. Plasma vitamin d and prostate cancer risk:  results from the selenium and vitamin e cancer prevention trial. Cancer Epidemiol Biomarkers Prev, 23(8), 1494–​ 1504. PMCID: PMC4119495. Kushi LH, Doyle C, McCullough M, et  al. 2012. American Cancer Society Guidelines on nutrition and physical activity for cancer prevention: reducing the risk of cancer with healthy food choices and physical activity. CA Cancer J Clin, 62(1), 30–​67. PMID: 22237782. Kyro C, Olsen A, Bueno-​de-​Mesquita HB, et al. 2014a. Plasma alkylresorcinol concentrations, biomarkers of whole-​grain wheat and rye intake, in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Br J Nutr, 111(10), 1881–​1890. PMID: 24521535. Kyro C, Olsen A, Landberg R, et al. 2014b. Plasma alkylresorcinols, biomarkers of whole-​grain wheat and rye intake, and incidence of colorectal cancer. J Natl Cancer Inst, 106(1), djt352. PMCID: PMC3906988. Lampe JW. 2009. Is equol the key to the efficacy of soy foods? Am J Clin Nutr, 89(5), 1664S–​1667S. PMCID: PMC2677009. Lamprecht SA, and Lipkin M. 2001. Cellular mechanisms of calcium and vitamin D in the inhibition of colorectal carcinogenesis. Ann NY Acad Sci, 952, 73–​87. Lappe JM, Travers-​Gustafson D, Davies KM, Recker RR, and Heaney RP. 2007. Vitamin D and calcium supplementation reduces cancer risk: results of a randomized trial. Am J Clin Nutr, 85(6), 1586–​1591. Larsson SC, Bergkvist L, and Wolk A. 2009. Long-​term dietary calcium intake and breast cancer risk in a prospective cohort of women. Am J Clin Nutr, 89(1), 277–​282. PMID: 19056569. Larsson SC, Orsini N, and Wolk A. 2015. Urinary cadmium concentration and risk of breast cancer: a systematic review and dose-​response meta-​ analysis. Am J Epidemiol, 182(5), 375–​380. PMID: 26254432. Lauby-​Secretan B, Scoccianti C, Loomis D, et  al. 2016. Body Fatness and Cancer—​Viewpoint of the IARC Working Group. N Engl J Med, 375, 794–​798. Lee I, Cook NR, Manson JE, Buring JE, and Hennekens CH. 1999. B-​carotene supplementation and incideence of cancer and cardiovascular disease: the women’s health study. J Natl Cancer Inst, 91(24), 2102–​2106. Lee JE, Hunter DJ, Spiegelman D, et al. 2007. Intakes of coffee, tea, milk, soda and juice and renal cell cancer in a pooled analysis of 13 prospective studies. Int J Cancer. 121(10), 2246–​2253. PMID: 17583573. Lee JE, Li H, Chan AT, et  al. 2011a. Circulating levels of vitamin D and colon and rectal cancer:  the Physicians’ Health Study and a meta-​ analysis of prospective studies. Cancer Prev Res (Phila), 4(5), 735–​743. PMCID: PMC3412303. Lee JE, Mannisto S, Spiegelman D, et  al. 2009. Intakes of fruit, vegetables, and carotenoids and renal cell cancer risk: a pooled analysis of 13 prospective studies. Cancer Epidemiol Biomarkers Prev, 18(6), 1730–​1739. PMID: 19505906.

Lee JE, Wei EK, Fuchs CS, et al. 2012. Plasma folate, methylenetetrahydrofolate reductase (MTHFR), and colorectal cancer risk in three large nested case-​control studies. Cancer Causes Control, 23(4), 537–​545. PMCID: PMC3721151. Lee JE, Willett WC, Fuchs CS, et al. 2011b. Folate intake and risk of colorectal cancer and adenoma: modification by time. Am J Clin Nutr, 93(4), 817–​ 825. PMCID: PMC3057549. Lee SA, Shu XO, Li H, et al. 2009. Adolescent and adult soy food intake and breast cancer risk: results from the Shanghai Women’s Health Study. Am J Clin Nutr, 89(6), 1920–​1926. PMCID: PMC2683002. Li G, Ma D, Zhang Y, Zheng W, and Wang P. 2013. Coffee consumption and risk of colorectal cancer: a meta-​analysis of observational studies. Public Health Nutr, 16(2), 346–​357. PMID: 22694939. Li H, Stampfer MJ, Giovannucci EL, et al. 2004. A prospective study of plasma selenium levels and prostate cancer risk. J Natl Cancer Inst, 96(9), 696–​ 703. PMID: 15126606. Liese AD, Krebs-​Smith SM, Subar AF, et  al. 2015. The Dietary Patterns Methods Project:  synthesis of findings across cohorts and relevance to dietary guidance. J Nutr, 145(3), 393–​402. PMCID: PMC4336525. Linos E, Willett WC, Cho E, and Frazier L. 2010. Adolescent diet in relation to breast cancer risk among premenopausal women. Cancer Epidemiol Biomarkers Prev, 19(3), 689–​696. PMCID: PMC2866323. Lippman SM, Klein EA, Goodman PJ, et al. 2009. Effect of selenium and vitamin E on risk of prostate cancer and other cancers:  the Selenium and Vitamin E Cancer Prevention Trial (SELECT). JAMA, 301(1), 39–​51. PMCID: PMC3682779. Liu C, Russell RM, and Wang X. 2003. Exposing ferrets to cigarette smoke and a pharmacologic dose of beta-​carotene supplementation enhance in vitro retinoic acid catabolism in lungs via induction of cytochrome P450 enzymes. J Nutr, 133, 173–​179. Loomis D, Guyton KZ, Grosse Y, et  al. 2016. Carcinogenicity of drinking coffee, mate, and very hot beverages. Lancet, 17(7), 877–​ 878. PMID: 27318851 Lovegrove C, Ahmed K, Challacombe B, et  al. 2015. Systematic review of prostate cancer risk and association with consumption of fish and fish-​ oils:  analysis of 495,321 participants. Int J Clin Pract, 69(1), 87–​105. PMID: 25495842. Ma Y, Zhang P, Wang F, et al. 2011. Association between vitamin D and risk of colorectal cancer:  a systematic review of prospective studies. J Clin Oncol, 29(28), 3775–​3782. PMID: 21876081. Magalhaes B, Bastos J, and Lunet N. 2011. Dietary patterns and colorectal cancer:  a case-​control study from Portugal. Eur J Cancer Prev, 20(5), 389–​395. PMID: 21558858. Maiani G, Caston MJ, Catasta G, et  al. 2009. Carotenoids:  actual knowledge on food sources, intakes, stability and bioavailability and their protective role in humans. Mol Nutr Food Res, 53 Suppl 2, S194–​218. PMID: 19035552. Mannisto S, Smith-​Warner SA, Spiegelman D, et al. 2004. Dietary carotenoids and risk of lung cancer in a pooled analysis of seven cohort studies. Cancer Epidemiol Biomarkers Prev, 13, 40–​48. Manson JE, and Bassuk SS. 2015. Vitamin D research and clinical practice: at a crossroads. JAMA, 313(13), 1311–​1312. PMID: 25695911. Manson JE, Bassuk SS, Lee IM, et  al. 2012. The VITamin D and OmegA-​3 TriaL (VITAL):  rationale and design of a large randomized controlled trial of vitamin D and marine omega-​3 fatty acid supplements for the primary prevention of cancer and cardiovascular disease. Contemp Clin Trials, 33(1), 159–​171. PMCID: PMC3253961. Martin LJ, Li Q, Melnichouk O, et  al. 2011. A randomized trial of dietary intervention for breast cancer prevention. Cancer Res, 71(1), 123–​133. PMID: 21199800. McCann SE, Thompson LU, Nie J, et al. 2010. Dietary lignan intakes in relation to survival among women with breast cancer: the Western New York Exposures and Breast Cancer (WEB) Study. Breast Cancer Res Treat, 122(1), 229–​235. PMCID: PMC2883619. McCullough ML, Patel AV, Kushi LH, et al. 2011. Following cancer prevention guidelines reduces risk of cancer, cardiovascular disease, and all-​ cause mortality. Cancer Epidemiol Biomarkers Prev, 20(6), 1089–​1097. PMID: 21467238. McCullough ML, Weinstein SJ, Freedman DM, et al. 2010. Correlates of circulating 25-​hydroxyvitamin D: Cohort Consortium Vitamin D Pooling Project of Rarer Cancers. Am J Epidemiol, 172(1), 21–​35. PMCID: PMC2892536. Menkes MS, Comstock GW, Vuilleumier JP, et al. 1986. Serum beta-​carotene, vitamins A and E, selenium, and the risk of lung cancer. N Engl J Med, 315(20), 1250–​1254. PMID: 3773937. Menzel P. 2005. Hungry planet family food portraits.Available from: http://m ​ enzelphoto.photoshelter.com/​gallery/​Hungry-​Planet-​Family-​Food-​Portraits/​

 347



Diet and Nutrition

G0000zmgWvU6SiKM/​C0000k7JgEHhEq0w. Accessed on March 4, 2016. Messina MJ, and Wood CE. 2008. Soy isoflavones, estrogen therapy, and breast cancer risk:  analysis and commentary. Nutr J, 7, 17. PMCID: PMC2443803. Meyerhardt JA, Niedzwiecki D, Hollis D, et al. 2007. Association of dietary patterns with cancer recurrence and survival in patients with stage III colon cancer. JAMA, 298(7), 754–​764. PMID: 17699009. Miller JC. 1994. Importance of glycemic index in diabetes. Am J Clin Nutr, 59(3 Suppl), 747S-​752S. PMID: 8116560. Missmer SA, Smith-​Warner SA, Spiegelman D, et  al. 2002. Meat and dairy food consumption and breast cancer: a pooled analysis of cohort studies. Int J Epidemiol, 31, 78–​85. Moyer VA, and USPS Task Force. 2014. Vitamin, mineral, and multivitamin supplements for the primary prevention of cardiovascular disease and cancer: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med, 160(8), 558–​564. PMID: 24566474. Mozaffarian D, Hao T, Rimm EB, Willett WC, and Hu FB. 2011. Changes in diet and lifestyle and long-​term weight gain in women and men. N Engl J Med, 364(25), 2392–​2404. PMCID: PMC3151731. Mulholland HG, Murray LJ, Cardwell CR, and Cantwell MM. 2009. Glycemic index, glycemic load, and risk of digestive tract neoplasms:  a systematic review and meta-​ analysis. Am J Clin Nutr, 89(2), 568–​ 576. PMID: 19088152. Murphy N, Norat T, Ferrari P, et  al. 2012. Dietary fibre intake and risks of cancers of the colon and rectum in the European prospective investigation into cancer and nutrition (EPIC). PLoS One, 7(6), e39361. PMCID: PMC3382210. Nagata C. 2010. Factors to consider in the association between soy isoflavone intake and breast cancer risk. J Epidemiol, 20(2), 83–​89. PMCID: PMC3900805. Nair SC, Toshkov IA, Yaktine AL, et  al. 1995. Dietary energy restriction-​ induced modulation of protein kinase C zeta isozyme in the hamster pancreas. Mol Carcinog, 14(1), 10–​15. PMID: 7546218. Nan H, Lee JE, Rimm EB, et  al. 2013. Prospective study of alcohol consumption and the risk of colorectal cancer before and after folic acid fortification in the United States. Ann Epidemiol, 23(9), 558–​ 563. PMCID: PMC3749268. Neuhouser ML, Platz EA, Till C, et  al. 2013. Insulin-​like growth factors and insulin-​ like growth factor-​ binding proteins and prostate cancer risk:  results from the prostate cancer prevention trial. Cancer Prev Res (Phila), 6(2), 91–​99. PMCID: PMC3565024. Ng K, Wolpin BM, Meyerhardt JA, et al. 2009. Prospective study of predictors of vitamin D status and survival in patients with colorectal cancer. Br J Cancer, 101, 916–​923. Nomura AM, Lee J, Stemmermann GN, and Combs GF, Jr. 2000. Serum selenium and subsequent risk of prostate cancer. Cancer Epidemiol Biomarkers Prev, 9(9), 883–​887. PMID: 11008904. O'Keefe SJD, Li JV, Lahti L, et al. 2015. Fat, Fiber and Cancer Risk in African Americans and Rural Africans. Nat Commun, 6, 6342. Ollberding NJ, Wilkens LR, Henderson BE, Kolonel LN, and Le Marchand L. 2012. Meat consumption, heterocyclic amines and colorectal cancer risk:  the Multiethnic Cohort Study. Int J Cancer, 131(7), E1125–​1133. PMCID: PMC3553660. Omenn GS, Goodman GE, Thornquist MD, et al. 1996. Effects of a combination of beta carotene and vitamin A  on lung cancer and cardiovascular disease. N Engl J Med, 334, 1150–​1155. Ong TP, Moreno FS, and Ross SA. 2011. Targeting the epigenome with bioactive food components for cancer prevention. J Nutrigenet Nutrigenomics, 4(5), 275–​292. PMCID: PMC3388269. Orlich MJ, Singh PN, Sabate J, et  al. 2015. Vegetarian dietary patterns and the risk of colorectal cancers. JAMA Intern Med, 175(5), 767–​ 776. PMCID: PMC4420687. Panel on Dietary Antioxidants and Related Compounds. 2000. Dietary reference intakes for vitamin C, vitamin E, selenium, and carotenoids. Washington, DC: National Academies Press. Park Y, Hunter DJ, Spiegelman D, et al. 2005. Dietary fiber intake and risk of colorectal cancer: a pooled analysis of prospective cohort studies. JAMA, 294(22), 2849–​2857. PMID: 16352792. Parkin DM. 2011. The fraction of cancer attributable to lifestyle and environmental factors in the UK in 2010. Br J Cancer, 105(Suppl 2), S2–​5. PMCID: PMC3252063. Pelucchi C, La Vecchia C, Bosetti C, Boyle P, and Boffetta P. 2011. Exposure to acrylamide and human cancer: a review and meta-​analysis of epidemiologic studies. Ann Oncol, 22(7), 1487–​1499. PMID: 21239401.

347

Pereira MA, Jacobs DR, Jr, Pins JJ, et al. 2002. Effect of whole grains on insulin sensitivity in overweight hyperinsulinemic adults. Am J Clin Nutr, 75(5), 848–​855. Peterson JJ, Dwyer JT, Jacques PF, and McCullough ML. 2015. Improving the estimation of flavonoid intake for study of health outcomes. Nutr Rev, 73(8), 553–​576. PMCID: PMC4581481. Petrick JL, Freedman ND, Graubard BI, et al. 2015. Coffee consumption and risk of hepatocellular carcinoma and intrahepatic cholangiocarcinoma by sex:  The Liver Cancer Pooling Project. Cancer Epidemiol Biomarkers Prev, 24(9), 1398–​1406. PMCID: PMC4576990. Pierce JP, Natarajan L, Caan BJ, et  al. 2007. Influence of a diet very high in vegetables, fruit, and fiber and low in fat on prognosis following treatment for breast cancer:  the Women’s Healthy Eating and Living (WHEL) randomized trial. JAMA, 298(3), 289–​ 298. PMCID: PMC2083253. Prentice RL, Caan B, Chlebowski RT, et al. 2006. Low-​fat dietary pattern and risk of invasive breast cancer: the Women’s Health Initiative Randomized Controlled Dietary Modification Trial. JAMA, 295(6), 629–​ 642. PMID: 16467232 Prentice RL, Kakar F, Hursting S, et  al. 1988. Aspects of the rationale for the Women’s Health Trial. J Natl Cancer Inst, 80(11), 802–​814. PMID: 3292773. Produce for Better Health Foundation. State of the Plate. 2015 study on America’s consumption of fruit and vegetables. Produce for Better Health Foundation, 2015. Web. Qiao YL, Dawsey SM, Kamangar F, et al. 2009. Total and cancer mortality after supplementation with vitamins and minerals:  follow-​up of the Linxian General Population Nutrition Intervention Trial. J Natl Cancer Inst, 101(7), 507–​518. PMCID: PMC2664089. Rich-​Edwards JW, Ganmaa D, Pollak MN, et al. 2007. Milk consumption and the prepubertal somatotropic axis. Nutr J, 6, 28. PMCID: PMC2098760. Richman EL, Kenfield SA, Chavarro JE, et al. 2013. Fat intake after diagnosis and risk of lethal prostate cancer and all-​cause mortality. JAMA Intern Med, 173(14), 1318–​1326. PMCID: PMC3935610. Rimm EB, Giovannucci EL, Stampfer MJ, et  al. 1992. Reproducibility and validity of a expanded self-​administered semiquantitative food frequency questionnaire among male health professionals. Am J Epidemiol, 135, 1114–​1126. Rochester JR. 2013. Bisphenol A and human health: a review of the literature. Reprod Toxicol, 42, 132–​155. PMID: 23994667. Rock CL, Doyle C, Demark-​Wahnefried W, et al. 2012. Nutrition and physical activity guidelines for cancer survivors. CA Cancer J Clin, 62(4), 242–​ 274. PMID: 22539238. Romagnolo DF, and Selmin OI. 2012. Flavonoids and cancer prevention:  a review of the evidence. J Nutr Gerontol Geriatr, 31(3), 206–​ 238. PMID: 22888839. Romaguera D, Vergnaud AC, Peeters PH, et al. 2012. Is concordance with World Cancer Research Fund/​American Institute for Cancer Research guidelines for cancer prevention related to subsequent risk of cancer? Results from the EPIC study. Am J Clin Nutr, 96(1), 150–​163. PMID: 22592101. Ross MH, and Bras G. 1971. Lasting influence of early caloric restriction on prevalence of neoplasms in the rat. J Natl Cancer Inst, 47(5), 1095–​1113. PMID: 4330798. Rothwell JA, Perez-​Jimenez J, Neveu V, et  al. 2013. Phenol-​Explorer 3.0:  a major update of the Phenol-​Explorer database to incorporate data on the effects of food processing on polyphenol content. Database (Oxford), 2013, bat070. PMCID: PMC3792339. Rowlands MA, Gunnell D, Harris R, et al. 2009. Circulating insulin-​like growth factor peptides and prostate cancer risk: a systematic review and meta-​ analysis. Int J Cancer, 124(10), 2416–​2429. PMCID: PMC2743036. Saarinen NM, Tuominen J, Pylkkanen L, and Santti R. 2010. Assessment of information to substantiate a health claim on the prevention of prostate cancer by lignans. Nutrients, 2(2), 99–​115. PMCID: PMC3257165. Scalbert A, Brennan L, Manach C, et  al. 2014. The food metabolome:  a window over dietary exposure. Am J Clin Nutr, 99(6), 1286–​ 1308. PMID: 24760973. Schatzkin A, Lanza E, Corle D, et al. 2000. Lack of effect of a low-​fat, high-​ fiber diet on the recurrence of colorectal adenomas. N Engl J Med, 342(16), 1149–​1155. Schatzkin A, Mouw T, Park Y, et al. 2007. Dietary fiber and whole-​grain consumption in relation to colorectal cancer in the NIH-​AARP Diet and Health Study. Am J Clin Nutr, 85(5), 1353–​1360. PMID: 17490973. Schatzkin A, Park Y, Leitzmann MF, Hollenbeck AR, and Cross AJ. 2008. Prospective study of dietary fiber, whole grain foods, and small intestinal cancer. Gastroenterology, 135(4), 1163–​1167. PMID: 18727930.

348

348

PART III:  The Causes of Cancer

Schwartz GG, and Skinner HG. 2012. A prospective study of total and ionized serum calcium and time to fatal prostate cancer. Cancer Epidemiol Biomarkers Prev, 21(10), 1768–​1773. PMID: 22914529. Schwingshackl L, and Hoffmann G. 2015. Adherence to Mediterranean diet and risk of cancer:  an updated systematic review and meta-​ analysis of observational studies. Cancer Med, 4(12), 1933–​ 1947. PMID: 26471010. Setchell KD, Brown NM, Zhao X, et  al. 2011. Soy isoflavone phase II metabolism differs between rodents and humans:  implications for the effect on breast cancer risk. Am J Clin Nutr, 94(5), 1284–​1294. PMCID: PMC3192476. Shai I, Schwarzfuchs D, Henkin Y, et  al. 2008. Weight loss with a low-​ carbohydrate, Mediterranean, or Low-​fat diet. N Engl J Med, 359(3), 229–​241. Shin M, Holmes MD, Hankinson SE, et al. 2002. Intake of dairy products, calcium, and vitamin D and risk of breast cancer. J Natl Cancer Inst, 94(17), 1301–​1311. PMID: 12208895 Sieri S, Chiodini P, Agnoli C, et  al. 2014. Dietary fat intake and development of specific breast cancer subtypes. J Natl Cancer Inst, 106(5). PMID: 24718872. Sinha R, Knize MG, Salmon CP, et al. 1998a. Heterocyclic amine content of pork products cooked by different methods and to varying degrees of doneness. Food Chem Toxicol, 36(4), 289–​297. PMID: 9651045. Sinha R, Rothman N, Salmon CP, et al. 1998b. Heterocyclic amine content in beef cooked by different methods to varying degrees of doneness and gravy made from meat drippings. Food Chem Toxicol, 36(4), 279–​287. PMID: 9651044. Smith-​Warner SA, Spiegelman D, Adami HO, et  al. 2001. Types of dietary fat and breast cancer: a pooled analysis of cohort studies. Int J Cancer, 92(5), 767–​774. Smith-​Warner SA, Spiegelman D, Ritz J, et  al. 2006. Methods for pooling results of epidemiologic studies:  the Pooling Project of Prospective Studies of Diet and Cancer. Am J Epidemiol, 163(11), 1053–​ 1064. PMID: 16624970. Smith-​Warner SA, Spiegelman D, Yaun SS, et  al. 1998. Alcohol and breast cancer in women:  a pooled analysis of cohort studies. JAMA, 279(7), 535–​540. Stahelin HB, Rosel F, Buess E, and Brubacher G. 1984. Cancer, vitamins, and plasma lipids: prospective Basel study. J Natl Cancer Inst, 73(6), 1463–​ 1468. PMID: 6595456. State of California. 2015. Notice of intent to list:  tetrachlorvinphos, parathion, malathion, glyphosate. Available from: http://​oehha. ca.gov/ ​ p roposition- ​ 6 5/ ​ c rnr/ ​ n otice- ​ i ntent- ​ l ist- ​ t etrachlorvinphos​parathion-​malathion-​glyphosate. Stevens VL, McCullough ML, Sun J, et al. 2011. High levels of folate from supplements and fortification are not associated with increased risk of colorectal cancer. Gastroenterology, 141(1), 98–​ 105, 105 e101. PMID: 21586288. Stolzenberg-​Solomon RZ, Hayes RB, Horst RL, et al. 2009. Serum vitamin D and risk of pancreatic cancer in the prostate, lung, colorectal, and ovarian screening trial. Cancer Res, 69(4), 1439–​1447. PMCID: PMC3052200 Sun Z, Chen T, Thorgeirsson SS, et al. 2013. Dramatic reduction of liver cancer incidence in young adults: 28 year follow-​up of etiological interventions in an endemic area of China. Carcinogenesis, 34(8), 1800–​1805. PMCID: PMC3731800. Szymanski KM, Wheeler DC, and Mucci LA. 2010. Fish consumption and prostate cancer risk: a review and meta-​analysis. Am J Clin Nutr, 92(5), 1223–​1233. PMID: 20844069. Tannenbaum A. 1942. The genesis and growth of tumors. III: Effects of a high fat diet. Cancer Res, 2, 468–​475. Tannenbaum A, and Silverstone H. 1953. Nutrition in relation to cancer. Adv Cancer Res, 1, 451–​501. Tannour-​Louet M, Lewis SK, Louet JF, et  al. 2014. Increased expression of CYP24A1 correlates with advanced stages of prostate cancer and can cause resistance to vitamin D3-​based therapies. FASEB J, 28(1), 364–​372. PMID: 24081904. Teegarden D, Romieu I, and Lelievre SA. 2012. Redefining the impact of nutrition on breast cancer incidence: is epigenetics involved? Nutr Res Rev, 25(1), 68–​95. PMCID: PMC4090752. The Alpha-​Tocopherol Beta-​Carotene Cancer Prevention Study Group. 1994. The effect of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers. N Engl J Med, 330, 1029–​1035. The Women’s Health Initiative Study Group. 1998. Design of the Women’s Health Initiative clinical trial and observational study. Control Clin Trials, 19(1), 61–​109. Thomson CA, McCullough ML, Wertheim BC, et  al. 2014a. Nutrition and physical activity cancer prevention guidelines, cancer risk, and mortality

in the Women’s Health Initiative. Cancer Prev Res (Phila), 7(1), 42–​53. PMID: 24403289. Thomson CA, Van Horn L, Caan BJ, et al. 2014b. Cancer incidence and mortality during the intervention and postintervention periods of the Women’s Health Initiative dietary modification trial. Cancer Epidemiol Biomarkers Prev, 23(12), 2924–​2935. PMCID: PMC4257879. Thun MT, Peto R, Lopez AD, et al. 1997. Alcohol consumption and mortality among middle-​aged and elderly U.S. adults. N Engl J Med, 337(24), 1705–​1714. Tobias DK, Chen M, Manson JE, et al. 2015. Effect of low-​fat diet interventions versus other diet interventions on long-​term weight change in adults: a systematic review and meta-​analysis. Lancet Diabetes Endocrinol, 3(12), 968–​979. PMCID: PMC4667723. Toledo E, Salas-​Salvado J, Donat-​Vargas C, et  al. 2015. Mediterranean diet and invasive breast cancer risk among women at high cardiovascular risk in the PREDIMED Trial: a randomized clinical trial. JAMA Intern Med, 175(11), 1752–​1760. PMID: 26365989. Trichopoulou A, Costacou T, Bamia C, and Trichopoulos D. 2003. Adherence to a Mediterranean diet and survival in a Greek population. N Engl J Med, 348(26), 2599–​2608. Trichopoulou A, Martinez-​Gonzalez MA, Tong TY, et al. 2014. Definitions and potential health benefits of the Mediterranean diet:  views from experts around the world. BMC Med, 12, 112. PMCID: PMC4222885. Trock B, Lanza E, and Greenwald P. 1990. Dietary fiber, vegetables, and colon cancer: critical review and meta-​analyses of the epidemiologic evidence. J Natl Cancer Inst, 82, 650–​661. US Department of Health and Human Services and US Department of Agriculture. 2015. 2015–​2020 dietary guidelines for Americans (8th ed.). December 2015. Available at http://​health.gov/​dietaryguidelines/​2015/​ guidelines/​. US Dietary Guidelines Advisory Committee. 2015. Scientific report of the 2015 US Dietary Guidelines Advisory Committee. USDA. Available at: https://​ health.gov/​ d ietaryguidelines/​ 2 015-​ s cientific-​ r eport/​ p dfs/​ s cientific-​ report-​of-​the-​2015-​dietary-​guidelines-​advisory-​committee.pdf Van Blarigan EL, Kenfield SA, Yang M, et al. 2015. Fat intake after prostate cancer diagnosis and mortality in the Physicians’ Health Study. Cancer Causes Control, 26(8), 1117–​1126. PMCID: PMC4499006. Van Blarigan EL, and Meyerhardt JA. 2015. Role of physical activity and diet after colorectal cancer diagnosis. J Clin Oncol, 33(16), 1825–​1834. PMCID: PMC4438267. Van Dam RM, and Hunter DJ. 2013. Biochemical indicators of dietary intake. In:  Willett WC (Ed.), Nutritional epidemiology (3rd ed.). New York: Oxford University Press. Vartanian LR, Schwartz MB, and Brownell KD. 2007. Effects of soft drink consumption on nutrition and health: a systematic review and meta-​analysis. Am J Public Health, 97(4), 667–​675. PMCID: PMC1829363. Vastag B. 2007. Soy and prostate cancer study results mixed. J Natl Cancer Inst, 99(18), 1364–​1365. PMID: 17848664. Velentzis LS, Cantwell MM, Cardwell C, et al. 2009. Lignans and breast cancer risk in pre-​and post-​menopausal women: meta-​analyses of observational studies. Br J Cancer, 100(9), 1492–​1498. PMCID: PMC2694431. Vergnaud AC, Romaguera D, Peeters PH, et al. 2013. Adherence to the World Cancer Research Fund/​American Institute for Cancer Research guidelines and risk of death in Europe: results from the European Prospective Investigation into Nutrition and Cancer cohort study. Am J Clin Nutr, 97(5), 1107–​1120. PMID: 23553166. Virk-​Baker MK, Nagy TR, Barnes S, and Groopman J. 2014. Dietary acrylamide and human cancer: a systematic review of literature. Nutr Cancer, 66(5), 774–​790. PMCID: PMC4164905. Visvanathan K, Mondul A, Zeleniuch-​Jacquotte A, et  al. 2015. Circulating vitamin D concentrations and breast cancer risk: a pooled analysis of 17 cohorts [abstract]. Paper presented at the Thirty-​Seventh Annual CTRC-​ AACR San Antonio Breast Cancer Symposium, 2014, San Antonio, TX. Vogt TM, Ziegler RG, Graubard BI, et al. 2003. Serum selenium and risk of prostate cancer in U.S. blacks and whites. Int J Cancer, 103(5), 664–​670. PMID: 12494476. Wactawski-​Wende J, Kotchen JM, Anderson GL, et al. 2006. Calcium plus vitamin D supplementation and the risk of colorectal cancer. N Eng J Med, 354(7), 684–​696. Wang L, Lee IM, Zhang SM, et al. 2009. Dietary intake of selected flavonols, flavones, and flavonoid-​rich foods and risk of cancer in middle-​aged and older women. Am J Clin Nutr, 89(3), 905–​912. PMCID: PMC2667658. Wang XD, and Russell R. 1999. Procarcinogenic and anticarcinogenic effects of beta-​carotene. Nutr Rev, 57, 263–​272. WCRF International. 2015. Cancer preventability estimates for diet, nutrition, body fatness, and physical activity. Available at:  http://​www.wcrf.org/​

 349



Diet and Nutrition

int/​cancer-​facts-​figures/​preventability-​estimates/​cancer-​preventability-​ estimates-​diet-​nutrition. Accessed on October 20, 2016. WCRF/​AICR. 1997. Food, nutrition, and the prevention of cancer:  a global perspective. Washington, DC: AICR. WCRF/​AICR. 2007. Food, nutrition, physical activity, and the prevention of cancer: a global perspective. Washington DC: AICR. WCRF/​AICR. 2010. Continuous update project report: food, nutrition, physical activity, and the prevention of breast cancer. London; Washington, DC: WCRF/​AICR. WCRF/​AICR. 2011. Continuous update project report: food, nutrition, physical activity and the prevention of colorectal cancer. London; Washington, DC:: WCRF/​AICR. WCRF/​AICR. 2013. Continuous update project report: food, nutrition, physical activity, and the prevention of endometrial cancer. London; Washington, DC: WCRF/​AICR. WCRF/​AICR. 2014a. Continuous update project report:  diet, nutrition, physical activity, and breast cancer survivors. London; Washington, DC: WCRF/​AICR. WCRF/​AICR. 2014b. Continuous update project report: diet, nutrition, physical activity, and prevention of prostate cancer. London; Washington, DC: WCRF/​AICR. WCRF/​AICR. 2015a. Continuous update project report:  diet, nutrition, physical activity and prevention of kidney cancer. London; Washington, DC: WCRF/​AICR. WCRF/​AICR. 2015b. Continuous update project report:  diet, nutrition, physical activity and prevention of liver cancer. London; Washington, DC: WCRF/​AICR. Webb A, and McCullough ML. 2005. Dietary lignans: potential role in cancer prevention. Nutr Cancer, 51(2), 117–​131. Weindruch R, and Walford RL. 1982. Dietary restriction in mice beginning at 1  year of age:  effect on life-​span and spontaneous cancer incidence. Science, 215(4538), 1415–​1418. PMID: 7063854. Wells HF, Hodan F, and Buzby JC. 2008. Dietary assessment of major trends in U.S. food consumption, 1970–​2005. Economic Information Bulletin No. 33. US Department of Agriculture. Welsh JA, Sharma AJ, Grellinger L, and Vos MB. 2011. Consumption of added sugars is decreasing in the United States. Am J Clin Nutr, 94(3), 726–​734. PMCID: PMC3155936. Willett W. 2013. Genetics in dietary analyses nutritional epidemiology. New York: Oxford University Press. Willett WC. 2010. The WHI joins MRFIT: a revealing look beneath the covers. Am J Clin Nutr, 91(4), 829–​830. PMID: 20181816. Willett WC. 2013. Nutritional epidemiology (3rd ed.). New  York:  Oxford University Press. Willett WC, and Colditz GA. 2013. Vitamin A and lung cancer. In: Willett WC (Ed.), Nutritional epidemiology (3rd ed.). New York: Oxford University Press. Willett WC, Sampson L, Stampfer MJ, et  al. 1985. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol, 122, 51–​65. Wilson KM, Kasperzyk JL, Rider JR et  al. 2011. Coffee consumption and prostate cancer risk and progression in the Health Professionals Follow-​up Study. J Natl Cancer Inst, 103(11), 876–​884. PMCID: PMC3110172. Wilson KM, Shui IM, Mucci LA, and Giovannucci E. 2015. Calcium and phosphorus intake and prostate cancer risk: a 24-​y follow-​up study. Am J Clin Nutr, 101(1), 173–​183. PMCID: PMC4266887.

349

Wilson KM, Vesper HW, Tocco P, et al. 2009. Validation of a food frequency questionnaire measurement of dietary acrylamide intake using hemoglobin adducts of acrylamide and glycidamide. Cancer Causes Control, 20(3), 269–​278. PMCID: PMC3147248. Wolever TMS, and Jenkins DJA. 1986. The use of the glycemic index in predicting the blood glucose response to mixed meals. Am J Clin Nutr, 43, 167–​172. Wolpin BM, Ng K, Bao Y, et al. 2012. Plasma 25-​hydroxyvitamin D and risk of pancreatic cancer. Cancer Epidemiol Biomarkers Prev, 21(1), 82–​91. PMCID: PMC3253914. Wu X, Zhu X, and Xie M. 2015. Association between dietary cadmium exposure and breast cancer risk:  an updated meta-​analysis of observational studies. Med Sci Monit, 21, 769–​775. PMCID: PMC4371715. Xie W, Yang M, Chan J, et al. 2016. Association of genetic variations of selenoprotein genes, plasma selenium levels, and prostate cancer aggressiveness at diagnosis. Prostate, 76(7), 691–​699, PMID: 26847995. Xu Y, Shao X, Yao Y, et al. 2014. Positive association between circulating 25-​ hydroxyvitamin D levels and prostate cancer risk: new findings from an updated meta-​analysis. J Cancer Res Clin Oncol, 140(9), 1465–​1477. PMID: 24838848. Yang M, Kenfield SA, Van Blarigan EL, et al. 2015. Dietary patterns after prostate cancer diagnosis in relation to disease-​specific and total mortality. Cancer Prev Res (Phila), 8(6), 545–​551. PMCID: PMC4452958. Yang TO, Crowe F, Cairns BJ, Reeves GK, and Beral V. 2015. Tea and coffee and risk of endometrial cancer: cohort study and meta-​analysis. Am J Clin Nutr, 101(3), 570–​578. PMCID: PMC4340062. Yin L, Grandi N, Raum E, et  al. 2010. Meta-​analysis:  serum vitamin D and breast cancer risk. Eur J Cancer, 46, 2196–​2205. PMID: 20456946 Yoshizawa K, Willett WC, Morris SJ, et al. 1998. Study of prediagnostic selenium level in toenails and the risk of advanced prostate cancer. J Natl Cancer Inst, 90(16), 1219–​1224. Yuan JM, Sun C, and Bustler LM. 2011. Tea and cancer prevention. Pharmacol Res, 64(2), 123–​135. PMID: 21419224 Zamora-​Ros R, Agudo A, Lujan-​Barroso L, et  al. 2012. Dietary flavonoid and lignan intake and gastric adenocarcinoma risk in the European Prospective Investigation into Cancer and Nutrition (EPIC) study. Am J Clin Nutr, 96(6), 1398–​1408. PMID: 23076618. Zamora-​Ros R, Fedirko V, Trichopoulou A, et al. 2013. Dietary flavonoid, lignan and antioxidant capacity and risk of hepatocellular carcinoma in the European prospective investigation into cancer and nutrition study. Int J Cancer, 133(10), 2429–​2443. PMID: 23649669. Zamora-​Ros R, Sacerdote C, Ricceri F, et al. 2014. Flavonoid and lignan intake in relation to bladder cancer risk in the European Prospective Investigation into Cancer and Nutrition (EPIC) study. Br J Cancer, 111(9), 1870–​1880. PMCID: PMC4453722. Ziegler RG, Smith-​Warner SA, Fedirko V, et al: Circulating vitamin D and risk of breast and colorectal cancer: an international pooling project of 21 cohorts, 2015 Vitamin D Workshop. J Steroid Biochem and Mol Biol, in press Zeisel SH. 2007. Nutrigenomics and metabolomics will change clinical nutrition and public health practice: insights from studies on dietary requirements for choline. Am J Clin Nutr, 86(3), 542–​548. PMCID: PMC2430757. Zhao M, Li X, Xing C, and Zhou B. 2013. Association of methylenetetrahydrofolate reductase C677T and A1298C polymorphisms with colorectal cancer risk: a meta-​analysis. Biomed Rep, 1(5), 781–​791. PMCID: PMC3917732. Zhou Y, Tian C, and Jia C. 2012. A dose-​response meta-​analysis of coffee consumption and bladder cancer. Prev Med, 55(1), 14–​22. PMID: 22564775.

350

 351

20

Obesity and Body Composition NANA KEUM, MINGYANG SONG, EDWARD L. GIOVANNUCCI, AND A. HEATHER ELIASSEN

OVERVIEW

MEASURES OF BODY FAT AND BODY COMPOSITION

In 2014, an estimated 1.9 billion adults were either overweight (BMI 25–​29.9) or obese (BMI ≥30), worldwide. The so-​called obesity epidemic began in high-​income, English-​speaking countries in the early 1970s, but soon spread globally; worldwide, more than one-​third (38%) of all adults and 600,000 children under age 5 are overweight or obese, as are two-​thirds (69%) of adults in the United States. Excessive body fat is a major cause of type 2 diabetes, hypertension, and cardiovascular and liver disease, among other disorders, and has been designated a definite cause of at least 14 cancer sites:  breast (postmenopausal), colorectum, endometrium, esophagus (adenocarcinoma), gallbladder, kidney (renal cell), pancreas, gastric cardia, liver, ovary, prostate (advanced tumors), multiple myeloma, thyroid, and meningioma. A  causal association is considered possible for leukemia and lymphoma. Estimates of the proportion of cancers attributable to overweight and/​or obesity globally range from 6% for rectal cancer to 33% for esophageal cancer in men and from 4% for rectal cancer to 34% for endometrial cancer in women. Mechanisms by which adipose tissue are thought to promote tumor growth include the endocrine and metabolic effects of fat on sex hormones, growth factors, and inflammation, as well as local chemical or mechanical injury of gastrointestinal organs. Because of the high prevalence of excess adiposity worldwide and the difficulty that most individuals have in losing substantial amounts of weight and maintaining that weight loss, population-​based approaches will require policies and community actions that help individuals avoid excessive weight gain throughout life.

Individuals vary in their size and shape. Because of this variation, it is not surprising that no single measure of adiposity and body composition is ideal in all circumstances. The so-​called gold s​tandard approaches for measuring the amount and distribution of body fat in clinical settings involve imaging with computed tomography (CT) or magnetic resonance imaging (MRI). While these provide the most accurate measurements, the devices are expensive to transport and operate, limiting their use in large-​scale population studies. Various anthropometric measures have been developed that use self-​reported or measured information on height and weight and/​or waist and hip circumference to assess different aspects of body composition. These are widely used because of their simplicity and low cost. It is important to understand the specific significance of these measures, as well as their limitations, since they serve as the exposure variable(s) in epidemiologic studies.

INTRODUCTION The hypothesis that excess energy intake may contribute to cancer risk originated from experimental studies of animals in the 1940s, in which mice fed ad libitum developed larger and more numerous mammary tumors than mice on energy-​restricted diets (Sellers et al., 2007; Tannenbaum, 1940a, 1940b). As noted in Chapter 19, this finding has consistently been replicated in a wide variety of mammary and other tumor models (Birt et al., 1992; Nair et al., 1995; Ross and Bras, 1971; Weindruch and Walford, 1982); a 30% restriction in energy intake since birth reduces mammary tumors in laboratory animals by as much as 90% (Boissonneault et al., 1986). Epidemiologic studies, beginning in the 1970s, reported that overweight women were more likely to develop breast and endometrial cancer (Blitzer et  al., 1976; de Waard and Baanders-​van Halewijn, 1974). Large prospective studies subsequently observed associations between excess body weight and increased mortality from multiple types of cancer (Calle et al., 2003; Lew and Garfinkel, 1979). During the last 15–​20 years, the literature on cancer in relation to obesity and body composition has expanded exponentially. This chapter will focus primarily on epidemiologic studies published in the last 10–​15 years and on advances in understanding the relationship of obesity and body composition to cancer since the publication of the corresponding chapter in the third edition of Cancer Epidemiology and Prevention (Ballard-​Barbash et al., 2006).

Body Mass Index (BMI) BMI is the most commonly used measure of overall body fatness. BMI is defined as weight in kilograms (kg) divided by height in meters squared (m2). The advantages of BMI are that it provides a single measure of adiposity that is independent of height and that can be compared across populations (Gallagher et al., 1996). BMI is highly correlated (r = 0.82–​0.91) with absolute fat mass, calculated from percent body fat measured by densitometry, even after accounting for extraneous variations caused by age and sex (Spiegelman et al., 1992; Willett, 2013). Densitometry has long been considered a gold standard for measuring total body fat. The World Health Organization (WHO) has established criteria to classify individuals according to their BMI (WHO, 2015). Adults (age >18 years) with a BMI 10% weight loss is needed to increase adiponectin levels significantly (Fabian et  al., 2013; Madsen et  al., 2008). Weight loss as low as 2.5%–​5% of body weight has been shown to lower insulin levels and improve insulin sensitivity (Fabian et al., 2013; Kaaks et al., 2003; Lofgren et  al., 2005; Mason et  al., 2011; Ryan and Nicklas, 2004). The evidence for IGF-​1 is inconsistent (Kaaks et  al., 2003; Mason et al., 2013).

MECHANISMS OF CARCINOGENESIS Various mechanisms have been proposed by which excess body fat could affect cancer risk (Figure 20–​3). These can be classified broadly as hormonal (endocrine) effects from both sex steroid hormones and peptide metabolic hormones and the effects of chronic inflammation. Systemic and local inflammation can result from the hormonal effects of adipose tissue, or it can be caused by the chemical or mechanical injury, as in the context of gastroesophageal reflux or chronic gallstones. The effects of obesity on metabolic hormones are thought to affect multiple types of cancers. Effects on steroid hormones are thought to be more specific, affecting breast and endometrial cancer (and perhaps prostate and colon as well); local inflammation has been implicated at some, but not all, cancer sites. Fat cells secrete and receive endocrine signals and are a major site for the metabolism of sex steroids. It is difficult to separate the metabolic and endocrine functions of adipose tissue, since these pathways overlap.

Metabolic Peptide Hormones Insulin and IGF-​1

Insulin and IGF-​1 are closely related metabolic protein (peptide) hormones that are thought to mediate some of the effects of excess body fat on cancer risk. Excess adiposity, particularly visceral fat (Donohoe et  al., 2011; Wajchenberg, 2000), induces insulin resistance. The resulting hyperinsulinemia suppresses hepatic production of hormone binding proteins (e.g., insulin-​ like growth factor binding proteins [IGFBPs]) (Kaaks et al., 2002a). This increases the concentration of bioavailable IGF-​1, especially that which is not bound to IGFBP-​3 (van Kruijsdijk et al., 2009). The net consequence of excess adiposity on metabolic hormones is to increase concentrations of circulating

360

360

PART III:  THE CAUSES OF CANCER Bioavailable Sex Hormones Aromatase SHBG

Excessive Calorie Intake

Free Fatty Acids Central Adiposity

Insulin

Insulin Resistance

Leptin

Proliferation

Apoptosis

Adiponectin IGFBPs Bioavailable IGF-1

Genomic Instability

Inflammation

Figure 20–​3.  Proposed mechanisms by which excess adiposity affects cancer risk.

insulin and bioavailable IGF-​1 (Kaaks et al., 2002a). Normal and cancerous cells express receptors for insulin and IGF-​1 (Pollak, 2008). Peptide hormones such as insulin and bioavailable IGF-​1 promote carcinogenesis by enhancing cell proliferation and inhibiting apoptosis (van Kruijsdijk et al., 2009). It should be noted that IGF-​1 elicits stronger mitogenic and anti-​apoptotic responses than insulin. Insulin may act primarily through enhancing the bioavailability of IGF-​1, rather than through direct ligand activity (Nimptsch and Giovannucci, 2012). Cancer cells may reactivate the expression of insulin receptor isoform A, which is normally expressed primarily in embryonic/​fetal tissue. The A  isoform is more mitogenic than the B isoform of the insulin receptor, which is typically expressed in adults (Pandini et al., 2003). Yu and Rohan have comprehensively reviewed the role of the IGF family on cancer development and progression (Yu and Rohan, 2000). Based on epidemiologic evidence, the insulin and/​or IGF-​1 pathways appear to be particularly relevant to cancers of the colorectum, breast, endometrium, and prostate. Indirect evidence linking IGF-​1 to colorectal cancer comes from the observation that taller individuals, who potentially had greater exposure to IGF-​1 during adolescent growth, are consistently associated with an increased risk of colorectal cancer (Green et al., 2011). Epidemiologic studies based on direct measurement of IGF-​1 are limited to measurements made in adults. Recent meta-​analyses of observational studies found that higher levels of circulating IGF-​1 levels in adulthood are associated with an increased risk of both colorectal cancer (Rinaldi et al., 2010) and high-​risk adenomas (Yoon et al., 2015a). Of note, adult IGF-​1 levels are an imperfect surrogate for adolescent IGF-​1 levels. Whereas height is strongly correlated with IGF-​1 levels in adolescence, it is only weakly correlated with levels in adults. However, Yoon and colleagues have inferred that both early-​life and later-​life IGF-​1 levels may contribute to the development and progression of colorectal cancer (Yoon et al., 2015b). Strong evidence supports a role for insulin in colorectal carcinogenesis. Studies of this issue have generally measured C-​peptide, a marker of insulin secretion, rather than insulin itself, because the longer half-​life of C-​peptide provides a more accurate marker of insulin activity. In a recent meta-​analysis of eight prospective studies, higher C-​peptide levels were associated with an approximately 40% increased risk of colorectal cancer (Chen et al., 2013). Furthermore, an elevated risk of colorectal neoplasia is associated with disorders that involve hyperinsulinemia, such as insulin resistance (as marked by high HOMA-​IR or by hypertriglyceridemia) and diabetes mellitus (Grundy, 1999; Yao and Tian, 2015; Yoon et al., 2015b). Interestingly, in some studies, individuals who have elevations in both C-​peptide and IGF-​1 levels have the same level of increased risk of colorectal cancer as those who have elevated levels of only one biomarker (Wei et al.,

2005; Wu et  al., 2011). This finding, if confirmed, suggests a common carcinogenic pathway shared by insulin and IGF-​1, whereby high levels of either insulin or IGF-​1 may be sufficient to achieve maximal effect. For example, colorectal cancer may possess hybrid insulin and IGF-​1 receptors, which can be stimulated and saturated by either insulin or IGF-​1. Adult height is also associated with breast cancer risk (Zhang B et al., 2015), potentially implicating the IGF-​1 pathway in the pathogenesis of breast as well as colorectal cancer. The Nurses’ Health Study was the first prospective study to examine prediagnostic concentrations of circulating IGF-​1 in relation to breast cancer risk. A  strong positive association was observed only in women who were premenopausal at the time of blood collection (Hankinson et al., 1998). In contrast, a pooled analysis of 17 prospective studies reported increased risk of breast cancer in women with higher circulating IGF-​1 concentrations, regardless of menopausal status at the time of blood collection. In fact, the positive association was stronger among postmenopausal than premenopausal women (Endogenous Hormones Breast Cancer Collaborative Group, 2010). Both this analysis and a subsequent study conducted in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort (Kaaks et  al., 2014)  observed the positive association with IGF-​1 only with ER+ breast cancer. This difference by ER status is supported by accumulating evidence of the presence of synergistic cross-​talk in mammary cells between IGF-​1 and estrogen signaling pathways (Yee and Lee, 2000). The role of insulin in breast cancer etiology remains unclear. Given that estrogens have a dominant influence on breast cancer (Colditz, 1998), it has been hypothesized that insulin may promote breast cancer by suppressing hepatic production of sex hormone binding globulins (SHBG), thereby increasing the bioavailability of estrogens (Singh et al., 1990). However, no overall association was observed between insulin or C-​peptide and breast cancer in a recent meta-​analysis of 10 prospective studies (Autier et al., 2013) Six of these studies measured insulin (Eliassen et al., 2007; Gunter et al., 2009; Kaaks et al., 2002b; Kabat et  al., 2009; Mink et  al., 2002; Sieri et  al., 2012); five measured C-​peptide (Cust et al., 2009; Eliassen et al., 2007; Keinan-​Boker et al., 2003; Toniolo et al., 2000; Verheus et al., 2006). Contrary to the hypothesis that estrogen mediates the increased risk, the most recent prospective study that examined the relationship between C-​peptide and subtypes of breast cancer found a stronger association with ER–​ than ER+ cancer (Ahern et al., 2013). Thus, the mechanism by which hyperinsulinemia increases breast cancer risk remains unclear. The relationship between excess body fat and endometrial cancer is thought to be driven by primarily by estrogen (Schmandt et al., 2011), although insulin may also play a role. Epidemiologic evidence supporting the link between insulin resistance and endometrial cancer

 361



Obesity and Body Composition

risk is comprehensively reviewed by Mu and Zhu (Mu et al., 2012). A recent meta-​analysis found an approximately 2-​fold increased risk of endometrial cancer among women with type 2 diabetes mellitus compared with non-​diabetic women (Liao et al., 2014). In a Mendelian randomization study that provides evidence for causality, single nucleotide polymorphisms (SNPs) for fasting insulin and SNPs for post-​ challenge insulin were linked to an elevated endometrial cancer risk (Nead et al., 2015). A recent meta-​analysis of studies of height and prostate cancer observed a modest positive association between adult attained height and prostate cancer risk (Zuccolo et  al., 2008). A  pooled study of 12 prospective studies that measured IGF-​1 concentrations in adults found a positive relationship between IGF-​ 1 and prostate cancer (Roddam et  al., 2008). The results are inconsistent, however, with respect to the association by stage and grade of prostate cancer. IGF-​ 1 appears to be more strongly associated with low-​grade than high-​ grade tumors (Chan et  al., 2002; Nimptsch et  al., 2011b; Roddam et  al., 2008), whereas the opposite is observed for both height and obesity in adulthood. Only a few prospective studies have examined the association between prediagnostic levels of circulating IGF-​1 and insulin and pancreatic cancer risk, and the results are conflicting (Giovannucci and Michaud, 2007). However, numerous studies have investigated the relationship between history of type 2 diabetes and pancreatic cancer risk. A recent meta-​analysis reported an elevated risk associated with type 2 diabetes, regardless of the duration of diabetes prior to cancer diagnosis (RR = 1.5, 95% CI: 1.3, 1.8 for 5–​9 years; RR = 1.5, 95% CI: 1.2, 2.0 for 10+ years) (Huxley et al., 2005).

Adiponectin

Another peptide hormone, adiponectin, possesses significant anti-​ inflammatory and insulin-​ sensitizing effects (Tilg and Moschen, 2006), and has been postulated to mediate the relationship between obesity and several cancers (Vansaun, 2013). Although predominantly secreted by white adipose tissue (Maeda et  al., 1996), adiponectin expression is reduced in obesity because of complex factors that suppress transcription, including adipocyte hypertrophy, inflammation, and oxidative stress (Kadowaki et  al., 2006; Kim et  al., 2015; Maury and Brichard, 2010). Lower levels of adiponectin have been linked to the development of metabolic syndrome and type 2 diabetes (Kadowaki et al., 2014; Li et al., 2009; Matsuzawa, 2006). The insulin-​ sensitizing effect of adiponectin may be mediated, at least in part, by an increase in fatty-​acid oxidation via activation of AMP-​activated protein kinase (AMPK) and peroxisome proliferator-​activated receptor (PPAR)-​ α, which bind to adiponectin receptors AdipoR1 and AdipoR2, respectively (Yamauchi et  al., 2002, 2003). Synthetic AdipoR agonist has been shown to ameliorate diabetes and prolong the life span in mice who are either genetically obese or become so on a high-​fat diet, suggesting that adiponectin can be a promising therapeutic target for metabolic disorders (Okada-​Iwabu et al., 2013). Given the well-​established role of inflammation and insulin resistance in carcinogenesis, adiponectin has been hypothesized to lower obesity-​ related cancer risk. In addition, experimental evidence suggests that adiponection may directly control malignant potential by regulating metabolic, inflammatory, and cell cycle signaling pathways (Fujisawa et  al., 2008; Moon et  al., 2013; Sugiyama et  al., 2009). Despite the strength of experimental data, epidemiologic studies have reported inconsistent findings on the relationship of prediagnostic adiponectin with risk of several specific cancers. Six prospective case-​control studies have studied the relationship between adiponectin and colorectal cancer. Three of these reported no association (Chandler et al., 2015; Lukanova et al., 2006; Stocks et al., 2008), whereas three others reported mixed results (Aleksandrova et al., 2012; Ho et al., 2012; Song et al., 2013b). In the Women’s Health Initiative, while high plasma adiponectin was associated with lower colorectal cancer risk, this association disappeared after adjusting for insulin levels, suggesting that insulin could be on the causal pathway (Ho et al., 2012). This notion was not supported by EPIC, however, in that the high molecular weight fraction of adiponectin, which is most strongly associated with insulin sensitivity and diabetes (Heidemann

361

et al., 2008; Oh et al., 2007), was not more strongly associated with colorectal cancer risk (Aleksandrova et al., 2012). Similarly intriguing, a sex difference has been observed in an analysis of the Nurses’ Health Study and the Health Professionals Follow-​up Study, where the inverse association between adiponectin and colorectal cancer risk was confined to men (Song et al., 2013b). Of note, women generally had higher levels of adiponectin than men (median:  8.2 vs. 5.3 μg/​mL). The gender difference in plasma levels appears to be independent of fat mass or distribution, and may result from the influence of sex steroid hormones (Cnop et al., 2003; Lanfranco et al., 2004; Yarrow et al., 2012). Thus, the sex heterogeneity in the adiponectin–​colorectal cancer association may reflect a distinct influence of altered estrogen and testosterone concentrations related to adiposity (Giovannucci, 2007; Koo and Leong, 2010). A Mendelian randomization study of SNPs associated with circulating adiponectin levels (Dastani et al., 2012; Ling et al., 2009; Richards et al., 2009) did not report consistent, strong associations with a variety of outcomes, including colorectal cancer (Song et al., 2015a), although the SNPs explain only a small fraction of variation in adiponectin levels. However, low circulating levels of adiponectin have been primarily linked to the increased risk of KRAS-​mutant cancer (Inamura et al., 2016), and a larger Mendelian randomization study would be needed to assess whether adiponectin-​related SNPs are associated with any particular subtypes of colorectal cancer. For breast cancer, although retrospective case-​control studies have generally reported lower adiponectin levels in cases than in controls (Macis et  al., 2014), this finding was replicated in only two (Macis et  al., 2012; Tworoger et  al., 2007)  of the seven prospective studies (Cust et al., 2009; Gaudet et al., 2010; Gunter et al., 2015; Macis et al., 2012; Ollberding et  al., 2013; Touvier et  al., 2013; Tworoger et  al., 2007). However, the two studies differ in that adiponectin was associated with breast cancer risk in premenopausal women in one study (Macis et  al., 2012), but only with postmenopausal breast cancer in the other (Tworoger et  al., 2007). Similarly, for endometrial cancer, prospective studies have produced inconsistent findings. While an early study reported that prediagnostic adiponectin level, independent of BMI and other metabolic biomarkers, was associated with a lower risk of postmenopausal endometrial cancer (Cust et  al., 2007), this finding was not replicated in three subsequent studies (Dallal et  al., 2013; Luhn et al., 2013; Soliman et al., 2011), except for an inverse association noted in one study only among non-​current MHT users only (Luhn et al., 2013). For prostate cancer, higher adiponectin levels were associated with lower risk of high-​grade or lethal prostate cancer in the Physicians’ Health Study (Li et  al., 2010), and this finding was further corroborated by genetic studies demonstrating that SNPs related to lower adiponectin levels were associated with higher prostate cancer risk (Dhillon et al., 2011). However, such findings have not been consistently replicated in other studies (Baillargeon et al., 2006; Moore et al., 2009; Touvier et al., 2013). For pancreatic cancer, a pooled analysis of five prospective cohort studies observed a common inverse association between adiponectin and pancreatic cancer risk among smokers and non-​smokers (Bao et al., 2013). Higher levels of circulating adiponectin were inversely associated with multiple myeloma risk (Hofmann et al., 2012), and also have been associated with a lower risk of progression from monoclonal gammopathy of undetermined significance (MGUS) to myeloma (Fowler et al., 2011; Reseland et al., 2009). For primary liver cancer, four prospective studies have consistently shown that high adiponectin is associated with increased risk among patients with chronic hepatitis C (Arano et al., 2011), among patients with hepatitis B or C infection (Chen CL et al., 2014; Michikawa et al., 2013), or among participants with low infection rate (Aleksandrova et  al., 2014). This positive association appears to be stronger for non-​HMW adiponectin (Aleksandrova et al., 2014; Michikawa et al., 2013). Although these findings seem to contradict those for other cancers, circulating adiponectin has been found to be elevated in patients with chronic hepatitis C or with high hepatitis B viral load (Chen CL et al., 2014; Wong et al., 2010), suggesting that adiponectin may rise as a result of reduced degradation by the liver due to virus-​induced inflammation and injury. Furthermore, higher adiponectin levels

362

362

PART III:  THE CAUSES OF CANCER

among patients with newly diagnosed or recurrent hepatocellular carcinoma have been linked to increased mortality (Siegel et al., 2015). Interestingly, the phenomenon of adiponectin-​ associated increased mortality, known as the “adiponectin paradox,” has also been reported in patients with colorectal cancer (Chong et al., 2015), prevalent cardiovascular disease (Cavusoglu et al., 2006), and heart failure (Kistorp et al., 2005), and among the elderly (Karas et al., 2014; Kistorp et al., 2005; Kizer et  al., 2012; Laughlin et  al., 2007; Poehls et  al., 2009; Wannamethee et al., 2007). Although there is evidence suggesting that increased adiponectin concentrations may represent a response to an underlying disease process, this paradox has not been explained and warrants further studies to elucidate the complex pathophysiology of adiponectin (Sattar and Nelson, 2008).

Sex Steroid Hormones Contributing to the role of adipose tissue as an endocrine organ, body fat is a major site for the synthesis of sex steroid hormones, and contributes to the bioavailability of these hormones. Steroid hormone synthesis occurs as a result of adipocyte expression of the aromatase enzyme. In men and postmenopausal women, the primary source of circulating estrogens is the conversion of androgens (testosterone and androstenedione) to estrogens (estradiol and estrone) by adipose tissue. Less active forms of these hormones (androstenedione, estrone) are converted to more active forms (testosterone, estradiol) by the enzyme 17B-​hydroxysteroid dehydrogenase, expressed in adipocytes. Adipose tissue contributes to increased bioavailability of sex hormones through the inhibition by insulin of the synthesis of SHBG in the liver (Crave et al., 1995). Lower SHBG levels result in higher concentrations of free estradiol and free testosterone, which are more readily taken up by cells in target organs. Combining the strengths of several nested case-​control studies of circulating hormones and breast cancer risk, a large collaborative study of BMI and circulating hormones was conducted recently, including more than 6000 postmenopausal women (Endogenous Hormones Breast Cancer Collaborative Group, 2011). Compared to thin women (BMI ≤22.5), obese women (BMI >30) had 47% higher circulating estradiol and estrone levels. In addition, obese women had 46% lower SHBG levels, contributing to significantly higher circulating free estradiol (89%) and free testosterone (72%) levels. Steroid hormones play an important etiologic role in cancers of the breast, endometrium, ovary, and prostate. Circulating hormone levels have been consistently associated with postmenopausal breast cancer (Kaaks et al., 2005; Key et al., 2002; Zhang X et al., 2013). In a pooled analysis of nine prospective studies, including 663 breast cancer cases and 1765 controls, higher circulating levels of estrogens, including estradiol, estrone, estrone sulfate, and free estradiol, were associated with higher subsequent risk of breast cancer (e.g., top vs. bottom quintile estradiol, RR = 2.0; 95% CI: 1.5, 2.7) (Key et al., 2002). Adipose tissue may also affect breast cancer risk locally, given that adipocytes in breast stromal tissue likely contribute to estrogen levels in the microenvironment. In endometrial carcinogenesis, the interplay between peptide and sex steroid hormones appears to be critical (Kaaks et al., 2002a). Higher insulin and IGF-​1 levels in obese women increase ovarian androgen production. In premenopausal women, ovarian androgen production interferes with ovulation, which decreases production of progesterone. Normally, progesterone provides a check on estrogen-​induced proliferation in the uterus (Key and Pike, 1988). Therefore, decreases in progesterone production in premenopausal women, and increases in bioavailable estrogens, create the environment for endometrial carcinogenesis. Sex steroid hormones play a role in colorectal cancer, and may partly account for the stronger association with obesity observed in men. In women, higher exposure to estrogen and progesterone, whether through pregnancy or the use of MHT, is associated with lower risk of colorectal cancer. For instance, in the Women’s Health Initiative trial, combined use of estrogen and progestin was associated with a 40% lower risk of colorectal cancer (Chlebowski et  al., 2004). Although associations with circulating levels of estradiol are inconsistent (Clendenen et  al., 2009; Gunter et al., 2008; Lin et al., 2013), a higher ratio of estradiol

to testosterone in women not using MHT has been associated with lower colorectal cancer risk after adjustment for BMI and C-​peptide levels (Lin et al., 2013). One potential explanation is that the increase in estrogens associated with obesity may partially offset the increased risk from obesity-​related metabolic factors, such as hyperinsulinemia. Adiposity in men is associated with lower levels of androgens (Kapoor et al., 2005); higher circulating levels of testosterone are associated with lower colorectal cancer (Lin et al., 2013). A higher ratio of estradiol to testosterone was associated with higher risk in men, which may reflect increasing aromatization and lower testosterone levels (Lin et al., 2013).

Inflammation Chronic inflammation creates a tissue microenvironment that stimulates cellular proliferation, suppresses apoptosis, and generates free radicals that can damage DNA (see Chapter 25). While it is difficult to measure local inflammation in population studies, there are numerous clinical examples in which chronic local inflammation predisposes to certain types of cancer (see Chapter 25). This section discusses several examples in which tissue damage caused by the effects of obesity fosters neoplastic growth and malignant transformation. It also discusses systemic markers of inflammation that have been measured in epidemiologic studies of cancer.

Examples of Local Inflammation Esophageal Adenocarcinoma. Esophageal adenocarcinoma

typically occurs in the lower one-​third of the esophagus, and has a greater propensity to occur in men than in women (see Chapter 30). Incidence rates have increased in parallel with the increasing prevalence of obesity. This malignancy usually arises from mucosa that has been damaged by chronic reflux of acid and bile from the stomach into the lower esophagus. A recent extensive meta-​analysis demonstrated the importance of central adiposity in increasing risk of esophageal inflammation, metaplasia (Barrett’s esophagus), and neoplasia (Singh et al., 2013). Obesity, particularly central obesity, is the strongest risk factor for acid reflux through disruption of the normal physiology of the gastroesophageal junction (Friedenberg et al., 2008). The relationship between obesity and acid reflux likely largely explains the association between obesity and esophageal adenocarcinoma. The male propensity for an abdominal pattern of adiposity may at least partly account for the higher rates of esophageal adenocarcinoma in men (Friedenberg et al., 2008; Singh et al., 2013).

Gallbladder Cancer. Gallbladder cancer is associated with

excess adiposity (Larsson and Wolk, 2007; Park et al., 2014; Tan et al., 2015), and with related metabolic abnormalities such as type 2 diabetes (Gu et al., 2015), lipid disorders, and elevated fasting blood glucose (Borena et al., 2014; Rapp et al., 2006) (see Chapter 34). Obese women are at particularly high risk of gallbladder cancer. The main risk factor for gallbladder cancer is a history of gallstones (Randi et al., 2006) (see Chapter 34). Obesity, hyperinsulinemia, and dyslipidemia are related to a cluster of factors that predispose to gallstones, including biliary supersaturation with cholesterol, inflammation, hypersecretion of mucin, slow colonic motility, and increased intestinal cholesterol absorption. Chronic inflammation from recurrent gallstones is the presumptive mediator of obesity-​related gallbladder cancer.

Liver Cancer.  Obesity has been suggested to be the second larg-

est contributor to the recent increase in the incidence of hepatocellular carcinoma in developed countries, after hepatitis C infection (Caldwell et al., 2004; Starley et al., 2010) (see Chapter 33). Obesity plays an important role in the progression from non-​alcoholic fatty liver disease to steatohepatitis to cirrhosis and hepatocellular carcinoma (Caldwell et al., 2004; Cohen et al., 2011; Hassan et al., 2015). As the most common form of chronic liver disease, non-​alcoholic fatty liver disease affects about 30% of the US general population and up to 90% of those who are morbidly obese (Torres and Harrison, 2008). The more aggressive form of non-​alcoholic fatty liver disease,

 36



Obesity and Body Composition

non-​alcoholic steatohepatitis, has a prevalence of 5%–​7%; in this subgroup, up to 9% can progress to cirrhosis (Ong and Younossi, 2007). An estimated 40%–​60% of patients with cirrhosis from non-​alcoholic steatohepatitis develop major complications, including hepatocellular carcinoma over 5–​7 years of follow-​up (Adams et al., 2005; Hui et al., 2003).

Systemic Markers of Inflammation

Epidemiologic studies can more easily measure systemic markers of inflammation than inflammation at the tissue level. It is often unclear whether the increase in a systemic marker derives from local inflammation or from a more generalized inflammatory process, perhaps involving the infiltration of immune cells into adipose tissue, especially visceral adipose. It is also uncertain whether a low-​grade generalized inflammatory state can predispose to cancers at remote sites. One of the most widely used inflammatory biomarkers is CRP, which is secreted from the liver and is an acute phase marker of inflammation. CRP is increased in obese individuals (Visser et al., 1999), but also is increased by smoking, alcohol, physical inactivity, and various inflammatory conditions (Chan et al., 2015). Recognizing the complex determinants of CRP, some studies have examined pre-​diagnostic CRP levels in blood in relation to risk of various cancers. In a meta-​analysis of circulating CRP and incident breast cancer risk, a statistically significant positive association was found; each doubling of the CRP concentration was associated with a 7% increase in incident breast cancer (Chan et al., 2015). In a meta-​analysis of colon cancer, circulating CRP was modestly associated with increased risk, although a closely related inflammatory marker, IL-​6, was not (Zhou et al., 2014). The interpretation of these results is unclear; some studies observed increased risk with higher levels of the inflammatory markers only in the early years after blood samples were provided (Song et al., 2013a), raising the possibility that the tumor itself may have affected the circulating levels (i.e., reverse causation). Because CRP and IL-​6 are affected by many factors besides excess body weight, it is difficult to ascribe any observed association strictly to obesity. This problem is illustrated in a large case-​control study of prostate cancer nested in a prospective cohort. As expected, obese men had substantially higher levels of CRP and IL-​6. However, IL-​ 6 was associated with increased risk of prostate cancer only among men with normal BMI (Stark et al., 2009). An inverse relationship was observed between IL-​6 and prostate cancer among men with increased BMI. This finding suggests that determinants of elevated inflammatory markers unrelated to adiposity accounted for the increased risk. A possible explanation is that high IL-​6 or CRP among the leaner men may be more indicative of inflammation within the prostate gland, which directly influences risk, whereas high concentrations of inflammatory markers among the heavier men reflected the greater amount of adipose tissue. Similarly, in colon cancer, an association with IL-​6 was observed only in leaner men, again suggesting that inflammation at the tissue level, rather than systemic inflammation related to adiposity, influences risk (Song et al., 2013a). Further, using a Mendelian randomization approach, genetic determinants of elevated levels of CRP were not associated with increased risk of cancer, possibly arguing against a causal role of systemic inflammation for cancer risk (Allin et al., 2010). Although it is not clear that obesity-​associated systemic inflammation would necessarily affect the risk of solid tumors, effects on immune system cancers are also plausible. Obesity has been associated with increased risk of multiple myeloma, a malignancy of plasma cells (Teras et al., 2014); IL-​6, a pro-​inflammatory cytokine, is a potent growth factor for plasma cells and may contribute to the risk of multiple myeloma (Ziakas et al., 2013).

Mechanisms Associated with Early Life Adiposity Excess body fat in childhood and/​or adolescence occurs during an important developmental period, and may influence lifelong metabolic and hormonal patterns, as well as behaviors. Early life obesity has been suggested to influence basal insulin levels, and thus contribute to

363

increased lifetime exposure to insulin (Caprio et al., 1995), which may contribute to subsequent risk of colon cancer. Mechanisms by which childhood/​adolescent obesity may influence breast cancer risk later in life could involve both the hormonal milieu in early life and the establishment of set points that persist with age. Although overweight and obese girls are likely to have earlier ages at menarche compared to normal weight girls, they also are more likely to have anovulatory menstrual cycles (Rich-​Edwards et  al., 1994; Stoll, 1998) and irregular cycle patterns (Tehard et al., 2005). Adiposity at young ages is associated with slower adolescent growth (Berkey et al., 1999). Excess adiposity in childhood is inversely associated with at least two breast cancer risk factors in adulthood: circulating IGF-​1 levels (Poole et  al., 2011), and breast density (Samimi et al., 2008; Sellers et al., 2007). Among 6520 women in the Nurses’ Health Study and Nurses’ Health Study II, those who were overweight at young ages had 14% lower adult plasma IGF-​1 levels, compared with leaner girls (Poole et al., 2011). The inverse associations observed between childhood body fatness and breast density suggest that the structure of breast tissue may be established as a result of early life adiposity and then may persist throughout life (Samimi et al., 2008; Sellers et al., 2007).

Microbiome The explosion of research in host-​microbe interaction over recent years supports a potential role of the gut microbiota in human health and disease through regulation of host metabolism and immune function (Blumberg and Powrie, 2012; Brestoff and Artis, 2013; Guarner and Malagelada, 2003; Kau et al., 2011; Sommer and Backhed, 2013; Tremaroli and Backhed, 2012). Several lines of evidence indicate a bidirectional association. While perturbations in the gut microbiota may predispose individuals to obesity (Cox and Blaser, 2013), obesity also can alter gut microbiota, potentially contributing to obesity-​ linked carcinogenesis. Maintenance of body weight strongly predicts microbiota stability (Faith et al., 2013); obesity has been associated with lower diversity and altered composition of gut microbiota, specifically a decrease in the phylum Bacteroidetes and an increase in Firmicutes (Furet et  al., 2010; Le Chatelier et  al., 2013; Ley et  al., 2005, 2006). An opposite change has been observed with weight loss (Aron-​Wisnewsky et  al., 2012; Furet et  al., 2010; Kong et  al., 2013; Ley et  al., 2006; Zhang et  al., 2009). Furthermore, growing evidence implicates the altered microbial community and enrichment of certain microbes in cancer development (e.g., Fusobacterium nucleatum in colorectal cancer) (Castellarin et al., 2012; Kostic et al., 2013). Microbes may amplify or mitigate carcinogenesis by affecting genomic stability (Cuevas-​Ramos et  al., 2010; Guerra et  al., 2011; Nougayrede et  al., 2006), altering the balance of host cell proliferation and death (Rubinstein et  al., 2013; Sears, 2009), regulating pro-​inflammatory or immunosuppressive responses (Hu et al., 2013; Irrazabal et  al., 2014; Kostic et  al., 2013; Warren et  al., 2013), and influencing host metabolism (Belcheva et al., 2014; Donohoe et al., 2014; Singh et al., 2014). Therefore, it has been hypothesized that the gut microbiota may affect multiple pathways by which obesity influences cancer risk (Ohtani et al., 2014). Indeed, studies in liver cancer have suggested that obesity-​induced alteration of the gut microbiota resulted in enhanced production of DNA-​damaging secondary bile acid (e.g., deoxycholic acid) that promotes carcinogenesis through cellular senescence-​related pathways (Ohtani et al., 2014; Yoshimoto et al., 2013). Given the potential role of secondary bile acids, further studies are needed to investigate whether metabolic change due to microbial imbalance also mediates obesity-​related tumor promotion in other organs.

Site-​Specific Mechanisms: Hypertension in Renal Cell Cancer Obesity is a risk factor for hypertension. Hypertension is associated with higher risk of renal cell cancer in many studies (Chow et  al., 2000, 2010; Sanfilippo et  al., 2014; Weikert et  al., 2008), and has

364

364

PART III:  THE CAUSES OF CANCER

been postulated as a mechanism linking obesity to renal cell cancer. However, epidemiologic data lend little support to this hypothesis. Although individuals who are both obese and hypertensive have a higher risk of renal cell cancer than those who have only one of these conditions, the increased risk of renal cell cancer associated with BMI persists even after adjustment for diagnosis of hypertension or measurement of blood pressure (Adams et  al., 2008; Chow et  al., 2000; Sanfilippo et  al., 2014; Setiawan et  al., 2007). This indicates that hypertension-​independent mechanisms may mediate renal cell cancer development in obese individuals.

ESTIMATED ATTRIBUTABLE FRACTION FOR  INCIDENCE AND SURVIVAL Several studies have estimated the population attributable fraction (PAF), which incorporates both the prevalence of the exposure and the strength of the association between the exposure and disease. While the effects of obesity are relatively modest for specific types of cancer (Table 20–​1), the prevalence of overweight and obesity are high, especially in the United States and other high-​income countries. Therefore the number of cancer cases attributable to overweight and obesity is likely to be substantial. The WCRF/​AICR has estimated the PAF for cancer sites designated as definitely caused by obesity (WCRF/​AICR, 2010b). The estimates for the United States range from 11% for gallbladder cancer to 34% for pancreatic cancer in men, and from 3% for colorectal to 49% for endometrial cancer in women. PAF estimates for the world were published recently by the IARC (Arnold et al., 2015); the estimated fractions in men ranged from 6% for rectal cancer to 33% for esophageal adenocarcinoma; for women, they ranged from 4% for rectal cancer to 34% for endometrial cancer. Combining PAF estimates for specific cancer sites into an overall PAF for obesity and cancer, the IARC estimated that 481,000 (3.6%) of all new cancers in 2012 in adults were attributable to high BMI (Arnold et al., 2015). By sex, the PAF were 1.9% in men and 5.4% in women. Further, one-​quarter of these cases in 2012 were attributed to the global increase in BMI since 1982. If the population had maintained the prevalence of overweight and obesity in 1982, instead of increasing to today’s prevalence, an estimated 118,000 cases of cancer could have been avoided. The magnitude of the PAF also varies by human developmental indices (HDIs), since obesity is more common in countries with higher HDIs. For example, the IARC estimates that 5.4% of all incident cancers could be attributed to obesity in countries with very high HDI, 4.8% in those with high HDI, 1.6% in moderate HDI countries, and 1.0% in those with low HDI. In contrast to the estimates for obesity alone, the WCRF/​AICR Continuous Update Project estimates the PAF for the combined effects of diet, physical activity, and body mass (WCRF/​AICR, 2009). The estimates range from 19% to 26% in low-​to high-​income countries, respectively. While this approach would exaggerate the PAF due to body mass alone, these factors inherently overlap, since weight gain reflects both diet and activity level. For example, the association between colorectal cancer and a dietary pattern that stimulates insulin secretion appears to be highly dependent on BMI and physical activity level (Fung et al., 2012). Thus, at least for some cancers, the effects of obesity cannot be isolated from those related to diet and physical inactivity. The published PAFs (besides those from the WCRF/​ AICR Continuous Update Project) for cancer are likely to be substantial underestimates of the true impact of excess adiposity on cancer risk for numerous reasons. First, despite the categorical terms such as “overweight” and “obesity,” adiposity is a continuous variable. Thus, selection of the cutpoint is arbitrary. For example, a cutpoint of 25 kg/​m2 is typically used because it is the lower bound of overweight, and may be considered a potential target for population-​level interventions. Yet, the association may be linear even below 25 kg/​m2. Moreover, most studies have relied on a single measure of adiposity, whereas BMI and other measures of adiposity may vary over time. The potential importance of the cutpoint and of obtaining repeated measures of BMI are

illustrated in the Health Professionals Follow-​up Study, where the PAF for colon cancer in men was 14% based on a single measure of BMI and a cutpoint of 25, but increased to 30% when the BMI cutpoint was reduced to 22.5 kg/​m2 and was based on an average of multiple measures of BMI (Thygesen et al., 2008). Other methodologic factors also influence the estimates of PAF. Although BMI is widely used for PAF estimates because of its availability, it is not an ideal measure of adiposity, as previously discussed. More direct measures of visceral or central adiposity, such as VAT, might be expected to provide stronger associations of the relative risks and higher estimates of the PAF. For example, VAT based on CT scan was strongly associated with risk of colorectal adenoma, even adjusting for BMI (Keum et  al., 2015). In study populations where the majority of participants had BMI levels below 25 kg/​m2, VAT was strongly and linearly associated with the risk of colorectal adenoma (Keum et al., 2015). To date, there are limited data by which to estimate PAF using measures other than the traditional cutpoints for BMI. Relying on a single measure of BMI during adulthood conflates exposures that occur at different times of life. For example, a study of breast cancer based on adult BMI cannot separate early life adiposity (which decreases breast cancer risk) from adult weight gain (which increases risk). On a per kilogram increment basis, adult weight gain is a much stronger risk factor for breast cancer than is adult BMI (Keum et al., 2015). Thus, the use of a single measure of BMI may not accurately reflect the relevant measure of adiposity at different time points. The presence of effect modifiers may also differ between populations and may change over time, further constraining the estimates of PAF. For example, as described earlier, MHT may modify the effect of obesity on breast cancer, just as smoking modifies the relation of adiposity to smoking-​related cancers. The prevalence and timing of these modifying factors in the population will affect the strength of the association and the estimated PAF. For example, as the fraction of perimenopausal women who use MHT in the United States decreases, the estimated PAF for obesity and breast cancer would be expected to increase. A particularly important modifier to consider is smoking. As discussed earlier, the complex relationship between smoking and adiposity makes it difficult to interpret the effects of adiposity. In the Cancer Prevention Study II, a prospective mortality study begun by the American Cancer Society in 1982, the PAF of total cancer mortality in the United States due to high BMI was 4.2% in men and 14.3% in women in the entire population, but these estimates increased to 14.2% and 19.8%, respectively, in the population limited to healthy never-​smokers (Calle et  al., 2003). The inclusion of smokers qualitatively changes the association observed among never smokers, either because of reverse causation (as hypothesized for lung cancer) or intractable confounding (see earlier disucssion). Regardless of the reason, given the strong effect modification, the implications are that the overall PAF in a population is arbitrary because it depends largely on the proportion of smokers in the population, and would not apply to either the subgroup of smokers or to non-​smokers (Song and Giovannucci, 2016). In summary, while the PAF is a widely used measure for communicating the impact of obesity on cancer, the calculation of PAF is not straightforward, and the exact value can be influenced by many factors. Nevertheless, the high prevalence of obesity and its consistent associations with multiple types of cancer indicate that the disease burden is high for cancer, as for other major diseases.

EXCESS ADIPOSITY AND CANCER SURVIVAL The association between obesity and survival of patients with cancer is a relatively recent area of research, and one that is methodologically difficult for several reasons. If adiposity is measured before diagnosis, one cannot separate an effect on cancer incidence from an effect on disease progression. However, if adiposity is measured after diagnosis, it becomes impossible to exclude the possibility that weight loss resulted from reverse causation. Weight loss is a common and generally unfavorable prognostic indicator in patients with advanced or metastatic cancer. Particularly for cancers that tend to have short

 365



Obesity and Body Composition

survival time (e.g., pancreatic cancer), the effects of local progression or metastatic disease may cause weight loss some time before diagnosis. Therefore, a measure of body mass at the time of diagnosis or shortly thereafter is likely to be misleading. Further, low body weight may reflect loss of lean body mass prior to diagnosis. The relationship of obesity to survival from breast cancer has been more extensively studied than its relationship to survival of other cancers, because of the large number of cases and generally long survival. A recent meta-​analysis summarized the results of 82 studies, encompassing 213,075 breast cancer survivors and 41,477 deaths (23,182 from breast cancer). Increased adiposity was associated with poorer overall and site-​specific survival for both pre-​and postmenopausal breast cancer, regardless of when BMI was ascertained (Chan et al., 2014). Relative to normal weight women, the summary RRs for all-​ cause mortality were as follows: 1.41, 95% CI: 1.29, 1.53 for obese (BMI >30.0); 1.07, 95% CI:  1.02, 1.12 for overweight (BMI 25.0–​ 500 cases in prospective studies Evidence for a relationship Biologically plausible Some possibility of confounding Evidence for a relationship Results seem likely to be confounded Sensitivity analyses do not rule out confounding Results vary by study type or over time Reasons for variability are unclear > 2000 cases in prospective studies Aggregate RR of 0.95 to 1.05

Cancers at Evidence Level Colon, Female breast

Kidney, Endometrium

Bladder, Liver, Esophageal adenocarcinoma, Gastric cardia, Myeloma, Myeloid leukemia, Non-​Hodgkin lymphoma, Head and neck Lung, Melanoma (increased risk) Pancreas, Prostate, Rectum, Brain, Thyroid Ovary, Lymphocytic leukemia

 38



383

Physical Activity, Sedentary Behaviors, and Risk of Cancer Category

# of Studies

RR (95% Cl)

Overall

52

0.76 (0.72, 0.81)

Cohort

28

0.83 (0.78, 0.88)

Case-control

24

0.69 (0.65, 0.74)

Men

30

0.76 (0.71, 0.82)

Women

30

0.79 (0.71, 0.88)

Occupational

32

0.78 (0.74, 0.83)

LTPA

26

0.77 (0.72, 0.82)

0

1 Relative risk (95% Cl) High vs. low physical activity

2

Figure 21–​2.  Colon cancer and its association with physical activity.

of proximal colon cancer was 27% lower among the most active people compared to the least active (random effects summary RR = 0.73; 95% CI: 0.66, 0.81). This reduction in risk was indistinguishable from the 26% lower risk of distal colon cancer (RR = 0.74; 95% CI: 0.68, 0.80). As with the other analyses described earlier, the findings were similar in men and in women, and the inverse relationship was stronger in case-​control than cohort studies. The results of the pooled analysis in 2016 strongly support the findings of Wolin et  al. (2009) with respect to the overall reduction in risk. The relative risk and 95% confidence interval estimates for high versus low levels of leisure-​time activity were 0.84 (95% CI:  0.77, 0.91) for colon cancer. This is nearly identical to the hazard ratio of 0.83 observed in the meta-​analysis of prospective studies by Wolin et al. (2009). In summary, based on the available evidence, there is convincing support for a causal relationship between physical activity and reduced risk of developing colon cancer. This conclusion is consistent with those of several expert panels (IARC Handbooks of Cancer Prevention, 2002; Kushi et al., 2012; World Cancer Research Fund and American Institute for Cancer Research, 2007).

Female Breast Cancer The only meta-​analysis that formally quantified the association between physical activity and breast cancer risk was by Wu et  al. (2013). It included 63,786 breast cancer cases drawn from 31 prospective cohorts in North America, Europe, and Asia. Overall, the most active women had a 13% lower risk of developing breast cancer than the least active women (RR = 0.87; 95% CI: 0.83, 0.92; see Figure 21–​3). Associations were similar for recreational activity (RR = 0.87; 95% CI: 0.83, 0.91; 25 studies) and occupational activity (RR = 0.84; 95% CI: 0.73, 0.96; 7 studies). The analysis of recreational activity used restricted cubic spline models to show that the dose–​response association was approximately linear (P for non-​linearity = 0.45); each additional increment of 10 MET-​hours/​week correlated with a 3% lower risk of breast cancer. The associations were stronger for vigorous-​intensity physical activity (RR = 0.85; 95% CI: 0.80, 0.90; 21 studies) than for moderate-​intensity activity (RR = 0.95; 95% CI: 0.90, 0.99; 16 studies). Wu et  al. (2013) also examined whether the inverse association between breast cancer and physical activity was modified by

menopausal status, body mass index, or period of life during which activity occurred. The point estimates were lower for women with premenopausal breast cancer (RR  =  0.77; 95% CI:  0.69, 0.86) than postmenopausal disease (RR = 0.87; 95% CI: 0.82, 0.92), and in those with a BMI < 25  kg/​m2 (RR  =  0.72; 95% CI:  0.65, 0.90) than in women who were overweight or obese (BMI ≥ 25  kg/​m2) (RR  =  0.93; 95% CI:  0.87, 0.98). Finally, the pooled RR estimate associated with physical activity was slightly lower for women age ≥ 50 years (RR = 0.83; 95% CI: 0.76, 0.91) than for women age < 25 years (RR = 0.90; 95% CI: 0.81, 1.02). However, no formal tests of heterogeneity were reported, and the 95% CI overlapped in most subgroup analyses. Associations also varied according to estrogen receptor (ER) and progesterone receptor (PR) positivity, with lower point estimates for ER–​/​PR–​cancers (RR = 0.77; 95% CI: 0.65, 0.90) (8 studies) than for ER+/​PR+ cancers (RR  =  0.93; 95% CI:  0.87, 0.98) (7 studies). The significance of this is uncertain, given the limited number of studies in these analyses and the lack of a mechanistic basis for the finding. Relative risk estimates in studies were appropriately adjusted for potential confounders, including body mass index (BMI), menopausal status and hormone therapy, reproductive history, and family history. Where available, the authors also presented relative risks from analyses adjusted only for age and found substantively similar results (RR  =  0.85; 95% CI:  0.79, 0.90), indicating that adjustment for the available confounders had little overall effect on the results. Friedenrich et al. (2011) systematically reviewed the association of physical activity with breast cancer risk in 75 cohort and case control studies conducted worldwide. In this review, women who were the most physically active had, on average, 25% lower risk of developing breast cancer as compared with the least active women. These inverse relationships were slightly stronger in the case control studies (RR 0.7) than in the cohort studies (RR 0.8). Overall, the results were generally consistent with those of Wu et al. (2013). The 2016 pooled analysis reported a relative risk estimate of 0.90 (95% CI: 0.87, 0.93) comparing breast cancer risk among women with the highest versus lowest level of leisure-​time activity. This is similar to the hazard ratio of 0.87 reported from the meta-​analysis by Wu et al. (2013). In summary, the evidence is convincing that higher levels of physical activity are associated with lower risk of breast cancer. Additional data on the benefits of vigorous-​intensity versus moderate-​intensity

384

384

PART III:  THE CAUSES OF CANCER Category

# of Studies

RR (95% Cl)

Overall

31

0.87 (0.83, 0.92)

Premenopausal

6

0.77 (0.69, 0.86)

Postmenopausal

17

0.87 (0.82, 0.92)

ER-/PR-

7

0.77 (0.65, 0.90)

ER+/PR+

8

0.93 (0.87, 0.98)

Occupational

7

0.84 (0.73, 0.96)

LTPA

25

0.87 (0.83, 0.91)

0

1 Relative risk (95% Cl) High vs. low physical activity

2

Figure 21–​3.  Female breast cancer and its association with physical activity.

activities, and on how associations vary according to ER/​PR and other breast cancer subtypes, would help to refine understanding of this relationship.

Kidney Cancer Behrens and Leitzmann (2013) identified a total of 19 studies (11 cohort, 8 case control) from North America, Europe, and Asia that examined the association between physical activity and renal cell carcinoma, which accounts for approximately 90% of all kidney cancers. The analysis was based on 10,756 cases. Overall, physical activity was inversely associated with risk of renal cell cancer (RR = 0.88; 95% CI: 0.79, 0.97). Results were similar for the prospective studies (RR = 0.87; 95% CI: 0.79, 0.97) and case-​control studies (RR  =  0.89; 95% CI:  0.74, 1.06). The point estimate was somewhat lower (RR = 0.78; 95% CI: 0.66, 0.92) in studies judged to be of higher quality according to a predetermined rubric than in the overall analysis. The dose–​response relationship was not assessed. The strength of association was similar for recreational and occupational physical activity, and did not appear to vary according to the age when the activity occurred. Adjustments for smoking, BMI, hypertension, and diabetes had little effect on the association. In the 2016 pooled analysis, leisure-​ time physical activity was inversely associated with kidney cancer (RR  =  0.77; 95% CI:  0.70, 0.85). This association was stronger than that observed by Behrens and Leitzmann (2013) when comparing high versus low levels of overall activity (RR = 0.88; 95% CI: 0.79, 0.97), and recreational activity (RR = 0.88; 95% CI: 0.77, 1.00). The findings from the 2016 pooled analysis strengthen the evidence of an inverse association between physical activity and kidney cancer.

Endometrial Cancer A meta-​analysis by Schmid et  al. (2015) identified 18 cohort studies, 1 case-​cohort study, and 14 case-​control studies that examined the association between physical activity and endometrial cancer risk. Collectively, these studies included 19,558 endometrial cancer cases.

The risk of endometrial cancer risk was 20% lower for women in the highest compared to the lowest category of physical activity, as shown by the summary estimate of relative risk (RR = 0.80; 95% CI: 0.75, 0.85) in Figure 21–​4. Exposure assessment in studies of endometrial cancer and physical activity was based on self-​administered questionnaires (16 studies), interview-​assisted questionnaires (10 studies), occupational job titles (4 studies), or a combination of job titles and interview-​assisted questionnaires (3 studies). Physical activity was inversely associated with endometrial cancer risk across all domains of activity. In 22 studies of recreational physical activity there was a 16% reduction in the risk of endometrial cancer (RR = 0.84; 95% CI: 0.78, 0.91) comparing the highest to the lowest category of activity; in 19 studies of occupational activity there was a 19% reduction in risk (RR = 0.81; 95% CI: 0.75, 0.87), and in 7 studies of household activity there was a 30% reduction in risk (RR = 0.70; 95% CI:  0.47, 1.02). Ten studies specifically examined walking and reported an 18% reduction in risk (RR = 0.82; 95% CI: 0.69, 0.97). No differences in risk were observed among studies with different intensity of physical activity. Two studies included light activity and reported a 35% risk reduction (RR = 0.65; 95% CI: 0.49, 0.86), eight assessed moderate-​to-​vigorous physical activity and reported a 17% reduction (RR = 0.83; 95% CI: 0.71, 0.96) and eight included vigorous activity and reported a 20% reduction (RR = 0.80; 95% CI: 0.72, 0.90). The dose–​ response relationship was examined separately in nine studies that expressed activity in terms of MET-​hours per week. The inverse association between the level of physical activity and risk of endometrial cancer appeared to be non-​linear, with no further decrease in risk observed at activity levels higher than 20 MET-​hours per week. As with other end points, the inverse relationship with endometrial cancer appeared to be stronger in the case-​control studies (RR = 0.72; 95% CI: 0.64, 0.80) than in the prospective cohort studies (RR = 0.84; 95% CI: 0.78, 0.91), suggesting some influence from reporting bias. Most studies (N = 29) were conducted in North America, or Europe (N = 29); three studies were from Asia and one from South America. No significant differences were observed in the associations among geographic regions, based on the limited number of studies currently available in middle and low-​income countries. There was no evidence that the association differed by study size or quality.

 385



385

Physical Activity, Sedentary Behaviors, and Risk of Cancer Category

# of Studies

RR (95% Cl)

Overall

33

0.80 (0.75, 0.85)

Cohort

19

0.84 (0.78, 0.91)

Case-control

14

0.72 (0.64, 0.80)

Premenopausal

4

0.74 (0.49, 1.13)

Postmenopausal

7

0.81 (0.67, 0.97)

Occupational

19

0.81 (0.75, 0.87)

LTPA

22

0.84 (0.78, 0.91)

Adjusted for adiposity

24

0.79 (0.73, 0.85)

Not adjusted for adiposity

9

0.83 (0.72, 0.97)

0

1

2

Relative risk (95% Cl) High vs. low physical activity

Figure 21–​4.  Endometrial cancer and its association with physical activity.

One potential confounder in the analysis of endometrial cancer is unopposed estrogen therapy for menopausal symptoms. Adjustment for menopausal hormone therapy did not change the association between physical activity and endometrial cancer. There was no evidence of effect modification by parity, use of oral contraceptives, menopausal status, or the time period of physical activity (i.e., adolescence, midlife, or older age). Increased body weight is strongly associated with the risk of endometrial cancer. Most of the studies (24 of 33)  in the meta-​analysis of physical activity controlled for BMI as a potential confounder, but nine studies did not, including some that considered BMI a potential mediator that should not be adjusted for. In the meta-​analysis, there was no difference in the summary estimate between studies that controlled for BMI (RR = 0.79; 95% CI: 0.73, 0.85) and those that did not (RR 0.83; 95% CI: 0.72, 0.97). In the 2016 pooled analysis, the relative risks and 95% confidence intervals for a high versus low level of leisure-​time activity were 0.79 (95% CI:  0.68, 0.92) for endometrial cancer, similar to the meta-​ analysis estimate. However, in contrast to the meta-​analysis, the pooled study found that adjustment for BMI essentially eliminated the inverse association between physical activity and endometrial cancer (relative risk of 0.98). When compared with previous studies, the pooled analysis used a more thorough approach in adjusting for BMI, including each of the subcategories of obesity (30.0–​34.9, 35.0–​39.9, 40+ kg/​m2) separately. Finer adjustment may explain the greater attenuation in the 2016 pooled analysis. Replication is needed to resolve this difference in results. In summary, higher levels of physical activity are associated with an approximately 20% reduction in the risk of endometrial cancer. It is presently unclear, however, whether this association is confounded and/​or mediated by BMI.

Bladder Cancer A meta-​analysis by Keimling et  al. (2014) identified 15 studies (11 cohort, 4 case control) from North America, Europe, and Asia that examined the association between physical activity and bladder cancer risk. The analysis, based on 27,784 cases, found an inverse association between physical activity and bladder cancer risk (RR = 0.85; 95% CI: 0.74, 0.98). The findings were not significantly

different between the prospective studies (RR = 0.89; 95% CI: 0.80, 1.00) and the case-​control studies (RR = 0.71; 95% CI: 0.43, 1.16) (Pheterogeneity  =  0.11). Statistical modeling suggested a linear dose–​ response relationship. There was no evidence of heterogeneity by gender or BMI, nor was there obvious confounding by BMI or smoking. Associations did not differ for recreational versus occupational activity. The 2016 pooled analysis found an inverse relationship between bladder cancer risk and leisure-​time physical activity (RR = 0.87; 95% CI: 0.82, 0.92), with a point estimate similar to that observed in the meta-​analysis (Keimling et al., 2014). However, when stratifying on smoking, the association was stronger in former than in never smokers, raising the possibility of residual confounding. Thus, despite the large number of cases in the published studies, the level of evidence was considered limited (suggestive).

Liver Cancer and Gallbladder Cancer The association between physical activity and risk of liver cancer and gallbladder cancer has been studied only recently. No meta-​analyses have been conducted. The largest published study is based on the National Institutes of Health (NIH)-​AARP Diet and Health Study, a prospective cohort of more than 500,000 men and women residing in North America (Behrens et al., 2013). A total of 628 cases of liver cancer and 123 cases of gallbladder cancer were identified in the cohort over 10 years of follow-​up. Physical activity, defined by the number of days per week that subjects performed vigorous physical activity continuously for at least 20 minutes, was inversely associated with liver cancer (RR  =  0.64; 95% CI:  0.49, 0.84). The association with gallbladder cancer was similar but not statistically significant (RR = 0.63; 95% CI: 0.33, 1.21). The dose–​response relationship between physical activity and liver cancer was approximately linear, with no apparent effect modification by age, gender, BMI, diabetes, alcohol intake, or coffee consumption. The 2016 pooled analysis, which included the NIH-​AARP cohort, also found an inverse association between leisure-​time physical activity and risk of liver cancer (RR = 0.73; 95% CI: 0.55, 0.98). The number of cases in the pooled study (N  =  1384) more than doubled the number in the AARP study (N = 628) (Behrens et al., 2013). The overall evidence for causality was considered to be “limited–​suggestive.”

386

386

PART III:  THE CAUSES OF CANCER

Gastroesophageal Cancers Behrens et al. (2014) identified 24 studies (9 cohorts, 15 case control) from North America, Europe, Asia, and the Middle East that examined associations between physical activity and subtypes of gastroesophageal cancer. The analyses included 965 cases of esophageal adenocarcinoma, 844 cases of gastric cardia cancer, 1444 cases of esophageal squamous cell carcinoma, and 1571 cases of non-​cardia stomach cancers. Physical activity was inversely associated with risk of esophageal adenocarcinoma (RR = 0.79; 95% CI: 0.66, 0.94) and gastric cardia cancer (RR = 0.83; 95% CI: 0.69, 0.99), but not esophageal squamous cell carcinoma (RR = 0.94; 95% CI: 0.41, 2.16). In an analysis that combined all subtypes of gastroesophageal cancer, the point estimate for physical activity was lower in the case control studies (RR = 0.77; 95% CI: 0.67, 0.89) than in the prospective studies (RR = 0.89; 95% CI: 0.78, 1.01). The dose–​response relationship was non-​linear, with diminishing benefits at high levels of physical activity. The inverse association was somewhat stronger in women than in men, and more strongly associated with early-​life physical activity than that occurring later in life. There was no apparent confounding by BMI, smoking, or alcohol use. In the 2016 pooled analysis, leisure-​ time physical activity was inversely associated with risk of esophageal adenocarcinoma (RR = 0.58; 95% CI: 0.38, 0.89) and cancer of the gastric cardia (RR = 0.78; 95% CI:  0.64, 0.95). These results were similar to, but somewhat stronger than, those of the Behrens et al. (2014) meta-​analysis. The analysis was based on 899 cases of esophageal adenocarcinoma and 790 cases of gastric cardia cancer. The evidence for these two cancer endpoints was considered to be “limited–​suggestive.” Evidence for other gastroesophageal endpoints was too sparse to draw conclusions.

Hematologic Cancers A meta-​analysis by Jochem et al. (2014) included 23 studies (15 cohort, 8 case control) from North America, Europe, and Asia that examined the relationship between physical activity and risk of hematologic malignancies. The analysis was based on 1362 cases of myeloma, 1736 cases of leukemia (limited data available on subtypes), and 9447 cases of non-​ Hodgkin lymphoma. No statistically significant association was observed between physical activity and all hematologic cancers Category

combined. The RR estimates comparing high versus low physical activity were 0.86 (95% CI: 0.68, 1.09) for multiple myeloma, 0.97 (95% CI: 0.84, 1.13) for leukemia, and 0.91 (95% CI: 0.82, 1.00) for non-​ Hodgkin lymphoma. Results were generally similar between cohort and case-​control studies, and for recreational and occupational activity. In the 2016 pooled analysis, leisure-​ time physical activity was inversely associated with risk of myeloma (RR = 0.83; 95% CI: 0.72, 0.95), and myeloid leukemia (RR  =  0.80; 95% CI:  0.70, 0.92), and there was a borderline significant association for non-​Hodgkin lymphoma (RR = 0.91; 95% CI: 0.83, 1.00). No association was observed with lymphocytic leukemia. The point estimates for these hematologic cancers were similar but statistically more precise than those in the meta-​analysis by Jochem et al. (2014). Based on these analyses, the evidence for a causal relationship with myeloid leukemia, myeloma, and non-Hodgkin lymphoma was considered to be “limited–​suggestive.” Although the evidence for non-​Hodgkin lymphoma was based on more than 10,000 cases in the prospective studies, the association reached only borderline significance (p = 0.05) in both the meta-​analysis and the 2016 pooled analysis. For lymphocytic leukemia, the null findings indicate that a substantial effect is unlikely.

Head and Neck Cancer No meta-​ analyses have examined physical activity in relation to head and neck cancer risk. In the NIH-​AARP Diet and Health Study (Leitzmann et al., 2008), during up to 8 years of follow-​up, 1249 cases of head and neck cancer were diagnosed. In age-​and gender-​adjusted analyses, physical activity was inversely associated with head and neck cancer risk (RR  =  0.62; 95% CI:  0.52, 0.74) comparing the top and bottom quintiles. The association became statistically non-​significant (RR  =  0.89; 95% CI:  0.74, 1.06), however, after multivariate adjustment for smoking, alcohol consumption, and other factors. The authors stated that the association in age-​and gender-​adjusted analyses likely reflected some degree of residual confounding by smoking status in particular. In the pooled analysis, the risk of head and neck cancers was inversely associated with leisure-​time physical activity (RR  =  0.85; 95% CI:  0.78, 0.93). This association was statistically significant even after adjusting for smoking status, and was similar for never

# of Studies

RR (95% Cl)

Overall

27

0.76 (0.69, 0.85)

Cohort

21

0.79 (0.70, 0.89)

Case-control

6

0.64 (0.55, 0.74)

Men

17

0.82 (0.74, 0.90)

Women

10

0.83 (0.69, 0.99)

Never smokers

4

0.96 (0.78, 1.18)

Current smokers

2

0.66 (0.49, 0.89)

Adenocarcinoma

6

0.80 (0.72, 0.88)

Squamous

6

0.80 (0.71, 0.90)

Small cell

6

0.79 (0.66, 0.94)

0

1

Relative risk (95% Cl) High vs. low physical activity

Figure 21–​5.  Lung cancer and its association with physical activity.

2

 387



Physical Activity, Sedentary Behaviors, and Risk of Cancer

(RR = 0.83; 95% CI: 0.68, 1.02), former (RR = 0.90; 95% CI: 0.79, 1.04), and current smokers (RR  =  0.85; 95% CI:  0.65, 1.10). These results do not support residual confounding by smoking. The evidence was designated “limited–​suggestive.”

Lung Cancer Brenner et  al. (2016) conducted a meta-​analysis of 27 studies (21 cohort, 6 case control) from North America, Europe, and Asia on the association of recreational physical activity with risk of lung cancer. In total, the meta-​analyses included 20,169 cases. A significant 24% lower risk of being diagnosed with lung cancer was noted (RR = 0.76; 95% CI:  0.69, 0.85), comparing adults with higher versus lower levels of physical activity (see Figure 21–​5). A  subset of studies in the meta-​analysis stratified the analysis by smoking status. Physical inactivity was inversely associated with lung cancer in two studies of current smokers (RR = 0.66; 95% CI: 0.49, 0.89) but not in four studies that restricted the analysis to never smokers (RR = 0.96; 95% CI: 0.78, 1.18). In the 2016 pooled analysis, higher levels of leisure-​time physical activity were associated with lower risk of lung cancer in the overall population (RR  =  0.74, 95% CI:  0.71, 0.77), but not in an analysis restricted to never smokers (RR = 1.03; 95% CI: 0.89, 1.20). The null findings in never smokers underscore concern about residual confounding by smoking in the overall analysis. In summary, higher levels of physical activity are consistently associated with 25% lower risk of developing lung cancer in prospective studies. However, this association is susceptible to confounding by smoking, which is associated with both physical inactivity and a large (approximately 25-​fold) increase in lung cancer risk. The inverse association has not been observed in studies of never smokers. Unless it can be shown that physical activity decreases lung cancer risk in never smokers, or changes the natural history of lung cancer in smokers, the likelihood of confounding cannot be excluded. The overall level of evidence for a causal association between physical activity and lung cancer risk was designated “uncertain–​may be confounded.”

Malignant Melanoma No prior meta-​analyses or prospective cohort studies have examined the association between physical activity and risk of melanoma. Only one case-​control study with 386 cases had previously examined this association (Shors et al., 2001); this study reported that people with a high versus low level of physical activity had a 30% lower risk of melanoma. In contrast, the 2016 pooled analysis, based on 12,348 cases, found that leisure-​time physical activity was associated with increased risk of melanoma (RR = 1.27; 95% CI: 1.16, 1.40). Not only was the pooled analysis much larger than the single previous case-​control study, but separate analyses of 8 of the 12 cohorts found at least a 20% higher risk among those who were physically active. No biological mechanisms have been proposed to explain how physical activity might increase the risk of melanoma. An alternative hypothesis is that physically active individuals spend more time outdoors and have greater exposure to the sun. Physical activity is frequently done outdoors in light clothing, increasing the risk of sunburn (Holman et al., 2014). The overall level of evidence for a causal relationship between physical activity and melanoma risk is “uncertain–​ may be confounded.”

Pancreatic Cancer As reviewed by Behrens et al. (2015), a total of 30 studies (22 cohort, 8 case control) from North America, Europe and Asia have examined the association between physical activity and risk of incident pancreatic cancer. Taken together, these studies included 10,501 cases. The results differ markedly between prospective studies and case-​control studies and so are presented separately. The prospective studies,

387

which comprise most of the data on this issue, suggest that physically active participants have a 7% lower risk of pancreatic cancer (RR  =  0.93; 95% CI:  0.88, 0.98) than inactive participants. Results from case-​control studies show a larger reduction; risk was 22% lower (RR = 0.78; 95% CI: 0.66, 0.94) among the most active versus least active participants. A dose–​response analysis suggests that the association departs somewhat from linearity, reaching a plateau at high activity levels. The association had no evident heterogeneity by gender, smoking, or BMI, and no difference in results for recreational as opposed to occupational activity. The strength of the association did not appear to vary in relation to the timing of physical activity over the life span, though it was suggested that consistent physical activity over time may be more important than recent physical activity or physical activity during any given past period of life. Level of adjustment for different confounders also had little effect. Overall, these results indicate that the link between physical activity and lower risk of pancreatic cancer—​if a true association exists—​is comparatively weak. This represents a downgrading of the evidence since prior reviews (Schottenfeld and Fraumeni, 2006; World Cancer Research Fund and American Institute for Cancer Research, 2007), largely owing to the weak associations observed in more recent cohort studies. In the 2016 pooled analysis, leisure-​time physical activity was not significantly associated with the risk of pancreatic cancer (RR = 0.95; 95% CI:  0.83, 1.08), although the point estimate was similar to that from the prospective studies in the Behrens et  al. (2015) meta-​ analysis (RR = 0.93). These findings differ markedly from the results of case-​control studies (RR = 0.78). Consequently, the evidence level for an association between physical activity and pancreatic cancer is “inconsistent.”

Prostate Cancer A total of 43 studies (19 cohort, 24 case control) from North America and Europe have examined the association between physical activity and prostate cancer (see Figure 21–​6), as summarized by Liu et  al. (2011). The studies included a total of 88,294 cases. This meta-​analysis found that a high level of total physical activity (leisure-​time physical activity and occupational activity together) was inversely associated with prostate cancer risk (RR = 0.90; 95% CI: 0.84, 0.95). The association was weaker in prospective studies (RR = 0.94; 95% CI: 0.90, 0.98) than in case-​control studies (RR  =  0.86; 95% CI:  0.75, 0.97), however. Additionally, the associations were not consistent across domains of physical activity, with weaker associations for leisure-​time physical activity (RR  =  0.95; 95% CI:  0.89, 1.00) than for occupational physical activity (RR = 0.81; 95% CI: 0.73, 0.91). The reasons for the heterogeneity of associations across study designs and domains of physical activity are not well understood. Some evidence suggests that men who engage in a high level of leisure-​time physical activity are more frequently screened for prostate cancer, either by PSA or digital rectal exam, than men who engage in little leisure-​time physical activity (Moore et al., 2008). This may lead to higher detection rates among active men, and the resulting bias may obscure inverse associations that would have otherwise been present. The extent of detection bias could plausibly be different for leisure-​time physical activity than for occupational physical activity, leading to heterogeneity by type of activity. Fewer than half of the studies reported whether there was a significant trend in prostate cancer risk with increasing physical activity levels. Taken together with the disparate physical activity types and measurements, it was not possible to determine whether a dose–​ response association exists. Associations were stronger in studies where the mean age of men was less than 65 years than in studies where the mean age was higher, perhaps reflecting the greater range in activity levels among younger men. Whether this difference by age subgroup was statistically significant was not formally tested in the meta-​analysis, leaving open the possibility of a chance finding. Associations were also much stronger in studies with longer follow-​up time (i.e., greater than 10 years) than in studies with less follow-​up time, perhaps reflecting that a relatively

38

388

PART III:  THE CAUSES OF CANCER Category

# of Studies

RR (95% Cl)

Overall

43

0.90 (0.84, 0.95)

Cohort

19

0.94 (0.90, 0.98)

Case-control

24

0.86 (0.75, 0.97)

Localized cases

14

0.96 (0.87, 1.05)

Advanced cases

14

0.94 (0.80, 1.10)

Occupational

27

0.81 (0.73, 0.91)

LTPA

34

0.95 (0.89, 1.00)

0

1 Relative risk (95% Cl) High vs. low physical activity

2

Figure 21–​6.  Prostate cancer and its association with physical activity.

long period of physical activity may be required to decrease prostate cancer risk. In analyses of prostate cancer subtypes, physical activity was associated with approximately the same decrease in risk for both localized and advanced subtypes. The studies in the analysis included a wide range of potential confounders. These confounders appeared to have little effect on the association, as results did not vary between models that were more fully adjusted and those that were minimally adjusted. In contrast to previous studies, the 2016 pooled analysis showed that higher levels of leisure-​time physical activity were associated with increased, rather than decreased, risk of prostate cancer (RR = 1.06; 95% CI: 1.03, 1.08). Inconsistency in the results for prostate cancer is unsurprising given the concerns about detection bias and the heterogeneity of associations across study designs and domains of physical activity noted above in the Liu et al. (2011) meta-​analysis. In summary, while physical activity was associated with a modestly lower risk of prostate cancer, especially in case-​control studies, this was not confirmed by the pooled analysis of prospective studies. Consequently, the evidence that physical activity protects against prostate cancer risk is considered “inconsistent” rather than “probable” or “convincing.”

Rectal Cancer A meta-​ analysis by Robsahm et  al. (2013) compiled findings for 11 cohort studies (case-​control studies were excluded) from North America, Europe, Asia, and Australia that examined physical activity in relation to rectal cancer risk. There were 8698 rectal cancer cases overall. In the meta-​analysis, the authors found that the relative risk of rectal cancer comparing most active with least active participants in cohort studies was 0.98 (95% CI = 0.88, 1.08). No association was observed between physical activity and risk of developing rectal cancer. In the 2016 pooled analysis, leisure-​ time physical activity was inversely associated with risk of rectal cancer (RR  =  0.87; 95% CI:  0.80, 0.95). This finding contradicts the null association previously reported by Robsahm et al. (2013), leaving considerable uncertainty about the true relationship. We consider the current evidence for an association between physical activity and rectal cancer to be “inconsistent.”

Brain Cancer As reviewed by Niedermaier et  al. (2015), a total of six studies (4 cohort, 2 case control) from North America, Europe, and Australia (have examined physical activity in relation to risk of glioma (a heterogeneous cancer that comprises approximately 80% of all brain cancers). There were a total of 3057 glioma cases. The authors found that high versus low physical activity levels were inversely associated with risk of glioma (RR = 0.86; 95% CI: 0.76, 0.97). Preliminary evidence indicates that the magnitude of the association may vary according to the timing of physical activity during the life span, with physical activity during adolescence associated with greater reductions in risk than activity later in life (Moore et al., 2009). The 2016 pooled analysis found no association between leisure-​ time physical activity and brain cancer risk (RR = 1.06; 95% CI: 0.93, 1.20). This contradicts the prior finding of an inverse association for gliomas (Niedermaier et  al., 2015), leading us to conclude that the overall evidence for an association is “inconsistent.”

Thyroid Cancer As reviewed by Schmid et al. (2013), a total of eight studies (5 cohort, 3 case control) from North America, Europe, and Asia have examined the association between physical activity and risk of thyroid cancer. In total, their meta-​analysis included 2250 cases of thyroid cancer. The authors found that high versus low levels of physical activity had no association with thyroid cancer risk (RR = 1.06; 95% CI: 0.79, 1.42). In prospective studies, those with high versus low levels of physical activity had increased risk of thyroid cancer (RR = 1.28; 95% CI: 1.01, 1.63), whereas in case-​control studies, they had a statistically non-​ significant decrease in risk (RR = 0.70; 95% CI: 0.48, 1.03). In the 2016 pooled analysis, no association was observed between thyroid cancer and leisure-​time physical activity (RR = 0.92; 95% CI: 0.81, 1.06). The evidence for this association is considered to be inconsistent.

Ovarian Cancer Zhong et al. (2014) identified a total of 19 studies (9 cohort, 10 case control) from North America, Europe, Asia, and Australia on the

 389



Physical Activity, Sedentary Behaviors, and Risk of Cancer

association between non-​occupational (mostly recreational) physical activity and ovarian cancer risk. There were 9459 cases in the analysis. Overall, there was a borderline inverse association between physical activity and ovarian cancer risk (RR = 0.92; 95% CI: 0.84, 1.00). In prospective studies, no association was observed between physical activity and ovarian cancer risk (RR = 1.03; 95% CI: 0.87, 1.20), in contrast to the findings of case-​control studies (RR = 0.86; 95% CI: 0.80, 0.93). In the 2016 pooled analysis, leisure-​time physical activity was not associated with ovarian cancer risk (RR = 1.01; 95% CI: 0.91, 1.13). Given that results from cohort studies are afforded greater weight than case-​control studies due to their lower potential for recall and selection biases, the balance of evidence indicates that physical activity is unlikely to have a substantial effect on ovarian cancer risk.

All Other Cancers Other cancers that have been examined with regard to an association with physical activity include esophageal squamous cell carcinoma (Behrens et al., 2014), gallbladder cancer (Behrens et al., 2013), cancers of the small intestine (Cross et al., 2013), and soft tissue (Moore et  al., 2016). Data on these cancers are too sparse to draw conclusions about whether physical activity is associated with decreases or increases in risk.

SEDENTARY BEHAVIORS AND RISK OF DEVELOPING CANCER Sedentary behavior (i.e., sitting) has features that differ from general physical inactivity, and has been independently associated with various risk factors for chronic disease such as weight gain, high cholesterol, and high fasting insulin levels, as well as other biomarkers (Ford et al., 2005; Fung et al., 2000; Healy et al., 2008; Jakes et al., 2003). Sitting time has also been associated with premature mortality, cardiovascular disease, type 2 diabetes mellitus, and obesity (Biswas et al., 2014). Few studies have examined associations between sitting time and cancer with mixed results. A recent meta-​analysis, including 21 prospective cohorts and 22 case-​control studies, examined sitting time in relation to site-​specific cancer risk (Schmid and Leitzmann, 2014). Evidence was limited for most individual cancer sites, however. Only a few studies examined any given site, often in relation to different domains of sitting time (e.g., occupational, television, leisure-​time, or total). Approximately half of those studies examined occupational sitting time. Overall, the meta-​analysis found that sitting time was positively associated with colon and endometrial cancer risk. Following this meta-​ analysis, a recent study comprehensively examined leisure time spent sitting in relation to total and site-​specific cancer incidence and found that women who reported sitting for more than 6 hours/​day versus fewer than 3 hours/​day during their leisure time had a 10% higher risk of all cancers combined (Patel et al., 2015). The overall association in women was driven by site-​specific associations with invasive breast cancer (RR = 1.10; 95% CI: 1.00, 1.21), ovarian cancer (RR = 1.43; 95% CI: 1.10, 1.87), and multiple myeloma (RR = 1.65; 95% CI: 1.07, 2.54). Leisure-​time sitting was not associated with cancer risk in men overall, although among obese men there was an 11% higher risk associated with high levels of sitting time. Most previous studies reporting a positive association with colon cancer specifically examined television time (Howard et  al., 2008; Steindorf et  al., 2000)  or occupational sitting (Boyle et  al., 2010; George et al., 2010; Gerhardsson et al., 1986; Weiderpass et al., 2003), and only one study (Howard et al., 2008) reported an association with total sitting time. Associations appear to be strongest with television watching, but it is not clear whether this may reflect the greater precision of measuring television time versus overall sitting, or effect modification by unhealthy behaviors, such as excess snacking, that may correlate with television watching. While some previous studies have reported positive associations with endometrial cancer, these associations are attenuated by adjustment for BMI (Friberg et  al., 2006;

389

Gierach et  al., 2009; Patel et  al., 2015). Collectively, these findings suggest that the association between sitting time and endometrial cancer incidence may be confounded or mediated by BMI. Seven other studies have examined leisure-​time sitting in relation to cancer mortality (Dunstan et al., 2010; Katzmarzyk et al., 2009; Kim et al., 2013; Matthews et al., 2012a, 2014; Patel et al., 2010; Wijndaele et  al., 2010). Of these, three reported statistically significant positive associations between sitting time and cancer mortality in women (Kim et al., 2013; Matthews et al., 2012a; Patel et al., 2010), whereas only two reported positive associations in men (Kim et al., 2013; Matthews et al., 2012a). Evidence linking sitting time to other individual cancer sites is limited, although some associations have been reported. Two prospective studies reported modestly higher risk of invasive breast cancer associated with total sitting (George et  al., 2010)  or leisure-​time sitting (Patel et al., 2015). One prospective cohort (Patel et al., 2015) and one case-​control study (Zhang et al., 2004) reported a higher risk between leisure-​time sitting and ovarian cancer, whereas two studies of occupational sitting (Dosemeci et al., 1993; Lee et al., 2013) and one of total sitting time (Xiao et  al., 2013)  were null. A  statistically significant association has been reported with multiple myeloma and borderline associations with head and neck and gallbladder cancer in men and women; associations with esophageal cancer in women and pancreatic cancer in men have also been reported (Patel et al., 2015). In contrast, two prospective cohort studies have reported an inverse association with total prostate cancer, although there was no dose–​response relationship, and the results were attenuated when restricted to advanced disease (Lynch et  al., 2014; Patel et  al., 2015). In general, physical activities of a moderate to vigorous intensity, BMI, and age have not been shown to modify associations of sitting time with cancer risk. In summary, there is some, but not sufficient evidence to support a positive association between sitting time and some types of cancer, including colon, endometrium, breast, and ovary. The exact amount or domain of sitting time that may be associated with increased risk remains unclear; whether associations are modified by exercise or body weight requires further investigation. The role of body weight on the causal pathway for some cancers, such as endometrial cancer, also warrants further exploration.

PHYSICAL ACTIVITY, SEDENTARY BEHAVIORS, AND CANCER SURVIVORSHIP Currently, there are approximately 14.5  million individuals in the United States living after a diagnosis of cancer. While cancer treatments have improved, traditional therapies have significant toxicity, and targeted therapies are expensive. In the late 1980s researchers began to investigate whether exercise training during active treatment could mitigate the side effects of treatment and help to preserve physical function (MacVicar et al., 1989). This research has accelerated in the last decade. Courneya and Friedenreich (2007) have elegantly described where and how physical activity fits within the cancer control framework (Figure 21–​7). Here we briefly describe the evidence regarding the benefits of physical activity following a diagnosis of cancer. Points of intervention include pre-​treatment, treatment, and survivorship (Figure 21–​7). Exercise prior to surgery, chemotherapy, or radiation is prescribed as “prehab” to enhance physical fitness and coping prior to initiating cancer treatment. Cancer rehabilitation and health promotion programs also use exercise during and after treatment to reduce fatigue, enhance function, and improve the quality of life (Schmitz et al., 2010). There is ample evidence that exercise can be safe during this period. The American College of Sports Medicine has developed exercise guidelines to enhance the safety and effectiveness of exercise (Schmitz et al., 2010) during and after cancer treatment. The American Cancer Society has published recommendations for physical activity (Rock et al., 2012), and practical information for cancer survivors (Demark-​ Wahnefried et al., 2015). In the last decade, researchers have begun to investigate whether physical activity before or after diagnosis can improve survival and

390

390

PART III:  THE CAUSES OF CANCER DIAGNOSIS CANCER CONTROL CATEGORIES

Prevention

Detection

Prescreening

Screening

Treatment Preparation/ Coping

Treatment Effectiveness/ Coping

Pretreatment

Treatment

Recovery/ Rehabilitation

Disease Prevention/ Health Promotion

Survivorship

Palliation

Survival

End of Life

POSTDIAGNOSIS

PREDIAGNOSIS

CANCER-RELATED TIME PERIODS

Figure 21–​7.  Physical activity and cancer control framework.

reduce cancer recurrence. Recently, Schmid and Leitzmann (2014) reviewed 16 survivorship studies of breast and colorectal cancer encompassing 49,095 participants and 8129 total deaths. They examined physical activity performed before and after diagnosis in relation to cause-​specific mortality rates. Physical activity before diagnosis was associated with significantly lower death rates from breast cancer (HR = 0.77; 95% CI: 0.66, 0.90), and colorectal cancer (HR = 0.75; 95% CI: 0.62, 0.91). For post-​diagnosis activity, meeting currently recommended amounts of moderate to vigorous physical activity during the post-​diagnosis period was associated with 24% lower mortality for breast cancer survivors, and 28% lower risk for those previously diagnosed with colorectal cancer. For cancer-​specific death, post-​diagnosis physical activity was associated with a 28% lower risk for breast cancer mortality (HR = 0.72; 95% CI: 0.60, 0.85), as well as a 39% lower risk for colorectal cancer mortality (HR = 0.61; 95% CI: 0.40, 0.92). There is much less evidence for prostate cancer, although in some studies physical activity is associated with a lower risk of dying from prostate cancer, and recent studies have suggested that physical activity post-​diagnosis can improve survival among men diagnosed with prostate cancer (Bonn et al., 2014); still, much more work is needed. This emerging body of evidence suggests significant mortality benefits from physical activity for cancer survivors; however, some caveats must be considered. Many of the observational studies reporting these results were not originally designed to address this question, and there is often limited information about surgical procedures and outcomes, or detailed treatment information throughout the course of chemotherapy and radiation treatments, or information about compliance with long-​term hormonal treatment (e.g., aromatase inhibitors), all of which can have an impact on survival and may influence physical activity levels of study participants. Thus, the potential for residual confounding and/​or reverse causation calls for caution in interpreting these results. In summary, the number of individuals who will live for many years after their cancer diagnosis has increased dramatically in recent decades, and physical activity has emerged as an important potential component of our cancer control framework during the active treatment period and in the post-​treatment rehabilitation and health-​ promotion phases of cancer survivorship.

OPPORTUNITIES FOR PREVENTION Opportunities to prevent cancer by increasing physical activity levels and decreasing sedentary behaviors are discussed in detail in Chapter 62.2 of this volume.

FUTURE RESEARCH The relationship of physical activity with cancer risk is an active area of research, with many aspects of this relationship being investigated

simultaneously. We highlight two research areas of particular importance for improving our understanding of physical activity and cancer in the future.

Ascertaining the Type, Intensity, and Amount of Physical Activity Needed to Reduce Cancer Risk and Establishing the Role of Sedentary Behaviors in Cancer Incidence Studies of physical activity and cancer risk have typically quantified the relative risk for high versus low levels of physical activity. However, to develop clear recommendations with respect to cancer risk, it is also important to define the type, intensity, and amount of physical activity required for each level of benefit. For example, is there a particular kind of physical activity (e.g. aerobic vs. strength-​ oriented) that is especially beneficial with respect to reducing cancer risk? Are the benefits comparable for light versus moderate versus vigorous intensity activity? Especially important are efforts to define the dose–​response relationships between sedentary behavior, physical activity, and cancer risk. The relationship of physical activity with health and longevity is typically characterized as curvilinear, with the greatest benefits resulting from the transition from inactivity to the recommended minimum activity levels (Arem et al., 2015; Moore et al., 2012). As activity levels increase, the benefits are thought to diminish, ultimately reaching a plateau at activity levels two to three times that of the US recommended minimum levels. For cancer incidence, however, the dose–​response association may differ; preliminary evidence suggests that benefits may accrue in a more linear relationship with activity levels (Moore et al., 2016). If replicated, this would have important implications for the framing of physical activity recommendations in cancer prevention and control efforts. Most of the existing studies of sedentary behaviors and cancer risk have focused on relatively few types of cancer. Future studies should examine a broader range of cancer sites. In addition, not all studies explicitly account for the potential for sedentary behaviors to displace physical activity, especially light-​intensity physical activity. Future studies should evaluate the separate effects of sedentary behavior and physical inactivity, while accounting for potential displacement of active behaviors. The effects of prolonged sitting, independent of total sitting time, deserve further study.

Application of Novel Technologies to Improve Measurement of Physical Activity To date, most of the research on sedentary behavior, physical activity, and cancer has used self-​reported rather than objective measurement. Recall of activity levels over months or years in the past is difficult, in the absence of contextual cues. Misclassification of exposure due to

 391



Physical Activity, Sedentary Behaviors, and Risk of Cancer

errors in recall can attenuate relative risk estimates in epidemiologic studies. Newer technological approaches aim to minimize such error by using Internet-​based 24-​hour recalls, or accelerometry to measure movement objectively. Over the last 5  years, several automated Internet-​based 24-​hour recalls of physical activity behavior (Gomersall et al., 2011; Matthews et  al., 2012b) have been validated with respect to measuring sitting time, including different activity intensities and the location and purpose of the activity (Gomersall et al., 2011; Hunt et al., 2013; Keadle et al., 2014; Mace et al., 2014; Matthews et al., 2013). These recalls have been shown to have less reporting error than many questionnaires (Gomersall et al., 2011; Matthews et al., 2013), and can be administered multiple times per year in order to assess habitual or usual sitting and physical activity levels (Matthews et al., 2012b). To date, 24-​hour physical activity recalls have not been employed in large-​scale population studies due to logistical concerns. However, as epidemiologic studies move toward collecting participant data using mobile technologies, it will become increasingly feasible to integrate 24-​hour recalls into ongoing protocols for population-​based studies. Accelerometers are wearable motion sensors that measure accelerations in up to three different directions multiple times per second. These data are then integrated into estimates of overall energy expenditure per day or week, time in distinct activity intensity categories (e.g., sedentary, moderate), and/​or time seated versus standing. Accelerometers have several advantages over self-​report, including that they minimize recall and social desirability biases. Accelerometers perform well for estimating time in sedentary and light-​intensity activities, which is particularly difficult to measure from self-​reported data. They also place no cognitive demands on participants, thus making the technology attractive for studies in children. However, accelerometers are expensive, their application to large-​scale studies can be logistically complicated, and they yield complex data that are challenging to process (Lee and Shiroma, 2014). Despite these challenges, at least two studies with 10,000 or more participants with prospective follow-​up for disease outcomes are currently underway (Howard et al., 2015; Lee and Shiroma, 2014). In addition to research-​based accelerometers, commercial monitors to assess activity, such as the Fitbit and Apple Watch, are now widely available in consumer markets and are used by millions of people. Data from such commercial monitors could potentially be used to conduct large-​scale epidemiologic studies at a relatively low cost, provided that their reliability and validity can be confirmed.

CONCLUSION In the last 5 years, the publication of numerous prospective studies and meta-​analyses, as well the large consortium-​based analysis, have shed considerable light on the relationship between physical activity and cancer risk. In addition to the associations for colon, postmenopausal breast, and endometrial cancers previously identified, physical activity has now been associated with lower risk of multiple types of cancer, including, for example, gastroesophageal and hematologic malignancies. The extensive breadth of findings highlights a key role for physical activity in population-​level cancer prevention and control efforts. Although much has been learned in recent years about physical activity and cancer risk, many questions remain. Newly identified associations should be confirmed, and the type, intensity, and amount of physical activity needed to reduce cancer risk should be clearly defined. Newer technological methods should be implemented in cancer etiology studies. Finally, whether sedentary behaviors independently contribute to cancer risk should be thoroughly investigated in large prospective studies and consortia-​based studies. References Abbenhardt C, McTiernan A, Alfano CM, et al. 2013. Effects of individual and combined dietary weight loss and exercise interventions in postmenopausal women on adiponectin and leptin levels. J Intern Med, 274(2), 163–​175. PMCID: PMC3738194.

391

Arem H, Moore SC, Patel A, et  al. 2015. Leisure time physical activity and mortality:  a detailed pooled analysis of the dose-​response relationship. JAMA Intern Med, 175(6), 959–​967. PMCID: PMC4451435. Behavioral Risk Factor Surveillance System. Available from: http://​www.cdc. gov/​brfss/​. Accessed June 15, 2015. Behrens G, Jochem C, Keimling M, et al. 2014. The association between physical activity and gastroesophageal cancer:  systematic review and meta-​ analysis. Eur J Epidemiol, 29(3), 151–​170. PMID: 24705782. Behrens G, Jochem C, Schmid D, et al. 2015. Physical activity and risk of pancreatic cancer: a systematic review and meta-​analysis. Eur J Epidemiol, 30(4), 279–​298. PMID: 25773752. Behrens G, and Leitzmann MF. 2013. The association between physical activity and renal cancer: systematic review and meta-​analysis. Br J Cancer, 108(4), 798–​811. PMCID: PMC3590672. Behrens G, Matthews CE, Moore SC, et  al. 2013. The association between frequency of vigorous physical activity and hepatobiliary cancers in the NIH-​AARP Diet and Health Study. Eur J Epidemiol, 28(1), 55–​66. PMCID: PMC3670752. Biswas A, Oh P, Faulkner G, et  al. 2014. Sedentary time and its association with risk for disease incidence, mortality, and hospitalization in adults: a systematic review and meta-​analysis. Ann Intern Med, 162(2), 123–​132. PMID: 25599350. Bonn SE, Sjölander A, Lagerros YT, et  al. 2014. Physical activity and survival among men diagnosed with prostate cancer. Cancer Epidemiol Biomarkers Prev, 24(1), 57–​64. PMID: 25527697. Boyle T, Fritschi L, Heyworth J, and Bull F. 2010. Long-​ term sedentary work and the risk of subsite-​specific colorectal cancer. Am J Epidemiol, 173(10), 1183–​1191. PMID: 21421743. Boyle T, Keegel T, Bull F, Heyworth J, and Fritschi L. 2012. Physical activity and risks of proximal and distal colon cancers: a systematic review and meta-​analysis. J Natl Cancer Inst, 104(20), 1548–​1561. PMID: 22914790. Brenner DR, Yannitsos DH, Farris MS, Johansson M, and Friedenreich CM. 2016. Leisure-​time physical activity and lung cancer risk:  a systematic review and meta-​analysis. Lung Cancer, 95, 17–​27. PMID: 27040847. Brown SB, and Hankinson SE. 2015. Endogenous estrogens and the risk of breast, endometrial, and ovarian cancers. Steroids, 99(Pt A), 8–​ 10. PMID: 25555473. Calle EE, and Kaaks R. 2004. Overweight, obesity and cancer: epidemiological evidence and proposed mechanisms. Nat Rev Cancer, 4(8), 579–​591. PMID: 15286738. Campbell KL, Campbell PT, Ulrich CM, et al. 2008. No reduction in C-​reactive protein following a 12-​month randomized controlled trial of exercise in men and women. Cancer Epidemiol Biomarkers Prev, 17(7), 1714–​1718. PMCID: PMC2493470. Campbell PT, Campbell KL, Wener MH, et al. 2009. A yearlong exercise intervention decreases CRP among obese postmenopausal women. Med Sci Sports Exerc, 41(8), 1533–​1539. PMCID: PMC3850754. Carlson SA, Densmore D, Fulton JE, Yore MM, and Kohl HW, 3rd. 2009. Differences in physical activity prevalence and trends from 3 U.S.  surveillance systems: NHIS, NHANES, and BRFSS. J Phys Act Health, 6 Suppl 1, S18–​S27. PMID: 19998846. Caspersen C, Powell K, and Christenson G. 1985. Physical activity, exercise, and physical fitness:  definitions and distinctions for health related research. Public Health Rep, 100, 12–​131. PMCID: PMC1424733. Centers for Disease Control and Prevention. 2013. Adult participation in aerobic and muscle-​strengthening physical activities:  United States, 2011. MMWR Morb Mortal Wkly Rep, 62(17), 326–​330. PMID: 23636025. Chlebowski RT, Hendrix SL, Langer RD, et  al. 2003. Influence of estrogen plus progestin on breast cancer and mammography in healthy postmenopausal women: the Women’s Health Initiative Randomized Trial. JAMA, 289(24), 3243–​3253. PMID: 12824205. Courneya KS, and Friedenreich CM. 2007. Physical activity and cancer control. Semin Oncol Nursing, 23(4), 242–​252. PMID: 18022052. Coussens LM, and Werb Z. 2002. Inflammation and cancer. Nature, 420(6917), 860–​867. PMCID: PMC2803035. Cross AJ, Hollenbeck AR, and Park Y. 2013. A  large prospective study of risk factors for adenocarcinomas and malignant carcinoid tumors of the small intestine. Cancer Causes Control, 24(9), 1737–​ 1746. PMCID: PMC3737280. Demark-​Wahnefried W, Rogers LQ, Alfano CM, et al. 2015. Practical clinical interventions for diet, physical activity, and weight control in cancer survivors. CA Cancer J Clin, 65(3), 167–​189. PMID: 25683894. Dosemeci M, Hayes R, Vetter R, et  al. 1993. Occupational physical activity, socioeconomic status, and risks of 15 cancer sites in Turkey. Cancer Causes Control, 4(4), 313–​321. PMID: 8347780.

392

392

PART III:  THE CAUSES OF CANCER

Dunstan DW, Barr ELM, Healy GN, et  al. 2010. Television viewing time and mortality:  The Australian Diabetes, Obesity, and Lifestyle Study (AusDiab). Circulation, 121(3), 384–​391. PMID: 20065160. Endogenous Hormones and Breast Cancer Collaborative Group. 2010. Insulin-​ like growth factor 1 (IGF1), IGF binding protein 3 (IGFBP3), and breast cancer risk:  pooled individual data analysis of 17 prospective studies. Lancet Oncol, 11(6), 530–​542. PMID: 20472501 Ennour-​Idrissi K, Maunsell E, and Diorio C. 2015. Effect of physical activity on sex hormones in women:  a systematic review and meta-​analysis of randomized controlled trials. Breast Cancer Res, 17(1), 139. PMCID: PMC4635995. Ferrari P, Friedenreich C, and Matthews CE. 2007. The role of measurement error in estimating levels of physical activity. Am J Epidemiol, 166(7), 832–​840. PMID: 17670910. Ford ES, Kohl HW, 3rd, Mokdad AH, and Ajani UA. 2005. Sedentary behavior, physical activity, and the metabolic syndrome among U.S. adults. Obes Res, 13(3), 608–​614. PMID: 15833947. Frank LL, Sorensen BE, Yasui Y et al. 2005. Effects of exercise on metabolic risk variables in overweight postmenopausal women: a randomized clinical trial. Obes Res, 13(3), 615–​625. PMID: 15833948. Friberg E, Mantzoros CS, and Wolk A. 2006. Physical activity and endometrial cancer: a population-​based prospective cohort study. Cancer Epidemiol Biomarkers Prev, 15(11), 2136–​2140. Friedenreich CM. 2011. Physical activity and breast cancer: review of the epidemiologic evidence and biologic mechanisms. Recent Results Cancer Res, 188, 125–​139. PMID: 21253795. Friedenreich CM, Neilson HK, and Lynch BM. 2010. State of the epidemiological evidence on physical activity and cancer prevention. Eur J Cancer, 46(14), 2593–​2604. PMID: 20843488. Friedenreich CM, Neilson HK, Woolcott CG, et al. 2011. Changes in insulin resistance indicators, IGFs, and adipokines in a year-​long trial of aerobic exercise in postmenopausal women. Endocr Relat Cancer, 18(3), 357–​ 369. PMCID: PMC3111235. Fung TT, Hu FB, Yu J, et  al. 2000. Leisure-​time physical activity, television watching, and plasma biomarkers of obesity and cardiovascular disease risk. Am J Epidemiol, 152(12), 1171–​1178. PMID: 11130623. Garabrant DH, Peters JM, Mack TM, and Bernstein L. 1984. Job activity and colon cancer risk. Am J Epidemiol, 119(6), 1005–​ 1014. PMID: 6731427. George S, Irwin M, Matthews C, et al. 2010. Beyond recreational physical activity: examining occupational and household activity, transportation activity, and sedentary behavior in relation to postmenopausal breast cancer risk. Am J Public Health, 100(11), 2288–​2295. PMCID: PMC2951936. Gerhardsson M, Norell S, Kiviranta H, Pedersen N, and Ahlbom A. 1986. Sedentary jobs and colon cancer. Am J Epidemiol, 123(5), 775–​780. PMID: 3962961 Gierach GL, Chang S, Brinton LA, et  al. 2009. Physical activity, sedentary behavior, and endometrial cancer risk in the NIH-​AARP Diet and Health Study. Int J Cancer, 124(9), 2139–​2147. PMCID: PMC2845165. Godsland IF. 2010. Insulin resistance and hyperinsulinaemia in the development and progression of cancer. Clinical Science, 118(5-​6), 315–​332. PMCID: PMC2782313. Gomersall SR, Olds TS, and Ridley K. 2011. Development and evaluation of an adult use-​of-​time instrument with an energy expenditure focus. J Sci Med Sport, 14(2), 143–​148. PMID: 20932797. Gunter MJ, Hoover DR, Yu H, et al. 2008. A prospective evaluation of insulin and insulin-​like growth factor-​I as risk factors for endometrial cancer. Cancer Epidemiol Biomarkers Prev, 17(4), 921–​929. PMCID: PMC3090086. Gunter MJ, Hoover DR, Yu H, et al. 2009. Insulin, insulin-​like growth factor-​I, and risk of breast cancer in postmenopausal women. J Natl Cancer Inst, 101(1), 48–​60. PMCID: PMC2639294. Gunter MJ, Wang T, Cushman M, et  al. 2015. Circulating adipokines and inflammatory markers and postmenopausal breast cancer risk. J Natl Cancer Inst, 107(9). PMCID: PMC4651104. Hallal PC, Bauman AE, Heath GW, et al. 2012. Physical activity: more of the same is not enough. Lancet, 380(9838), 190–​191. PMID: 22818932. Hankinson AL, Daviglus ML, Bouchard C, et  al. 2010. Maintaining a high physical activity level over 20  years and weight gain. JAMA, 304(23), 2603–​2610. PMCID: PMC3864556. Hawley JA, Hargreaves M, Joyner MJ, and Zierath JR. 2014. Integrative biology of exercise. Cell, 159(4), 738–​749. PMID: 25417152. Healy GN, Clark BK, Winkler EA, et al. 2011. Measurement of adults’ sedentary time in population-​based studies. Am J Prev Med, 41(2), 216–​227. PMCID: PMC3179387. Healy GN, Wijndaele K, Dunstan DW, et al. 2008. Objectively measured sedentary time, physical activity, and metabolic risk. Diabetes Care, 31(2), 369–​371. PMID: 18000181.

Ho GY, Wang T, Gunter MJ, et  al. 2012. Adipokines linking obesity with colorectal cancer risk in postmenopausal women. Cancer Res, 72(12), 3029–​3037. PMCID: PMC3790260. Holman DM, Berkowitz Z, Guy GP, Jr., Hartman AM, and Perna FM. 2014. The association between demographic and behavioral characteristics and sunburn among U.S. adults: National Health Interview Survey, 2010. Prev Med, 63, 6–​12. PMCID: PMC4535173. Howard RA, Freedman DM, Park Y, et  al. 2008. Physical activity, sedentary behavior, and the risk of colon and rectal cancer in the NIH-​AARP Diet and Health Study. Cancer Causes Control, 19(9), 939–​953. PMID: 18437512. Howard VJ, Rhodes JD, Mosher A, et al. 2015. Obtaining accelerometer data in a national cohort of black and white adults. Med Sci Sports Exerc, 47(7), 1531–​1537. PMCID: PMC4404169. Hunt T, Williams MT, and Olds TS. 2013. Reliability and validity of the multimedia activity recall in children and adults (MARCA) in people with chronic obstructive pulmonary disease. PLoS One, 8(11), e81274. PMCID: PMC3842291. IARC Handbooks of Cancer Prevention. 2002. Volume 6: Weight Control and Physical Activity. Lyon, France: IARC Press. Ish-​Shalom D, Christoffersen CT, Vorwerk P, et  al. 1997. Mitogenic properties of insulin and insulin analogues mediated by the insulin receptor. Diabetologia, 40(Suppl 2), S25–​S31. PMID: 9248698. Jakes RW, Day NE, Khaw KT, et al. 2003. Television viewing and low participation in vigorous recreation are independently associated with obesity and markers of cardiovascular disease risk:  EPIC-​Norfolk population-​ based study. Eur J Clin Nutr, 57(9), 1089–​1096. PMID: 12947427. Jochem C, Leitzmann MF, Keimling M, Schmid D, and Behrens G. 2014. Physical activity in relation to risk of hematologic cancers: a systematic review and meta-​analysis. Cancer Epidemiol Biomarkers Prev, 23(5), 833–​846. PMID: 24633143. Judd HL, Shamonki IM, Frumar AM, and Lagasse LD. 1982. Origin of serum estradiol in postmenopausal women. Obstet Gynecol, 59(6), 680–​686. PMID: 7078905. Kaaks R, and Lukanova A. 2001. Energy balance and cancer:  the role of insulin and insulin-​like growth factor-​I. Proc Nutr Soc, 60(1), 91–​106. PMID: 11310428. Kahn BB, and Flier JS. 2000. Obesity and insulin resistance. J Clin Invest, 106(4), 473–​481. PMCID: PMC380258. Katzmarzyk PT, Church TS, Craig CL, and Bouchard C. 2009. Sitting time and mortality from all causes, cardiovascular disease, and cancer. Med Sci Sports Exerc, 41(5), 998–​1005. PMID: 19346988. Keimling M, Behrens G, Schmid D, Jochem C, and Leitzmann MF. 2014. The association between physical activity and bladder cancer: systematic review and meta-​analysis. Br J Cancer, 110(7), 1862–​1870. PMCID: PMC3974090. Kelley GA, and Kelley KS. 2006. Effects of aerobic exercise on C-​reactive protein, body composition, and maximum oxygen consumption in adults: a meta-​analysis of randomized controlled trials. Metabolism, 55(11), 1500–​ 1507. PMID: 17046553. Key TJ, Appleby PN, Reeves GK, et al. 2003. Body mass index, serum sex hormones, and breast cancer risk in postmenopausal women. J Natl Cancer Inst, 95(16), 1218–​1226. PMID: 12928347. Key TJ, Appleby PN, Reeves GK, et al. 2015. Steroid hormone measurements from different types of assays in relation to body mass index and breast cancer risk in postmenopausal women: Reanalysis of eighteen prospective studies. Steroids, 99(Pt A), 49–​55. PMCID: PMC4502556. Kim Y, Wilkens L, Park S-​Y, et  al. 2013. Association between various sedentary behaviours and all-​ cause, cardiovascular disease, and cancer mortality:  the Multiethnic Cohort. Int J Epidemiol, 42(4), 1040–​1056. PMCID: PMC3781003. Kozey Keadle S, Lyden K, Hickey A, et al. 2014. Validation of a previous day recall for measuring the location and purpose of active and sedentary behaviors compared to direct observation. Int J Behav Nutr Phys Act, 11, 12. PMCID: PMC3922158. Kushi LH, Doyle C, McCullough M, et  al. 2012. American Cancer Society Guidelines on nutrition and physical activity for cancer prevention: reducing the risk of cancer with healthy food choices and physical activity. CA Cancer J Clin, 62(1), 30–​67. PMID: 22237782. Lee A, Su D, Pasalich M, Wong Y, and Binns C. 2013. Habitual physical activity reduces risk of ovarian cancer: a case-​control study in southern China. Prev Med, 57, S31–​S33. Lee IM, Djousse L, Sesso HD, Wang L, and Buring JE. 2010. Physical activity and weight gain prevention. JAMA, 303(12), 1173–​ 1179. PMCID: PMC2846540. Lee IM, and Oguma Y. 2006. Physical activity. In:  Schottenfeld D, and Fraumeni JF, Jr. (Eds.), Cancer Epidemiology and Prevention (3rd ed., pp. 449–​467). New York: Oxford University Press.

 39



Physical Activity, Sedentary Behaviors, and Risk of Cancer

Lee IM, and Shiroma EJ. 2014. Using accelerometers to measure physical activity in large-​scale epidemiological studies: issues and challenges. Br J Sports Med, 48(3), 197–​201. PMCID: PMC3947179. Leitzmann MF, Koebnick C, Freedman ND, et al. 2008. Physical activity and head and neck cancer risk. Cancer Causes Control, 19(10), 1391–​1399. PMCID: PMC3507337. Liu Y, Hu F, Li D, et al. 2011. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-​analysis. Eur Urol, 60(5), 1029–​ 1044. PMID: 21802197. Lynch B, Friedenreich C, Kopciuk K, et  al. 2014. Sedentary behavior and prostate cancer risk in the NIH-​AARP Diet and Health Study. Cancer Epidemiol Biomarkers Prev, 23(5), 882–​889. PMCID: PMC4012010. Lynch BM, Dunstan DW, Healy GN, et al. 2010. Objectively measured physical activity and sedentary time of breast cancer survivors, and associations with adiposity:  findings from NHANES (2003–​2006). Cancer Causes Control, 21(2), 283–​288. PMID: 19882359. Mace CJ, Maddison R, Olds T, and Kerse N. 2014. Validation of a computerized use of time recall for activity measurement in advanced-​aged adults. J Aging Phys Act, 22(2), 245–​254. PMID: 23752221. MacVicar MG, Winningham ML, and Nickel JL. 1989. Effects of aerobic interval training on cancer patients’ functional capacity. Nursing Res, 38(6), 348–​351. PMID: 2587289. Mantovani A, Allavena P, Sica A, Balkwill F. 2008. Cancer-​related inflammation. Nature, 454(7203), 436–​444. PMID: 18650914. Matthews C, Cohen S, Fowke J, et  al. 2014. Physical activity, sedentary behavior, and cause-​specific mortality in black and white adults in the Southern Community Cohort Study. Am J Epidemiol, 180(4), 394–​405. PMCID: PMC4128770. Matthews CE. 2002. Techniques for physical activity assessment:  self-​report instruments. In: Welk G, and Dale D (Eds.), Physical Activity Assessments for Health-​ Related Research (pp. 107–​ 123). Champaign, IL:  Human Kinetics. Matthews CE, Ainsworth BE, Hanby C, et al. 2005. Development and testing of a short physical activity recall questionnaire. Med Sci Sports Exerc, 37(6), 986–​994. PMID: 15947724. Matthews CE, Chen KY, Freedson PS, et  al. 2008. Amount of time spent in sedentary behaviors in the United States, 2003–​2004. Am J Epidemiol, 167(7), 875–​881. PMCID: PMC3527832. Matthews CE, George SM, Moore S, et  al. 2012a. Amount of time spent in sedentary behaviors and cause-​specific mortality in US adults. Am J Clin Nutr, 95(2), 437–​445. PMCID: PMC3260070. Matthews CE, Keadle SK, Sampson J et al. 2013. Validation of a previous-​day recall measure of active and sedentary behaviors. Med Sci Sports Exerc, 45(8), 1629–​1638. PMCID: PMC3717193. Matthews CE, Moore SC, George SM, Sampson J, and Bowles HR. 2012b. Improving self-​reports of active and sedentary behaviors in large epidemiologic studies. Exerc Sport Sci Rev, 40(3), 118–​126. PMCID: PMC3388604. McTiernan A. 2008. Mechanisms linking physical activity with cancer. Nat Rev Cancer, 8(3), 205–​211. PMID: 18235448. McTiernan A, Sorensen B, Yasui Y et al. 2005. No effect of exercise on insulin-​ like growth factor 1 and insulin-​like growth factor binding protein 3 in postmenopausal women:  a 12-​month randomized clinical trial. Cancer Epidemiol Biomarkers Prev, 14(4), 1020–​1021. PMID: 15824183. Moore SC, Lee IM, Weiderpass E et al. 2016. Association of leisure-​time physical activity with risk of 26 types of cancer in 1.44 million adults. JAMA Intern Med, 176(6), 816–​825. PMID: 27183032. Moore SC, Patel AV, Matthews CE, et al. 2012. Leisure time physical activity of moderate to vigorous intensity and mortality:  a large pooled cohort analysis. PLoS Med, 9(11), e1001335. PMCID: PMC3491006. Moore SC, Peters TM, Ahn J, et al. 2008. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev, 17(9), 2458–​2466. PMCID: PMC3443883. Moore SC, Rajaraman P, Dubrow R, et al. 2009. Height, body mass index, and physical activity in relation to glioma risk. Cancer Res, 69(21), 8349–​ 8355. PMCID: PMC2783605. Moradi T, Nyren O, Bergstrom R et al. 1998. Risk for endometrial cancer in relation to occupational physical activity:  a nationwide cohort study in Sweden. Int J Cancer, 76(5), 665–​670. PMID: 9610723. Mu N, Zhu Y, Wang Y, Zhang H, and Xue F. 2012. Insulin resistance: a significant risk factor of endometrial cancer. Gynecol Oncol, 125(3), 751–​757. PMID: 22449736. National Center for Health Statistics. (2015). Health, United States, 2014: With Special Feature on Adults Aged 55–​64. Hyattsville, MD: NCHS. National Health and Nutrition Examination Survey. Available from:  http://​ www.cdc.gov/​nchs/​nhanes.htm Accessed June 15, 2015. National Health Interview Survey. Available from:  http://​www.cdc.gov/​nchs/​ nhis.htm Accessed June 15, 2015.

393

Neilson HK, Robson PJ, Friedenreich CM, and Csizmadi I. 2008. Estimating activity energy expenditure:  how valid are physical activity questionnaires? Am J Clin Nutr, 87(2), 279–​291. PMID: 18258615. Niedermaier T, Behrens G, Schmid D, et  al. 2015. Body mass index, physical activity, and risk of adult meningioma and glioma: a meta-​analysis. Neurology, 85(15), 1342–​1350. PMID: 26377253. Nimptsch K, Aleksandrova K, Boeing H, et al. 2015. Association of CRP genetic variants with blood concentrations of C-​reactive protein and colorectal cancer risk. Int J Cancer, 136(5), 1181–​1192. PMID: 25043606. Ollberding NJ, Kim Y, Shvetsov YB, et al. 2013. Prediagnostic leptin, adiponectin, C-​reactive protein, and the risk of postmenopausal breast cancer. Cancer Prev Res (Phila), 6(3), 188–​195. PMCID: PMC3595121. Patel AV, Bernstein L, Deka A, et al. 2010. Leisure time spent sitting in relation to total mortality in a prospective cohort of US adults. Am J Epidemiol, 172(4), 419–​429. PMCID: PMC3590043. Patel AV, Hildebrand JS, Campbell PT, et al. 2015. Leisure-​time spent sitting and site-​specific cancer incidence in a large US cohort. Cancer Epidemiol Biomarkers Prev, 24(9), 1350–​1359. PMID: 26126627. Pedersen BK, and Febbraio MA. 2012. Muscles, exercise and obesity:  skeletal muscle as a secretory organ. Nat Rev Endocrinol, 8(8), 457–​465. PMID: 22473333. Pereira MA, Fitzgerald SJ, Gregg EW, et  al. 1997. A  collection of physical activity for health-​ related research. Med Sci Sports Exerc, 29(6 Supplement), S1–​S203. Physical Activity Guidelines Advisory Committee. 2008. Physical Activity Guidelines Advisory Committee Report (pp. A1–​ A10). Washington, DC: U.S. Department of Health and Human Services. Pugeat M, Crave JC, Elmidani M, et al. 1991. Pathophysiology of sex hormone binding globulin (SHBG): relation to insulin. J Steroid Biochem Mol Biol, 40(4–​6), 841–​849. PMID: 1958579. Robsahm TE, Aagnes B, Hjartåker A, Langseth H, Bray FI, Larsen IK. 2013. Body mass index, physical activity, and colorectal cancer by anatomical subsites: a systematic review and meta-​analysis of cohort studies. Eur J Cancer Prev, 22(6), 492–​505. PMID: 23591454. Rock CL, Doyle C, Demark-​Wahnefried W, et al. 2012. Nutrition and physical activity guidelines for cancer survivors. CA Cancer J Clin, 62(4), 243–​ 274. PMID: 22539238. Sallis JF. 1997. A collection of physical activity questionnaires for health-​ related research: seven-​day physical activity recall. In: Kriska AM, and Casperson CJ (Eds.), A Collection of Physical Activity Questionnaires. Med Sci Sports Exerc (Vol. 29, pp. S89–​S103). Sallis JF, and Saelens BE. 2000. Assessment of physical activity by self-​report: limitations and future directions. Res Q Exercise Sport, 71(Suppl 2), S1–​S14. Schmid D, Behrens G, Jochem C, Keimling M, and Leitzmann M. 2013. Physical activity, diabetes, and risk of thyroid cancer: a systematic review and meta-​analysis. Eur J Epidemiol, 28(12), 945–​958. PMID: 24243033. Schmid D, Behrens G, Keimling M, et al. 2015. A systematic review and meta-​ analysis of physical activity and endometrial cancer risk. Eur J Epidemiol, 30(5), 397–​412. PMID: 25800123. Schmid D, and Leitzmann M. 2014. Television viewing and time spent sedentary in relation to cancer risk: a meta-​analysis. J Natl Cancer Inst, 106(7), dju098. PMID: 24935969. Schmid D, and Leitzmann MF. 2014. Association between physical activity and mortality among breast cancer and colorectal cancer survivors: a systematic review and meta-​analysis. Ann Oncol, 25(7), 1293–​1311. PMID: 24644304. Schmitz KH, Courneya KS, Matthews C, et  al. 2010. American College of Sports Medicine roundtable on exercise guidelines for cancer survivors. Med Sci Sports Exerc, 42(7), 1409–​1426. PMID: 20559064. Schottenfeld D, and Fraumeni JF, Jr. (Eds.) 2006. Cancer Epidemiology and Prevention (3rd ed.). New York: Oxford University Press. Shafie SM, and Hilf R. 1981. Insulin receptor levels and magnitude of insulin-​ induced responses in 7,12-​dimethylbenz(a)anthracene-​induced mammary tumors in rats. Cancer Res, 41(3), 826–​829. PMID: 6780189. Shors AR, Solomon C, McTiernan A, and White E. 2001. Melanoma risk in relation to height, weight, and exercise (United States). Cancer Causes Control, 12(7), 599–​606. PMID: 11552707. Steindorf K, Tobiasz-​Adamczyk B, Popiela T, et  al. 2000. Combined risk assessment of physical activity and dietary habits on the development of colorectal cancer:  a hospital-​based case-​control study in Poland. Eur J Cancer Prev, 9(5), 309–​316. PMID: 11075883. Stolzenberg-​Solomon RZ, Graubard BI, Chari S et  al. 2005. Insulin, glucose, insulin resistance, and pancreatic cancer in male smokers. JAMA, 294(22), 2872–​2878. PMID: 16352795. Toriola AT, Cheng TY, Neuhouser ML, et  al. 2013. Biomarkers of inflammation are associated with colorectal cancer risk in women but are not suitable as early detection markers. Int J Cancer, 132(11), 2648–​2658. PMCID: PMC3609926.

394

394

PART III:  THE CAUSES OF CANCER

Troiano RP, Berrigan D, Dodd KW, et al. 2008. Physical activity in the United States measured by accelerometer. Med Sci Sports Exerc, 40(1), 181–​188. PMID: 18091006. Tucker JM, Welk GJ, and Beyler NK. 2011. Physical activity in U.S. adults: compliance with the physical activity guidelines for Americans. Am J Prev Med, 40(4), 454–​461. PMID: 21406280. US Department of Health and Human Services. 2008 Physical Activity Guidelines for Americans. Available from: http://​www.health.gov/​paguidelines/​Accessed September 6, 2013. van Poppel MNM, Chinapaw MJM, Mokkink LB, van Mechelen W, and Terwee CB. 2010. Physical activity questionnaires for adults: a systematic review of measurement properties. Sports Med, 40(7), 565–​600. PMID: 20545381. Wang T, Rohan TE, Gunter MJ, et  al. 2011. A  prospective study of inflammation markers and endometrial cancer risk in postmenopausal hormone nonusers. Cancer Epidemiol Biomarkers Prev, 20(5), 971–​977. PMCID: PMC3096873. Watson KB, Carlson SA, Carroll DD, and Fulton JE. 2014. Comparison of accelerometer cut points to estimate physical activity in US adults. J Sports Sci, 32(7), 660–​669. PMCID: PMC4589136. Weiderpass E, Adami HO, Baron JA, et  al. 1999. Risk of endometrial cancer following estrogen replacement with and without progestins. J Natl Cancer Inst, 91(13), 1131–​1137. PMID: 10393721. Weiderpass E, Vainio H, Kauppinen T, et al. 2003. Occupational exposures and gastrointestinal cancers among Finnish women. J Occup Environ Med, 45(3), 305–​315. PMID: 12661188. Wijndaele K, Brage S, Besson H, et al. 2010. Television viewing time independently predicts all-​cause and cardiovascular disease mortality: the EPIC Norfolk Study. Int J Epidemiol, 40(1), 150–​159. PMID: 20576628.

Wolin KY, Lee IM, Colditz GA, et al. 2007. Leisure-​time physical activity patterns and risk of colon cancer in women. Int J Cancer, 121(12), 2776–​ 2781. PMID: 17722094. Wolin KY, Yan Y, Colditz GA, and Lee IM. 2009. Physical activity and colon cancer prevention: a meta-​analysis. Br J Cancer, 100(4), 611–​616. PMID: 19209175. World Cancer Research Fund, and American Institute for Cancer Research. 2007. Food, nutrition, physical activity, and the prevention of cancer: a global perspective. Available from:  http://​www.dietandcancerreport.org/​ Accessed June 15, 2015. World Health Organization. 2010. Global Recommendations on Physical Activity for Health. Geneva: WHO. Wu Y, Zhang D, and Kang S. 2013. Physical activity and risk of breast cancer: a meta-​analysis of prospective studies. Breast Cancer Res Treat, 137(3), 869–​882. PMID: 23274845. Xiao Q, Yang H, Wentzensen N, Hollenbeck A, and Matthews C. 2013. Physical activity in different periods of life, sedentary behavior and the risk of ovarian cancer in the NIH-​AARP Diet and Health Study. Cancer Epidemiol Biomarkers Prev, 22(11), 2000–​2008. PMID: 23966580. Yager JD, and Davidson NE. 2006. Estrogen carcinogenesis in breast cancer. N Engl J Med, 354(3), 270–​282. PMID: 16421368. Zapf J, Schmid C, and Froesch ER. 1984. Biological and immunological properties of insulin-​like growth factors (IGF) I and II. Clin Endocrinol Metab, 13(1), 3–​30. PMID: 6202445. Zhang M, Xie X, Lee AH, and Binns CW. 2004. Sedentary behaviors and epithelial ovarian cancer risk. Cancer Causes Control, 15(1), 83–​89. PMID: 14970738. Zhong S, Chen L, Lv M, et al. 2014. Nonoccupational physical activity and risk of ovarian cancer: a meta-​analysis. Tumour Biol, 35(11), 11065–​11073. PMID: 25095977.

 395

22

Hormones and Cancer ROBERT N. HOOVER, AMANDA BLACK, AND REBECCA TROISI

OVERVIEW Hormones are highly biologically active endogenous compounds that control the growth, development, physiology, and homeostasis of numerous organ systems. Because of this, they have long been thought likely to play key roles in both normal and abnormal (malignant) growth. They are also noteworthy for being produced away from the tissues that they control, and are thus secreted into circulating blood to reach their target organs. This combination of potent, targeted agents of growth and development that can be measured in available biologic fluids has made them particularly susceptible and relevant to epidemiologic investigation. In addition, as pharmacologic science has progressed, medications containing hormones and hormone antagonists have come into widespread use, providing further opportunities for epidemiologic insights into hormonal carcinogenesis. Based on the development of increasingly more accurate assays to measure sex hormones and their metabolites, this has been the most productive subject matter area to date, with major advances in understanding the hormonal etiology of breast and gynecologic malignancies. It has also taught us the challenges and complexities of these studies, such as our continuing failure to uncover a role for androgens in prostate cancer etiology. One major lesson learned is that because of the biological importance of hormones, there are multiple mechanisms to regulate them and their effects. These include some things that can be measured and accounted for, such as circulating binding proteins that inhibit the hormone’s ability to bind to the tissues it regulates. However, there are also potent effects of other regulatory mechanisms that are not easily accounted for in epidemiologic studies. Perhaps the most important of these are hormone receptors to which virtually all hormones bind in order to initiate their actions. A  variety of homeostatic mechanisms control the activation and availability of these receptors, including increasing activation in the presence of declining hormone levels, and decreasing activation with rising levels. With the major and rapid advances in molecular biology and molecular epidemiology, we are seeing a surge of new opportunities and creative new investigations across the entire range of human hormones, and anticipate major advances in our understanding of human hormonal carcinogenesis in the near future. To take full advantage of these opportunities, we will also need to learn how to account for the relevant regulatory mechanisms that evolution has developed to control the effects of these potent biologic agents.

INTRODUCTION Hormones are chemical signals secreted into the bloodstream that act on distant tissues, usually in a regulatory fashion (Melmed et al., 2016). In humans, hormones have important roles in growth, development, homeostasis, and maintenance of normal physiology. As such, it would seem that they might also have a central role, for better or worse, in the process and regulation of abnormal growth and development, as in malignancies. Despite this, with one notable exception, hormones have garnered little interest in cancer epidemiology. The exception is the area of sex hormones, which for decades has been a major focus of investigations of cancers of the reproductive system. Currently, we may be seeing the beginnings of a broader epidemiologic interest in hormonal carcinogenesis in the context of the recent revolution in molecular science, opening possibilities for molecular epidemiology

to shed light on the role of metabolic pathways in cancer—​pathways that often involve hormonal control. The recognition of the potentially central roles that two such hormones—​insulin and IGF-​1—​may play in multiple malignancies is one example of this trend (Renehan et al., 2004). Indeed, the very existence of this chapter is further evidence of an expanded interest beyond sex hormones. The first two editions of this text had no such chapter, and the third edition had one, restricted to “Exogenous Hormones.” The current chapter is not intended to be a compilation of the state of our current knowledge of which hormones are, or might be, related to specific cancers or of cancers of hormone-​producing glands, as these will be summarized, along with the supporting data, in the relevant chapters on site-​specific cancers. Instead, we will discuss some of the history of the attempts to identify and understand hormonal carcinogenesis via epidemiologic investigations, and what we might learn from these efforts. We will begin with the sex hormones, since as noted, they remain by far the most extensively investigated hormones in relation to cancer risk. Following this will be a brief discussion of other hormones that have come under exploration more recently, and likely represent the vanguard of a more concerted effort in the near future to exploit the opportunities provided by the various “-​omics” and other new molecular epidemiologic tools now being developed. Finally, we will also discuss how lessons learned thus far might guide us in designing new investigations in this era of interdisciplinary studies and rapid expansion of our understanding of the underlying biology of cancer. It is our hope that when the time for the fifth edition of this text arrives, the authorship of this chapter will need to be extensively expanded in order to cover the new knowledge brought about by using these new tools to study hormones responsible for controlling so much of human physiology.

SEX HORMONES Estrogen, progesterone, and androgens (particularly testosterone) are hormones that have specific roles in reproduction and sexual development, and have thus been labeled as sex hormones. Studies of their effects on cancer risks have consequently been focused on reproductive organs—​specifically, the roles of these hormones in breast, endometrial, and ovarian cancers in women, and in prostate and testicular cancers in men. While appropriate, this focus is also somewhat myopic since all of these hormones are biologically active in more organs than simply the reproductive ones, as evidenced by the presence of estrogen receptors in the central nervous, cardiovascular, respiratory, gastrointestinal, urinary, and other organ systems (Levin, 2015; Millas and Liquidato, 2009). The observation of a possible protective effect of menopausal estrogen therapy on colon cancer risk (Chan and Giovannucci, 2010), and the recent appreciation that the risk of breast cancer associated with circulating levels of testosterone likely may not be wholly explained by its role as a substrate for estrogen synthesis indicate the need for broader and more biologically based assessments.

Breast Cancer As noted in the introduction, the history of our past attempts at using epidemiologic approaches to understand hormonal carcinogenesis can teach us valuable lessons about productive ways to pursue these goals in the future. Perhaps the most instructive example is the study of breast cancer,

395

396

396

PART III:  THE CAUSES OF CANCER

which has a long history, filled with innovative methods and creative insights, as well as missed opportunities and flawed interpretations. In 1700, Bernardino Ramazzini noted that “… cancerous tumors are very often generated in the woman’s breast, and tumors of this sort are found in nuns more often than in other women. Now these are not caused by suppression of the menses but rather, in my opinion, by their celibate life” (Ramazzini, 1700). It was over 100  years later that Rigoni-​Stern quantified this observation by noting that the proportional mortality rates for breast cancer in Verona, Italy, and its surrounding area were six times higher in nuns compared to women in the general population (Rigoni-​ Stern, 1842). One of the first to link breast cancer risk to ovarian function was Beatson, who in 1896 posed the question, “Is cancer of the mamma due to some ovarian irritation, as from some defective steps in the cycle of ovarian changes; and, if so, would the cell proliferation be brought to a standstill … were the ovaries to be removed” (Beatson, 1896). The concept of chemical “messengers” in the body had been recognized in the mid-​nineteenth century (Bernard, 1949), and persons with certain disorders were being treated with extracts of various endocrine organs by late in the century (Borell, 1976). However, laboratory enthusiasm to pursue studies of these compounds did not really take hold until the early 1900s, when Ernest Starling coined the term “hormone”: “These chemical messengers, however, or hormones (from the Greek word óρμáω, I  excite or arouse), as we might call them, have to be carried from the organ where they are produced to the organ which they affect by means of the blood stream and the continually recurring physiological needs of the organism must determine their repeated production and circulation through the body” (Henderson, 2005). Even with this stimulus, it was not until 1929 that the estrogen molecule was identified and characterized (Tata, 2005). However, this was quickly followed by laboratory animal experiments linking estrogen exposure to increased breast cancer risk (Kordon, 2008). While this led to a substantial and sustained laboratory commitment to understanding the role of estrogens in breast and other cancers, epidemiologic advances had to wait for the maturation of the newly developing field of chronic disease epidemiology.

Hormonally Linked Risk Factors

In the 1950s and 1960s, Brian MacMahon (MacMahon et al., 1970), Abraham Lilienfeld (Lilienfeld, 1956), Ernst Wynder (Wynder et al., 1960), and several other pioneers in this field turned their attention to breast cancer and discovered multiple important risk factors, including many associated with reproduction—​ nulliparity, parity within the parous, age at menarche, age at first birth, lactation, and age at menopause. About the same time, obesity among postmenopausal women was also identified (Kelsey, 1979). Given the hormone-​related

Duration of use and time since last use Never-user

Cases/Controls 12467/23568

nature of these risk factors, and the compelling findings from animal studies (Mceuen, 1939; Nandi et  al., 1995), estrogens were indicted as the most likely responsible carcinogen. Geographic pathologists also contributed, identifying major international differences in breast cancer risk, most notably between East Asian and Western countries (Haenszel and Kurihara, 1968), differences that progressively disappeared over two to three generations of migrants from low-​to high-​ risk countries (Ziegler et al., 1993). These geographic differences and changes with migration were also attributed to hormonal differences, specifically circulating estrogen levels that were thought to be driven by major differences in dietary fat consumption (Lea, 1966). The belief that hormones, and specifically estrogens, explain these risk factors was so strong that little was done to test these hypotheses. Rather, they were combined to support a unifying hypothesis that virtually all established breast cancer risk factors were consistent with an explanation of cumulative lifetime estrogen exposure being the cause of breast cancer (Henderson and Feigelson, 2000; Marshall, 1993). Over time, many observations have been made that seem inconsistent with this hypothesis, and we will return to this point later.

Exogenous Hormones

Surprisingly understudied by the mid-​1970s were estrogen drugs. The first synthetic estrogens were produced in 1938 (Dodds et al., 1938), and shortly thereafter were administered to women to treat various disorders of pregnancy, menopausal symptoms, and other conditions. Assessments of the long-​term consequences of these therapies were few, and most were methodologically flawed. Indeed, a publication in 1971 (Defares, 1971)  summarized four small follow-​up studies that included 1130 women treated with estrogen drugs for menopause, and noted that 74 cases of cancer would have been expected, but only 2 were observed, concluding that exogenous estrogens were profoundly protective for all cancer. Beginning in 1976 (Hoover et al., 1976), and over the course of the next 15 years, a series of epidemiologic studies of estrogen therapy (ET) for menopause was conducted. Most, but not all, demonstrated a positive association of breast cancer with duration of ET use. A combined analysis of studies conducted by the early 1990s found an overall association with breast cancer that was particularly consistent and strong among those studies rated as having high-​quality methods (Steinberg et al., 1991). Larger studies and pooled analyses added to and refined our understanding of the role of exogenous estrogen in carcinogenesis. Particularly noteworthy was the finding that stopping ET resulted in a rapid and marked decline in risk (Beral and Million Women Study Collaborators, 2003; Collaborative Group on Hormonal Factors in Breast Cancer, 1997). In fact, as

RR (FSE)*

RR and 99% FCl*

1.00 (0.021)

Last use < 5 years before diagnosis Duration < 1 year Duration 1–4 years Duration 5–9 years Duration 10–14 years Duration ≥ 15 years

368/860 891/2037 588/1279 304/633 294/514

0.99 (0.085) 1.08 (0.060) 1.31 (0.079) 1.24 (0.108) 1.56 (0.128)

Last use ≥ 5 years before diagnosis Duration < 1 year Duration 1–4 years Duration 5–9 years Duration ≥ 10 years

437/890 566/1256 151/374 93/233

1.12 (0.079) 1.12 (0.068) 0.90 (0.115) 0.95 (0.145) 0

0.5 1.0 1.5 2.0

Figure 22–​1.  Relative risk (RR) of breast cancer for duration of use within categories of time since last use of HRT. *Source: Collaborative Group on

Hormonal Factors in Breast Cancer. Breast cancer and hormone replacement therapy: collaborative reanalysis of data from 51 epidemiological studies of 52,705 women with breast cancer and 108 411 women without breast cancer. Lancet 1997;350:1047–1059.

 397



397

Hormones and Cancer

depicted in Figure 22–​1 within only a few years of stopping, long-​term users had risks equivalent to those who never used ET (Figure 22–​1). The addition of progestins to ET (EPT) was also a learning experience. A year before the 1976 publication linking ET to breast cancer risk, two studies (Ziel and Finkle, 1975; Smith et al., 1975) found a very strong increased risk of endometrial cancer with use of ET. The combination of strong associations with two common cancers appearing over a short time period led to a decline in ET prescriptions. Given that progestins were known to differentiate the endometrium, stopping estrogen-​driven proliferation, formulations combining both an estrogen and a progestin were introduced and became increasingly popular. Suggestions were made that in addition to lowering the risk of endometrial cancer related to ET, combined formulations would likely also ameliorate any increase in breast cancer (Gambrell, 1984), even though it had been established that progesterone had different biologic activity in the breast versus the endometrial epithelium (Key and Pike, 1988). In 1989, data from a Swedish cohort (Bergkvist et  al., 1989; Persson et  al., 1999)  suggested that while endometrial cancer risk was substantially lower among the EPT users versus ET users, breast cancer was not, and might actually be higher among EPT users. This finding of increased breast cancer risk was subsequently replicated in larger data sets (Schairer et al., 2000), indicating significant heterogeneity in risk between EPT and ET use. In the mid-​1990s, the Women’s Health Initiative (Rossouw et  al., 1995), a randomized clinical trial of hormone therapy in menopausal women, was initiated. The main goal of the trial was to determine whether the suggestion from observational studies that hormone therapy in menopausal women might provide protection from cardiovascular disease was correct, and if so, whether the benefit outweighed the risks of treatment, particularly the increase in breast cancer. Women with an intact uterus were randomized to EPT or placebo, and women who had undergone hysterectomy were randomized to ET or placebo. The EPT trial was stopped when women had been on the medications for an average of 5.2 years. The primary reason for stopping was an excess risk of cardiovascular disease, rather than an anticipated deficit, among those taking the active drugs (Manson et al., 2003; Writing Group for WHI, 2002). At the time of stopping, a significant increase in breast cancer with increasing duration of EPT use, similar to that seen in the observational studies, was observed. Since stopping, the EPT group has experienced a marked reduction in excess breast cancer risk (Chlebowski et al., 2009), also similar to the findings in the observational studies. The ET trial was also stopped early, primarily because of an excess risk of strokes in the exposed group. For the women in the ET group, however, there was no evidence of increased breast cancer risk (Anderson et al., 2012). In fact, at the time of stopping, this group had a non-​significant deficit of 20%; early in the post-​intervention period, this deficit remained and became statistically significant (Manson et al., 2013). Indeed, this association had a rather unusual pattern, as depicted in Figure 22–​2 with the deficit appearing almost immediately after trial

initiation, and remaining at about the same level during the trial and also for several years after cessation of treatment (Figure 22–​2). Based on the results of observational studies described in the preceding, this finding of apparent protection in the ET group was unexpected. On closer inspection, the reduced risk appeared to be driven by subjects in the trial who began use 5 or more years after their menopause, whereas those who began their use closer in time to menopause had risks similar to the placebo group (Prentice, 2009). It has been noted that this pattern of reduced risk could be consistent with the observation that among women with breast cancer, treatments resulting in long-​term estrogen deprivation can cause “the eventual development and evolution of anti-​hormonal resistant cell populations that emerge with a vulnerability, as estrogen is no longer a survival signal, but is an apoptotic trigger” (Lewis-​Wambi and Jordan, 2009; Obiorah et al., 2014). In a very large cohort study that had enough women to assess those who had not begun use of ET until 5 years or more after menopause, these women were not at a reduced risk of breast cancer, but also were not at increased risk, while those who began use within 5  years of menopause were at a significantly increased risk (Beral et al., 2011). There are also a number of other, more mundane reasons that the trial may not be able to optimally address the observational findings. One relates to dose–​response. In the ET trial, the average follow-​up time at stopping was 6.8 years (Anderson et al., 2004). In several of the larger observational studies, the excess risk does not appear until at least 5 to 8 years of exposure, and rises with further use (Collaborative Group on Hormonal Factors, 1997; Schairer et  al., 2000). Also, as noted earlier, most of the observational studies have found evidence of a strong interaction of estrogen use and breast cancer risk with body mass index (BMI), with the risks being higher in, and frequently limited to, those with a BMI under 25 (Collaborative Group on Hormonal Factors, 1997; Schairer et al., 2000). With only 1110 women with a BMI under 25 in the ET arm of the WHI, and an expectation of no, or a very low level of, excess risk at 6 years of follow-​up, there is likely little statistical power to adequately address this question.

Endogenous Hormones

With the plethora of seemingly hormone-​related risk factors for breast cancer, and with the prime explanatory candidate hypothesized to be estrogen, one would think that an early response would have been an assessment of circulating levels of estrogen. This did not happen for some time, likely due to concern over the quality of assays available, and particularly with the paucity of prospective cohort studies with biospecimen collections. For whatever reason, it was not until 30  years after the elucidation of many hormonally related risk factors, and 20  years after the first assessments of exogenous estrogen exposure, that reasonable studies of prospectively collected endogenous estrogens began to appear. Most of the initial studies, although relatively small, did show evidence of a positive association between

No. (%) of Events CEE

Invasive breast cancer Intervention Postintervention Overall

104 (0.28) 47 (0.26) 151 (0.27)

Placebo

135 (0.35) 64 (0.75) 199 (0.35)

HR (95% Cl)

0.79 (0.61–1.02) 0.75 (0.51–1.09) 0.77 (0.62–0.95)

P value for Difference

Favors CEE

Favors Placebo

.76

0.50

0.67

1.00 HR (95% Cl)

1.50

2.00

Figure 22–​2.  Cumulative hazard of breast cancer associated with Conjugated Equine Estrogen (CEE), overall and post-​intervention. The hazard ratios (HRs) are derived from proportional hazards models stratified by prior disease (for outcomes in which women were eligible for enrollment with and without the prevalent condition), age, and dietary modification randomization group. The P values for differences between the intervention and post-​ intervention phases were calculated from models for the overall mean follow-​up period that also included a time-​dependent term for trial phase. For the intervention and overall phases, time to event equals 0 on date of randomization. For the post-​intervention phase, time to event equals 0 on February 29, 2004. Source: JAMA 2011;305(13):1305–1314.

398

398

PART III:  THE CAUSES OF CANCER

estradiol and estrone levels and elevated breast cancer risk in postmenopausal women that became convincing in larger studies and pooled analyses (Key et  al., 2002). More recently, circulating estrogens in premenopausal women have been linked to increased breast cancer risk (Endogenous Hormones and Breast Cancer Collaborative Group et al., 2013), although at lower levels of risk than for postmenopausal women—​somewhat surprising given the much higher absolute estrogen concentrations in premenopausal women. There is also some evidence that the magnitude of this risk may differ by time in the menstrual cycle when estrogen was measured (Fortner et al., 2013). Thus far, these observations on premenopausal estrogen levels have come from assessing all breast cancers following hormone measurement. Very recently, one study was able to assess the association of premenopausal estrogen levels separately for breast cancers occurring before and after menopause (Fortner et al., 2013). A doubling of premenopausal estradiol resulted in a 10% increase in risk for breast cancers occurring in premenopausal women, but a 40% reduced risk for those occurring in postmenopausal women. It is clear that what is needed is much more focus on studies of the effects of premenopausal hormonal profiles in order to understand the underlying biology. Before there were even prospective epidemiologic studies of circulating estrogens, both epidemiologists and laboratory investigators were interested in a potential role for estrogen metabolism, and thus metabolites in carcinogenesis. The most prominent early hypothesis from epidemiologists was the estrogen-​fraction hypothesis (Cole and MacMahon, 1969). Noting that estriol, a metabolite of the estrogens estrone and estradiol, was a much “weaker” estrogen than its parents, it speculated that women with larger ratios of estriol to estradiol and estrone would be at lower breast cancer risk. Since there were no prospective studies in which to test such a hypothesis, they tried to assess its credibility by comparing measurements in populations with varying breast cancer risks, including women living in high-​and low-​risk countries, and among women by their parity, ages at first birth, at menarche, and menopause, and other risk factors (MacMahon et al., 1970). While some of these comparisons were consistent with the hypothesis, some were not, and there were some differences between pre-​and postmenopausal studies. In addition, as one of the originators of the hypothesis commented, “… as knowledge of the biology of the estrogen fractions developed, the idea that the estrogen fractions compete for binding sites became less attractive” (MacMahon, 2006). At about the same time as the estrogen fraction hypothesis emerged, laboratory investigators, after decades of work on potential mechanisms of carcinogenesis, had also turned their attention to estrogen metabolism. It was noted that the parent estrogens were rapidly metabolized, with the initial step being hydroxylation at the 16, 2, or 4 position of the steroid ring, followed by a cascade of metabolites down each of these pathways. It was found that the 16-​hydroxy metabolites had an affinity for the estrogen receptor and thus were strong mitogens stimulating cell proliferation. On the other hand, metabolites down the 2 and 4 pathways (known as the catechol estrogens) were not strong mitogens, but could be oxidized to form quinones that interact directly with DNA to form a variety of adducts that could enhance mutational events (Yager and Davidson, 2006). This was particularly true for the 4-​OH pathway. As a result, two distinct hypotheses arose for how estrogen could cause breast cancer and maybe other malignancies as well—​a mitogenic hypothesis and a mutagenic hypothesis (Yager and Davidson, 2006). Each of these hypotheses developed a constituency within the laboratory community, leading to several decades of experiments and spirited debates on which hypothesis was correct, or at least more important than the other. Until recently, for the most part, the epidemiology community had not been involved in assessing these hypotheses. Indeed, for about a decade, many epidemiologists were largely unaware of the relevant laboratory work. Even after becoming aware of it, there was very little investigation in human populations because there were no reliable assays to measure the relevant metabolites in humans. Eventually, two of the metabolites within these pathways were measured (2-​hydroxyestrone and 16-​alpha-​hydroxyestrone), with inconsistent results with respect to risk (Cauley et al., 2003; Muti et al., 2000; Ursin et al., 1999). Several years later, a comprehensive method to assess a range of metabolites across these pathways was finally developed (Xu et al., 2007). As noted

in Chapter  45 of this volume, the first studies to use this method to comprehensively evaluate these pathways in prospective cohort studies of breast cancer (Dallal et al., 2014; Falk et al., 2013; Fuhrman et al., 2012; Moore et al., 2016) found a strong relationship of decreasing risk of postmenopausal breast cancer with an increasing proportion of estrogen being metabolized down the 2-​OH pathway. This supported the mitogenic hypothesis for estrogen-​induced breast cancer. The fact that low breast cancer–​risk populations in Asia have relatively high ratios of 2-​OH to 16-​OH metabolites that decline in Asian migrants to the West (Falk et al., 2005; Moore et al., 2016), along with more favorable ratios being associated with increased levels of exercise (Dallal et al., 2016; Matthews et al., 2004), suggest that this risk factor may be modifiable.

Unifying Hypothesis?

As previously noted, very early in the pursuit of the epidemiology of breast cancer, risk factors seemed to indict hormones, specifically estrogens, leading to the unifying hypothesis that cumulative lifetime exposure to estrogen explained the disease. Unfortunately, this was so widely accepted that it seemed to discourage any analytic attempts to robustly test the hypothesis. Over many decades, a number of well-​ replicated observations have challenged this assumption. • Obesity, associated with increased levels of estrogen, is a postmenopausal breast cancer risk factor, but only near the time of diagnosis, and not when it only existed, even for long periods, at some time in the past (Ziegler, 1997). • Long-​term use of exogenous estrogen therapy by postmenopausal women increases risk, but only among current or very recent users, with even very long-​ term users who have stopped for several years having the same risk as never users (Collaborative Group on Hormonal Factors in Breast Cancer, 1997). • The anticipated additional increased risk associated with exogenous hormones is not apparent in obese women who take these drugs (Collaborative Group on Hormonal Factors in Breast Cancer, 1997). • Circulating estrogens in premenopausal women (when estrogen levels are at their highest) are associated with only modest increases in breast cancer risk (Endogenous Hormones and Breast Cancer Collaborative Group et al., 2013) compared to effects in postmenopausal women (Key et al., 2002). • With respect to the international differences in risk, the actual differences in circulating estrogens between low-​and high-​risk populations do not appear to explain much of the observed differences in breast cancer rates (Hoover, 2012). Moreover, in new data from a previously unstudied Asian population (Mongolians), we see very low rates of breast cancer, but higher estrogen levels than those in the West (Troisi et al., 2014). • Doubt has even been raised about the risk factor that was thought to most clearly show a long-​ term estrogen dose effect—​ age at menarche—​where seemingly a few years more of menstrual cycling would end up contributing to risk many years later. Breast cancer incidence in Bangladesh is five times lower than in the United Kingdom, and South Asians in Britain have a rate double that in South Asia, a pattern that has also been observed with East Asian populations studied in the past (Houghton et al., 2014). However, in a recent study of Bangladeshi migrants, the age at menarche was the same for girls living in Bangladesh as for those who had migrated to Britain, and both were the same as for British girls (Houghton et al., 2014). What was different was the pubertal transition (Figure 22–​3). As depicted in Figure 22–​3 the interval from adrenarche (when the adrenal cortex secretes increased levels of androgens) to thelarche (the onset of breast development) decreased, and that from thelarche to menarche increased with increasing Western acculturation. This is just one study, and conclusions may change with further investigations, but it has stimulated investigators to speculate about whether the historic age at menarche observation might not operate by increasing lifetime estrogen exposure, but through early molecular events impacting cancer susceptibility. This is the direction that much of the search for the underlying mechanism of the effect of another risk factor, age at first birth, has been taking recently. Initially, the focus was on possible hormonal changes associated

 39



Hormones and Cancer

White British

British-Bangladeshi

1.6

3.9

2.2

Bangladeshi

2.5

3.5

1.8

Years between milestones Adrenarche to Thelarche

Thelarche to Menarche

Figure 22–​3.  Juvenile and pubertal tempo.

with this risk factor (MacMahon et al., 1970), but most recent efforts have turned to studies of significant molecular changes within the breast (Sivaraman and Medina, 2002). While the described findings are not consistent with a cumulative lifetime exposure hypothesis for estrogen carcinogenesis, they do provide impressive support for estrogen acting relatively late in breast carcinogenesis, likely as a tumor promoter. In addition, some of these findings—​lack of additive effects of obesity and menopausal hormone therapy, smaller effects for pre-​versus postmenopausal circulating estrogen—​could be viewed as reflecting a potential (upper) threshold for estrogen’s influence. This would be consistent with estrogen acting mainly via receptors, with saturation of available receptors at some level of concentration. But if we have learned anything from the last 60 years of breast cancer research, perhaps we simply should be working to more fully characterize and understand the known risk factors, rather than attempting to develop comprehensive explanatory hypotheses.

Breast Cancer Subtypes

It should be noted that the discussions of hormones and breast cancer in this chapter have focused on the development of our knowledge of its etiology over many decades, and therefore we have described studies of overall breast cancer. More recently, molecular markers in tumors have been used to form subgroups that have distinct therapeutic, and likely etiologic, implications. The first and most widely used categorization distinguishes between tumors that express estrogen receptors and those that do not. As might be expected, most etiologic studies that address both types find that the estrogen-​related risk factors tend to be much stronger in, and sometimes limited to, the estrogen receptor positive (ER+) malignancies (Press et al., 2011; discussed comprehensively in the reproductive factors section of Chapter 45 in this volume). Studies of recent time trends in incidence have noted a steady decline in incidence of ER–​tumors and an increase in ER+ tumors, trends that are predicted to continue at least through 2030 (Anderson et al., 2011; Rosenberg et al, 2015). Discerning the reasons for these patterns would seem to be a high priority. Until recently, the classification of subtypes for therapeutic purposes has used classical immunochemistry (e.g., estrogen, progesterone, and HER2 receptors) in combination with clinical and pathologic variables. The recent introduction of high-​throughput gene-​expression analyses has introduced a whole new and complex method for predicting tumor response to treatments (Blows et al., 2010; Dai et al., 2015). How many prognostic categories this will result in and how relevant any of this will be to etiology are unknown, but worth keeping track of.

Future Prospects

Some of the seemingly conflicting evidence for hormonal effects in breast cancer risk could be clues to a deeper understanding of hormonal carcinogenesis, but for the most part has not been aggressively pursued with this in mind. A good example of this is the striking risk

399

factor differences for pre-​versus postmenopausal disease. Circulating estrogen levels, a well-​ established risk factor for postmenopausal breast cancer, is likely associated, but to a much smaller degree, with premenopausal disease. Likewise, obesity is associated with increased risk of postmenopausal breast cancer, but actually appears to be protective for premenopausal disease (Hunter and Willett, 1993). In fact, the expectation that pre-​and postmenopausal breast cancers should have a common epidemiology and share the same risk factors could be the problem. Certainly all aspects of reproductive biology and lactation have been driven by powerful evolutionary pressures over millennia, while postmenopausal hormonal physiology was likely not subject to such pressures. Thus, perhaps one should actually expect there to be consequential differences between pre-​and postmenopausal hormonal epidemiology, and exploiting these differences may give us clues to etiology. For example, normal postmenopausal breast lobules contain 4–​5 times the number of cells expressing estrogen receptors than those present in premenopausal lobules (Bernstein and Press, 1998). This likely at least partially reflects the frequently observed phenomenon of an increase in activation of receptors in response to a decline in the presence of their ligands (De Meyts and Rousseau, 1980). If this relatively simple explanation is responsible for the shift in ER status of tumors with age, some focus on a more comprehensive understanding of the epidemiology of ER expression might be useful. Another emerging possibility is one that has a more general implication as well. That is, as our knowledge of hormones and their actions expand, we are seeing the breadth of hormones that contribute and interact in regulation of specific organs and their functions. One of the more recent of these to be appreciated for the breast is the interrelationship between estrogens and their receptors and the insulin/​IGF system of hormones and their receptors. Recent prospective studies have shown stronger associations of circulating levels of IGF-​1 with postmenopausal breast cancer risk in the presence of elevated androgens or estrogens (Tworoger et al., 2011). In addition, there is impressive laboratory evidence of “cross-​talk” between these two hormonal systems, suggesting that assessment of interactions between these two systems in human risk might be fruitful (De Marco et al., 2015). However, to do this comprehensively will likely require not only measurement of the hormones, but also their receptors, for which practical methods of measurement in large epidemiologic studies will need to be developed. Receptors are not the only likely significant modifiers of estrogenicity. Sex hormone binding globulin (SHBG) binds these hormones, particularly androgens and estrogens, and can thus modify their impact on multiple biologic actions (Thaler et al., 2015). Fortunately, SHBG can be measured reasonably well in serum and can be evaluated for its impact on the effects of hormones. Indeed, as our colleagues in basic science pursue advances in molecular biology, it has become clear that steroid hormones can be bound by numerous proteins in the cell membrane (Caldwell et al., 2016). As is frequently the case, it will be up to epidemiologists to determine which aspects of this continuing elucidation of the complexity of biologic processes need to be accommodated in order to come to meaningful conclusions relevant to disease risk and prevention opportunities at the population level. While exploring new possibilities for hormones that might play a role in breast cancer etiology, we need to also include other sex hormones that seem obvious, but have not yet been thoroughly investigated. The most notable of these are progesterone, testosterone and prolactin. Given that the proliferative action of progesterone in breast duct tissue has been known for decades (Graham and Clarke, 1997), and its carcinogenic action as a component of menopausal hormone therapy has been known for some time (Schairer et al., 2000), remarkably little has been done to assess its role at physiologic levels in breast cancer etiology (Endogenous Hormones and Breast Cancer Collaborative Group, 2013; Fortner et al., 2013; Kaaks et al., 2014). With respect to testosterone, while not pursued as extensively as estrogen, prospective studies have fairly consistently found circulating levels to be positively associated with both pre-​(Kaaks et al., 2014) and postmenopausal breast cancer (Key et al., 2002). Until very recently, the association in postmenopausal women has frequently been written off as only reflecting testosterone’s role as the molecular source for the synthesis of estrogen in peripheral fat. However, this seems

40

400

PART III:  THE CAUSES OF CANCER

Table 22–​1. The Effect on Basal Serum Prolactin Levels of Pregnancy, Parity, and the Elapsed Time after the Last Delivery Variable

No.

Serum Prolactin ng/​ml (mean = SE)

P Value*

19 29

8.85 ß 1.39 4.77 ß 0.43

< 0.01

10 14 5

5.18 ß 0.47 4.61 ß 0.48 4.40 ß 0.35

NS

10 10 9

4.15 ß 0.42 4.87 ß 0.53 4.83 ß 0.25

NS

Pregnancy Nulliparous Parous No. of pregnancies 1 2 3 Time since last delivery (mo) 12–​36 37–​78 79–​150

underlying mechanisms in human populations in order to determine if and how we can use this information for prevention. We have been able to recommend no or minimal use of menopausal hormone therapy, have emphasized the need to reduce obesity, and have developed certain hormone-​targeted chemoprevention interventions for women whose risk/​benefit ratio supports such treatment. The last example is unfortunately also an example of how, even when we understand enough about a specific aspect of hormonal carcinogenesis to develop a preventive intervention, the pervasive level of biologic activities of hormones can discourage meaningful uptake of such an intervention. Clinical trials of tamoxifen and raloxifene, medications that can block estrogen receptors, have documented reductions in breast cancer risk of 40%–​50% (Fisher et al., 1998; Vogel et al., 2006). They also have documented consequential adverse biologic effects. Statistical modeling of benefit/​risk ratios have identified groups of women for whom benefits outweigh risks (Gail, 2012), and such prevention efforts have been widely promoted for these groups. These efforts have been widely rejected by the women involved, who have made quite different personal risk/​benefit judgments (Hoerger et al., 2016; Smith et al., 2016). Because the hormonal component for breast cancer is so profound, there has been hope that with a deeper understanding of the relevant underlying normal and abnormal hormonal mechanisms, we might learn how to use this knowledge for more effective preventive strategies.

NS = not significant.

less likely to explain its association with premenopausal breast cancer, where circulating estrogen levels primarily reflect ovarian production. For postmenopausal disease, in studies that include both estrogen and testosterone in their models, both hormones remain associated, but at reduced levels (Key et al., 2002). Clarifying the role and mechanism(s) of action of testosterone’s effect on breast cancer risk would seem to be a high priority for both epidemiology and laboratory investigators. Prolactin, with its major role in normal breast physiology, has also long been thought of as a candidate hormone for impact on breast cancer risk. This enthusiasm was enhanced when it was determined that normal levels of circulating prolactin were reduced by about 50% following a first birth, and further reduced by smaller amounts with each successive birth (Musey et al., 1987; Table 22–​1). This led to speculation that this lowering might explain the age at first birth and parity risk factors for breast cancer. Unfortunately, as noted in Chapter 45, studies to date have not consistently linked circulating prolactin to breast cancer risk (Tikk et al., 2014). Some of this inconsistency may relate to measurement issues. There are in fact a number of different isoforms of prolactin, with substantial differences in size and shape. From a teleological perspective, this would seem to imply that these prolactin variants might have different biologic roles. Some have attempted to identify “bioactive” forms of prolactin using various immunoassays. The most recent effort found some evidence of increased risk with a bioactivity assay in a subgroup of postmenopausal women whose blood was sampled relatively close in time to breast cancer diagnosis (Tworoger et al., 2015). A more definitive resolution of the role of prolactin will probably require a comprehensive, agnostic assessment of all of its isoforms, a technology not yet available.

Summary

While the saga of attempts to identify hormonal risk factors for breast cancer, and the underlying hormones and mechanisms involved in carcinogenesis, has spanned centuries, revealing associations that have markedly increased our understanding of some of the elements of breast cancer etiology, as yet only a few provide practical opportunities for prevention. On the other hand, much remains unknown, despite many provocative observations. This is itself a commentary on the level of knowledge that we need in order to translate findings in hormonal carcinogenesis, as well as in other metabolic processes, into practical prevention. In much of chemical carcinogenesis, it is sufficient to identify the carcinogen and design ways of decreasing exposure to it. Since hormones are essential for life and generally are regulated in a fairly tight manner, for the most part we need to understand their

Prostate Cancer Given the role of testosterone, or androgens generally, in the growth, function, and maintenance of the prostate gland, as well as how prominently androgens are associated with tumor activity once the disease develops, it has been surprising and mystifying to clinicians, epidemiologists, and laboratory scientists alike that circulating levels of these hormones have not been consistently related to prostate cancer risk (Roddam et al., 2008). Attention has turned to the possibility that other hormones such as estrogen, or the balance of estrogen and testosterone, may be important. In men, most testosterone is converted to dihydrotestosterone (DHT) by 5a reductase, but may also be enzymatically converted to estrogens by aromatase (CYP19). While it has been hypothesized that higher levels of intraprostatic DHT may be associated with increased prostate cancer risk (Platz and Giovannucci, 2004), it is also possible that if 5a reductase activity is inhibited, excess testosterone may instead be metabolized to estradiol. The bulk of the evidence shows no association of estradiol concentrations and prostate cancer risk, and findings for the association of the estradiol:testosterone ratio and prostate cancer risk have been mixed. A recent study of aggressive prostate cancer in the PLCO cohort study simultaneously assessed the effects of testosterone and the estradiol:testosterone ratio and found a marked attenuation of an association with testosterone, but minimal effect on the inverse association for the estradiol:testosterone ratio (Black et al., 2014). Estrogen’s metabolites have drawn attention recently in prostate cancer epidemiology, as they have in breast cancer epidemiology, because of improvements in assay methodologies used to measure them. A recent meta-​analysis (Roddam et al., 2008) showed an increased risk of prostate cancer with higher urinary 16a-​ hydroxyestrone and a protective effect of excretion of urine with a higher 2:16a-​hydroxyestrone ratio, neither of which was observed in a recent study of aggressive disease (Black et al., 2014). In contrast, the latter study reported a positive association between serum 2:16a-​ hydroxyestrone ratio and aggressive prostate cancer (Black et al., 2014). The inverse association for estradiol:testosterone ratio, together with the increasing risk with an increasing proportion of estrogen being metabolized down the 2-​OH pathway, may indicate a potential role for estrogen metabolism in the development of aggressive prostate cancer. In breast cancer model systems, 16a-​hydroxyestrone can bind covalently to the ER and has been shown to induce abnormal cell proliferation (Bradlow et al., 1985; Suto et al., 1993), whereas 2-​ hydroxyestrone has a weak binding capacity to the estrogen receptor and has been associated with normal cell differentiation and apoptosis (Gupta et al., 1998; Telang et al., 1997).

 401



Hormones and Cancer

In contrast, strong estrogenic effects (relative to androgen or antagonistic metabolites of estrogen) may give protection against aggressive prostate cancer. Previous work to establish independent associations of androgens and estrogens with prostate cancer risk may have resulted in inconclusive findings because these hormones interact as part of a complex pathway. Future work should focus on mechanistic studies in prostate cancer, which are currently lacking. Given the track record for studies of this malignancy—​of positive findings that are not replicable—​these recent findings clearly require replication before any further pursuit. The accumulation of increasing numbers of studies focused on various hypotheses to explain the lack of any consistent association between circulating androgens and prostate cancer risk has not yet led to an answer. This has resulted in speculation that perhaps circulating levels of testosterone may not reflect its concentration in the prostate. As previously noted, testosterone is one of the hormones for which paracrine and autocrine production is particularly noteworthy. While the textbook examples of this often focus on testosterone’s role in the testes, paracrine production also occurs in the prostate. It would seem to be a high priority, using modern measurement techniques, to assess the levels of correlation between circulating and prostate tissue levels of androgens and estrogens overall, and for their various metabolic subtypes. In addition, as was discussed for the estrogens, it is likely that an adequate understanding of the role of androgens in prostate carcinogenesis is also going to require much more information about, and measurement of, androgen receptors. In fact, the complexity of receptor activity is enhanced in the prostate gland. Both epithelial and stromal tissues have receptors that have differential expression that can result in different growth factors expressed by the stroma. These actions can mediate the effects of androgen on the epithelium, potentially influencing malignant initiation. Similar interactions can also affect the progression of an established malignancy, including the induction of direct apoptotic effects, effectively moving androgens from a cause to a potential treatment of prostate cancer (Berry et al., 2008; Nieto et al., 2014; Stamatiou and Pierris, 2013). Part of our difficulty in understanding the role, if any, of hormones in prostate cancer risk may also lie in our ignorance of almost any risk factors for prostate carcinogenesis. It is often pointed out that many decades ago we knew of three well-​established risk factors—​age, race, and family history—​and this describes our level of knowledge today as well. For a long period of time, part of the enthusiasm for pursuing a hormonal etiology was not only the central role of hormones in the normal function of the prostate, but also the belief that prostate cancer showed substantial international variation, similar to breast and other hormonally related cancers. This belief was based on major international differences in prostate cancer incidence and mortality, with low rates in Asian and African men, higher rates in Asian migrants to the United States, substantially higher rates in white men in the United States, and the highest prostate cancer rates in the world in African-​ American men. Recent observations have called these conclusions into question. With the improvement of cancer registration in Africa, estimates of prostate cancer there have risen (Korir et al., 2015). Indeed, a systematic prostate cancer screening study in Ghana produced prostate cancer rates similar to those among African-​Americans (Hsing et al., 2014). Coupled with these observations has been an increasing appreciation of the major role of genetics in prostate cancer etiology. While a family history of prostate cancer has been a well-​recognized risk factor, the magnitude of the role of genetics in this tumor was not appreciated until recently. A seminal study within the Nordic Twin Cohort reported in 2000 that the estimate of heritability for prostate cancer was 42%, higher than for any other common cancer (Lichtenstein et  al., 2000). A  more recent update of the study (Hjelmborg et  al., 2014)  with increased numbers of cases and improved methods estimated heritability at 58% (95% CI: 52, 63). Concurrently, over the past 10 years, with the new capability to agnostically interrogate the entire genome for common genetic variants associated with cancer risk, the number of such identified variants has risen from less than 6 to more than 600—​and counting (Gusev et al., 2016). Initially surprising, but in line with the heritability estimates, the cancer with the most such

401

variants identified has been prostate cancer. Currently, over 100 such susceptibility loci have been identified, likely explaining a meaningful proportion of the familial risk. These numbers are also likely to rise further with ongoing genetic studies. As for the vast majority of all the newly identified genetic variants associated with cancer risks, we know little about the biologic pathways through which the variants related to prostate cancer risk operate. It will take some time for our laboratory colleagues to address these questions. However, in the interim, we will have the ability to investigate how this major role of genetics in prostate cancer etiology controls, relates to, and/​or interacts with other potential risk factors in prostate carcinogenesis. Certainly, investing in such efforts with candidate hormonal exposures may help to clarify the current enigmatic state of our knowledge. Our attempts to understand the hormonal etiology of prostate cancer provide a valuable counterpoint to the epidemiologic elucidation of the role of estrogen and other hormones in breast and gynecologic malignancies. Hormones are central to normal growth, development, and homeostasis, and thus are also likely to be biologically involved in pathologic alterations of these processes. In some instances, these might be easily identified using epidemiologic tools at hand. In other instances, we may need to have a much more detailed understanding of the possible underlying biology involved in order to develop the tools and methods required to assess effects in human populations. With the recent marked advances in molecular science, and the development of molecular epidemiologic methods, we are hopefully entering an era where this will be possible.

ENDOCRINE DISRUPTORS There is one area of hormonal carcinogenesis that has garnered substantial attention in both scientific and lay communities—​ the potential impact of endocrine-​disrupting agents in the environment and the risk of various cancers (Soto and Sonnenschein, 2010). Of particular concern are exposures to these agents in very early life, when most hormone-​driven development of many organs is prominent. To date, most of the concerns have come from laboratory animal experiments (Fenton et al., 2012; Newbold et al., 1987), augmented by various malformations and malignancies identified in wildlife raised in water sources highly polluted with endocrine-​disrupting chemicals (Bernanke and Köhler, 2009; Hamlin and Guillette, 2010). Epidemiologically robust studies of cancer risks in humans following high-​level exposure to suspect endocrine disruptors have been few and limited, and have generally not found consistently compelling evidence of risk (Soto and Sonnenschein, 2010; Teitelbauma et al., 2015). The one notable exception has been in utero exposure to high doses of the synthetic estrogen drug diethylstilbestrol (DES). The first evidence of adverse effects of in utero exposure to DES was the identification of a cluster of seven cases of the rare tumor clear-​cell adenocarcinoma of the vagina in adolescent girls whose mothers had used the drug in their pregnancies (Herbst et al. 1971; Ulfelder, 1980). More systematic studies have estimated the attack rate of this tumor in the exposed through age 39 to be about 1.6/​1000 women (Melnick et al., 1987; Troisi et al., 2007). Other studies of this exposure have also identified evidence of increased risk of pre-​malignant lesions of the uterine cervix (Hatch et al., 2001) and adenocarcinomas of the breast (Troisi et al., 2007). While these are important leads, it should also be recognized that the overwhelming adverse impact on women of this fetal exposure has been in anatomic and physiologic abnormalities resulting in very high rates of infertility and abnormal conditions and complications of pregnancy, as depicted in Figure 22–​4 (Hoover et al., 2011). Both the neoplastic and adverse gynecologic outcomes risks are higher among women whose mothers received higher doses and/​or received them early in gestation, as evidenced by the increased prevalence of vaginal epithelial changes in these women, an anatomic marker of high-​dose and/​or early exposure. Overall it is estimated that 27.5% of women in these high-​risk categories experienced one or more serious adverse outcomes by age 55 years as a result of their exposure (Hoover et al., 2011).

402

402

PART III:  THE CAUSES OF CANCER Infertility§§ Spontaneous abortion†§§ Ectopic†§§ Second trimester loss†

Preterm birth†§§ Pre-eclampsia† Stillbirth‡† Neonatal death‡§§ Early menopause

ClN2+§ VEC + VEC –

Breast cancer ≥ 40§ 0.4

1

10

40

HR and 95% Cl (log-scale)

Figure 22–​4.  Health Outcomes According to Diethylstilbestrol (DES) Exposure Status and, in the DES-​Exposure Group, the Presence or Absence of Vaginal Epithelial Changes (VEC) at a Young Age. Hazard ratios differed significantly between VEC-​positive and VEC-​negative subgroups of the DES-​ exposed group for infertility, spontaneous abortion, ectopic pregnancy, preterm delivery, and neonatal death (P < 0.01), as well as for grade 2 or higher cervical intraepithelial neoplasia (CIN 2+ and invasive breast cancer at an age of 40 years or older (P < 0.05). Total numbers of women vary among outcomes, primarily reflecting whether all, pregnant, or parous women were included in the analysis, but also owing to some missing responses to the questionnaires ascertaining the outcome and, to a lesser extent, to specific questions. Data for spontaneous abortion, ectopic pregnancy, and loss of second-​trimester pregnancy were computed with age as the time metric and with adjustment for date of birth and cohort. Data for preterm delivery, pre-​eclampsia, stillbirth, and neonatal death were restricted to pregnant women and were adjusted for number of pregnancies. † The analysis was restricted to gravid women and adjusted for number of pregnancies. ‡ The analysis was restricted to parous women and adjusted for number of births. § p-​value < 0.05. §§ p-​value 30 pack-​year smoking history, who currently smoked or had quit in the last 15 years. Bach et  al. used data on over 18,000 subjects enrolled in the Beta-​ Carotene and Retinol Efficacy Trial (CARET) to derive a cancer risk prediction model that included age, gender, asbestos exposure, and smoking history (Bach et al., 2003). The authors concluded that the one-​quarter of smokers at highest risk will account for about half of the lung cancer cases (Bach et al., 2003; Etzel and Bach, 2011). The Spitz model was based on a sample of 1851 lung cancer cases and 2001 age-​, sex-​, race-​, and smoking-​status-​(never, former, or current smokers) matched controls from an ongoing lung cancer case-​ control study at the University of Texas MD Anderson Cancer Center (MDACC) in Houston, Texas (Spitz et  al., 2007). The authors used logistic regression to derive log-​odds models separately for never, former, and current smokers and estimated the relative risk of developing lung cancer. Spitz et  al. followed the methods of Gail et  al. (1989) and Fears et al. (2006) to combine the baseline relative risk from the log-​odds model, age-​and gender-​specific incidence rates (corrected for smoking status) with all-​cause mortality rates (excluding lung cancer) to estimate X-​year absolute risk of lung cancer. In their original paper, Spitz et al. presented 1-​year probabilities of lung cancer, based on national incidence rates (Spitz et al., 2007). The Liverpool Lung Project (LLP) model was developed using 579 lung cancer cases and 1157 age-​and sex-​matched population-​based controls from a case-​control study in Liverpool, England (Cassidy et al., 2008; Etzel and Bach, 2011). The authors then used a conversion method similar to that of Gail et al. (1989) and Chen et al. (2006) to combine the relative risk estimates with lung cancer incidence rates and presented 5-​year absolute risks for lung cancer. These three models

536

536

PART IV:  Cancers by Tissue of Origin

differed primarily in whether they included lung-​related comorbidities (emphysema or pneumonia), history of prior malignancies, and definition of family cancer history (Etzel and Bach, 2011). The inclusion of comorbidities in the Spitz and LLP models, compared with the Bach model, produced a small gain in discriminatory power (to 0.69 and 0.66, respectively), but was still less than the 0.75 considered by some to be the lower limit for clinical utility (Janssens et al., 2007). A fourth model, called the Pittsburgh Predictor, was developed using four risk factors (duration of smoking, smoking status, smoking intensity, and age) to predict 6-​year lung cancer incidence (Wilson and Weissfeld, 2015). Although this model had slightly less area under the receiver curve than the Bach or PLCO m2012 model, it was proposed as a simpler lung cancer risk prediction model that might facilitate standardized procedures for advising and selecting patients with respect to lung cancer screening (Wilson and Weissfeld, 2015). Several studies have evaluated these models in terms of calibration and discriminatory power (Etzel and Bach, 2011; Li et al., 2015). In the original paper by Bach et al. (2003), the authors completed an internal validation of their model with a 10-​fold cross-​validated calibration plot and observed that the model was consistent with excellent calibration. Later, investigators at the National Cancer Institute performed an external validation of the Bach model (Cronin et  al., 2006). Li et  al. (2015) and D’Amelio et al. (2010) have further evaluated these models. D’Amelio and colleagues compared the discriminatory power and clinical utility of the Bach, Spitz, and LLP models in predicting 5-​year absolute risk of lung cancer in a population of 3197 lung cancer patients and 1703 cancer-​free controls recruited to an ongoing case-​control study of lung cancer at the Harvard School of Public Health and Massachusetts General Hospital (D’Amelio et al., 2010). The LLP and Spitz models had comparable discriminatory power (0.69), whereas the Bach model had a slightly lower area under the curve (AUC) of 0.66. The clinical utility of the Spitz, Bach, and LLP models was evaluated using the Search Partition Analysis (SPAN, Auckland, New Zealand) program (Marshall, 2001, 2005, 2009), with a definition of “high risk for lung cancer” as 5-​year cumulative risk greater than 7.5%. At this cutoff for risk, the Bach model correctly identified five true lung cancer cases for every non–​lung cancer case. The LLP model performed slightly less well, with a ratio of four lung cancer patients correctly identified for every one incorrectly identified. In contrast, only true lung cancer cases were identified at this level, and no controls were incorrectly categorized by the Spitz model; however, the Spitz model did have very low sensitivity (2.2%) compared with the Bach and LLP models.

PREVENTION

that improve ventilation (Chapter 13). The World Health Organization recommends that homes be tested for radon, and that repeated measurements be performed above the action level of 2.7 picocuries per liter (pCi/​L) or 100 Bq/​m3 (WHO, 2005). The corresponding action level from the US Environmental Protection Agency is 4 pCi/​L or 148 Bq/​m3 (US Environmental Protection Agency, 2017). Smoke-​free regulations can prevent exposure to secondhand smoke in public buildings, workplaces, restaurants, bars, casinos, airplanes, and mass transit in countries at all levels of economic development. Exposure to secondhand smoke in private homes and cars can be reduced by education to discourage active smoking. Indoor air pollution from the use of coal and biomass for cooking and heating can be reduced by properly maintained cookstoves, improved ventilation, and the use of cleaner fuels, such as compressed natural gas.

Occupational Exposures Exposure to respiratory carcinogens in occupational settings has decreased in high-​ income countries but has increased in LMICs (Chapter 16). These exposures can be reduced by engineering changes such as substitution, containment, ventilation, and improved waste disposal (Chapter  61). Reductions in occupational exposures generally occur because of environmental regulations and health and safety guidelines in the workplace (Chapter 62.6).

Ambient Air Regulatory measures can reduce outdoor air pollution and reduce or eliminate exposure to carcinogens (including environmental tobacco smoke) in public and occupational settings.

Screening While screening cannot prevent lung cancer from occurring, the National Lung Screening Trial demonstrated that yearly screening with low-​dose helical CT scans decreased mortality risk by 20% among people at high risk for developing lung cancer. The US Preventive Services Task Force recommends yearly lung cancer screening with low-​dose CT for people who (a)  have a history of heavy smoking (30 pack-​years or more); (b) currently smoke or have quit within the past 15 years; and (c) are between 55 and 80 years old. The evidence regarding low-​dose CT screening for lung cancer is covered in Chapter 63.

Tobacco Control

RESEARCH DIRECTIONS

More than 80% of lung cancers in most Western populations could be prevented by the elimination of active smoking. Comprehensive tobacco control measures that effectively reduce active smoking and prevent involuntary exposure to tobacco smoke are discussed in Chapter 62.1. Even the partial application of these measures has substantially reduced lung cancer rates in men and has caused a downturn at younger ages in women in high-​and some middle-​income countries. Smoking cessation at any age avoids much of the future risk from continued smoking (Chapter 11). The relative and absolute benefits are greatest when cessation occurs at an early age but are substantial even when stopping occurs by age 50 or 60 years (Peto et al., 2000). There is growing evidence that patients who quit smoking after diagnosis of early-​stage lung cancer have reduced risk of a recurrent primary tumor or development of a second primary, as well as experiencing higher survival compared with people who continue to smoke (Warren et al., 2010).

The causal relationships between lung cancer, tobacco smoking, and other airborne carcinogens discussed in this chapter are well-​ established. Besides the dominant relationship of lung cancer with active tobacco smoking, exposures to other airborne carcinogens in ambient, indoor, and occupational settings have been definitively shown to increase risk. The effectiveness of primary prevention is evidenced by the large decrease in lung cancer incidence and death rates among men in many high-​and middle-​income countries, and by the leveling off or initial downturn of the rates among women. Despite this encouraging progress, ongoing research is needed to support cancer control, monitor population trends, refine dose–​ response relationships, elucidate molecular pathogenesis in well-​ defined populations, and evaluate the effectiveness of screening in real-​life settings.

Reductions in Indoor Air Pollution

Population-​based surveillance of lung and other cancers is essential for cancer control. Surveillance research monitors risks and disease burden, evaluates temporal trends, measures progress (or lack thereof), and estimates current and future health care needs (see Chapter  8)

Indoor exposure to radon gas in homes, schools, and workplaces can be reduced by a variety of relatively simple and inexpensive strategies

Surveillance Research

 537



Lung Cancer

(Anderson, 2009; Parkin, 2008). In Western populations, the incidence and death rates from lung cancer also serve as sentinel indicators of the adverse health effects of tobacco smoking. Lung cancer mortality has been used to estimate the stage of the tobacco epidemic (Lopez et al., 1994; Thun et al., 2012), the number of deaths caused by smoking in developed countries (Peto et al., 2015), and the impact of state tobacco control programs in the United States (Centers for Disease Control and Prevention [CDC], 2000; Jemal et al., 2008). The use of lung cancer as an indicator of tobacco smoking is not applicable in all populations, however, as evidenced by the high risk of lung cancer among the predominantly non-​smoking women in China. Temporal trends in lung cancer incidence and/​or mortality rates are especially informative when examined by age and birth cohort as well as by age-​standardized rates. For example, trends in age-​specific rates at ages 35–​39 years—​often the youngest age group in which the lung cancer rate can be measured with reasonable stability—​reflect changes in smoking behavior among adolescents and young adults during the previous 20 years. As mentioned earlier, age-​specific rates in the young are a more sensitive indicator of the impact of behavioral changes than age-​standardized rates, which are weighted toward older ages. Efforts to improve surveillance programs confront different challenges in different countries, generally reflecting the level of economic development. In LMICs the greatest problems involve the quality of diagnostic information and coverage of the registry. For example, in Africa and South America less than 10% of people are covered by population-​based tumor registries (Chapter  8). Countries that submit high-​quality mortality data to the WHO mortality database represent only about 30% of the world’s population. The remainder provide either lower-​quality cause-​of-​death data or, in 74 (mainly low-​ resource) countries, no such data at all. There is an urgent need to improve the quality and coverage of diagnostic information on cancer in many LMICs to monitor the rise in non-​communicable disease. In high-​income countries, cancer registries and mortality databases can usually obtain reliable diagnostic information on primary lung cancer, but without specificity with regard to the molecular characteristics of the tumor or clinical data on recurrence, metastases, and disease progression. The Institute of Medicine (IOM) has identified the lack of molecular data in population-​based tumor registries in the United States as a major gap that impedes efforts to evaluate quality of care (IOM, 2013). Cancer registries in the United States now have the capacity to capture pathology reports electronically on most cancer patients. This could assist with the collection of data on emerging indicators, such as biomarkers for mutations in EGFR and ALK that can be targeted with FDA-​approved drugs.

537

Multicenter consortia of prospective cohorts are well suited to assess differences in the etiology of various histologic and/​ or molecular subtypes of lung cancer. Large-​scale studies will be needed to assess exposure–​disease relationships and to detect interactions among exposures or between exposures and germline susceptibility variants. Given the speed of advances in molecular research, cohort and case-​control studies will need to collect and archive tissue in order to later characterize molecular endpoints in a uniform and standardized manner. Consortial efforts that share biospecimens, harmonize exposure and outcome data across studies, and facilitate pooling and replication will have the greatest statistical power for discovery and robust replication. The detection of signature mutations in lung cancer provides novel opportunities to identify the causes of cancer in individuals as well as a groups. The most widely recognized signature mutation associated with smoking is the G→T transversion in TP53 found in 30% of lung cancers from smokers (Alexandrov et  al., 2016). If other signature mutations can be identified for known or suspected lung carcinogens, this could improve the estimates of attributable fraction, especially in studies of lifelong non-​smokers. Collaboration between epidemiologists and genomicists could bring a population science perspective to genomic research on lung cancer. Sampling from well-​defined populations could reduce the variability of estimates of the frequency and spectrum of mutations. To date, genomic studies of lung cancer have usually been based on convenience samples of a few hundred cases, with little consideration of patient or population characteristics. Estimates of the frequency of various somatic mutations have not generally been stratified by demographic variables such as sex or race/​ethnicity, or exposures such as smoking. Collaborative studies that draw tumor samples from well-​ characterized populations such as cohort studies can utilize prospectively collected data on individual exposures. The molecular profiling of tumors is not without challenges, since clonal heterogeneity within a cancer and changes in tumor pathology during treatment are likely to complicate the classification of distinct tumor subtypes of lung cancer, as has been the case for breast cancer (Norum et al., 2014). Several unresolved issues concerning lung cancer could be addressed in existing data sets. For example, is the risk of adenocarcinoma slower to change after smoking cessation than other subtypes of lung cancer? Do patients diagnosed with lung cancer or other smoking-​related lung diseases lose substantial amounts of weight before diagnosis? If so, over what time period does this occur? Does abdominal obesity increase lung cancer risk? Does lung cancer progress more slowly in women than in men?

Screening Cohort and Case-​Control Studies Cohort and case-​control studies can address issues related to both cancer control and etiology. Analytic studies conducted in countries or populations where risk factors have not previously been evaluated can provide local data on the determinants of lung and other cancers in order to motivate and inform policy decisions. Serial analytic studies can monitor changes in relative risk or attributable fraction resulting from birth cohort patterns of smoking or changes in cigarette design. Changes in the distribution of histologic subtypes or topography can be evaluated in relation to particular exposures. Social mobility and changing patterns of energy use influence individual exposure to indoor and ambient air pollutants. As large numbers of people move from rural communities to the rapidly expanding and highly polluted cities of Asia and the Southern Hemisphere, there is a clear need for research on outdoor air pollution in the developing world (Chapter 17). Coal is still widely used for fuel in China and in the cities of the poorest developing countries. Ambient pollution from stationary and mobile sources in LMICs exceeds the levels commonly encountered in Europe and North America (Pandey et al., 2006). Indoor air pollution from combustion of coal and biomass in poorly ventilated stoves and heaters remains a common problem in LMICs. Critical questions remain to be answered about the specific pollutants that increase lung cancer risk and their separate and joint effects on risk.

Multiple professional societies in the United States now recommend helical CT screening for individuals at high risk of lung cancer. This is not yet the case in Europe, where national public health authorities are awaiting the findings of several other trials before making decisions (Briggs, 2014). As described in Chapter 63, the recommendations in the United States are based on one large randomized trial that found reduced all-​cause mortality in heavy and former smokers randomized to low-​dose CT screening (National Lung Screening Trial Research Team, 2011). Because this study was conducted primarily in academic centers, it is possible that the balance of benefits and harms could differ as screening disseminates into the community. Further studies of over-​diagnosis and harms related to false positive screens will also help refine estimates of the net benefit of screening. References Al-​Hashimi MM, and Wang XJ. 2014. Trend analysis of lung cancer incidence rates in Ninwa province, Iraq, from 2000 to 2010: decrease and recent stability. Asian Pac J Cancer Prev, 15(1), 385–​390. PMID: 24528061. Alexandrov LB, Ju YS, Haase K, et al. 2016. Mutational signatures associated with tobacco smoking in human cancer. Science, 354(6312), 618–​622. PMID: 27811275. American Cancer Society. 2016. Cancer Facts and Figures, 2016. http://​www. cancer.org/​acs/​groups/​content/​@research/​documents/​document/​acspc-​ 047079.pdf.

538

538

PART IV:  Cancers by Tissue of Origin

American Joint Commission on Cancer. 2016. AJCC Cancer Staging System. https://​cancerstaging.org/​Pages/​default.aspx Anderson WF. 2009. Cancer surveillance research. Cancer Epidemiol Biomarkers Prev, 18(6), 1669–​1671. PMCID: PMC2943641. Auerbach O, Forman JB, Gere JB, et al. 1957. Changes in the bronchial epithelium in relation to smoking and cancer of the lung: a report of progress. N Engl J Med, 256(3), 97–​104. PMID: 13388038. Auerbach O, and Garfinkel L. 1991. The changing pattern of lung carcinoma. Cancer, 68(9), 1973–​1977. PMID: 1655232. Auerbach O, Garfinkel L, and Parks VR. 1975. Histologic type of lung cancer in relation to smoking habits, year of diagnosis and sites of metastases. Chest, 67(4), 382–​387. PMID: 1122765. Auerbach O, Stout AP, Hammond EC, and Garfinkel L. 1961. Changes in bronchial epithelium in relation to cigarette smoking and in relation to lung cancer. N Engl J Med, 265, 253–​267. PMID: 13685078. Bach P. 2008. Computed tomography screening for lung cancer. Clin Cancer Res, 14(8), 2511; author reply 2511–​2512. PMID: 18413846. Bach PB, Kattan MW, Thornquist MD, et  al. 2003. Variations in lung cancer risk among smokers. J Natl Cancer Inst, 95(6), 470–​478. PMID: 12644540. Bailey-​Wilson JE, Amos CI, Pinney SM, et al. 2004. A major lung cancer susceptibility locus maps to chromosome 6q23–​25. Am J Hum Genet, 75(3), 460–​474. PMCID: 1182024. Bain C, Feskanich D, Speizer FE, et al. 2004. Lung cancer rates in men and women with comparable histories of smoking. J Natl Cancer Inst, 96(11), 826–​834. PMID: 15173266. Barone-​Adesi F, Chapman RS, Silverman DT, et al. 2012. Risk of lung cancer associated with domestic use of coal in Xuanwei, China:  retrospective cohort study. BMJ, 345, e5414. PMCID: 3431444. Berrington de Gonzalez A, Gilbert E, Curtis R, et al. 2013. Second solid cancers after radiation therapy:  a systematic review of the epidemiologic studies of the radiation dose-​response relationship. Int J Radiat Oncol Biol Phys, 86(2), 224–​233. PMCID: 3816386. Berrington de Gonzalez A, Kim KP, and Berg CD. 2008. Low-​dose lung computed tomography screening before age 55:  estimates of the mortality reduction required to outweigh the radiation-​induced cancer risk. J Med Screen, 15(3), 153–​158. PMCID: 2782431. Blot WJ, Cohen SS, Aldrich M, et  al. 2011. Lung cancer risk among smokers of menthol cigarettes. J Natl Cancer Inst, 103(10), 810–​ 816. PMCID: 3096798. Boffetta P, Clark S, Shen M, et al. 2006. Serum cotinine level as predictor of lung cancer risk. Cancer Epidemiol Biomarkers Prev, 15(6), 1184–​1188. PMID: 16775179. Bonjour S, Adair-​Rohani H, Wolf J, et al. 2013. Solid fuel use for household cooking: country and regional estimates for 1980–​2010. Environ Health Perspect, 121(7), 784–​790. PMCID: PMC3701999. Brennan P, Hainaut P, and Boffetta P. 2011. Genetics of lung-​cancer susceptibility. Lancet Oncol, 12(4), 399–​408. PMID: 20951091. Brenner DR, Boffetta P, Duell EJ, et  al. 2012. Previous lung diseases and lung cancer risk: a pooled analysis from the International Lung Cancer Consortium. Am J Epidemiol, 176(7), 573–​585. Brooks DR, Palmer JR, Strom BL, and Rosenberg L. 2003. Menthol cigarettes and risk of lung cancer. Am J Epidemiol, 158(7), 609–​616; discussion 617–​620. PMID: 14507595. Burns D, Lee L, Shen LZ, et  al. 1997. Cigarette smoking behavior in the United States. In: Shopland D, Burns D, Garfinkel L, and Samet J (Eds.), Smoking and Tobacco Control Monograph 8.  Changes in Cigarette Related Disease Risks and Their Implication for Prevention and Control (pp. 13–​112). Washington, DC: National Cancer Institute. Cancer Genome Atlas Research Network. 2012. Comprehensive genomic characterization of squamous cell lung cancers. Nature, 489(7417), 519–​525. PMCID: 3466113. Cancer Genome Atlas Research Network. 2014. Comprehensive molecular profiling of lung adenocarcinoma. Nature, 511(7511), 543–​ 550. PMCID: 4231481. Cassidy A, Myles JP, van Tongeren M, et  al. 2008. The LLP risk model:  an individual risk prediction model for lung cancer. Br J Cancer, 98(2), 270–​ 276. PMCID: 2361453. Centers for Disease Control and Prevention (CDC). 2000. Declines in lung cancer rates: California, 1988–​1997. MMWR Morb Mortal Wkly Rep, 49, 1066–​1969. PMCID: PMID: 11186612. Chai H, and Brown RE. 2009. Field effect in cancer-​an update. Ann Clin Lab Sci, 39(4), 331–​337. PMID: 19880759. Chang JW, Asamura H, Kawachi R, and Watanabe S. 2009. Gender difference in survival of resected non-​small cell lung cancer: histology-​related phenomenon? J Thorac Cardiovasc Surg, 137(4), 807–​812. PMID: 19327500.

Chen J, Pee D, Ayyagari R, et al. 2006. Projecting absolute invasive breast cancer risk in white women with a model that includes mammographic density. J Natl Cancer Inst, 98(17), 1215–​1226. PMID: 16954474. Chen LS, Hung RJ, Baker T et al. 2015. CHRNA5 risk variant predicts delayed smoking cessation and earlier lung cancer diagnosis: a meta-​analysis. J Natl Cancer Inst, 107(5). PMID: 25873736. Chen Z, Peto R, Zhou M, et  al. 2015. Contrasting male and female trends in tobacco-​ attributed mortality in China:  evidence from successive nationwide prospective studies. Lancet, 386(10002), 1447–​ 1456. PMC: 4691901. Cote ML, Liu M, Bonassi S, et al. 2012. Increased risk of lung cancer in individuals with a family history of the disease: a pooled analysis from the International Lung Cancer Consortium. Eur J Cancer, 48(13), 1957–​ 1968. PMCID: 3445438. Cronin KA, Gail MH, Zou Z, et al. 2006. Validation of a model of lung cancer risk prediction among smokers. J Natl Cancer Inst, 98(9), 637–​640. PMID: 16670389. D’Amelio AM, Jr, Cassidy A, Asomaning K, et al. 2010. Comparison of discriminatory power and accuracy of three lung cancer risk models. Br J Cancer, 103(3), 423–​429. PMCID: 2920015. DeLancey JO, Thun MJ, Jemal A, and Ward EM. 2008. Recent trends in Black-​ White disparities in cancer mortality. Cancer Epidemiol Biomarkers Prev, 17(11), 2908–​2912. PMID: 18990730. Denholm R, Schuz J, Straif K et al. 2014. Is previous respiratory disease a risk factor for lung cancer? Am J Respir Crit Care Med, 190(5), 549–​559. PMCID: 4214084. Ding L, Getz G, Wheeler DA, et  al. 2008. Somatic mutations affect key pathways in lung adenocarcinoma. Nature, 455(7216), 1069–​ 1075. PMCID: PMC2694412. Doll R, Peto R, Boreham J, and Sutherland I. 2005. Mortality from cancer in relation to smoking:  50  years observations on British doctors. Br J Cancer, 92(3), 426–​429. PMCID: 2362086. Egawa H, Furukawa K, Preston D, et al. 2012. Radiation and smoking effects on lung cancer incidence by histological types among atomic bomb survivors. Radiat Res, 178(3), 191–​201. PMCID: 4031371. El-​Zein M, Parent ME, Siemiatycki J, and Rousseau MC. 2014. History of allergic diseases and lung cancer risk. Ann Allergy Asthma Immunol, 112(3), 230–​236. PMID: 24439421. Etzel CJ, and Bach PB. 2011. Estimating individual risk for lung cancer. Semin Respir Crit Care Med, 32(1), 3–​9. PMID: 21500119. Fears TR, Guerry Dt, Pfeiffer RM, et al. 2006. Identifying individuals at high risk of melanoma:  a practical predictor of absolute risk. J Clin Oncol, 24(22), 3590–​3596. PMID: 16728488. Ferlay J, Soerjomataram I, Ervik M, et  al. 2013. GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality Worldwide. Available from: International Agency for Research on Cancer, http://​globocan.iarc.fr. Retrieved August 7, 2015. Freedman ND, Abnet CC, Caporaso NE, et al. 2016. Impact of changing US cigarette smoking patterns on incident cancer:  risks of 20 smoking-​ related cancers among the women and men of the NIH-​AARP cohort. Int J Epidemiol, 45(3), 846–​856. PMCID: PMC5005940. Freedman ND, Leitzmann MF, Hollenbeck AR, Schatzkin A, and Abnet CC. 2008. Cigarette smoking and subsequent risk of lung cancer in men and women: analysis of a prospective cohort study. Lancet Oncol, 9(7), 649–​ 656. PMCID: 2601691. Funatogawa I, Funatogawa T, and Yano E. 2013. Trends in smoking and lung cancer mortality in Japan, by birth cohort, 1949–​2010. Bull World Health Organ, 91(5), 332–​340. PMCID: 3646352. Furukawa K, Preston DL, Lonn S, et al. 2010. Radiation and smoking effects on lung cancer incidence among atomic bomb survivors. Radiat Res, 174(1), 72–​82. PMCID: 3857029. Gail MH, Brinton LA, Byar DP, et al. 1989. Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. J Natl Cancer Inst, 81(24), 1879–​1886. PMID: 2593165. Gandini S, Botteri E, Iodice S, et  al. 2008. Tobacco smoking and cancer:  a meta-​analysis. Int J Cancer, 122(1), 155–​164. PMID: 17893872. Gao C, Patel CJ, Michailidou K, et al. 2016. Mendelian randomization study of adiposity-​related traits and risk of breast, ovarian, prostate, lung and colorectal cancer. Int J Epidemiol, 45(3), 896–​908. PMID: 27427428. Gao YT, Blot WJ, Zheng W, et al. 1987. Lung cancer among Chinese women. Int J Cancer, 40(5), 604–​609. PMID: 2824385. Gilbert ES, Sokolnikov ME, Preston DL, et al. 2013. Lung cancer risks from plutonium: an updated analysis of data from the Mayak worker cohort. Radiat Res, 179(3), 332–​342. PMCID: 3661277. Global Burden of Disease Cancer, Collaboration. 2015. The Global Burden of Cancer 2013. JAMA Oncol, 1(4), 505–​527. PMCID: 4500822.

 539



Lung Cancer

Govindan R, Ding L, Griffith M, et al. 2012. Genomic landscape of non-​small cell lung cancer in smokers and never-​smokers. Cell, 150(6), 1121–​1134. PMCID: PMC3656590. Guo L, Liu S, Zhang S, et al. 2015. [A meta-​analysis of body mass index and the risk of lung cancer in the Chinese population]. Zhonghua Yu Fang Yi Xue Za Zhi, 49(7), 649–​653. PMID: 26310480. Han W, and Lo HW. 2012. Landscape of EGFR signaling network in human cancers: biology and therapeutic response in relation to receptor sub-­ cellular locations. Cancer Lett, 318(2), 124–​134. PMCID: 3304012. Harris JE. 1983. Cigarette smoking among successive birth cohorts of men and women in the United States during 1900–​80. J Natl Cancer Inst, 71(3), 473–​479. PMID: 6577223. Henley SJ, Flanders WD, Manatunga A, and Thun MJ. 2002. Leanness and lung cancer risk:  fact or artifact? Epidemiology, 13(3), 268–​276. PMID: 11964927. Henschke CI, and Miettinen OS. 2004. Women’s susceptibility to tobacco carcinogens. Lung Cancer, 43(1), 1–​5. PMID: 14698531. Hidayat K, Du X, Chen G, Shi M, and Shi B. 2016. Abdominal obesity and lung cancer risk: systematic review and meta-​analysis of prospective studies. Nutrients, 8(12). PMCID: PMC5188465. Holford TR, Levy DT, McKay LA, et  al. 2014. Patterns of birth cohort-​ specific smoking histories, 1965–​2009. Am J Prev Med, 46(2), e31–​e37. PMCID: 3951759. Hosgood HD, 3rd, Boffetta P, Greenland S, et  al. 2010. In-​home coal and wood use and lung cancer risk:  a pooled analysis of the International Lung Cancer Consortium. Environ Health Perspect, 118(12), 1743–​1747. PMCID: 3002194. Howlader N, Noone AM, Krapcho MD, et al. 2015. SEER Cancer Statistics Review, 1975–​2012. http://​seer.cancer.gov/​csr/​1975_​2012/​, based on November 2014 SEER data submission, posted to the SEER website, April 2015. Hsiung CA, Lan Q, Hong YC, et al. 2010. The 5p15.33 locus is associated with risk of lung adenocarcinoma in never-​smoking females in Asia. PLoS Genet, 6(8). PMCID: 2916850. Huxley R, Jamrozik K, Lam TH, et  al. 2007. Impact of smoking and smoking cessation on lung cancer mortality in the Asia-​Pacific region. Am J Epidemiol, 165(11), 1280–​1286. PMID: 17369610. Hwang SJ, Cheng LS, Lozano G, et  al. 2003. Lung cancer risk in germline p53 mutation carriers: association between an inherited cancer predisposition, cigarette smoking, and cancer risk. Hum Genet, 113(3), 238–​243. PMID: 12802680. International Agency for Research on Cancer (IARC). 2004. Tobacco smoke and involuntary smoking. IARC Monogr Eval Carcinogenic Risks Hum (Vol. 83), xiv. PMID: 6642935. IARC. 2010. Household use of solid fuels and high-​temperature frying. IARC Monogr Eval Carcinogenic Risks Hum (Vol. 95). Lyon, France: IARC. IARC. 2012a. Diesel and gasoline engine exhausts and some nitroarenes. IARC Monogr Eval Carcinogenic Risks Hum (Vol. 105). Lyon, France: IARC. IARC. 2012b. Household use of solid fuels and high-​temperature frying. IARC Monogr Eval Carcinogenic Risks Hum (Vol. 95). Lyon, France: IARC. IARC. 2012c. A review of human carcinogens. E. Personal habits and indoor combustions. IARC Monogr Eval Carcinogenic Risks Hum (Vol. 100E). Lyon, France: IARC. IARC. 2013. Diesel and gasoline engine exhausts and some nitroarenes. IARC Monogr Eval Carcinogenic Risks Hum (Vol. 105). Lyon, France: IARC. IARC. 2015. Outdoor air pollution. IARC Monogr Eval Carcinogenic Risks Hum (Vol. 109). Lyon, France: IARC. Imielinski M, Berger AH, Hammerman PS, et al. 2012. Mapping the hallmarks of lung adenocarcinoma with massively parallel sequencing. Cell, 150(6), 1107–​1120. PMCID: PMC3557932. International Early Lung Cancer Action Program, Henschke CI, Yip R, and Miettinen OS. 2006. Women’s susceptibility to tobacco carcinogens and survival after diagnosis of lung cancer. JAMA, 296(2), 180–​184. PMID: 16835423. Institute of Medicine (IOM). 2013. Delivering High-​Quality Cancer Care: Charting a New Course for a System in Crisis. Washington, DC: National Academy Press. James MA, Wen W, Wang Y, et  al. 2012. Functional characterization of CLPTM1L as a lung cancer risk candidate gene in the 5p15.33 locus. PLoS One, 7(6), e36116. PMCID: 3366984. Janssens AC, Moonesinghe R, Yang Q, et al. 2007. The impact of genotype frequencies on the clinical validity of genomic profiling for predicting common chronic diseases. Genet Med, 9(8), 528–​ 535. PMID: 17700391. Jemal A, Cokkinides VE, Shafey O, and Thun MJ. 2003. Lung cancer trends in young adults: an early indicator of progress in tobacco control (United States). Cancer Causes Control, 14(6), 579–​585. PMID: 12948289.

539

Jemal A, Thun MJ, Ries LA, et al. 2008. Annual report to the nation on the status of cancer, 1975–​2005, featuring trends in lung cancer, tobacco use, and tobacco control. J Natl Cancer Inst, 100(23), 1672–​1694. PMCID: 2639291. Jha P, and Peto R. 2014. Global effects of smoking, of quitting, and of taxing tobacco. N Engl J Med, 370(1), 60–​68. PMID: 24382066. Jin ZY, Wu M, Han RQ, et al. 2014. Household ventilation may reduce effects of indoor air pollutants for prevention of lung cancer:  a case-​control study in a Chinese population. PLoS One, 9(7), e102685. PMCID: 4097600. Kabat GC, and Wynder EL. 1984. Lung cancer in nonsmokers. Cancer, 53(5), 1214–​1221. PMID: 6692309. Kandoth C, McLellan MD, Vandin F, et  al. 2013. Mutational landscape and significance across 12 major cancer types. Nature, 502(7471), 333–​339. PMCID: 3927368. Kim C, Gao YT, Xiang YB, et al. 2015. Home kitchen ventilation, cooking fuels, and lung cancer risk in a prospective cohort of never smoking women in Shanghai, China. Int J Cancer, 136(3), 632–​638. PMCID: 4232458. Kim HS, Lee HJ, Jeon JH, et al. 2013. Natural history of ground-​glass nodules detected on the chest computed tomography scan after major lung resection. Ann Thorac Surg, 96(6), 1952–​1957. PMID: 24083798. Kong J, Xu F, He M, Chen K, and Qian B. 2014. The incidence of lung cancer by histological type:  a population-​based study in Tianjin, China during 1981–​2005. Respirology, 19(8), 1222–​1228. PMID: 25168588. Krewski D, Lubin JH, Zielinski JM, et al. 2006. A combined analysis of North American case-​control studies of residential radon and lung cancer. J Toxicol Environ Health A, 69(7), 533–​597. PMID: 16608828. Kreyberg L. 1962. Histological lung cancer types:  a morphological and biological correlation. Acta Pathol Microbiol Scand Suppl, Suppl 157, 1–​92. PMID: 14035775. Langer CJ, Besse B, Gualberto A, Brambilla E, and Soria JC. 2010. The evolving role of histology in the management of advanced non-​small-​cell lung cancer. J Clin Oncol, 28(36), 5311–​5320. PMID: 21079145. Lewis DR, Check DP, Caporaso NE, Travis WD, and Devesa SS. 2014. US lung cancer trends by histologic type. Cancer, 120(18), 2883–​ 2892. PMCID: 4187244. Li K, Husing A, Sookthai D, et  al. 2015. Selecting high-​risk individuals for lung cancer screening:  a prospective evaluation of existing risk models and eligibility criteria in the German EPIC cohort. Cancer Prev Res (Phila), 8(9), 777–​785. PMID: 26076698. Li X, Mutanen P, and Hemminki K. 2001. Gender-​specific incidence trends in lung cancer by histological type in Sweden, 1958–​1996. Eur J Cancer Prev, 10(3), 227–​235. PMID: 11432709. Lopez AD, Collishaw N, and Piha T. 1994. A descriptive model of the cigarette epidemic in developed countries. Tob Control, 3, 242–​247. Lortet-​Tieulent J, Soerjomataram I, Ferlay J, et al. 2014. International trends in lung cancer incidence by histological subtype: adenocarcinoma stabilizing in men but still increasing in women. Lung Cancer, 84(1), 13–​22. PMID: 24524818. Lozano R, Naghavi M, Foreman K, et al. 2012. Global and regional mortality from 235 causes of death for 20 age groups in 1990 and 2010:  a systematic analysis for the Global Burden of Disease Study 2010. Lancet, 380(9859), 2095–​2128. PMID: 23245604. Marshall RJ. 2001. Displaying clinical data relationships using scaled rectangle diagrams. Stat Med, 20(7), 1077–​1088. PMID: 11276037. Marshall RJ. 2005. Scaled rectangle diagrams can be used to visualize clinical and epidemiological data. J Clin Epidemiol, 58(10), 974–​ 981. PMID: 16168342. Marshall RJ. 2009. Cardiovascular risk can be represented by scaled rectangle diagrams. J Clin Epidemiol, 62(9), 998–​1000. PMID: 19635404. Matakidou A, Eisen T, and Houlston RS. 2005. Systematic review of the relationship between family history and lung cancer risk. Br J Cancer, 93(7), 825–​833. PMCID: 2361640. McKay JD, Hung RJ, Gaborieau V, et al. 2008. Lung cancer susceptibility locus at 5p15.33. Nat Genet, 40(12), 1404–​1406. PMCID: 2748187. Metayer C, Wang Z, Kleinerman RA, et al. 2002. Cooking oil fumes and risk of lung cancer in women in rural Gansu, China. Lung Cancer, 35(2), 111–​ 117. PMID: 11804682. Meza R, Meernik C, Jeon J, and Cote ML. 2015. Lung cancer incidence trends by gender, race and histology in the United States, 1973–​2010. PLoS One, 10(3), e0121323. PMCID: 4379166. Midha A, Dearden S, and McCormack R. 2015. EGFR mutation incidence in non-​small-​cell lung cancer of adenocarcinoma histology:  a systematic review and global map by ethnicity (mutMapII). Am J Cancer Res, 5(9), 2892–​2911. PMCID: PMC4633915. Mitsudomi T. 2014. Molecular epidemiology of lung cancer and geographic variations with special reference to EGFR mutations. Transl Lung Cancer Res, 3(4), 205–​211. PMCID: 4367697.

540

540

PART IV:  Cancers by Tissue of Origin

Myneni AA, Chang SC, Niu R, et al. 2013. Genetic polymorphisms of TERT and CLPTM1L and risk of lung cancer: a case-​control study in a Chinese population. Lung Cancer, 80(2), 131–​137. PMCID: 3627395. National Center for Health Statistics, Centers for Disease Control and Prevention. 2016a. US Mortality volumes 1930 to 1959 and US Mortality data 1960 to 2013. National Center for Health Statistics, Centers for Disease Control and Prevention. 2016b. National Health Interview Surveys 1965–2013. National Council on Radiation Protection and Measurements (NCRP). 2009. Ionizing Radiation Exposure of the Population of the United States. NCRP Report No. 160. Bethesda, MD:  National Council on Radiation Protection and Measurements. National Lung Screening Trial Research Team. 2011. Reduced lung-​cancer mortality with low-​dose computed tomographic screening. N Engl J Med, 365, 395–​409. PMCID: PMC4356534. National Research Council. 1999. Health Effects of Exposure to Radon: BEIR VI. Washington, DC: National Academy Press. Nik-​Zainal S, Kucab JE, Morganella S, et al. 2015. The genome as a record of environmental exposure. Mutagenesis, 30(6), 763–​ 770. PMCID: PMC4637815. Norum JH, Andersen K, and Sorlie T. 2014. Lessons learned from the intrinsic subtypes of breast cancer in the quest for precision therapy. Br J Surg, 101(8), 925–​938. PMID: 24849143. Paech DC, Weston AR, Pavlakis N, et  al. 2011. A systematic review of the interobserver variability for histology in the differentiation between squamous and nonsquamous non-​small cell lung cancer. J Thorac Oncol, 6(1), 55–​63. PMID: 21107286. Pandey KD, Wheeler D, Ostro B, et  al. 2006. Ambient Particulate Matter Concentrations in Residential and Pollution Hotspot Areas of World Cities: New Estimates Based on the Global Model of Ambient Particulates (GMAPS). Washington, DC:  The World Bank Development Economics Research Group and the Environment Department Working Paper, The World Bank. Pao W, and Hutchinson KE. 2012. Chipping away at the lung cancer genome. Nat Med, 18(3), 349–​351. PMID: 22395697. Pao W, Iafrate AJ, and Su Z. 2011. Genetically informed lung cancer medicine. J Pathol, 223(2), 230–​240. PMID: 21125677. Parkin DM. 2008. The role of cancer registries in cancer control. Int J Clin Oncol, 13(2), 102–​111. PMID: 18463952. Parkin DM, Ferlay J, Curado M, et  al. 2014. Cancer Incidence in Five Continents (CI5), Volumes I to X. IARC CancerBase No. 12 [Internet]. Available from: http://​ci5.iarc.fr. Accessed June 24, 2015. Pesch B, Kendzia B, Gustavsson P, et  al. 2012. Cigarette smoking and lung cancer:  relative risk estimates for the major histological types from a pooled analysis of case-​control studies. Int J Cancer, 131(5), 1210–​1219. PMCID: 3296911. Pesic M, Podolski-​Renic A, Stojkovic S, et  al. 2015. Anti-​cancer effects of cerium oxide nanoparticles and its intracellular redox activity. Chem Biol Interact, 232, 85–​93. PMID: 25813935. Peto R, Darby S, Deo H, et al. 2000. Smoking, smoking cessation, and lung cancer in the UK since 1950: combination of national statistics with two case-​control studies. BMJ, 321(7257), 323–​329. PMCID: 27446. Peto R, Lopez AD, Pan H, Boreham J, and Thun M. 2017. Mortality from smoking in developed countries, 1950–​2020. Available from: https://​ www.ctsu.ox.ac.uk/​.../​mortality-​from-​smoking. Accessed April 28, 2017. Peto R, Lopez AD, Boreham J, Thun M, and Heath C, Jr. 1992. Mortality from tobacco in developed countries:  indirect estimation from national vital statistics. Lancet, 339(8804), 1268–​1278. PMID: 1349675. Pfeifer GP, Denissenko MF, Olivier M, et  al. 2002. Tobacco smoke carcinogens, DNA damage and p53 mutations in smoking-​associated cancers. Oncogene, 21(48), 7435–​7451. PMID: 12379884. Pinsky PF, and Berg CD. 2012. Applying the National Lung Screening Trial eligibility criteria to the US population: what percent of the population and of incident lung cancers would be covered? J Med Screen, 19(3), 154–​156. PMID: 23060474. Pirie K, Peto R, Reeves GK, et al. 2013. The 21st century hazards of smoking and benefits of stopping: a prospective study of one million women in the UK. Lancet, 381(9861), 133–​141. PMCID: 3547248. Rajaraman P, Sigurdson AJ, Doody MM, et al. 2006. Lung cancer risk among US radiologic technologists, 1983–​1998. Int J Cancer, 119(10), 2481–​ 2486. PMID: 16858679. Rosenberger A, Bickeboller H, McCormack V, et al. 2012. Asthma and lung cancer risk: a systematic investigation by the International Lung Cancer Consortium. Carcinogenesis, 33(3), 587–​597. PMCID: 3291861. Rostron B. 2012. Lung cancer mortality risk for U.S. menthol cigarette smokers. Nicotine Tob Res, 14(10), 1140–​1144. PMID: 22387991.

Samet JM, Avila-​Tang E, Boffetta P, et al. 2009. Lung cancer in never smokers:  clinical epidemiology and environmental risk factors. Clin Cancer Res, 15(18), 5626–​5645. PMCID: 3170525. Sanders BM, Jay M, Draper GJ, and Roberts EM. 1989. Non-​ocular cancer in relatives of retinoblastoma patients. Br J Cancer, 60(3), 358–​365. PMCID: 2247167. SEER. 2015. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Incidence—​SEER 18 Regs Research Data + Hurricane Katrina Impacted Louisiana Cases, Nov 2014. Sub (1973–​2012 varying)—​Linked to County Attributes—​Total U.S., 1969–​2013 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2015, based on the November 2014 submission. Retrieved December 2015, from Surveillance, Epidemiology, and End Results (SEER) Program, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, www.seer.cancer.gov. Shiels MS, Katki HA, Freedman ND, et al. 2014. Cigarette smoking and variations in systemic immune and inflammation markers. J Natl Cancer Inst, 106(11). PMCID: 4200029. Shiels MS, Pfeiffer RM, Hildesheim A, et al. 2013. Circulating inflammation markers and prospective risk for lung cancer. J Natl Cancer Inst, 105(24), 1871–​1880. PMCID: 3888091. Siegel RL, Jacobs EJ, Newton CC, et al. 2015. Deaths due to cigarette smoking for 12 smoking-​related cancers in the United States. JAMA Intern Med, 175(9), 1574–​1576. PMID: 26076120. Spitz MR, Hong WK, Amos CI, et al. 2007. A risk model for prediction of lung cancer. J Natl Cancer Inst, 99(9), 715–​726. PMID: 17470739. Straif K, Baan R, Grosse Y, et al. 2006. Carcinogenicity of household solid fuel combustion and of high-​temperature frying. Lancet Oncol, 7(12), 977–​ 978. PMID: 17348122. Swanton C, and Govindan R. 2016. Clinical implications of genomic discoveries in lung cancer. N Engl J Med, 374(19), 1864–​1873. PMID: 27168435. Swift M, and Chase C. 1979. Cancer in families with xeroderma pigmentosum. J Natl Cancer Inst, 62(6), 1415–​1421. PMID: 286113. Takemiya M, Shiraishi S, Teramoto T, and Miki Y. 1987. Bloom’s syndrome with porokeratosis of Mibelli and multiple cancers of the skin, lung and colon. Clin Genet, 31(1), 35–​44. PMID: 3568432. Tessema M, Willink R, Do K, et al. 2008. Promoter methylation of genes in and around the candidate lung cancer susceptibility locus 6q23–​25. Cancer Res, 68(6), 1707–​1714. PMID: 18339850. Thun M, Peto R, Boreham J, and Lopez AD. 2012. Stages of the cigarette epidemic on entering its second century. Tob Control, 21(2), 96–​101. PMID: 22345230. Thun MJ, Carter BD, Feskanich D, et  al. 2013. 50-​year trends in smoking-​ related mortality in the United States. N Engl J Med, 368(4), 351–​364. PMCID: 3632080. Thun MJ, Hannan LM, Adams-​Campbell LL, et al. 2008. Lung cancer occurrence in never-​smokers: an analysis of 13 cohorts and 22 cancer registry studies. PLoS Med, 5(9), e185. PMCID: 2531137. Thun MJ, Henley SJ, and Calle EE. 2002. Tobacco use and cancer:  an epidemiologic perspective for geneticists. Oncogene, 21(48), 7307–​7325. PMID: 12379875. Thun MJ, Lally CA, Flannery JT, et al. 1997. Cigarette smoking and changes in the histopathology of lung cancer. J Natl Cancer Inst, 89(21), 1580–​1586. PMID: 9362155. Tjulandin S, Imyanitov E, Moiseyenko V, et al. 2015. Prospective cohort study of clinical characteristics and management patterns for patients with NSCLC in the Russian Federation: EPICLIN-​Lung. Curr Med Res Opin, 31(6), 1117–​1127. PMID: 25867537. Tobacco Atlas. 2015. The Tobacco Atlas, 5th edition. Eriksen M, Mackay J, Schluger N, Gomeshtapeh F I, and Drope J (Eds.). 5th ed. Atlanta, GA: American Cancer Society, Anerican Lung Association. Tokuhata GK, and Lilienfeld AM. 1963. Familial aggregation of lung cancer among hospital patients. Public Health Rep, 78, 277–​ 283. PMCID: 1915246. Torre LA, Bray F, Siegel RL, et al. 2015. Global cancer statistics, 2012. CA Cancer J Clin, 65(2), 87–​108. PMID: 25651787. Tortora G and Derrickson B. The Respiratory System, John Wiley and Sons, 2009. https://www.mscc.edu/webs/sedwards/Bio_2020/PowerPoints/ch23.ppt Travis WD. 2010. Advances in neuroendocrine lung tumors. Ann Oncol, 21 Suppl 7, vii65–​71. PMID: 20943645. Travis WD. 2014. Lung Cancer World Cancer Report 2014 (pp. 350–​361). Lyon, France: IARC. Travis WD, Brambilla E, Nicholson AG, et al. 2015. The 2015 World Health Organization classification of lung tumors: impact of genetic, clinical and radiologic advances since the 2004 classification. J Thorac Oncol, 10(9), 1243–​1260. PMID: 26291008.

 541



Lung Cancer

Travis WD, Brambilla E, Noguchi M, et  al. 2011. International Association for the Study of Lung Cancer/​ American Thoracic Society/​ European Respiratory Society: international multidisciplinary classification of lung adenocarcinoma. J Thorac Oncol, 6(2), 244–285. PMID: 21252716. Travis WD, Brambilla E, Noguchi M, et al. 2013a. Diagnosis of lung adenocarcinoma in resected specimens: implications of the 2011 International Association for the Study of Lung Cancer/​American Thoracic Society/​ European Respiratory Society classification. Arch Pathol Lab Med, 137(5), 685–​705. PMID: 22913371. Travis WD, Brambilla E, Noguchi M, et al. 2013b. Diagnosis of lung cancer in small biopsies and cytology:  implications of the 2011 International Association for the Study of Lung Cancer/​American Thoracic Society/​ European Respiratory Society classification. Arch Pathol Lab Med, 137(5), 668–​684. PMCID: PMC4509741. US Department of Health and Human Services. 2006. The Health Consequences of Involuntary Exposure to Tobacco Smoke:  A  Report of the Surgeon General. Washington, DC:  US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. US Department of Health and Human Services. 2010. How Tobacco Smoke Causes Disease:  The Biology and Behavioral Basis for Smoking-​ Attributable Disease. A  Report of the Surgeon General. Washington, DC: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. US Department of Health and Human Services. 2014. The Health Consequences of Smoking:  50 Years of Progress. A  Report of the Surgeon General. Washington, DC: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. US Department of Health, Education and Welfare. 1964. Smoking and Health: Report of the Advisory Committee to the Surgeon General of the Public Health Service. Washington, DC:  US Department of Health, Education and Welfare, Public Health Service, Centers for Disease Control. US Environmental Protection Agency. 2017. Recommended Residential Radon Mitigation Standard of Practice. http://​epa.gov/​iaq/​radon/​pubs/​mitstds. html. Vermeulen R, Silverman DT, Garshick E, et al. 2014. Exposure-​response estimates for diesel engine exhaust and lung cancer mortality based on data from three occupational cohorts. Environ Health Perspect, 122(2), 172–​ 177. PMCID: 3915263.

541

Wakai K, Inoue M, Mizoue T, et al. 2006. Tobacco smoking and lung cancer risk: an evaluation based on a systematic review of epidemiological evidence among the Japanese population. Jpn J Clin Oncol, 36(5), 309–​324. PMID: 16735374. World Health Organization (WHO). 1967. World Health Organization Classification of Malignant Lung Tumors. https://​www.bing.com/​ images/​ s earch?q=1967+who+classification+of+lung+tumors&v iew=detailv2&id=58CEAA05EF7ED900274D8588D107D0F54 53D7FDE&selectedindex=28&ccid=M5UM%2FmuB&simid=607991298228685311&thid=OIP.M5UM_​muB6o5760gqWzjcdwEsBL &mode=overlay&first=1. WHO. 2005. WHO Handbook on Indoor Radon: A Public Health Perspective. http://​www.who.int/​ionizing_​radiation/​env/​radon/​en/​index1.html. Wilson DO, and Weissfeld J. 2015. A simple model for predicting lung cancer occurrence in a lung cancer screening program: The Pittsburgh Predictor. Lung Cancer, 89(1), 31–​37. PMCID: 4457558. Yamanaka A, Hirai T, Ohtake Y, and Kitagawa M. 1997. Lung cancer associated with Werner’s syndrome: a case report and review of the literature. Jpn J Clin Oncol, 27(6), 415–​418. PMID: 9438005. Yang Y, Dong J, Sun K, et al. 2013. Obesity and incidence of lung cancer: a meta-​analysis. Int J Cancer, 132(5), 1162–​1169. PMID: 22777722. Yu IT, Chiu YL, Au JS, Wong TW, and Tang JL. 2006. Dose-​response relationship between cooking fumes exposures and lung cancer among Chinese nonsmoking women. Cancer Res, 66(9), 4961–​4967. PMID: 16651454. Zaric B, Perin B, Stojsic V, et al. 2013. Detection of premalignant bronchial lesions can be significantly improved by combination of advanced bronchoscopic imaging techniques. Ann Thorac Med, 8(2), 93–​ 98. PMCID: 3667452. Zhao P, Dai M, Chen W, and Li N. 2010. Cancer trends in China. Jpn J Clin Oncol, 40(4), 281–​285. PMID: 20085904. Zheng T, Holford TR, Boyle P, et al. 1994. Time trend and the age-​period-​cohort effect on the incidence of histologic types of lung cancer in Connecticut, 1960–​1989. Cancer, 74(5), 1556–​1567. PMID: 8062189. Zhong L, Goldberg MS, Gao YT, and Jin F. 1999. Lung cancer and indoor air pollution arising from Chinese-​ style cooking among nonsmoking women living in Shanghai, China. Epidemiology, 10(5), 488–​494. PMID: 10468420. Zugazagoitia J, Enguita AB, Nunez JA, Iglesias L, and Ponce S. 2014. The new IASLC/​ATS/​ERS lung adenocarcinoma classification from a clinical perspective: current concepts and future prospects. J Thorac Dis, 6(Suppl 5), S526–​S536. PMCID: PMC4209392.

542

 543

29

Oral Cavity, Oropharynx, Lip, and Salivary Glands MIA HASHIBE, ERICH M. STURGIS, JACQUES FERLAY, AND DEBORAH M. WINN

OVERVIEW

ANATOMIC CLASSIFICATION

Cancers of the oral cavity, oropharynx, lip, and salivary glands are malignancies of the head and neck. Some of these cancer sites share risk factors, although each has distinctive anatomic, epidemiologic, and clinical features. Oral cavity cancers arise on the inner lip and buccal mucosa, anterior two-​thirds of the tongue, gum, hard palate, and floor of mouth. These cancers are strongly associated with the use of smoked and smokeless tobacco products, heavy alcohol consumption, and chewing of betel quid or pan, but only minimally associated with prior infection with human papillomavirus (HPV). In contrast, oropharyngeal cancers affect the posterior one-​third (base) of the tongue, tonsils, soft palate, and other oropharyngeal tissues and are strongly associated with HPV 16 infection as well as the use of tobacco, alcohol, and betel quid. Cancers of the external lip and major salivary glands are uncommon malignancies that are causally related to radiation therapy (salivary gland tumors) and tobacco and sun exposure (lip cancer) but not to HPV infection. In principle, tumors of the oral cavity, oropharynx, and lip are among the most preventable forms of cancer. Despite this, the increasing incidence of HPV-​related oropharyngeal cancers among men in some high-​ income countries is offsetting the decrease in oral cavity cancers due to reductions in tobacco use. The incidence of oral cavity cancer is increasing among men in several emerging economies, notably India and Brazil.

The International Classification of Disease (ICD) codes for cancers of the oral cavity, oropharynx, external lip, and major salivary gland cancers are shown in Table 29–​1 (Fritz et al., 2012). These criteria were developed by clinicians and epidemiologists in the INHANCE Consortium (Hashibe et al., 2007). Oral cavity cancers include the inner lip, tongue excluding the base, gum, floor of mouth, hard palate, and other parts of the mouth (Table 29–​2). Oropharyngeal cancers include the base of tongue, lingual tonsil, soft palate, uvula, tonsils, vallecula, oropharynx, and branchial cleft. In this chapter the major salivary glands are discussed as a separate grouping, but minor salivary glands (located in the lips, buccal mucosa, and other linings) are classified with the anatomic site in which they occur in the oral cavity and oropharynx. For this chapter we assigned codes that overlap two anatomic groupings or subgroupings to the more frequently occurring category. For example, cancers of the external lip are more common than cancers arising on the inner lip; thus tumors coded as “lip, not otherwise specified” (NOS) can be grouped with cancers of the external lip and not included with oral cavity cancers. Interpreting etiologic findings from epidemiologic studies can be challenging for several reasons. First, etiologic studies have inconsistently grouped various anatomic sites and subsites, so that the disease endpoints vary across studies. Second, even recent etiologic studies continue to aggregate data for sites that differ anatomically, histologically, and in terms of embryologic origin. For example, most publications and cancer surveillance systems present aggregated data for tongue cancer, whereas cancer of the base of the tongue should be grouped with oropharyngeal cancers, which have a similar embryologic history and high density of lymphoid tissue. Similarly, cancer of the anterior tongue should be classified with the oral cavity, since the surface epithelium of these sites is the usual location for neoplasia. Whereas some forms of misclassification have a negligible effect on incidence rates, others can change the direction of an apparent temporal trend. For example, the inclusion of rare salivary gland tumors with oral cavity cancers has minimal effect on the observed incidence rate, whereas the classification of cancers at the base of the tongue (which account for approximately 70% of all tongue cancers) as oral cavity rather than oropharyngeal cancer substantially affects both groupings. A  third classification problem results from the assignment of non-​ specific codes for tumor location. Some misclassification is probably unavoidable due to the anatomic complexity of this region, particularly for tumors that bridge multiple anatomic sites and are of unknown origin.

INTRODUCTION This chapter discusses the epidemiology of cancers of the oral cavity, oropharynx, lip, and salivary glands. Until recently, tumors of the oral cavity and oropharynx were typically grouped together, with or without the addition of cancers in neighboring tissues, into a single category termed “oral cavity and pharyngeal cancer” or “oropharyngeal cancer.” The former approach was used in the previous 2006 edition of this text, when Mayne et al. described epidemiologic studies of these cancers through the year 2000 (Mayne et al., 2006). It has since been recognized that oropharyngeal (International Agency for Research on Cancer [IARC], 2012b), but not oral cavity cancers, are strongly associated with human papillomavirus (HPV) infection, and that patients with HPV-​positive oropharyngeal cancers have a better prognosis than HPV-​negative oropharyngeal cancers (Goodman et al., 2015). A standardized classification system recently proposed by the International Head and Neck Cancer Epidemiology (INHANCE) Consortium specifies the anatomic boundaries that separate oropharyngeal from oral cavity cancers. This chapter will adhere to the standardized nomenclature where possible. Where relevant, we will describe how inconsistent taxonomy complicates the interpretation of temporal trends, especially for oral cavity and oropharyngeal cancers. Other head and neck cancers are discussed elsewhere in this volume. Tumors of the nasopharynx are reviewed in Chapter 26, and cancers of the hypopharynx and larynx in Chapter 27. In terms of tumor histology, squamous cell carcinomas comprise 88%–​95% of cancers of the oral cavity, oropharynx, and lip, whereas adenocarcinomas account for 62% of tumors of the major salivary glands (SEER, 2016).

ORAL CAVITY, OROPHARYNX, LIP, AND SALIVARY GLANDS COMBINED Disease Burden In the United States, the disease burden from these cancers is often reported for the four sites combined. Approximately 42,000 new cases (30,000 males and 12,000 females) and 8,900 deaths (6,200 males and 2,600 females) occurred in the United States in 2013 (US Cancer

543

54

Table 29–​1.  INHANCE ICD Codes for Cancers of the Oral Cavity, Oropharynx, and Hypopharynx .0 C00 C01 C02 C03 C04 C05 C06 C07 C08 C09 C10 C11 C12 C13 C14

.1

External lip -​ Oral Oral Oral Oral Oral -​ Salivary gland Oropharynx Oropharynx Nasopharynx -​ Hypopharynx OOH NOS

External lip -​ Oral Oral Oral Oropharynx Oral -​ Salivary gland Oropharynx Larynx Nasopharynx -​ Hypopharynx -​

.2

.3

.4

External lip -​ Oral -​ -​ Oropharynx Oral -​ -​ -​ Oropharynx Nasopharynx -​ Hypopharynx OOH NOS

Oral -​ Oral -​ -​ -​ -​ -​ -​ -​ Oropharynx Nasopharynx -​ -​ -​

Oral -​ Oropharynx -​ -​ -​ -​ -​ -​ -​ Oropharynx -​ -​ -​ -​

.5

.6

Oral -​ -​ -​ -​ -​ -​ -​ -​ -​ -​ -​ -​ -​ -​

Oral -​ -​ -​ -​ -​ -​ -​ -​ -​ -​ -​ -​ -​ -​

.8 Oral -​ OOH NOS -​ Oral OOH NOS Oral -​ Salivary gland Oropharynx Oropharynx Nasopharynx -​ Hypopharynx OOH NOS

.9 Oral Oropharynx OOH NOS Oral Oral OOH NOS Oral Salivary gland Salivary gland Oropharynx Oropharynx Nasopharynx Hypopharynx Hypopharynx -​

Oral cavity-​oropharynx-​hypopharynx NOS (OOH NOS) Please note that C02.8, C02.9, C05.8, C05.9 were categorized with oropharyngeal cancer for the descriptive figures in this chapter.

Table 29–​2.  ICD Codes for Tissues Included as Cancers of the Oral Cavity, Oropharynx, and Oral Cavity/Oropharynx/Hypopharynx Not Specified Oral Cavity Cancers C00.3 C00.4 C00.5 C00.6 C00.8 C00.9 C02.0 C02.1 C02.2 C02.3

Oropharyngeal Cancers Base of tongue Base of tongue, NOS Lingual tonsil Soft palate, NOS Uvula Tonsil Tonsillar fossa Tonsillar pillar Overlapping lesion of tonsil Tonsil, NOS Oropharynx Vallecula Lateral wall of oropharynx Posterior wall of oropharynx Branchial cleft Overlapping lesion of oropharynx Oropharynx, NOS

C04.0 C04.1

Mucosa of upper lip Mucosa of lower lip Mucosa of lip, NOS Commissure of lip Overlapping lesion of lip Lip, NOS Dorsal surface of tongue, NOS Border of tongue Ventral surface of tongue, NOS Anterior 2/​3 of tongue, NOS Gum Upper gum Lower gum Gum, NOS Floor of mouth Anterior floor of mouth Lateral floor of mouth

C04.8

Overlapping lesion of floor of mouth

Oral cavity-​oropharynx-​hypopharynx NOS

C04.9 C05.0

Floor of mouth, NOS Hard palate Other and unspecified parts of mouth Cheek mucosa Vestibule of mouth Retromolar area Overlapping lesion of other and unspecified parts of mouth Mouth, NOS

C02.8 C02.9 C05.8 C05.9 C14.0 C14.2 C14.8

C03.0 C03.1 C03.9

C06.0 C06.1 C06.2 C06.8 C06.9

C01.9 C02.4 C05.1 C05.2 C09.0 C09.1 C09.8 C09.9 C10.0 C10.2 C10.3 C10.4 C10.8 C10.9

Overlapping lesion of tongue Tongue, NOS Overlapping lesion of palate Palate, NOS Pharynx, NOS Waldeyer’s ring Overlapping lesion of lip, oral cavity and pharynx

 54



Oral Cavity, Oropharynx, Lip, and Salivary Glands

Statistics [USCS], 2016). The lifetime risk of developing any of these cancers was 1.10% (Howlader et  al., 2015). Oropharyngeal cancers accounted for about half (52%) of these cases, oral cavity for 30%, external lip for 5%, and salivary gland cancers for 13% (Howlader et al., 2016). Collectively, these cancers accounted for approximately 2.2% of all incident cancers in the United States in 2013 (2.8% in men, 1.3% in women) (Howlader et al., 2016). The combination of external lip, oral cavity, oropharyngeal, and major salivary gland cancers ranked 9th in incidence for men and 14th in incidence for women (Howlader et al., 2016). The corresponding numbers for mortality were 15th for men and 20th for women. The number of new cases are estimated to have increased to 48,400 (34,800 among males, 13,600 in females) for 2016 (Siegel et al., 2016). The 2016 projections and lifetime risks also include rare cancers of the hypopharynx and nasopharynx. The International Agency for Research on Cancer (IARC) compiles global estimates that combine cancers of the hypopharynx and anterior surface of the epiglottis (larynx) with oropharyngeal cancers. By this definition, approximately 443,000 cases and 242,000 deaths from oral cavity and pharyngeal cancer were estimated to have occurred worldwide in 2012 (Ferlay et al., 2013). Men contribute almost two-​thirds of cases of both diagnoses (314,000 male, 129,000 female) and death (176,000 male, 66,000 female). Collectively these cancers accounted for 3.1% of incident cancers (4.3% in men and 1.9% in women) (Ferlay et al., 2013) and ranked 8th in the number of incident cases for males and 14th for females. For mortality they ranked 6th for men and 15th for women.

Incidence and Mortality Rates in the United States Based on data from the Surveillance, Epidemiology, and End Results Program (SEER, 2016) the incidence rate standardized to the US 2010 Census population for all four sites combined from 2009–​2013 was 9.6 per 100,000 persons per year in the United States (Table 29–​3). The incidence rate in men was 2.6 times higher than the rate in women (14.3 vs. 5.5). Among males and females combined, the rates were higher for whites (10.3 per 100,000 per year) than blacks (7.3 per 100,000). The incidence rates for American Indians/​Alaskan Natives and Asian/​Pacific Islanders were much lower (4.8 and 4.9 per 100,000, respectively). Incidence rates for Hispanics were 5.7 per 100,000. The mortality rates for 2009–​2013 combined were 1.6 per 100,000 per year (2.5 for males and 1.0 for females) (Howlader et al., 2016). Although the incidence rate is higher in white than in black men (Table 29–​3), the opposite is true for mortality (2.9 per 100,000 for black men vs. 2.7 per 100,000 for white men) (Howlader et al., 2016).

Temporal Trends by Subsite and Race/​Ethnicity in the United States Figure 29–​1 shows that the incidence rate of cancers of oral cavity, oropharynx, lip, and salivary gland combined from 1975 to 2013 decreased by almost half after peaking in 1990 among black Americans, but increased among whites since the early 2000s (SEER 2016). For oral cavity cancer, the incidence rate decreased among blacks more quickly than for whites; consequently the rate in whites has exceeded that in blacks since approximately 1990. For oropharyngeal cancer, the incidence rate decreased in blacks but increased in whites, crossing over in approximately 2005. Increases in oropharyngeal and to a lesser extent in oral cavity cancer were observed only for white men, and was driven by cancers at the base of tongue, tonsil not specified, and oropharynx not specified (Table 29–​3). Mortality rates for the four cancer sites combined have also increased among white men since the early 2000s after a long period of declining rates. Mortality rates have been stable for white women and decreasing for black men and women (Figures 29–​2 and 29–​3). In an analysis of age-​time-​cohort trends in anterior tongue, other oral cavity, and oropharyngeal cancer using SEER data up to 2008, the trends varied by sex, race, and ethnicity. Among white men, incidence rates for oral cavity and hypopharyngeal cancer declined, but increased for oropharyngeal cancer and anterior tongue (including

545

overlapping categories and tongue cancer NOS) (Brown et al., 2011, 2012). Among black males, rates decreased for all three cancer groups. Among women, black or white, only rates of oral tongue cancers were increasing, but not the others. When examined by birth cohorts, oropharyngeal cancer is increasing, particularly for the more recent year of birth cohorts among whites, with a suggestion of an increase among blacks in the youngest birth cohort (Brown et  al., 2012). An analysis combining multistage clonal expansion models with SEER data through 2012 suggests that anterior tongue cancer may have a different etiology than either the oral cavity cancers related to smoking and alcohol or the oropharyngeal cancers related to HPV (Brouwer et al., 2016). This issue deserves further study.

Geographic Distribution Maps showing mortality rates by state for 2009–​2013 show that among men death rates for oral cavity, oropharyngeal, lip and salivary gland are highest in the Southeastern United States, but the death rate patterns are less clear for women (Figure 29–​4 a–​b).

International Incidence Rates and Trends For oral cavity, lip, and salivary gland cancers combined (as grouped in GLOBOCAN 2012), the highest incidence rates (Age standardized rates [ASR], to the World Standard) are observed in Melanesia, South-​Central Asia, and Australia/​New Zealand (Figure 29–​5). For oropharyngeal and hypopharyngeal cancers combined (designated as pharyngeal cancers in GLOBOCAN 2012), the highest incidence rates are observed among men in Western Europe, South-​Central Asia, and Central and Eastern Europe (Figure 29–​6).

Tumor Progression Models The Cancer Genome Atlas (TCGA) Network comprehensively assessed the somatic genetic alterations in 279 head and neck squamous cell carcinoma tissue samples (TCGA, 2015). The majority of the samples were from oral cavity cancers (n = 172; 62%), while oropharyngeal (n = 33; 12%) and laryngeal cancer samples (n = 72; 26%) were also included. HPV positivity was detected in 64% of the oropharyngeal tumors and 6% of the non-​oropharyngeal tumors. Common genetic alterations in the HPV-​positive head and neck cancers included oncogene PIK3CA amplification, TRAF3 deletion and truncating mutations, and amplification of the cell cycle gene E2F1. HPV-​negative tumors were associated with loss of function TP53 mutations, CDKN2A inactivation, and amplification of 3q 26/​28 (TP63, SOX2, and PIK3CA), 11q13 (CCND1, FADD, CTTN), and 11q22 (BIRC2, YAP1). Other alterations included those in NSD1, AJUBA, Fat1, and NFE2L2. Another comprehensive analysis of somatic mutations was conducted on 130 head and neck squamous cell carcinoma samples (19.2% oral cavity, 57.5% oropharynx, 22.3% larynx, and 0.1% other) (Keck et al., 2015). The overall study included data from four discovery cohorts and four independent validations cohorts for a total of 938 head and neck cancer patients. A five subtype classification was proposed including (a) basal subtype (BA) with enrichment for hypoxia signaling, neuregulin signaling, and overexpression of epithelial markers; (b) classical subtype: CL-​HPV and CL-​nonHPV, with enrichment for the putrescine degradation pathway; (c)  inflamed/​mesenchymal subtypes (IMS-​HPV and IMS-​non-​HPV), with expression of immune response genes, increased expression of mesenchymal genes, and downregulation of epithelial markers.

Survival Five-​year relative survival for oral cavity, oropharynx, external lip, and major salivary glands combined has been improving over the last few decades from 58.2% in 1975–​1979 to 67.0% in 2008–​2012 (SEER 2016). Five-​year relative survival for oral and oropharyngeal cancer based on SEER data for cases diagnosed in 2006–​2012 was 66.7%,

546

Table 29–​3.  Incidence of Oral Cavity, Oropharyngeal, Lip, and Salivary Gland Cancers (Age-​Adjusted to 2000 US Standard Population) by Anatomical Site, Race, Hispanic Origin, and Sex (2009–​2013) 18 SEER Regions Incidence Per 100,000 Persons Total Oral Cavity, Oropharynx, Lip, and Salivary Gland Race White Black American Indian/​Alaskan Native Asian/​Pacific Islander Hispanic Origin Hispanic Anatomic Subgroups Oral Cavity Mucosa of lip Anterior tongue Floor of mouth Gum and other mouth Race White Black American Indian/​Alaskan Native Asian/​Pacific Islander Hispanic Origin Hispanic Oropharynx Base of tongue Soft palate and uvula Tonsil Other oropharynx Race White Black American Indian/​Alaskan Native Asian/​Pacific Islander Hispanic Origin Hispanic External Lip Race White Black American Indian/​Alaskan Native Asian/​Pacific Islander Hispanic Origin Hispanic Major Salivary Glands Race White Black American Indian/​Alaskan Native Asian/​Pacific Islander Hispanic Origin Hispanic

Males

Females

M:F Ratio

9.6

14.3

5.5

2.6

10.3 7.3 4.8 4.9

15.4 11.2 6.8 6.7

5.9 4.3 3.0 3.5

2.6 2.6 2.3 1.9

5.7

8.3

3.6

2.3

2.9 0.1 1.0 0.5 1.2

3.6 0.2 1.2 0.8 1.4

2.2 0.1 0.8 0.3 1.0

1.6 2.0 1.5 2.7 1.4

3.0 2.0 1.5 2.0

3.8 2.7 1.6 2.7

2.4 1.5 1.3 1.5

1.6 1.8 1.2 1.8

1.8 4.9 2.3 0.3 1.9 0.4

2.2 8.1 3.8 0.4 3.3 0.7

1.5 2.1 1.0 0.2 0.7 0.2

1.5 3.9 3.8 2.0 4.7 3.5

5.3 4.3 2.5 1.9

8.8 7.4 4.3 2.8

2.2 1.9 1.1 1.1

4.0 3.9 3.9 2.5

2.7 0.5

4.5 0.9

1.2 0.2

3.8 4.5

0.6 0.1 0.2 0.1

1.1 0.1 0.4 0.1

0.3 0.0 0.0 0.0

3.6

0.3 1.3

0.5 1.7

0.2 1.0

2.4 1.7

1.3 1.0 0.6 1.0

1.8 1.0 0.5 1.1

1.0 1.0 0.6 0.8

1.8 1.1 0.8 1.4

0.9

1.1

0.8

1.4

Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Incidence—​SEER 18 Regs Research Data + Hurricane Katrina Impacted Louisiana Cases, Nov 2015 Sub (2000–​2013) Katrina/​Rita Population Adjustment—​Linked to County Attributes—​Total U.S., 1969–​2014 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2016, based on the November 2015 submission.

 547

Oral Cavity, Oropharyngeal, External Lip, and Major Salivary Glands Cancers 16 14

Rate per 100,000

12 10 8

+++

6

+

+

+ +

+

+

+

+

+

+++

+

+

+++++

4 2 0 1975

1980

1985

1990 1995 2000 Year of Diagnosis

Oral Cavity Cancer

2005

2013

Oropharyngeal Cancer 9

7

8

6

7 Rate per 100,000

Rate per 100,000

5 4 3 2

6 5 4 3 2

1

1

0 1975 1980 1985 1990 1995 2000 2005 Year of Diagnosis

0 1975 1980 1985 1990 1995 2000 2005 Year of Diagnosis

2013

External Lip Cancer

2013

Major Salivary Gland Cancer

3 1.8 1.6

2.5

Rate per 100,000

Rate per 100,000

1.4 2 1.5 1

1.2 1 0.8 0.6 0.4

0.5

0.2 0 1975 1980 1985 1990 1995 2000 2005 Year of Diagnosis

2013

White

Black

0 1975 1980 1985 1990 1995 2000 2005 Year of Diagnosis API

AI/AN

2013

Hispanic

Figure 29–​1.  SEER incidence, both sexes, joinpoint analyses for whites and blacks (1975–​2013) and for Hispanics (1992–​2013).

548

548

Part IV:  Cancers by Tissue of Origin

(a)

(b)

Oral Cavity Cancer

Oropharyngeal Cancer

3

2 1.8 White Black API AI/AN Hispanic

Rate 100,000

1.4 1.2 1 0.8

Rate per 100,000

2.5

1.6

2

1.5

1

0.6 0.4

0.5

0.2 0 1975

1980

1985

(c)

1990 2000 1995 Year of Diagnosis

2005

0 1975

2013

1980

1985

1990

1995

2000

2005

2013

2005

2013

Year of Diagnosis

(d)

External Lip Cancer

0.11

Major Salivary Gland Cancer 0.5

0.45 0.4

0.09

Rate per 100,000

Rate per 100,000

0.35 0.07

0.05

0.3 0.25 0.2 0.15

0.03

0.1 0.5

0.01 1975 –0.01

1980

1985

1990 2000 1995 Year of Diagnosis

2005

2013

0 1975

1980

1985

1990

1995

2000

Year of Diagnosis

Figure 29–​2.  US mortality, both sexes joinpoint analyses for whites and blacks (1975–​2013) and for Asian/​Pacific Islanders, American Indians/​Alaska Natives and Hispanics (1992–​2013). (a) Oral cavity cancers, (b) Oropharyngeal cancer, (c) External lip cancer, (d) Major salivary gland cancer.

with marked differences by race (Table 29–​4). Relative survival among white males was 67.6% compared to 46.6% for black males, and the rate for white females was also much higher (68.9%) compared to 57.3% for black females. Persons diagnosed with localized disease had survival rates of 85.8%, higher than for regional 62.3%, and distant 36.6%. White females were most likely to be diagnosed at a localized stage (54%) and black males were least likely to be diagnosed at a localized stage (22%). However, at every stage, blacks had a poorer 5-​year survival than whites. Reasons for the difference in oral cavity cancer survival between blacks and whites may include differences in socioeconomic status and treatment (Arbes et al., 1999). Predictors for better prognosis include more recent year of diagnosis, localized stage, being female, white ethnicity, and younger age at diagnosis (Howlader et al., 2016).

Second Primary Cancers Among 39,501 oral cavity and pharyngeal cancer patients in the SEER data from 1973 to 2000, 7,076 (17.9%) were subsequently diagnosed with a second primary cancer (Curtis et  al., 2006). The relative risk of developing a second primary cancer was higher among individuals diagnosed with their first cancer before age 50 years, although the absolute risks were higher among individuals diagnosed at ages 50–​ 69. In a pooled analysis of 99,257 head and neck cancer patients drawn from cancer registries from Europe, Canada, Australia, and Singapore,

10,826 (10.9%) developed a second primary cancer (Chuang et  al., 2008). The 20-​year cumulative risk of second primary cancers after mouth and tongue cancer was 33%.

Etiologic Factors Although cancers of the oral cavity, oropharynx, and lip share some well-​established risk factors, there are also important etiologic differences. Tobacco use is causally related to cancers of the oral cavity, lip, and oropharynx; alcohol consumption to cancers of the oral cavity and oropharynx; HPV to oropharyngeal cancer; and sun exposure to lip cancer. Temporal trends in the incidence of these cancers reflect changes in the distribution of the underlying exposures.

Primary, Secondary, and Tertiary Prevention In addition to evidence that oral cavity and oropharyngeal cancer risks are lowest in non-​smokers and non-​drinkers of alcohol (Hashibe et al., 2007), the relative risk of these cancers decrease with time since cessation, when compared to the risk from continued use (Marron et al., 2010). Given that HPV vaccination strategies in the United States since 2010 are directed at adolescent boys and girls and the median age of presentation of HPV-​associated oropharyngeal cancers is 55 years of age, there is no evidence as yet of the impact of HPV vaccine uptake on trends in oropharyngeal cancer. The US Preventive Services Task

 549

(a)

25

20

20

Rate per 100,000

Rate per 100,000

25

15 10 5 0 1975

1980

1985 1990 1995 2000 Year of Diagnosis/Death

2005

Rate per 100,000

20 15 10 5

1980

10 5

1980

1985 1990 1995 2000 Year of Diagnosis/Death

1985 1990 1995 2000 Year of Diagnosis/Death

2005

2013 Incidence

2005

2013

2005

2013

Black Female

(d) 25

25

0 1975

15

0 1975

2013

Black Male

(c)

Rate per 100,000

White Female

(b)

White Male

20 15 10 5 0 1975

1980

1985

1990

1995

2000

Year of Diagnosis/Death Mortality

Figure 29–​3 a–​d.  Incidence and death rates (1975–​2013) for cancer of the oral cavity, oropharynx, external lip, major salivary glands. (a) White male; (b) White female; (c) Black male; (d) Black female. (a)

D.C

Rate per 100,000 Oral and Oropharyngeal Female 0.810–1.040 1.041–1.149 1.150–1.215 1.216–1.241 1.242–1.329 1.330–1.655 Suppressed

Underlying mortality data provided by NCHS (www.cdc.gov/nchs). Rates are per 100,000 and age-adjusted to the 2000 US Std Population (19 age groups - Census P25-1130 standard; Statistics are Suppressed due to fewer than 16 cases.

Figure 29–​4 a.  Age-​adjusted death rates by state (2009–​2013) for all races: oral, oropharyngeal, lip, and salivary gland cancer, female.

50

550

Part IV:  Cancers by Tissue of Origin (b)

D.C

Rate per 100,000 Oral and Oropharyngeal Male 1.730–2.751 2.752–3.062 3.063–3.194 3.195–3.527 3.528–3.949 3.950–5.770 Suppressed

Underlying mortality data provided by NCHS (www.cdc.gov/nchs). Rates are per 100,000 and age-adjusted to the 2000 US Std Population (19 age groups - Census P25-1130 standard; Statistics are Suppressed due to fewer than 16 cases.

Figure 29–​4 b.  Age-​adjusted death rates by state (2009–​2013) for all races: oral, oropharyngeal, lip, and salivery gland cancer, male.

Force (USPTF) rates evidence for oral cancer screening in the United States to be “insufficient to assess the balance of benefits and harms” (USPTF, 2015). Screening studies in India, where rates of oral cancer are high, have shown that oral cancer screening is effective in reducing oral cancer mortality when focused on individuals who smoke or chew tobacco and/​or drink alcohol (Sankaranarayanan et al., 2013). Treatment of these cancers, which may involve surgery, radiation, and chemotherapy, and the disease itself can lead to long-​term disability and symptoms. Newer treatments for oral and oropharyngeal cancer, including improvements in surgical and radiotherapy techniques, may partially account for improving survival rates over time (Chinn and Myers, 2015; Grégoire et al., 2015). Additionally, the dramatically better survival for HPV-​associated oropharyngeal cancer as compared to traditional smoking-​related oropharyngeal cancer also accounts for better survival rates over time for patients with oropharyngeal cancer. Nevertheless, especially in developing countries, a high proportion of oral and oropharyngeal cancer cases are diagnosed at an advanced stage. Among those surviving oral and oropharyngeal cancer, risks of recurrences and second primary cancers are common, though both of these risks appear much lower for patients with HPV-​associated oropharyngeal cancers.

ORAL CAVITY CANCER Disease Burden The number of new cases and deaths from oral cavity cancer (including the base of the tongue) is increasing in the United States. In 2012, an estimated 19,600 (13,000 men, 6600 women) were diagnosed and 3600 died (2300 men, 1300 women) (USCS, 2016). By 2016 the estimate had increased to 32,000 new cases and 6500 deaths (ACS, 2016); these counts include the base of the tongue and soft palate, and uvula

(which belong in oropharyngeal cancers), and do not include the inner lip or lip commissure. Worldwide in 2012, approximately 300,000 cases and 145,000 deaths from oral cavity cancer were estimated to have occurred (Ferlay et al., 2013). Of the 300,000 oral cavity cancer cases, 199,000 cases were diagnosed among men and 101,000 cases were diagnosed among women. Of the 145,000 deaths due to oral cavity cancer, 98,000 deaths were among men and 47,000 deaths were among women. Patients with oral cavity or oropharyngeal cancer experience a wide range of health concerns that affect their quality of life. These involve difficulty swallowing, oral pain, skin changes, dry mouth, dental health, trismus or open mouth, altered taste, excess or thick mucous, shoulder disability, hoarseness, social issues, and other problems (Chera et al., 2014).

Tumor Classification The ICD-​O3 codes for the oral cavity include those shown in Tables 29-​1 and 29-​2 (Fritz et  al., 2012):  C00.3 (mucosa of the upper lip), C00.4 (mucosa of the lower lip), C00.5 (mucosa of the lip, NOS), C00.6 (commissure of lip), C00.8 (overlapping lesion of lip), C00.9 (lip, NOS), C02.0 (dorsal surface of tongue, NOS), C02.1 (border of tongue), C02.2 (ventral surface of tongue, NOS), C02.3 (anterior two-​ thirds of tongue, NOS), C03.0 (upper gum), C03.1 (lower gum), C03.9 (gum, NOS), C04.0 (anterior floor of mouth), C04.1 (lateral floor of mouth), C04.8 (overlapping lesion of floor of mouth), C04.9 (floor of mouth, NOS), C05.0 (hard palate), C06.0 (cheek mucosa), C06.1 (vestibule of mouth), C06.2 (retromolar area), C06.8 (overlapping lesion of other and unspecified parts of mouth), and C06.9 (mouth, NOS). Based on data from cases diagnosed between 2009 and 2013 in US cancer registries, cancer of the anterior tongue accounts for about one-​third of oral cavity cancers (33%), with cancer of the floor of the

 51



551

Oral Cavity, Oropharynx, Lip, and Salivary Glands Oral Cavity, Lip, Salivary Gland ASR (W) per 100,000, all ages Male

Female

Melanesia

Incidence Mortality

South-Central Asia Australia/New Zealand Western Europe Central and Eastern Europe North America More developed regions Northern Europe Southern Africa World Southern Europe South America Less developed regions Eastern Africa Caribbean South-Eastern Asia Polynesia Middle Africa Micronesia Northern Africa Western Asia Central America Eastern Asia Western Africa 30

20

10

0

10

20

30

Figure 29–​5.  Oral cavity, lip, and salivary gland age-​adjusted cancer incidence and mortality rates in the world (standardized to the world population). GLOBOCAN 2012.

mouth accounting for 21% and mucosa of the lip 5%. The remaining 42% includes tumors of the gum, palate, and other mouth parts (Table 29-​5). Approximately 88.2% of oral cavity cancers are squamous cell carcinomas, with the remainder being other epidermoid cancers (2.9%), adenocarcinomas (6.7%), and other histologic types (2.2%) (SEER, 2016). As mentioned earlier, the available data do not consistently separate the subsites of head and neck cancer according to etiologically relevant categories. In particular, oral cavity is often grouped with oropharynx, and hypopharynx with nasopharynx. Many cases are categorized under generic terms such as “tongue” or “palate,” which include both oral cavity (oral tongue and hard palate) and oropharynx (base of tongue and soft palate). We encourage precise site documentation by clinicians.

Precancerous or Precursor Lesions Oral cavity cancers are sometimes preceded by “oral potentially malignant disorders” (OPMD), the most established of which include oral leukoplakia, erythroplakia, and oral submucous fibrosis

(Warnakulasuriya et  al., 2007). OPMDs have also been called “precancer, precursor lesions, premalignant lesions, intraepitheial neoplasia and potentially malignant.” Other OPMDs include proliferative verrucous leukoplakia, erythroleukoplakia, palatal lesions in reverse smokers, actinic keratosis, lichen planus, discoid lupus erythematosus, dyskeratosis congenita, and epidermoloysis bullosa. The nomenclature and classification were reviewed in a WHO Collaborating Centre workshop in 2005, and the term OPMD was proposed at this meeting (Warnakulasuriya et al., 2007). Oral leukoplakia is a clinical term for “white plaques of questionable risk having excluded (other) known diseases or disorders that carry no increased risk for cancer” (Warnakulasuriya et  al., 2007). Analogously, oral erythroplakia is “a fiery red patch that cannot be characterized clinically or pathologically as any other definable disease” (Warnakulasuriya et al., 2007). Plaques with both white and red areas are called erythroleukoplakia (Warnakulasuriya et  al., 2007). Oral submucous fibrosis is a “chronic disorder characterized by fibrosis of the lining mucosa of the upper digestive tract involving the oral cavity, oropharynx and frequently the upper third of the esophagus” (Warnakulasuriya et al., 2007).

52

552

Part IV:  Cancers by Tissue of Origin Oropharynx and hypopharynx ASR (W) per 100,000, all ages Male

Female Incidence Mortality

Western Europe South-Central Asia Central and Eastern Europe More developed regions South Africa North America Caribbean Northern Europe World Australia/New Zealand Southern Europe Melanesia Polynesia Less developed regions South America South-Eastern Asia Middle Africa Eastern Africa Eastern Asia Northern Africa Central America Western Asia Western Africa Micronesia 10

5

0

5

10

Figure 29–​6.  Oropharyngeal and hypopharyngeal age-​adjusted cancer incidence and mortality rates in the world (standardized to the world population). GLOBOCAN 2012.

OPMDs have a higher rate of malignant transformation than normal oral mucosa and are therefore considered precancerous. In a study of all OPMDs combined, the malignant transformation rate was estimated to be 2.6% annually, based on 1,357 OPMDs diagnosed in South East England between 1990 and 1999 (Warnakulasuriya et al., 2011). Erythroplakia is thought to have the highest malignant potential. The probability of malignant transformation is much higher in those leukoplakic lesions characterized clinically with a red, speckled, verrucous, or nodular component (non-​homogenous leukoplakia) and in those with a histopathologic diagnosis of epithelial dysplasia (Mayne et al., 2006). A pooled analysis across six studies estimated that the malignant transformation rate for oral leukoplakia was 1.36% (05% CI: 0.69, 2.03) (Petti, 2003). A recent review of 24 oral leukoplakia studies reported that the mean malignant transformation rate was 3.5% (Warnakulasuriya and Ariyawardana, 2016). Estimates of the malignant transformation rate for oral submucous fibrosis are limited;

however, one Indian study reported a rate of 7.6% over a median observation period of 10 years (Murti et al., 1985). In a multicenter study in the Chinese population including 424 oral cavity cancer cases and 106 oropharyngeal cancer cases, increased risks of these cancers were observed for individuals with a history of repetitive dental ulcers (OR = 4.92; 95% CI: 2.88, 8,42), oral leukoplakia (OR = 3.63; 95% CI: 2.06, 6.37) and oral submucous fibrosis (OR = 32.70; 95% CI: 9.63, 111.11) (Li et al., 2015). Increased risks of hypopharyngeal and laryngeal cancers were not observed for these precancers, suggesting that it is specific for oral cavity cancers. The prevalence of oral premalignant lesions and conditions varies by geographic region, population exposure patterns, and the case definition employed (Mayne et  al., 2006). The reported prevalence of oral leukoplakia in adult populations is generally within the range of less than 1%–​5%, although substantially higher estimates have been reported for populations engaging in high-​risk behaviors.

 53

Table 29–​4.  Relative Survival (Percent) for Oral Cavity, Oropharyngeal, External Lip, and Major Salivary Gland Cancers Diagnosed 2006–​2012, SEER Regions, by Stage, Sex, and Race By Stage White Male

Site Oral and Oropharynx Oral Cavity Oropharynx External Lip Major Salivary Gland

All Stages 5-​Year Survival 66.7 62.0 65.4 90.3 72.5

Localized

Regional

Distatn

85.8 82.2 75.7 93.5 93.0

62.3 42.2 68.7 51.9 68.1

36.3 25.8 39.6 27.5 35.6

All Stages 5-​ Unstaged Year Survival 54.5 54.5 44.8 84.6 49.6

67.6 60.6 68.6 91.1 62.7

White Female

Black Male

Black Male

Black Female

Black Female

All Stages 5-​Year Survival

Percent Localized

All Stages 5-​Year Survival

Percent Localized

All Stages 5-​Year Survival

Percent Localized

68.9 65.6 63.1 89.1 82.6

54% 62% 36% 86% 55%

22% 28% 9% 76% 41%

57.3 55.2 45.5 100.0 82.7

39% 42% 15% 83% 55%

White Male White Female Percent Localized 37% 56% 15% 89% 36%

46.6 42.4 44.6 67.2 67.5

Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Incidence—​SEER 18 Regs Research Data + Hurricane Katrina Impacted Louisiana Cases, Nov 2015 Sub (1973–​2013 varying)—​Linked to County Attributes—​Total U.S., 1969–​2014 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2016, based on the November 2015 submission.

54

Table 29–​5.  Annual Percentage Change (APC) in First Primary Oral Cavity, Oropharyngeal, Lip, and Salivary Gland Cancer Incidence Rates in Adults (2000–2013) Major Grouping and Codes

APC

Lower CI

Upper CI

Lip cancers C00.0: External upper lip C00.1: External lower lip C00.2: External lip, NOS C00.8: Overlapping lesion of lip C00.9: Lip, NOS Oral cavity cancers C00.3: Mucosa of upper lip C00.4: Mucosa of lower lip C00.5: Mucosa of lip, NOS C00.6: Commissure of lip C02.0: Dorsal surface of tongue, NOS C02.1: Border of tongue C02.2: Ventral surface of tongue, NOS C02.3: Anterior 2/​3 of tongue, NOS C03.0: Upper gum C03.1: Lower gum C03.9: Gum, NOS C04.0: Anterior floor of mouth C04.1: Lateral floor of mouth C04.8: Overlapping lesion of floor of mouth C04.9: Floor of mouth, NOS C05.0: Hard palate C06.0: Cheek mucosa C06.1: Vestibule of mouth C06.2: Retromolar area C06.8: Overlapping lesion of other and unspecified mouth C06.9: Mouth, NOS Oropharyngeal cancers C01.9: Base of tongue, NOS C02.4: Lingual tonsil C02.8: Overlapping lesion of tongue C02.9: Tongue, NOS C05.1: Soft palate, NOS C05.2: Uvula C05.8: Overlapping lesion of palate C05.9: Palate, NOS C09.0: Tonsillar fossa C09.1: Tonsillar pillar C09.8: Overlapping lesion of tonsil C09.9: Tonsil, NOS C10.0: Vallecula C10.2: Lateral wall of oropharynx C10.3: Posterior wall of oropharynx C10.4: Branchial cleft C10.8: Overlapping lesion of oropharynx C10.9: Oropharynx, NOS Salivary gland C07.9: Parotid gland C08.0: Submandibular gland C08.1: Sublingual gland C08.8: Overlapping lesion of major salivary glands C08.9: Major salivary gland, NOS

−​3.4 −​2.6* −​4.4* 2.3 −​8.7* −​4.4* −​0.9* −​3.3* −​3.1* ~ −​2.2 1.4* −​3.3* 1.3 2.5* 0.6 −​0.3 −​3 −​4.0* −​7.2* −​7.3* −​3.4* −​0.3 −​1 0.4 −​4.4* −​0.1 0.5 1.8* 2.4* −​3.8* −​2.3 0.6 −​4.5* −​7.2* −​2.6 0 −​4.2* −​0.5 −​0.9 4.4* −​0.8 −​1.9 2.2 −​3.7 −​0.4 4.1* −​0.6* −​0.4 −​0.4 2.2 ~ 3.7*

−​4.2 −​4.6 −​5.5 −​2.3 −​13.4 −​7.4 −​1.5 −​6.2 −​4.8 ~ −​6.3 0.1 −​4.2 −​0.1 1.1 −​0.9 −​1.4 −​6.1 −​5.1 −​9.5 −​10.3 −​4.2 −​1.3 −​2.3 −​4 −​5.5 −​4.5 −​1.3 1.4 1.7 −​7.3 −​4.7 −​0.8 −​5.7 −​9.9 −​5.5 −​3.1 −​5.3 −​1.8 −​4 3.7 −​4.2 −​5.6 −​0.2 −​11.1 −​2.7 3.1 −​1.1 −​0.9 −​2 −​3.2 ~ 1.2

−​2.6 −​0.5 −​3.3 7 −​3.7 −​1.2 −​0.4 −​0.2 −​1.4 ~ 2.1 2.7 −​2.3 2.6 3.9 2.1 0.8 0.2 −​2.8 −​4.7 −​4.2 −​2.6 0.7 0.3 5 −​3.4 4.4 2.3 2.2 3.1 −​0.1 0.2 2.1 −​3.3 −​4.5 0.4 3.3 −​3 0.8 2.2 5.1 2.7 1.8 4.6 4.3 2 5.2 −​0.2 0.1 1.3 7.9 ~ 6.2

Subsite within Major Grouping

Gum and other mouth Gum and other mouth Gum and other mouth Gum and other mouth Anterior tongue Anterior tongue Anterior tongue Anterior tongue Gum and other mouth Gum and other mouth Gum and other mouth Floor of mouth Floor of mouth Floor of mouth Floor of mouth Gum and other mouth Gum and other mouth Gum and other mouth Gum and other mouth Gum and other mouth Gum and other mouth Base of tongue Base of tongue Base of tongue Base of tongue Other oropharynx Other oropharynx Other oropharynx Other oropharynx Tonsil Tonsil Tonsil Tonsil Other oropharynx Other oropharynx Other oropharynx Other oropharynx Other oropharynx Other oropharynx

n

% Distribution

93,934 5646 718 4306 145 56 421 26,578 183 997 58 116 677 3263 1846 2910 1013 1996 348 1769 466 166 3129 1813 2428 202 1904 185 1109 48,042 15,474 353 747 4561 2056 309 340 436 3040 1302 220 15,627 468 258 320 73 410 2048 13,668 10,982 1943 131 12 600

6.0% 0.8% 4.6% 0.2% 0.1% 0.4% 28.3% 0.2% 1.1% 0.1% 0.1% 0.7% 3.5% 2.0% 3.1% 1.1% 2.1% 0.4% 1.9% 0.5% 0.2% 3.3% 1.9% 2.6% 0.2% 2.0% 0.2% 1.2% 51.1% 16.5% 0.4% 0.8% 4.9% 2.2% 0.3% 0.4% 0.5% 3.2% 1.4% 0.2% 16.6% 0.5% 0.3% 0.3% 0.1% 0.4% 2.2% 14.6% 11.7% 2.1% 0.1% 0.0% 0.6%

Rates are per 100,000 and age-​adjusted to the 2000 US Std Population (single ages to 84—​Census P25-​1130) standard; Confidence intervals are 95% for rates (Tiwari mod) and trends. APCs were calculated using weighted least squares method. ~ Statistic could not be calculated. *The APC is significantly different from zero (p 0–​3 cigarettes per day, 2.23 (95% CI: 1.45, 3.42) for > 3–​5 cigarettes per day and 2.18 (95% CI: 1.68, 2.83) for > 5–​10 cigarettes per day (1327 oral cavity cancer cases and 13,416 controls from 23 case-​control studies) (Berthiller et al., 2015). Cessation of tobacco smoking rapidly lowers the risk of oral cavity cancer, with a 35% reduction > 1–​4 years after quitting (OR = 0.64; 95% CI: 0.52, 0.80) (Marron et al., 2010). This analysis included 3302 oral cavity cancer cases and 16,337 controls from 16 case-​control studies. Longer years since quitting tobacco smoking resulted in lower risk of oral cavity cancer (p for trend < 0.001) and quitting for 20 or more years resulted in ORs similar to those of never smokers. Similar findings of lower risks in the first 4 years after quitting and decrease with time since quitting have also been observed in the NIH-​AARP cohort (Freedman et al., 2007a). The proportion of oral cavity cases that tobacco is responsible for ranges from 49% to 65%. In the INHANCE pooled data, the population attributable risk for tobacco was 64.7% in an analysis including 2993 oral cavity cancer cases and 16,152 controls from 18 case-​control studies (Hashibe et al., 2009). For head and neck cancer, the population attributable risk was lower for tobacco in North America compared to South America and Europe. In a cohort study including 177 head and neck cancer cases from the United States, the attributable fraction estimate for cigarette smoking was estimated to be 48.8% (Hashibe et al., 2013). In a Western European study including 485 oral cavity cancer cases and 1993 controls, 61% of oral cavity cancer cases were attributable to tobacco smoking (Anantharaman et al., 2011). Oral cavity, pharynx, and larynx cancer survivors who continue to smoke have poorer survival rates than those survivors who quit (Mayne et al., 2009). A tobacco–​alcohol interaction on the multiplicative scale has been reported for oral cavity cancer in a pooled analysis including 2992 cases and 16,152 controls, with a multiplicative interaction parameter of 3.09 (95% CI:  1.82, 5.23) (Hashibe et  al., 2009). A  multiplicative interaction parameter greater than 1 with a confidence interval excluding the null value indicates an interaction on the multiplicative scale. The highest risk was observed for individuals who smoked > 20 cigarettes per day and drank 3 or more alcoholic drinks per day (OR = 15.49; 95% CI: 7.24, 33.14). The attributable fraction for tobacco and/​or alcohol in the US population according to a cohort study was 66.6% (Hashibe et al., 2013). These studies of cigarette smoking confirm previous findings (Mayne, 2006) and extend earlier studies by providing more precise estimates of risks because of larger sample sizes, identifying duration of use, identifying smokers with even low frequency of use at increased risk of oral cancer, demonstrating that quitting has immediate benefits, and replicating findings that alcohol and cigarette smoking have synergistic effects, and that cigarette smoking continues to account for a substantial proportion of oral cavity cancer in the United States and elsewhere.

Oropharyngeal Cancer

Tobacco smoking has been established as a human carcinogen for pharyngeal cancers (including both oropharyngeal and hypopharyngeal cancers) by the IARC monographs in the 1986, 2004, and 2012 expert review panels (IARC, 1998, 2012a, 2012b). Cigarette smoking confers a risk of 3.01 for oropharyngeal cancer (95% CI: 2.71, 3.35) according to a pooled analysis including 3,834 cases and 19,114 controls (Wyss et al., 2013).

 563



Oral Cavity, Oropharynx, Lip, and Salivary Glands

563

Cessation of tobacco smoking is associated with a reduction in the relative risk of pharyngeal cancer within 1–​4 years after quitting (Marron et al., 2010). A pooled analysis of 3989 pharyngeal cancer cases and 16,337 controls from 17 case-​control studies reported an OR of 0.72 (95% CI:  0.51–​1.00) for individuals who had quit for > 1–​4  years compared to current smokers. Dose–​response relations were observed, with longer years since quitting tobacco associated with greater reductions in risk (p < 0.01). There was a 70% reduction in the OR after quitting for 20 or more years (OR = 0.28; 95% CI: 0.19, 0.43). The ratio of oropharyngeal to hypopharyngeal cancer cases was approximately 4:1 in this pooled analysis. At low frequencies of cigarette smoking, the oropharyngeal cancer risks conferred were 1.57 (95% CI: 1.10, 2.23) for > 0–​3 cigarettes per day, 2.17 (95% CI: 1.53, 3.06) for > 3–​5 cigarettes per day, and 2.85 (95% CI: 1.89, 4.08) for > 5–​10 cigarettes per day (1179 oropharyngeal cancer cases and 13,416 controls from 16 case-​control studies) (Berthiller et al., 2015). A tobacco–​alcohol interaction on the multiplicative scale has been reported for pharyngeal cancer in a pooled analysis including 4038 cases and 16,152 controls, with a multiplicative interaction parameter of 1.90 (95% CI: 1.41, 2.56) (Hashibe et al., 2009). The highest risk was observed for individuals who smoked > 20 cigarettes per day and drank 3 or more alcoholic drinks per day (OR = 14.29; 95% CI: 7.26, 28.15). Together, tobacco and alcohol accounted for approximately 71.5% of pharyngeal cancer cases (Hashibe et al., 2009).

oropharyngeal cancer cases, the odds ratios were 2.31 (95% CI: 1.54, 3.45) for cigar smoking and 1.65 (95% CI: 1.04, 2.60) for pipe smoking (Wyss et al., 2013). Cigar and pipe smoking are established risk factors for both oral cavity and oropharyngeal cancer.

Bidis

Oropharyngeal Cancer

Oral Cavity Cancer

Bidis are local tobacco products, mainly smoked in South Asia, and are composed of coarse and uncured tobacco and are often smoked without filters (IARC, 2012a). The IARC monographs reviewed 10 case-​ control studies that investigated the association between bidi smoking and the risk of oral cavity cancer in India (IARC, 2012a). Most of the studies focused on men since very few women were bidi smokers. The studies were consistent in showing an increased risk of oral cavity cancer for bidi smoking. Dose–​response relations for frequency and duration of bidi smoking were observed in two of the studies. Although some studies have been published on bidi smoking and oral cavity cancers since the monograph review, they did not separate out cigarettes and bidis or did not adjust on cigarette smoking. However, bidi smoking is an established risk factor for oral cavity cancers.

Oropharyngeal Cancer

In the IARC monograph review, four case-​control studies investigated the association between bidi smoking and the risk of oropharyngeal cancers (IARC, 2012a). Increased risks of oropharyngeal cancer were observed in all four studies due to bidi smoking, and dose–​response relations were observed in two of the studies.

Cigars and Pipes Assessment of cigar and pipe smoking as a risk factor for head and neck cancers requires careful consideration of the potential confounding effect of cigarette smoking. Since some of the pipe and cigar smokers may also be cigarette smokers, the ideal individuals to study are never cigarette smokers. This approach was taken by the INHANCE investigators with their pooled data. In a pooled analysis of 4110 oral cavity cancer cases who were never cigarette smokers, the odds ratios were 2.83 (95% CI:  1.91, 4.17) for cigar smoking and 2.51 (95% CI:  1.68, 3.75) for pipe smoking (Wyss et  al., 2013). Although the dose–​response relationships were estimated for head and neck cancer overall (including cancers of the oral cavity, oropharynx, hypopharynx and larynx), they were strong for cigars per day (p < 0.0001) and for years of cigar smoking (p for trend < 0.0001) (Wyss et  al., 2013)  among the never cigarette smokers. Similarly, dose–​response relationships were observed for pipe smoking frequency (pipes per day; p for trend = 0.0001) and pipe duration (years; p for trend < 0.0001) for head and neck cancer risk among never cigarette smokers. In a pooled analysis of approximately 384 never cigarette smoking

Smokeless Tobacco Oral Cavity Cancer

The IARC monograph concluded that there is sufficient evidence to support a causal relationship between the use of smokeless tobacco and oral cavity cancer (IARC, 2007, 2012a). Of the six cohort studies that investigated smokeless tobacco and oral cavity/​pharyngeal cancer risk, four did not show clear associations. A cohort in the United States reported an association but did not adjust for tobacco smoking. Another cohort in Sweden reported an increased risk of oral cavity cancer due to ever snuff use, with adjustment for tobacco smoking. Approximately 40 case-​control studies have investigated the association between smokeless tobacco and oral cavity cancer and contributed to evidence for an association. A  pooled analysis of 11 case-​control studies including 2034 oral cavity cancer cases and 8375 controls reported increased risks of oral cavity cancer among never cigarette smokers for smokeless tobacco use (OR = 3.01; 95% CI: 1.63, 5.55) (Wyss et al., 2015). Most of the studies conducted on smokeless tobacco combined oral cavity cancers and oropharyngeal cancer estimates. The IARC monograph did not conclude that there was sufficient evidence for smokeless tobacco specifically for pharyngeal cancers, although the evidence for oral cavity cancers was considered sufficient (IARC, 2007, 2012a). In an INHANCE pooled analysis of 481 oropharyngeal cancer cases and 3118 controls who were never cigarette smokers, snuff use was not associated with oropharyngeal cancer risk (OR = 1.07; 95% CI: 0.55, 2.08) (Wyss et al., 2015).

Betel Quid Oral Cavity Cancer

Betel quid, also known as paan, typically contains a mixture of areca nut, catechu, and slaked lime wrapped in a Piper betel leaf (IARC, 2012a). The IARC monographs classified betel quid with and without tobacco as carcinogenic for oral cavity cancers (Group 1) (IARC, 2012a). A  meta-​analysis of 25 studies in India on betel quid with tobacco reported a summary estimate of 8.47 (95% CI: 6.49, 11.05) for oral cavity cancers (Guha et al., 2014). For 13 studies on betel quid without tobacco in India, the oral cavity cancer risks were also elevated according to the summary estimate (RR = 2.41; 95% CI: 1.82, 3.19) (Guha et al., 2014). In Taiwan, betel quid is chewed without tobacco. Thirteen studies were identified for a pooled estimate of 10.98 (95% CI: 4.86, 24.84) for betel quid chewing in Taiwan (Guha et al., 2014).

Oropharyngeal Cancer

While betel quid with tobacco is classified as a Group 1 carcinogen for pharyngeal cancer, betel quid without tobacco is not established as a Group  1 carcinogen for pharyngeal cancer (IARC, 2012a). A  meta-​ analysis of eight oropharyngeal cancer studies in India on betel quid with tobacco reported a summary estimate of 4.36 (95% CI: 2.23, 8.53) (Guha et  al., 2014). For six studies on betel quid without tobacco in India, the oropharyngeal cancer risks were also elevated according to the summary estimate (RR = 2.61; 95% CI: 1.74, 3.92) (Guha et al., 2014).

Alcohol Oral Cavity Cancer

Alcohol is an established risk factor for oral cavity cancer according to the IARC monographs panels in 1988, 1996, and 2012 (IARC, 1988, 2010, 2012a). In a pooled analysis of 383 oral cavity cancer

564

564

Part IV:  Cancers by Tissue of Origin

cases and 5775 controls who were never tobacco users, alcohol drinking increased the risk of oral cavity cancer, with dose–​response trends for both the frequency (drinks per day; p for trend = 0.032) and duration (years) of alcohol drinking (p for trend < 0.001) (Hashibe et al., 2007). Similar dose–​response patterns were also observed for men and women in the NIH-​AARP study (Freedman et al., 2007b). In terms of the type of alcoholic beverage, a pooled analysis of 15 case-​control studies (572 oral cavity cancer cases and 4473 controls) reported that beer-​only drinkers had an increased risk of oral cavity cancer (OR = 2.0; 95% CI: 1.4, 2.8 for ≤ 15 beers/​week; and OR = 6.4; 95% CI: 3.9, 10.3 for > 15 beers/​week) (Purdue et al., 2009). The ORs were adjusted for age, sex, race/​ethnicity, center, education, pack-​ years of smoking, years of cigar smoking, and years of pipe smoking. The corresponding estimates for liquor-​only drinkers were 1.7 (95% CI: 0.9, 3.3) for ≤ 15 liquor drinks/​week and 3.2 (95% CI: 1.6, 6.4) for > 15 liquor drinks/​week. For wine-​only drinkers, the ORs were 1.3 (95% CI: 0.7, 2.2) for ≤ 15 wine glasses/​week and 5.9 (95% CI: 2.3, 15.4) for > 15 wine glasses/​week. Although previous studies had suggested that the predominant alcoholic beverage type confers the greatest risk of head and neck cancers in that region, this pooled analysis reported fairly similar levels of risk for the alcoholic beverage types. The analysis was also stratified by geographic regions (North America, South America, Europe), but differences in head and neck cancer risk conferred by alcoholic beverage types were not observed. Dose–​ response relationships with the frequency of these alcoholic beverage types were observed with head and neck cancer risk (p for trend < 0.0001 for wine, beer, and liquor). A multicenter study in Western Europe including 489 oral cavity cancer cases reported odds ratios for men of 2.59 (95% CI: 0.94, 7.17) for drinking only wine, 3.58 (95% CI: 1.33, 9.61) for drinking only beer, and 1.64 (95% CI: 0.34, 7.86) for drinking only liquor (Marron et al., 2012). Predominant drinkers of liquor had an increased risk of oral cavity cancer (OR  =  2.97; 95% CI:  1.03, 8.58). The ORs were adjusted for center, age, education level, smoking duration, smoking status by frequency and years since quitting, fruit and vegetable intake, as well as cumulative alcohol consumption, in an effort to estimate the role of the alcoholic beverage types on oral cavity cancer risk. For women, the corresponding ORs were 0.65 (95% CI:  0.30, 1.39) for drinking only wine, 0.85 (95% CI: 00.28, 2.59) for drinking only beer, and 0.77 (95% CI: 0.22, 2.70) for drinking only liquor. In the NIH-​AARP cohort, drinking more than three beers conferred an elevated risk of oral cavity cancer among men and women, and more than three liquor drinks conferred an elevated risk among women, adjusting for other types of alcohol (Freedman et al., 2007b). Cessation of alcohol drinking was associated with a reduced risk of oral cavity cancer after 10–​19 years since cessation (OR = 0.66; 95% CI: 0.47, 0.92, p for trend = 0.05) (Marron et al., 2010). This analysis included 2478 oral cavity cases and 12,033 controls from 13 case-​ control studies. The reduced risk of oral cavity cancer was similar for individuals who had quit drinking for 10–​19 years and those who had never had alcohol. Studies continue to support the independent effect of alcohol drinking on risk of oral cavity cancer, with strong dose–​response relationships measured by frequency or duration. Cessation reduces risks, with risks returning to those of non-​users after 10 years. All types of alcoholic beverages confer an excess risk. Survivors who continue alcohol drinking after oral cavity and pharyngeal cancer diagnosis have poorer survival rates than those survivors who quit (Mayne et al., 2009).

Oropharyngeal Cancer

Alcohol drinking increases the risk of oropharyngeal cancer in a dose-​dependent manner. In a pooled analysis of 369 oropharyngeal and hypopharyngeal cancer cases and 5775 controls who were never tobacco users in the INHANCE consortium, alcohol drinking increased the risk of pharyngeal cancer (ever drinker OR = 1.38; 95% CI: 0.99, 1.94), with dose-​response trends for both the frequency (drinks per day; p for trend < 0.001) and duration (years) of alcohol drinking (p for trend = 0.003) (Hashibe et al., 2007). Individuals who drank five or more alcohol drinks per day had a greater than five-​fold increase in pharyngeal cancer risk (OR = 5.50; 95% CI: 2.26, 13.36). Cessation of

alcohol drinking was not strongly associated with a pharyngeal cancer risk reduction, in contrast to cessation of tobacco smoking (Marron et al., 2010). In terms of alcoholic beverage types, all beverage types are associated with oropharyngeal cancer risk. A  pooled analysis of 15 case-​ control studies (648 pharyngeal cancer cases and 4473 controls) reported an increased risk of pharyngeal cancer for beer-​only drinkers (OR = 2.3; 95% CI: 1.7, 3.1 for ≤ 15 beers/​week; and OR = 4.3; 95% CI: 2.7, 6.8 for > 15 beers/​week), liquor-​only drinkers (OR = 2.0; 95% CI:  0.9, 4.6 for ≤ 15 liquor drinks/​week; and OR  =  3.6, 95% CI: 2.0, 6.3 for > 15 liquor drinks/​week) and for wine-​only drinkers (OR = 1.4; 95% CI: 0.9, 2.2 for ≤ 15 wine glasses/​week; and OR = 4.4; 95% CI: 2.9, 9.6 for > 15 wine glasses/​week) (Purdue et al., 2009). A multicenter study in Europe including 623 pharyngeal cancer cases reported ORs for men of 1.71 (95% CI: 0.70, 4.22) for drinking only wine, 2.45 (95% CI: 1.07, 5.63) for drinking only beer, and 3.16 (95% CI:  1.05, 9.44) for drinking only liquor (Marron et  al., 2012). Total exposure and exposure rate effects for alcohol drinking were assessed in another INHANCE pooled analysis of 3693 pharyngeal cancer cases and 15,589 controls (Lubin et al., 2009).

Human Papillomaviruses Oral Cavity Cancer

There is general consensus that HPV is more important for oropharyngeal cancers than for oral cavity cancers (Hübbers and Akgül, 2015; Combes and Franceschi, 2014). The IARC monographs concluded that there is sufficient evidence for HPV 16 to be a cause of oral cavity cancers, and HPV 18 is also possibly an oral cavity risk factor (IARC, 2012b). However, issues on classification of oral cavity cancers both in the research studies and in the clinical setting contributed to misclassification of sites, such as the base of the tongue, as oral cavity cancers instead of oropharyngeal cancer. In other words, some of the studies reporting associations between HPV and oral cavity cancer risk had included some oropharyngeal cancer subsites into the oral cavity cancer group. Another issue is that the method of HPV testing such as HPV DNA detection by PCR used in previous studies may have contributed to false positives (Mirghani et al., 2015).

Oropharyngeal Cancer

The IARC monographs have concluded that there is sufficient evidence in humans for the carcinogenicity of HPV 16 for oropharyngeal cancer (IARC, 2012b). The first large study on the association of HPV and oral cavity and oropharyngeal cancers reported a prevalence of HPV DNA in 4% of specimens from the oral cavity and 18% of specimens from the oropharynx (Herrero et al., 2003). However, the investigators grouped cancers of the base of tongue and soft palate (both oropharyngeal sites) with cancers of the oral cavity, thus making the ratio of oral cavity cancers to oropharyngeal cancers (> 5:1) in this study unusual. The tumor positivity rate of oral cavity cancers is likely overstated, while that of oropharyngeal cancers is underestimated, although the degree of impact of this misclassification is unknown because the actual numbers of cancers of the base of tongue and soft palate in the study were not provided. When cases were compared with controls, a strongly increased risk was observed for antibodies against HPV 16 E6 and E7 proteins, for cancer of the oropharynx (OR = 9.2; 95% CI: 4.8, 17.7). In a landmark nested case-​ control serologic analyses within a Nordic cohort of almost 900,000 individuals, there was an increased frequency of HPV 16 infection among subjects who later (on average 10 years after collection of the blood sample) developed oropharyngeal cancer or “tongue” cancer (a mixed grouping of base of tongue and oral tongue), but no evidence of an association with oral cavity cancer (Mork et al., 2001). Because of the sequential nature of these data (exposure years before cancer development), this study provided strong evidence on the causal nature of the association of HPV with oropharyngeal cancer. In another landmark study including 100 oropharyngeal cancer cases and 200 controls, risk of oropharyngeal

 56



Oral Cavity, Oropharynx, Lip, and Salivary Glands

cancer was increased with HPV 16 oral infection (OR  =  14.6; 95% CI:  6.3, 36.6) and with oral infection with any of 27 types of HPV (OR = 12.3; 95% CI: 5.4, 26.4) (D’Souza et al., 2007). A meta-​analysis for HPV prevalence including 53 studies reported that HPV DNA was detected in 45.8% (95% CI:  38.9%, 52.9%) of oropharyngeal cancer cases (Ndiaye et al., 2014). Within the oropharynx, the prevalence of HPV was 53.9% for the tonsil, 47.8% for the base of tongue, and 42.6% for palate not otherwise specified. A meta-​ analysis of five studies reported a combined estimate of 4.3 (95% CI: 2.1, 8.9) for HPV 16 and oropharyngeal cancer risk (Hobbs et al., 2006). For tonsil cancer, the combined estimate across eight studies was 15.1 (95% CI: 6.8, 33.7) for HPV 16 (Hobbs et al., 2006).

565

pooled analysis of 225 pharyngeal cancer cases (206 oropharynx and 19 hypopharynx) who had never smoked tobacco reported an increased risk of pharyngeal cancer for exposure to involuntary smoking at home for more than 15 years (OR = 4.13; 95% CI: 1.43, 11.89; p for trend = 0.02) (Lee et al., 2008). When further restricted to never tobacco users and never alcohol drinkers (n = 92 pharyngeal cancer cases), the association persisted for involuntary smoking at home (p for trend < 0.01), and an association was also observed for involuntary smoking for more than 15 years at work (OR = 3.99; 95% CI: 1.06, 15.08), although a dose–​response relationship was not present (p for trend = 0.13).

Body Mass and Related Measures of Obesity Marijuana Smoking Oral Cavity Cancer

Marijuana smoking has been suspected to be a potential risk factor for head and neck cancers, since marijuana shares many carcinogens with tobacco and is inhaled without filters (Huang et  al., 2015). For head and neck cancer, eight case-​control studies and two pooled data analyses have been conducted, with inconsistent results that may be due to differences in risk due to subgroups of head and neck cancer patients by HPV status (Huang et al., 2015). Most of the head and neck cancer studies on marijuana included oral cavity cancer cases, but the estimates were usually reported for combined head and neck cancer sites (Huang et al., 2015). The pooled analysis in the INHANCE consortium on marijuana use included 981 oral cavity cancer cases and 5015 controls (Berthiller et al., 2009). Associations with the risk of oral cavity cancer were not observed for the frequency or duration of marijuana use. When focusing on cancers of the oral tongue (356 cases and 7639 controls), a reduced risk was observed among never tobacco users and never alcohol drinkers (Marks et al., 2014). Three of the studies had stratified on HPV status and observed possible effect modification of the association between marijuana use and oral cavity cancer due to HPV status (Huang et  al., 2015). Further analysis will be necessary before the role of marijuana for oral cavity cancer can be established.

Oropharyngeal Cancer

While many of the studies on marijuana and the risk of head and neck cancer included oropharyngeal cancers, most did not report risk estimates specifically for oropharyngeal cancer. A pooled analysis in the INHANCE consortium on 1921 oropharyngeal cancer cases reported an increased risk of oropharyngeal cancer for marijuana smokers (OR  =  1.24; 95% CI:  1.06, 1.47), with dose–​response relations for both frequency (p for trend < 0.001) and for duration of use (p for trend < 0.001) (Marks et al., 2014). When restricted to never tobacco smokers and never drinkers, the association between marijuana use and the risk of oropharyngeal cancer was borderline, but the dose–​response relationship with cumulative marijuana exposure was persistent (p for trend = 0.037).

Involuntary Smoking Oral Cavity Cancer

A pooled analysis of 146 oral cavity cancer cases and 1939 controls who had never smoked tobacco reported no association with exposure to involuntary smoking at home or at work (Lee et al., 2008). A multicenter study in Western Europe reported an association between involuntary smoking at work and oral cavity/​oropharyngeal cancer risk (OR = 2.46; 95% CI: 1.24, 4.87) (Lee et al., 2009). A dose–​response relationship was observed with the duration of involuntary smoking exposure at work (p for trend = 0.025), with a risk of 1.92 (95% CI: 1.12, 3.28) for more than 15 years of exposure (Lee et al., 2009). The role of involuntary smoking for oral cavity cancer is not clear at present.

Oropharyngeal Cancer

A positive association of involuntary smoking with oropharyngeal cancer risk among never smokers was reported in one study. The

Oral Cavity Cancer

Studies have examined height and body mass measured by body mass index (BMI) (weight divided by the square of height) and risk of oral cavity cancer. A  pooled analysis from the INHANCE consortium including 3740 oral cavity cancer cases and 12,995 controls reported OR estimates for oral cavity of 0.46 (95% CI: 0.39, 0.54) for overweight individuals and 0.39 (95% CI: 0.29, 0.53) for obese individuals compared to normal BMI (Gaudet et  al., 2010). The decreased oral cavity cancer risks were observed also for BMI at age 20–​30 years (p for trend = 0.05). In the analysis including all head and neck cancer patients, these inverse associations were stronger among ever tobacco users and ever alcohol drinkers; thus it is possible that these associations are also stronger for oral cavity cancer among these subgroups. Decreased head and neck cancer risks also were observed with increasing BMI (per 5  kg/​m2 RR  =  0.94; 95% CI:  0.90, 0.98) in an analysis of data pooled from multiple cohorts including 3760 head and neck cancer cases (Gaudet et al., 2015). For never smoking oral cavity cancer cases, the association with body mass index was not observed (n = 298). In a study in the United Kingdom in a cohort of 5.24 million adults, the hazard ratio associated with a 5 kilogram per meter squared of the BMI was 0.81 for oral cavity cancer, but there was no association among the never smokers, suggesting that the effect among current and ex-​smokers and overall was due to confounding by smoking since smokers tend to have lower BMI (Bhaskaran et al., 2014). It should be noted that the digestive tract cancers that are inversely associated with BMI are those that may impair chewing or swallowing, raising the possibility of inverse causation (Chapters 11 and 20). Other anthropometric measures have been examined in relation to oral cavity cancer risk. A pooled analysis from the INHANCE consortium of 4714 oral cavity cancer cases reported on inverse associations between height and cancer risk for both men and women (Leoncini, 2014). In the pooled analysis of cohort studies including 3760 head and neck cancer cases, increased risks of head and neck cancer were observed for increasing height (per 5 cm RR = 1.02; 95% CI: 1.00-​], 1.05), increasing waist circumference (per 5 cm RR = 1.02; 95% CI: 1.00, 1.04), and increasing waist-​to-​hip ratio (per 0.1 unit RR = 1.06; 95% CI: 1.04, 1.09) (Gaudet et al., 2015). In addition, increased risks of oral cavity cancer among never smokers were observed for increasing waist circumference, hip circumference, and waist-​to-​hip ratio. In the same cohort studies analysis, increased risks of head and neck cancer were observed for increasing height (per 5 cm RR = 1.02; 95% CI: 1.00, 1.05) among non-​smokers, but no association was observed for oral cavity cancers. The relationship of BMI, other anthropometric factors, and oral cavity cancers is inconsistent across studies, and the interpretation is difficult to unravel. Alcohol drinkers and tobacco users are more likely than non-​users to be leaner, making it important to examine BMI risks in non-​smokers and non-​drinkers. BMI levels modify the effect of years of alcohol drinking times frequency of drinking on oral cavity and pharyngeal cancer risk, but not for the larynx, in a pooled case-​control study suggesting that cancer site–​specific factors may be involved (Lubin et al., 2010). However, the increased risks associated with measures of central adiposity in one cohort study (Gaudet et al., 2015) suggest a possible role for inflammation.

56

566

Part IV:  Cancers by Tissue of Origin

Oropharyngeal Cancer

A pooled analysis from the INHANCE consortium including 2882 oropharyngeal cancer cases and 14,186 controls reported decreased oropharyngeal cancer risks of 0.49 (95% CI:  0.40, 0.66) for overweight individuals and 0.42 (95% CI: 0.32, 0.56) for obese individuals (Gaudet et  al., 2010). These inverse associations were persistent for BMI 2–​5  years before interview, as well as for BMI at age 20–​30. A pooled analysis from the INHANCE consortium of 4663 oropharyngeal cancer cases reported on inverse associations between height and cancer risk for both men and women (Leoncini, 2014). In the pooled analysis of cohort studies, oropharyngeal cancer risk among never smokers was increased by height (p for trend = 0.0013) and reduced by waist-​to-​hip ratio (p for trend = 0.04) (Gaudet et al., 2015). Among never smokers, BMI, waist circumference, and hip circumference were not associated with oropharyngeal cancer risk. In summary, magnitudes and directions of the findings on anthropometric measures were not consistent for oropharyngeal cancer.

Physical Activity Oral Cavity Cancer

A pooled analysis of four case-​control studies and an additional two cohort studies have reported on the potential association between physical activity and the risk of oral cavity cancers. The pooled analysis study included 612 oral cavity cancer cases and 5580 controls, and reported a decreased risk of oral cavity cancer for individuals who reported moderate recreational physical activity (OR  =  0.74; 95% CI: 0.56, 0.97) and for individuals who reported high physical activity (OR = 0.53; 95% CI: 0.32, 0.88) (Nicolotti et al., 2011). The ORs were adjusted for age, sex, study, race/​ethnicity, education level, occupational physical activity, duration of cigarette smoking, number of cigarettes smoked per day, alcohol consumption (ml/​day), and years of drinking. The four studies were from Italy, Japan, and the United States. Although there is a suggestion of a reduced risk of oral cavity cancer due to physical activity, it is not conclusive at present. In contrast to the inverse association of physical activity in the pooled case-​control study, in the NIH-​AARP Diet and Health Study prospective cohort including 525 oral cavity cancer cases, physical activity was not associated with oral cavity cancer risk after adjustment for age, sex, race/​ethnicity, BMI, marital status, education, fruit and vegetable intake, red meat intake, cigarette smoking (status, frequency, duration, years since quitting) and alcohol (servings/​ day) (Leitzmann et al., 2008). One study found that the timing of vigorous physical activity in relation to time of diagnosis was important. In the Prostate, Lung, Colorectal, and Ovarian (PLCO) cohort study that included 177 head and neck cancer cases, physical activity measured as hours spent in vigorous activity at age 40 was not associated with the risk of head and neck cancer (Hashibe et al., 2013). However, hours spent in vigorous activity at the baseline interview was associated with a reduced risk of head and neck cancer (p for trend = 0.0312) (Hashibe et al., 2013). The reductions in risk were 0.57 (95% CI: 9.32, 1.00) for 1–​2 hours per week of vigorous activity, and 0.58 (95% CI: 0.35, 0.96) for 3 or more hours per week of vigorous activity (Hashibe et  al., 2013). In summary, the relationship between physical activity and oropharyngeal cancer risk is inconsistent.

Oropharyngeal Cancer

The INHANCE pooled analysis including four case-​control studies with 645 oropharyngeal cancer cases, 173 hypopharyngeal cancer cases, and 5580 controls reported a decreased risk of pharyngeal cancer for individuals who reported moderate recreational physical activity (OR = 0.67; 95% CI: 0.53, 0.85) and for individuals who reported high physical activity (OR = 0.58; 95% CI: 0.38, 0.89) (Nicolotti et al., 2011). The ratio of oropharyngeal to hypopharyngeal cancer cases in this study was 3.7:1. In the PLCO cohort study that included 177 head and neck cancer cases (of which 43 were oropharyngeal cancers), hours spent in vigorous activity at the baseline interview was associated with

a reduced risk of head and neck cancer (p for trend = 0.0312) (Hashibe et al., 2013). In the NIH-​AARP Diet and Health Study prospective cohort including 236 pharyngeal cancer cases, physical activity was not associated with pharyngeal cancer risk after adjustment for demographic and lifestyle factors (Leitzmann et al., 2008).

Mouthwash Use and Dental Hygiene Oral Cavity Cancer

Commercial mouthwash typically contains alcohol that results in measureable levels of the carcinogen acetaldehyde in saliva (Lachenmeier et al, 2009). Individual case-​control studies have yielded inconsistent findings regarding risks associated with mouthwash use (La Vecchia 2009). A meta-​analysis of 18 studies reported that regular mouthwash use and oral cavity cancer risk were not associated (Gandini et al., 2012). In contrast, in an INHANCE analysis of 2790 oral cavity cancer cases and 10,020 controls from 12 case-​control studies, ever mouthwash use was associated with an increased risk of oral cavity cancers (OR = 1.11; 95% CI: 1.00, 1.23) (Boffetta et al., 2015). Increased oral cavity cancer risks were not observed until the duration of mouthwash use reached 36 or more years (p for trend = 0.08). For increasing frequency of mouthwash use, increasing oral cavity cancer risks were observed (p for trend = 0.02). However, mouthwash use was not associated with oral cavity cancer risk among never tobacco smokers or never alcohol drinkers. The prevalence of mouthwash use was higher among heavy smokers; thus tobacco use may act as a potential confounder. An increased risk of upper aerodigestive tract (UADT, including oral cavity, oropharynx, hypopharynx, larynx, and esophagus) cancers for using mouthwash 3 or more times per day (OR  =  3.53; 95% CI: 1.65, 7.57) (Ahrens et al., 2014) also was reported in a large Western European multicenter study of 934 oral cavity and oropharyngeal cancers not included in the INHANCE analysis. Various specific indicators of poor oral health have been associated with increased odds ratios for oral cavity cancer. In the Western European study, other oral health factors that were associated with UADT cancer risk (1963 total UADT cases, of which 934 were oral cavity/​oropharynx) included having complete dentures, younger age at starting to wear dentures, less frequent teeth cleaning, and never visiting a dentist. Another meta-​analysis of 18 case-​control studies including 7068 cases and 9990 controls reported that low frequency of tooth brushing resulted in a two-​fold increase in head and neck cancer (OR = 2.08; CI 95%: 1.65, 2.62) (Zeng et al., 2015). Two studies in Asia also reported on associations between poorer dental health status and oral cavity cancer. In a study in Japan including 261 oral cavity cancer cases, individuals who did not brush their teeth had a six-​ fold increase in oral cavity cancer risk (OR = 6.11; 9% CI: 1.35, 27.6; p for trend = 0.046) (Sato et al., 2011). The ORs were adjusted for age, sex, smoking and alcohol consumption, intake of vegetables, fruits, hot beverages, BMI, occupation, and number of remaining teeth. For UADT cancer, having 21 or more teeth appeared to be protective (OR  =  0.41; 95% CI:  0.33, 0.52) compared to individuals with 1–​8 teeth. In a study in Taiwan including 212 oral cavity cancer cases and 296 controls, increased oral cavity cancer risk was observed for frequent gum bleeding, greater number of missing teeth, and less frequent dental visits (Chang et al., 2013). In this study, dental floss and mouthwash use were not associated with oral cavity cancer risk. An INHANCE pooled analysis study used an index of oral health based on denture wearing, gum disease, missing teeth, regular tooth brushing, and regular dental visits (Hashim et al., 2016). Poorer oral health on this scale was associated with a higher odds ratio for oral cavity cancer, with gum disease, fewer than five missing teeth, annual dental visit, and daily tooth brushing as individual factors. Investigators have used a wide range of self-​reported indicators of poor oral health, and not all of them are associated with oral cavity cancer risk within or across studies. However, the frequent findings of elevated risks with poorer oral health for individual oral health characteristics and an index of poor oral health adjusting for confounders, coupled with several possible biologically plausible explanations,

 567



Oral Cavity, Oropharynx, Lip, and Salivary Glands

suggest that poor oral health is an important risk factor for oral cancer (Hashim et al., 2016).

Oropharyngeal Cancer

Findings regarding mouthwash use and oropharyngeal cancer have been inconsistent. In a pooled analysis of 2632 oropharyngeal cancer cases and 10,090 controls, ever mouthwash use was associated with an increased risk of oropharyngeal cancers (OR = 1.22; 95% CI: 1.10, 1.35) (Boffetta et al., 2015). Mouthwash use was not associated with head and neck cancer risk among never tobacco smokers or never alcohol drinkers. The Western European multicenter study including 934 oral cavity and oropharyngeal cancer cases combined reported increased risks due to mouthwash use (OR = 3.53; 95% CI: 1.65, 7.57 for using mouthwash ≥ 3 times per day) (Ahrens et al., 2014). Other dental hygiene factors were examined in two studies from Asia and one international data pooling study. The Japanese study including 61 oropharyngeal cancer cases reported no association between tooth brushing and the risk of oropharyngeal cancer (Sato et al., 2011). The study in Taiwan including 75 pharyngeal cancer cases and 296 controls reported increased pharyngeal cancer risks with less frequent tooth brushing, frequent gum bleeding, and greater number of missing teeth (Chang et al., 2013). The INHANCE pooled analysis study found that missing teeth, not daily tooth brushing, and not having annual dental visits, were associated with a higher oropharyngeal risk (Hashim et  al., 2016). These patterns of poorer oral health status and elevated oropharyngeal cancer risks are similar to that for the oral cavity.

Indoor Air Pollution A multicenter study in Central Europe including 295 oral cavity cancer cases reported no associations with heating/​cooking fuel sources, including wood and coal (Sapkota et al., 2013). In contrast, dose–​response relationships were observed for the years of wood as a heating/​cooking fuel source for all pharyngeal cancer cases (p for trend = 0.05) and for pharyngeal cancer cases who had never used coal (p for trend = 0.04). The IARC monographs concluded that there is not enough evidence to evaluate the carcinogenicity of indoor air pollution, particularly coal emissions, for cancers other than the lung, such as laryngeal and hypopharyngeal cancers (IARC, 2012a). Additional studies are necessary to assess the possible association between indoor air pollution and oral cavity and oropharyngeal cancer risk.

Diet/​Nutrition Oral Cavity Cancer

Numerous studies have reported that diets low in fruits, non-​starchy vegetables, or carotenoid intake are associated with an elevated risk for head and neck cancer (World Cancer Research Fund [WCRF], 2007). An INHANCE pooled analysis of 14,520 head and neck cancer cases (patients with oral cavity, oropharyngeal, hypopharyngeal, and laryngeal cancer combined) and 22,737 controls from 22 case-​control studies reported decreased risks of head and neck cancer for fruit intake and vegetable intake (Chuang et al., 2012). Red meat intake and processed meat intake were associated with increased head and neck cancer risks. Of the 14,520 head and neck cancer cases, 3859 were oral cavity cancers. When dietary patterns were explored by principal component factor analysis, three patterns were identified as “animal products and cereals,” “antioxidant vitamins and fibers,” and “fats” (Edefonti et al., 2012). Oral cavity cancer risk was reduced with higher antioxidant vitamins and fibers. In the AARP cohort study, the Healthy Eating Index-​2005, which assesses compliance with the six components (total grains, whole grains, total vegetables, dark-​green and orange vegetables and legumes, total fruit, and whole fruit) of the 2005 Dietary Guidelines for Americans, and the Mediterranean Diet Score, to identify study participants above the median for dietary components (vegetables, legumes, fruit, nuts, whole grains, and fish, and below the median ratio of mono-​unsaturated to saturated fat and

567

red and processed meat), were used (Li et al., 2014). Dose-​dependent lower risks were reported for both of the healthier dietary patterns and head and neck cancers. Regarding studies examining the intake of individual nutrients, in other INHANCE pooled data analysis, decreased oral cavity and pharyngeal cancer combined risks were observed for vitamin E (Edefonti et  al., 2015a), vitamin C (Edefonti et  al., 2015b), and carotenoids (Leoncini et al., 2015). Folate intake was also associated with reduced oral cavity cancer risk (Galeone et al., 2015a). A cohort study in the United States including 147 head and neck cancer cases reported an interaction between alcohol and folate intake, particularly among women (Shanmugham et al., 2010). Another cohort study found that diets high in grains and fiber were associated with reduced head and neck cancer risk, and that the associations were similar for cancers of the oral cavity, oropharynx, and hypopharynx (Lam et al., 2011).

Oropharyngeal Cancer

Studies of dietary components and nutrients and oropharyngeal cancer have been reported. The INHANCE pooled analysis for diet of 14,520 head and neck cancer cases that reported reduced risks due to fruit intake and vegetable intake had included 4755 pharyngeal cancer cases (Chuang et al., 2012), and was consistent with the finding for oral cavity cancer in the same study. Garlic intake was associated with a reduced risk of pharyngeal cancer (OR for high intake = 0.62; 95% CI: 0.40, 0.97; p for trend = 0.07; n = 1544 pharyngeal cancer cases) (Galeone et  al., 2015b). Pharyngeal cancer risk was reduced with higher “antioxidant vitamins and fibers” dietary patterns in a principal component factor analysis (Edefonti et al., 2012). Folate intake was not associated with pharyngeal cancer risk (Galeone et al., 2015a), although it was associated with a reduced risk of oral cavity cancer in the same study.

Coffee and Tea Oral Cavity Cancer. An INHANCE pooled analysis includ-

ing 1191 oral cavity cancer cases and 9028 controls from nine studies reported reduced oral cavity cancer risk with caffeinated coffee drinking (p for trend < 0.01) but not for decaffeinated coffee (p for trend = 0.17) (Galeone et al., 2010). The dose-​dependent odds ratios for oral cavity cancer were 0.65 (95% CI: 0.42, 1.02) for < 3 coffee cups/​day, 0.52 (95% CI:  0.27, 0.98) for drinking 3–​4 cups/​day, and 0.46 (95% CI:  0.30, 0.71) for > 4 cups per day. A  meta-​analysis of nine case-​control studies and one cohort study reported a protective summary risk estimate for oral cavity and pharyngeal cancer of 0.64 (95% CI: 0.51, 0.80) for the highest versus lowest category of coffee drinking (Turati et al., 2011). More recent studies not included in the pooled analysis or meta-​ analysis also have for the most part supported an inverse association between coffee intake and oral cavity cancer risk. A population-​based case-​control study of 689 oral cavity cancer cases and 3481 controls in France reported that ever coffee drinking was not significant (OR  =  0.63; 95% CI:  0.37, 1.05) but dose–​response relations were observed for both frequency (cups/​day) and cumulative consumption (cup-​years) (Radoï et al., 2013b). The Cancer Prevention Study II cohort in the United States reported that the risk of oral/​pharyngeal cancer death was reduced for individuals who drank > 4 cups/​day of caffeinated coffee (RR  =  0.51; 95% CI: 0.40, 0.64), with dose–​response relations for frequency (cups/​day; n = 868 deaths) (Hildebrand et al., 2013). In another US cohort study including 145 head and neck cancer cases, coffee was not associated with head and neck cancer risk (Hashibe et al., 2015). In a European study of 2304 upper aerodigestive tract cancer cases and 2227 controls, coffee was not associated with UADT cancer risk (Lagiou et al., 2009). Of the 2304 UADT cancer cases, 534 were oral cavity cancer cases. Perhaps the latter two studies did not detect associations since the analysis was not focused on oral cavity or pharyngeal cancer cases, and included laryngeal and/​or esophageal cancers. Findings regarding tea drinking are inconsistent. The French case-​ control study reported reduced oral cavity cancer risks for tea intake (OR  =  0.70; 95% CI:  0.53, 0.92) with dose–​response relations for

568

568

Part IV:  Cancers by Tissue of Origin

frequency (cups/​day), duration (years), and lifetime cumulative consumption (cup-​years) (Radoï et al., 2013b). In a case-​control study in China including 723 oral cavity cancer cases and 857 controls, green tea was associated with a reduced oral cancer risk among men, but black tea consumption was not associated with cancer risk (Fu et al., 2013). Tea drinking was not clearly associated with oral cavity cancer risk in the INHANCE pooled analysis (Galeone et al., 2010) and head and neck cancer risk in a US cohort study (Hashibe et al., 2015). Green tea intake also was not associated with oral cavity cancer risk in a cohort study in Japan including 37 oral cavity cancer cases (Ide et al., 2007). It seems likely that coffee intake is inversely associated with oral cavity cancer risk, while the potential association between tea intake and oral cavity cancers will require further investigation.

Oropharyngeal Cancer. Coffee and tea drinking and the

risk of oropharyngeal cancer were reported in a few studies. The INHANCE pooled analysis including 2112 oropharynx and hypopharyngeal cancer cases and 9028 controls from nine studies reported reduced pharyngeal cancer risk with caffeinated coffee drinking (p for trend = 0.02) but not for decaffeinated coffee (p for trend = 0.75) (Galeone et al., 2010). The reduced risk was observed for individuals who drank > 4 coffee cups per day (OR = 0.58; 95% CI: 0.41,0.82). Tea intake was not associated with pharyngeal cancer risk in this study. The other studies on coffee or tea did not report risk estimates specific to pharyngeal or oropharyngeal cancers.

Occupation Several studies have examined the role of occupation in the etiology of oral cavity cancers, and working in certain occupational groups has been associated with increased risks of oral cavity cancers. Occupational groups where excess risk has been noted include butchers, male carpet installers, machinists, male leatherworkers, textile workers, women in the electronics industry, sugar-​cane farmers, and a variety of other occupations wherein blue collar workers are exposed to dusts, inhaled organic agents, or inhaled inorganic agents (Mayne et al., 2006). Recent studies have further explored occupations and the risk of oral cavity cancer in larger studies. A  multicenter study in Western Europe including 915 oral cavity and oropharyngeal cancer cases reported increased risks of oral cavity and oropharyngeal cancers for the following occupations: painters, construction, bricklayers, stonemasons and tile setters, bricklayer-​construction, reinforced concreter (general), building of complete constructions or part thereof, general construction of buildings and civil engineering works, construction of motorways, roads, airfield and sport facilities, and other retail sale in non-​specialized stores (Richiardi et al., 2012). The ratio of oral cavity to oropharyngeal cancer cases was approximately 1 in this study. A hospital-​based case-​control study in the United States including 241 oral cavity and oropharyngeal cancer cases and 1522 controls reported that wood dust exposure was not associated with an increased risk of oral cavity and oropharyngeal cancers (Jayaprakash et al., 2008). There is generally a lack of replication of findings of occupation and oral cancer risks, and it is unclear whether confounding by tobacco, alcohol, and other risk factors has been adequately addressed in most studies. Occupational studies of oral cavity and oropharyngeal cancer have not reported estimates separately for oropharyngeal cancers.

Asbestos The Institute of Medicine (IOM) reviewed 16 cohort studies and 6 case-​ control studies on the possible association between asbestos and pharyngeal cancer, but few of these studies assessed asbestos exposure well or adequately adjusted for smoking and alcohol as potential confounders (Institute of Medicine, 2006). Although the overall meta-​analysis estimate suggested an association between asbestos exposure and pharyngeal cancer, dose–​response relations were not evident. The IOM concluded that the evidence is “suggestive but not sufficient to infer a causal relationship between asbestos exposure and pharyngeal cancer.”

HOST FACTORS: ORAL CAVITY AND OROPHARYNGEAL CANCER Predisposing Medical Conditions and Immune Function A number of medical conditions in addition to oral premalignant lesions have been associated with an increased risk of oral cavity and oropharyngeal cancer. Among them are alcohol-related conditions including alcoholism, cirrhosis of the liver, Fanconi anemia and psoriasis (Mayne et al., 2006).

Oral Cavity Cancer

In the PLCO cohort study including 177 head and neck cancer cases, increased risks of head and neck cancer were observed for having had a history of chronic bronchitis (RR = 2.13; 95% CI: 1.24, 3.63) and cirrhosis (RR = 4.71; 95% CI: 1.17, 19.03) (Hashibe et al., 2013). Although these risk ratio estimates were adjusted for alcohol drinking frequency and tobacco pack-​years, residual confounding cannot be ruled out, since bronchitis is a smoking-​related disease and cirrhosis is an alcohol-​related disease. In a pooled analysis of 12 case-​control studies, diabetes diagnosis was associated with an increased risk of head and neck cancer among never smokers (OR = 1.59; 95% CI: 1.22, 2.07), but did not appear to be associated specifically with oral cavity cancer risk (Stott-​Miller et al., 2012). Other case control studies in Europe and China have not found consistent patterns of oral cavity cancer risk with chronic disease conditions reported by study participants (Li et al., 2015; Macfarlane et al., 2012; Radoï et al., 2013a). In cohort studies of immunocompromised individuals, increased risks of oral cancer were observed in HIV-​positive individuals (Clifford et al., 2005) and in organ transplant patients (Öhman et al., 2015).

Oropharyngeal Cancer

In a cohort study in Taiwan of individuals diagnosed with gastro-​ esophageal reflux disease, an excess risk of oropharyngeal cancer was observed (SIR  =  3.58; 95% CI:  1.85, 6.25) (Kuo et  al., 2015). Two other studies reported that oropharyngeal cancer risk was not associated with various other medical conditions and diseases (Macfarlane et al., 2012; Stott-​Miller et al., 2012).

Inherited Genetic Susceptibility Familial Aggregation Oral Cavity Cancer. The INHANCE consortium reported that

family history of head and neck cancer increased the risk of oral cavity cancer (OR = 1.53; 95% CI: 1.11, 2.11; 2332 cases and 12,741 controls) with adjustment for multiple factors including tobacco smoking and alcohol drinking (Negri et al., 2009). Approximately, 5%–​10% of head and neck cancer patients had a family history of head and neck cancer according to this pooled data of studies, largely from Europe, the United States, and South America. The risk of oral cavity cancer was not associated with family history of smoking-​related cancers or family history of cancer. Individuals who were 45 years or younger had an even higher risk of head and neck cancer if their family history of cancer was in young relatives (OR = 2.27; 95% CI: 1.26, 4.10) according to the INHANCE data (Toporcov et al., 2015). A population-​based case-​control study in France including 689 oral cavity cancer cases and 3481 controls reported an increased risk of oral cavity cancer for individuals with family history of head and neck cancer (OR  =  1.9; 95% CI:  1.2, 2.8) with adjustment for tobacco and alcohol (Radoï et al., 2013a). The oral cavity cancer risk was also elevated for family history of all cancers when the individual had three or more relatives with cancer (OR = 1.5; 95% CI: 1.0, 2.1). Other studies using population-​ based genealogical resources in Utah and Sweden have also reported on an increased the risk of oral cavity cancer due to family history of cancer (Teerlink et  al., 2012; Li and Hemminki, 2003). The limitation in these large-​scale database studies is that there was no information on tobacco and alcohol; thus it is difficult to assess whether the increased risk due to family history

 569



Oral Cavity, Oropharynx, Lip, and Salivary Glands

is because of shared genetics or shared lifestyle factors or both. In Sweden, for parents who had tongue cancer, excess risk among children was observed for lung cancer and skin cancer (Li and Hemminki, 2003). For parents with mouth cancer, excess risk was observed for cervical, female genital, and kidney cancers among children (Li and Hemminki, 2003). In the Utah genealogical database, first-​degree relatives of tongue cancer cases had an increased risk of tongue cancer (RR = 6.0; 95% CI: 2.37, 12.65) but increased risks were not observed for second-​or third-​degree relatives (Teerlink et al., 2012). In summary, the two case-​control studies involving study participants’ reports about the cancer experience of family members reported an increased risk of oral cavity cancer among their relatives (Negri et al., 2009; Radoi et al., 2013), even with adjustment for tobacco and alcohol behaviors. Although potential important confounders could not be controlled for, European and US genealogical studies also suggest a pattern of increased risk associated with a family history of types of oral cavity cancer (Li and Hemminki, 2003; Teerlink et al., 2012).

Oropharyngeal Cancer.  Similar to the findings for oral cavity

cancer, a family history of head and neck cancer was more common among oropharyngeal or pharyngeal cancer cases than controls. The International Head and Neck Cancer Epidemiology (INHANCE) consortium reported that family history of head and neck cancer increased the risk of oropharyngeal cancer (OR = 1.55; 95% CI: 1.16, 2.07; 2069 cases and 12,741 controls) with adjustment for multiple factors including tobacco smoking and alcohol drinking (Negri et  al., 2009). For family history of smoking-​related cancers, the risk of oropharyngeal cancer was suggestive of an increase (OR = 1.15; 95% CI: 0.99, 1.34). Family history of other cancers was not associated with oropharyngeal cancer risk. In Sweden, for parents who had pharyngeal cancer, excess risk among children was observed for stomach cancer, connective tissue cancer, and cancer overall (Li and Hemminki, 2003). In the Utah genealogical database, first-​and second-​degree relatives of pharyngeal cancer cases had an increased risk of pharyngeal cancer (RR = 7.9; 95% CI: 3.46, 15.67 for first-​degree relatives; and RR = 3.3; 95% CI: 1.12, 15.70 for second-​degree relatives) (Teerlink et al., 2012).

Genetic Variants Oral Cavity Cancer.  Common genetic variants have been studied

extensively for oral cavity cancers, with a focus on genes in pathways including tobacco carcinogen metabolism, alcohol metabolism, DNA repair, cell cycle, and inflammation (Wang LE, et  al., 2010). In the INHANCE consortium data were pooled on 28 common sequence variants on 1901 oral cavity cancer cases and 1751 pharyngeal cancer cases from 14 European, North America, Central American, and Asian case-​control studies (Chuang et al., 2011). The sequence variants MGMT Leu84Phe, ADH1B Arg48His, ADH1C Ile350Arg, and the GSTM1 null genotype were associated with head and neck cancer risk, with fairly low false-​positive report probabilities. The associations for ADH1B Arg48His, ADH1C Ile350Arg, and the GSTM1 null genotype were observed specifically with oral cavity cancer risk. In a large study focusing on six sequence variants in the ADH gene family, the strongest associations were observed for ADH1B Arg48His (OR = 0.45; 95% CI: 0.35, 0.57) and ADH7 G92A (OR = 0.70; 95% CI:  0.59, 0.84) with the risk of oral cavity and pharyngeal cancer (Hashibe et  al., 2008). This study of 3876 upper aerodigestive tract cancers included 1790 oral cavity and pharyngeal cancer cases. There was an inverse dose–​response relationship for the association of these genetic variants across drinking intensity categories. In other words, the odds ratios were smaller for individuals who drank more alcohol. The top hit from a genome-​wide association study (GWAS) of 2091 upper aerodigestive tract cancers and 3513 controls was a common genetic variant in ADH7 (rs971074) (McKay et  al., 2011), which is highly correlated with the previously reported ADH7 G92A (Hashibe et  al., 2008). The top 19 genetic variants were genotyped in 6514 UADT cancer cases and 7892 controls. Six of the variants that were significantly associated with UADT cancer risk were in the alcohol metabolism genes (ADH1B, ADH1C, ADH7, ALDH2). An additional significantly associated variant was in the HEL308 gene. The

569

odds ratios for oral cavity cancers were 0.72 (95% CI: 0.64, 0.82) for ADH7 G92A; 1.13 (95% CI: 1.05, 1.22) for ADH1C Ile350Arg; 1.12 (95% CI: 1.04, 1.21) for ALDH2 rs4767364 (in the intron); 1.13 (95% CI: 1.05, 1.22) for HEL308 V306I. The strongest evidence for the role of genetic susceptibility on the risk of oral cavity cancer thus far have been for genetic variants in the aldehyde and acetaldehyde dehydrogenase genes. This is consistent with the role of alcohol increasing the risk of oral cavity cancers.

Oropharyngeal Cancer.  Most of the studies on common genetic

variants with a candidate gene or genome-​wide association examined head and neck cancers (oral cavity, oropharynx, hypopharynx, larynx) or upper aerodigestive tract cancers (head and neck cancer and esophageal cancer). As mentioned in the oral cavity section, candidate gene approaches focused on genes on the tobacco carcinogen metabolism, alcohol metabolism, DNA repair, cell cycle, and inflammation pathways (Wang LE, 2010). In the INHANCE consortium pooled analysis of 28 common sequence variants including 1751 pharyngeal cancer cases from 14 European, North America, Central American, and Asian case-​control studies (Chuang et  al., 2011), ADH1B Arg48His was associated with pharyngeal cancer risk (OR = 5.77; 95% CI: 3.22, 10.1 for Arg/​Arg vs His/​His). The associations with ADH1C Ile350Arg and GSTM1 null genotypes were less clear for pharyngeal cancer. For the GWAS, the odds ratios for oropharyngeal cancers were 0.73 (95% CI: 0.64, 0.83) for ADH7 G92A; 1.12 (95% CI: 1.03, 1.21) for ADH1C Ile350Arg; 1.13 (95% CI: 1.04, 1.22) for ALDH2 rs4767364 (in the intron); 1.12 (95% CI: 1.04, 1.20) for HEL308 V306I (McKay et al., 2011). These studies of common sequence variants identified associations between alcohol metabolism variants and oropharyngeal cancer that were similar to those for oral cavity cancers.

EXTERNAL LIP CANCERS Lip cancer was diagnosed in 1,880 persons in the United States in 2013 (1,380 males and 500 females; includes interior lip) (USCS, 2016). Mortality is uniformly very low in all races and in Hispanic populations, with a US total of 70 persons dying from this cancer in 2013 (USCS, 2016). The ICD-​O-​3 topography code for the lip, C00, includes malignant neoplasms arising in the external and inner (mucosal) lip, on the vermilion border, commissure, and labial mucosa, but excludes cancers originating on the skin of the lip (C44.0; Fritz et al., 2000). This section presents data whenever possible for the external lip, which includes codes C00.0, C00.1, C00.2.; C00.8, and C00.9), whereas the lip mucosa, commissure, and overlapping lip tumors (C00.3–​C00.6) are included in the oral cavity cancer section of this chapter (most epidemiologic studies group all lip sites in ICD codes C00 together). The lower lip is most frequently affected (Table 29–​5). External lip cancers are primarily squamous cell carcinomas (92.1%), with the remainder including 3.6% other epidermoid carcinomas, and 2.4% adenocarcinomas, and 1.9% other types (SEER, 2016). Overall external lip cancer occurs at a rate of 0.5 per 100,000 persons per year in the years 2009–​2013 (Table 29–​3), increases with age (Figure 29–​7), and is higher among men (0.9 per 100,000) compared to women (0.2 per 100,000) (SEER, 2016). People of white race have the highest incidence of external lip cancer (0.6 per 100,000) compared to 0.1–​0.3 for any other race/​Hispanic ethnicity group. Lip cancer incidence rates have been declining (Figure 29–​1 and Table 29–​5). Mortality rates in the United States have been very low and stable for all races (Figures 29–​2 and 29–​18). Among men, the highest age-​ adjusted incidence rates (ASR, World) for lip cancer (C00) in the world around 2005 were observed in Granada, Spain (8.6 per 100,000), Tasmania, Australia (7.6 per 100,000), and Azores, Portugal (6.2 per 100,000) (Forman et  al., 2013). The lowest age-​adjusted incidence rates for lip cancer in the world in men are observed in the Philippines, Costa Rica, Japan, and among black populations in the U.S. registries (at ~0 per 100,000 with ~1 case diagnosed in a five-​year period). Among women, the highest age-​adjusted incidence rates in the world are observed in Tasmania,

570

570

Part IV:  Cancers by Tissue of Origin White Male

7

6

6

5 4 3 2 1 0 1975

1980

1985

1990 1995 2000 Year of Diagnosis/Death

2005

5 4 3 2 1 0 1975

2013

(d)

7

6

6

Rate per 100,000

7 5 4 3 2 1 0 1975

1980

1985

1990

1995

2000

1980

1985

1990

1995

2000

2005

2013

2005

2013

Year of Diagnosis/Death

Black Male

(c)

Rate per 100,000

White Female

(b)

7 Rate per 100,000

Rate per 100,000

(a)

2005

2013

Black Female

5 4 3 2 1 0 1975

1980

1985

1990

1995

2000

Year of Diagnosis/Death

Year of Diagnosis/Death Incidence Mortality

Figure 29–​18 a–​d.  Incidence and death rates for cancer of the lip. (a) White male; (b) White female; (c) Black male; (d) Black female.

Australia (2 per 100,000), Khon Kaen, Thailand (1.6 per 100,000), and Aracaju, Brazil (1.4 per 100,000). The lowest rates in the world among women are observed in countries such as Japan, Costa Rica, and Singapore (at ~0 per 100,000 with ~1 case diagnosed per five-​year period). Survival is generally very good; based on data from 2006–​2012, 90.3% of lip cancer patients survive 5 or more years (Table 29–​4; SEER, 2016). Among 8188 lip cancer patients diagnosed during 1973–​2000 in the SEER data, 1814 (22.2%) were diagnosed with a second primary cancer (Curtis et al., 2006). According to the IARC monographs, there are no carcinogenic agents with sufficient evidence in humans for lip cancer. Hydrochlorothiazide (IARC, 2016)  and solar radiation (IARC, 2012c) are carcinogenic agents with limited evidence in humans. Lip cancer is associated with low socioeconomic status, rural residence, and outdoor occupation, particularly fishing and farming (Mayne et al., 2006). Evidence also suggests that both solar radiation and tobacco are important risk factors for lip cancer. Solar radiation has long been considered a probable risk factor for lip cancer (IARC, 2012c). The fact that both outdoor occupation and rural residence are associated with an increased risk of lip cancer is consistent with a solar component, although it does not preclude the possibility that other outdoor elements or associated factors such as smoking are involved. The IARC monograph reviewed five case-​control studies on solar radiation and lip cancer, with the three earlier studies showing increased lip cancer risk, but without taking into account tobacco as a potential confounder. The latter two studies accounted for tobacco, with one study in women showing dose–​response relations (Pogoda and Preston-​Martin, 1996) and the other study in men not showing a dose–​response relationship (Perea-​Milla López et al., 2003). Another link in the relationship between sunlight exposure and lip cancer risk is chronic actinic cheilitis, a precancerous condition of the lip that presents as a hyperkeratosis interspersed with areas of erythema on the vermilion border. Actinic cheilitis is generally attributed to solar damage and may harbor epithelial dysplasia, carcinoma in situ, or squamous cell carcinoma (Kaugars et al., 1999; Nicolau and Balus,

1964). As with lip cancer, actinic cheilitis predominates on the lower lip and is seen most often in males, light-​skinned persons, and outdoor workers. Hydrochlorothiazide, an antihypertensive drug, appears to increase the risk of lip cancer, according to a study of 712 lip cancer patients and 22,904 controls (Friedman et al., 2012). Individuals who had taken the drug for 5 years had a 4.22-​fold increase in lip cancer risk (95% CI: 2.82, 6.31). Evidence for a relationship between smoking tobacco and lip cancer risk is based largely upon findings from case-​control studies. Most investigations conducted in the early through mid-​twentieth century implicated primarily pipe smoking (Mayne et al., 2006). Further evidence of a link between smoking and lip cancer is provided by the observation that lip cancer cases are at an elevated risk of second primary cancers known to be associated with smoking (lung, larynx) and vice versa (Mayne et al., 2006). There is little consistent evidence that alcohol consumption, syphilis, or herpetic lesions are important etiologic factors for lip cancer (Mayne et al., 2006). In cohort studies of immunocompromised individuals, an increased risk of lip cancer was observed in organ transplant patients (López-​ Pintor et al., 2011; Öhman et al., 2015). Several studies using genealogical databases generally find that first-​degree relatives of persons with lip cancer have higher risks of lip cancer, perhaps due to shared risk factors such as outdoor occupation or tobacco use. First-​degree relatives of lip cancer patients had an increased risk of lip cancer (RR = 5.04; 95% CI: 2.75, 9.52) according to the genealogical database in Iceland (Amundadottir et  al., 2004). Second-​, third-​, and fourth-​degree relatives did not have an increased risk of lip cancer. In Sweden, for parents who had lip cancer, excess risk among their children for various cancer was not observed (Li and Hemminki, 2003). In the Utah genealogical database, first-​, second-​, and third-​degree relatives of lip cancer cases had an increased risk of lip cancer (RR  =  3.4; 95% CI:  2.72, 4.24 for first-​degree relatives; RR = 2.2; 95% CI: 1.75, 4.24 for second-​degree relatives; and RR  =  1.3; 95% CI:  1.07, 1.58 for third-​degree relatives) (Teerlink et al., 2012).

 571



571

Oral Cavity, Oropharynx, Lip, and Salivary Glands

(Forman D et  al., 2013). The lowest age-​adjusted incidence rates in men were reported the Eastern Cape province in South Africa, Prince Edward Island in Canada and in Kingston, jamaica (at ~0 per 100,000 with ~1 case diagnosed in a five-​year period). Among women, the highest age-​adjusted incidence rates in the world were observed in Hawaii, Filipino (1.8 per 100,000), American Indian in Montana (1.7 per 100,000), and Manizales, Colombia (1.1 per 100,000). The lowest rates in the world among women were observed in countries such as China, Algeria, and South Africa (at ~0 per 100,000 with ~1 case diagnosed per 5-​year period). The 5-​year relative survival rate for major salivary gland cancers is 72.5% (Table 29–​4; SEER, 2016). Survival rates vary by stage from 93.0% for localized disease, 68.1% for regional disease, and 35.6% for distant disease, and are better for white and black females (82.6% and 82.7%, respectively) than for white and black males (62.7% and 67.5%, respectively). Among 5551 salivary gland cancer patients diagnosed during 1973–​2000 in the SEER data, 11.6% were diagnosed with a second primary cancer (Curtis et al., 2006). According to the IARC monographs, the agents with sufficient evidence of carcinogenicity in humans for salivary gland cancers are X-​radiation, and gamma-​radiation (IARC, 2012c). Radioiodines, including iodine-​131, are carcinogenic agents with limited evidence in humans. Radiation is an established risk factor for salivary gland cancer, with elevated risks and dose–​response relationships observed among atomic bomb survivors and persons exposed to prior therapeutic head or neck irradiation (IARC, 2012b). Salivary gland cancer risk has also been associated with diagnostic radiation directed to the head or neck, most notably among persons exposed to frequent full-​mouth dental X-​rays, and particularly for those exposed prior to the 1960s when substantially higher doses were used (Mayne et al., 2006). Previous studies linked ultraviolet light treatments to the head and neck, used primarily to treat acne, with the risk of salivary gland cancer, but these studies did not exclude squamous cancers of the parotid gland from their studies. Squamous cancer almost never arises from the parotid gland. It is possible that in these studies some of the cancers classified as arising in the parotid gland were skin squamous cell

MAJOR SALIVARY GLAND CANCER Approximately 4,400 persons developed major salivary gland cancer in 2013 (2,500 in males and 1,900 in females) (USCS, 2016). Nine hundred people died in 2013 from this cancer (600 males and 300 females). Salivary gland cancer can arise in either the major or minor salivary glands. Current topography codes (ICD-​O-​3, ICD-​10) for malignant neoplasms of the major salivary glands (C07–​C08) include the parotid, submandibular, and sublingual glands as well as their associated ducts, while cancers of the minor salivary glands are classified separately according to their anatomical site. Based largely upon these coding practices, population-​based reports on the incidence of salivary gland cancer are generally restricted to malignant neoplasms of the major salivary glands, of which the parotid is the most frequently affected, accounting for 81% of the total (SEER, 2016). The majority of salivary gland cancers are adenocarcinomas (62.0%). Other histologic types are squamous cell carcinomas (21.2%) and other epidermoid carcinomas (1%), with various types of unspecified carcinomas accounting for the rest (SEER, 2016). These minor salivary gland cancers probably account for nearly all of the non-​squamous cell cancers in the other anatomic regions covered in this chapter. The incidence rates of major salivary gland cancers is 1.3 per 100,000 persons per year and are higher for males (1.7 per 100,000) than for females (1.0 per 100,000)(SEER) (Table 29–​3). Rates increase with age (Figure 29–​7). Rates for whites (1.3 per 100,000) are higher than for other race and ethnic groups (0.6–​1.0 per 100,000). Incidence rates have been stable over the past several decades in all race and ethnic groups (SEER, 2016) (Figure 29–​1 and Table 29–​5). The mortality rate for the period 2009–​2013 was 0.2 per 100,000 per year (0.4 for males and 0.1 for females). Mortality rates have been fairly stable over the last few decades (Figures 29–​2 d and 29–​19). Among men, the highest age-​adjusted incidence rates (World standard) for major salivary glands (C07-​08) in the world around 2005 were observed in the indigenous population of the Australian Northern Territory (3.2 per 100,000), in the Northwest Territories of Canada (2.3 per 100,000), and among Filipino, Hawaii (2.2 per 100,000)

2.5

2 1.5 1 0.5 0 1975

1980

1985

1990

1995

White Female

(b)

White Male

2.5

Rate per 100,000

Rate per 100,000

(a)

2000

2005

2 1.5 1 0.5 0 1975

2013

1980

1985

Year of Diagnosis/Death Black Male

(c)

(d)

2

Rate per 100,000

Rate per 100,000

2.5

1.5 1 0.5 0 1975

1980

1985

1990

1995

2000

1990

1995

2000

2005

2013

2005

2013

Year of Diagnosis/Death

2005

2013

Black Female

2.5 2 1.5 1 0.5 0 1975

Year of Diagnosis/Death

1980

1985

1990

1995

2000

Year of Diagnosis/Death Incidence Mortality

Figure 29–​19 a–​d.  Incidence and death rates for cancer of the salivary gland. (a) White male; (b) White female; (c) Black male; (d) Black female.

572

572

Part IV:  Cancers by Tissue of Origin

carcinomas, a common cancer in the elderly, misclassified as having originated in the parotid gland leading to a spurious association in previous studies incorrectly with UV radiation. There is evidence to link some occupational groups to increased risks of salivary gland cancer. Elevated risks have been identified among women working in beauty shops, male woodworkers employed in automobile plants, rubber industry workers, persons occupationally exposed to nickel compounds or alloys, radioactive materials, or silica dust, as well as among employees of an Australian underground colliery and residents of asbestos-​mining counties in Quebec (Mayne et al., 2006). These findings need to be replicated. While alcohol and tobacco use are clearly related to oral cavity and oropharyngeal cancers, their relationship with salivary gland cancer is equivocal. Approximately eight epidemiologic studies have investigated the association between cigarette smoking and the risk of salivary gland cancers, but the results are inconsistent (IARC, 2012a). With regard to alcohol consumption, one case-​control study found that drinking doubled salivary gland cancer risk among women, but not men (Spitz et  al., 1990), while another investigation reported an OR of 2.5 for heavy drinking and an associated dose–​response trend for salivary gland cancer among men, but not women (Horn-​Ross et al., 1997). Other studies, however, provide little additional support for a link between alcohol consumption and salivary gland cancer (Mayne et al., 2006). Diet has received little attention in relation to salivary gland cancer risk; however, in a Chinese study, the consumption of dark yellow vegetables and liver was inversely related to risk (Zheng et al., 1996), while a US study found protective effects associated with a high, relative to low, intake of fiber from bean sources and total vitamin C, but an increased risk with high cholesterol consumption (Horn-​Ross et al., 1997). A more recent study in Canada did not identify any associations with dietary factors and salivary gland cancer risk (Forrest et al., 2008). In general, viruses have also received relatively little consideration in relation to salivary gland cancer risk, although a number of case reports and series suggest a strong link between Epstein-​Barr virus (EBV) and lymphoepithelial carcinomas of the salivary glands, cancers observed primarily in Eskimos and the southern Chinese (Mayne et  al., 2006). The possible etiologic role of HPV for salivary gland cancer has been discounted by the observation that the prevalence of HPV is very low or negligible in salivary gland tumors (Brunner et al., 2012; Skálová et al., 2013). A possible hormonal link to salivary gland cancer was supported by early reports of an excess risk of secondary breast cancer after salivary gland cancer and vice versa; however, most subsequent studies involving larger cohorts revealed little, if any, increased breast cancer risk (Mayne et al., 2006). Although parity was not associated with salivary gland cancer in one US-​based case-​control study (Spitz et al., 1984), age at menarche, number of births, age at first full-​term delivery, and length of oral contraceptive use were all inversely related to salivary gland cancer risk in another (Horn-​Ross et al., 1999). In the Utah genealogical database, first-​, second-​, and third-​degree relatives of salivary gland cancer cases did not have an increased risk of salivary gland cancer (Teerlink et  al., 2012). A  GWAS was conducted for 309 salivary gland cancer patients and 535 cancer-​free controls (Xu et al., 2015). Common variants in the CHRNA2 (cholinergic receptor, nicotinic, alpha 2), OR4F15 (olfactory receptor, family 4, subfamily F, member 15), ZNF343 (zinc finger protein 343), and PARP4 (poly(ADP-​ribose) polymerase family, member 4) genes were associated with salivary gland cancer risk.

OPPORTUNITIES FOR PREVENTION Very large geographic differences in incidence across countries suggest that acquired exposures are important in the etiology of oral cavity and oropharyngeal cancers. The most important of these are tobacco in all forms, alcohol consumption, and human papillomavirus, and for lip cancer, solar radiation that collectively contribute to the high incidence of these cancers in many areas of the world. For example,

tobacco and alcohol account for the majority of oral cavity and oropharyngeal cancers, thus suggesting the importance of these factors. Increasing exposures over time to these risk factors have been linked to corresponding increases in oral cavity and oropharyngeal cancer incidence some years later, such as subsequent to increasing uptake in cigarette consumption by men and later by women in the early and mid-​twentieth century and increases later in the century and into the twenty-​first century in oropharyngeal cancer associated with increasing extent of exposure to human papillomavirus. Other factors, such as diet and energy balance, may also be important. Tobacco control is discussed at greater length in Chapters  61 and 62.1; alcohol in Chapter 12; infectious agents in Chapters 24 and 62.2; and screening for oral cancer in Chapter 61. Epidemiologic studies have established that modifying exposure to some of these risk factors can reduce the risk of oral and oropharyngeal cancer; tobacco and alcohol cessation leads to lowered risks of oral cavity cancer, in the case of tobacco cessation reaching almost to the level of non-​smokers after 10–​20 years of abstinence from tobacco. Examination of trends over time demonstrates that the reductions in tobacco use generally observed in high-​income countries has led to decreases over time in oral cavity cancers (Lee and Hashibe, 2014), suggesting that successful interventions to reduce population levels of tobacco and alcohol consumption could lead to significant reductions in the population burden of oral cavity cancers. Despite declining rates of these cancers in some populations, wide disparities in the incidence of oral and oropharyngeal cancers exist worldwide geographically, with the highest rates in Southeast Asia and among lower socioeconomic subpopulations within countries due to multiple factors, including high rates of tobacco and alcohol use, use of tobacco (smoked and unsmoked forms) and alcohol products containing varying levels of carcinogens, use of both alcohol and tobacco or multiple forms of tobacco, and poorer nutritional status. Innovative and effective approaches to prevention and behavior modifications to promote cessation of risk behaviors in these populations are needed to make progress in reducing the burden of oral cavity and oropharyngeal cancers. Also important will be efforts to identify subgroups of populations with the highest risk based on risk-​factor profiles for intensive preventive interventions. HPV vaccination has the potential to reduce the incidence of oropharyngeal cancers, although uptake of the vaccination in the United States is not optimal (Reagan-​Steiner et al., 2015) and is not yet widely available in low-​resource countries (Bruni et al., 2016). Studies have shown that less than three doses is efficacious in preventing infection (Kreimer et al., 2015); this likely will help vaccination programs increase population coverage substantially. Risk prediction models have the potential to help better identify subgroups of the population at highest risk or oral cavity and oropharyngeal cancer and who might most benefit from interventions to prevent the disease or to target for potential screening approaches. There have also been improvements in treatment over time that have increased 5-​ year relative survival. As mentioned, survival is better for cancers diagnosed at earlier stages. Nevertheless, in many parts of the world, most patients with oral cavity and oropharyngeal cancer are diagnosed only at an advanced stage. Various tests, including conventional oral physical exam, special techniques for rinsing or staining lesions, and light-​based detection, have been evaluated for use in screening asymptomatic, apparently healthy adults (Walsh et  al., 2013). Among these tests, an oral examination by a health worker was the best. The current evidence is limited, however, due to the heterogeneity of study populations and methods.

FUTURE RESEARCH Both etiologic research and translational research are needed to improve measures of primary, secondary, and tertiary prevention. More appropriate grouping of oral and oropharyngeal cancer subtypes is also essential, as is standardization of exposure measurements.

 573



Oral Cavity, Oropharynx, Lip, and Salivary Glands

573

Etiologic Research

References

Although GWAS have consistently implicated loci in the aldehyde dehydrogenase and acetaldehyde pathways in studies of oral cavity cancer, the size of these studies and statistical significance of the findings have been limited. This is particularly true for less common variants and for salivary gland cancers, the etiology of which is poorly understood. Larger studies using next-​generation sequencing technology for less common variants could be informative for oral cavity and oropharyngeal cancers. Studies that characterize the somatic mutations of oral and oropharyngeal tumors suggest that there may be subtypes of these cancers. Pooled studies that examine the epidemiologic associations in relation to the molecular subtypes could assess whether various mutations and subtypes have different etiology. Genomic sequencing of premalignant lesions to identify those most strongly associated with progression to malignancy would be clinically useful. New tobacco and nicotine products are constantly being introduced that contain known or suspected carcinogens and that involve a wide variety of new delivery systems. Electronic nicotine delivery systems (ENDS) in particular are being marketed as emerging products that purport to reduce cancer risk. Since the epithelium of the oral cavity and oropharynx are directly exposed to vapor from these products, there may be opportunities to identify histopathologic changes and biomarkers that may relate to cancer risk, long before the development of invasive cancers. Similarly, a better understanding is needed of the effects of other potential risk factors such as hookah smoking (El-​Zaatari et al., 2015; Warnakulasuriya, 2011)  and marijuana use. These exposures are difficult to study because of joint use of multiple products (e.g., marijuana and various forms of tobacco), potential recall bias in retrospective studies where marijuana use is still illegal, rapid changes in the use of these products, the challenges of measuring dose (Berthiller et al., 2009; Hashibe et  al., 2006; Huang et  al., 2015; Marks et  al., 2014), and the difficulty of enrolling study participants with sufficient size and statistical power. In-​depth genomic characterization and long-​term follow-​up of the clinical course of premalignant lesions are also needed. Since the definitions of OPMDs have changed over time, this will involve using standardized diagnostic criteria for classifying and archiving specimens collected in different time periods. More epidemiologic studies of head and neck cancer and risk factors for cancers of the oral cavity and oropharynx are needed for populations in East Asia. Many studies of inherited susceptibility genes have involved European and American populations.

ACS (American Cancer Society) Cancer Statistics Center. http://​cancerstatisticscenter.cancer.org. Accessed May 11, 2016. Ahrens W, Pohlabeln H, Foraita R, et al. 2014. Oral health, dental care and mouthwash associated with upper aerodigestive tract cancer risk in Europe: the ARCAGE study. Oral Oncol, 50(6), 616–​625. PMID: 24680035. Amundadottir LT, Thorvaldsson S, Gudbjartsson DF, et al. 2004. Cancer as a complex phenotype: pattern of cancer distribution within and beyond the nuclear family. PLoS Med, 1(3), e65. PMCID: PMC539051. Anantharaman D, Marron M, Lagiou P, et al. 2011. Population attributable risk of tobacco and alcohol for upper aerodigestive tract cancer. Oral Oncol, 47(8), 725–​731. PMID: 21684805. Arbes SJ, Olshan AF, Caplan DJ, Schoenbach VJ, Slade GD, and Symons MJ. 1999. Factors contributing to the poorer survival of black Americans diagnosed with oral cancer (United States). Cancer Causes Control, 10(6), 513–​523. PMID: 10616821. Benard VB, Johnson CJ, Thompson TD, et al. 2008. Examining the association between socioeconomic status and potential human papillomavirus-​ associated cancers. Cancer, 113(10 Suppl), 2910–​2918. PMID: 18980274. Berthiller J, Lee YC, Boffetta P, et  al. 2009. Marijuana smoking and the risk of head and neck cancer:  pooled analysis in the INHANCE consortium. Cancer Epidemiol Biomarkers Prev, 18(5), 1544–​ 1551. PMCID: PMC3046921. Berthiller J, Straif K, Agudo A, et al. 2015. Low frequency of cigarette smoking and the risk of head and neck cancer in the INHANCE consortium pooled analysis. Int J Epidemiol, 45(3), 835–​845. PMID: 26228584. Bhaskaran K, Douglas I, Forbes H, dos-​Santos-​Silva I, Leon DA, and Smeeth L. 2014. Body-​mass index and risk of 22 specific cancers: a population-​ based cohort study of 5·24 million UK adults. Lancet, 384(9945), 755–​ 765. PMCID: PMC4151483. Boffetta P, Hayes RB, Sartori S, et al. 2015. Mouthwash use and cancer of the head and neck: a pooled analysis from the International Head and Neck Cancer Epidemiology Consortium. Eur J Cancer Prev, 25(4), 344–​348. PMID: 26275006. Brouwer AF, Eisenberg MC, and Meza R. 2016. Age effects and temporal trends in HPV-​related and HPV-​unrelated oral cancer in the United States: a multistage carcinogenesis modeling analysis. PLoS One, 11(3), e0151098. PMCID: PMC4786132. Brown LM, Check DP, and Devesa SS. 2011. Oropharyngeal cancer incidence trends:  diminishing racial disparities. Cancer Causes Control, 22(5), 753–​763. PMID: 21380619. Brown LM, Check DP, and Devesa SS. 2012. Oral cavity and pharynx cancer incidence trends by subsite in the United States:  changing gender patterns. J Oncol, Article ID 649498, 10 pages. PMCID: PMC3345247. Bruni L, Diaz M, Barrionuevo-​Rosas L, et al. 2016. Global estimates of human papillomavirus vaccination coverage by region and income level: a pooled analysis. Lancet Glob Health, 4(7), e453–​463. PMID: 27340003. Brunner M, Koperek O, Wrba F, et al. 2012. HPV infection and p16 expression in carcinomas of the minor salivary glands. Eur Arch Otorhinolaryngol, 269(10), 2265–​2269. PMID: 22207527. Chang JS, Lo HI, Wong TY, et al. 2013. Investigating the association between oral hygiene and head and neck cancer. Oral Oncol, 49(10), 1010–​1017. PMID: 23948049. Chaturvedi AK, Engels EA, Anderson WF, and Gillison ML. 2008. Incidence trends for human papillomavirus-​related and -​unrelated oral squamous cell carcinomas in the United States. J Clin Oncol, 26(4), 612–​619. PMID: 18235120 Chera BS, Eisbruch A, Murphy BA, et al. 2014. Recommended patient-​reported core set of symptoms to measure in head and neck cancer treatment trials. J Natl Cancer Inst, 106(7): dju127. PMCID: PMC4192043. Chinn SB, and Myers JN. 2015. Oral cavity carcinoma: current management, controversies, and future directions. J Clin Oncol, 33(29), 3269–​3276. PMID: 26351335. Chuang SC, Agudo A, Ahrens W, et al. 2011. Sequence variants and the risk of head and neck cancer: pooled analysis in the INHANCE Consortium. Front Oncol, 1, 13. PMCID: PMC3356135. Chuang SC, Jenab M, Heck JE, et al. 2012. Diet and the risk of head and neck cancer: a pooled analysis in the INHANCE consortium. Cancer Causes Control, 23(1), 69–​88. PMCID: PMC3654401. Chuang SC, Scelo G, Tonita JM, et al. 2008. Risk of second primary cancer among patients with head and neck cancers: a pooled analysis of 13 cancer registries. Int J Cancer, 123(10), 2390–​2396. PMID: 18729183. Clifford GM, Polesel J, Rickenbach M, et al. 2005. Cancer risk in the Swiss HIV Cohort Study:  associations with immunodeficiency, smoking, and highly active antiretroviral therapy. J Natl Cancer Inst, 97(6), 425–​432. PMID: 15770006.

Methodologic Issues As mentioned, it is essential that future etiologic studies and descriptive surveillance databases present data separately for cancers of the oral cavity, oropharynx, and hypopharynx. Where it is necessary to combine these sites for comparisons with older data, the results should also be presented separately. Studies should report the exact ICD codes included in each grouping. A more standardized approach to characterizing the use of tobacco products, areca nut, and other regional exposures (e.g., by the amount of product, manufacture and preparation methods, or usual patterns of usage) would make it possible to pool data from studies of varying size. Similarly, a standardized approach to adjusting for HPV would help in controlling for this important risk factor.

Translational Epidemiologic Research Methodologically rigorous evaluations of potential screening and other interventions are needed. For example, betel quid with or without tobacco is widely used in some Southeast Asian and Pacific Island countries and contributes to risk of oral cancer; more research is needed in approaches to prevent uptake and promote cessation.

574

574

Part IV:  Cancers by Tissue of Origin

Combes JD, and Franceschi S. 2014. Role of human papillomavirus in non-​ oropharyngeal head and neck cancers. Oral Oncol, 50(5), 370–​ 379. PMID: 24331868. Conway DI, Brenner DR, McMahon AD, et al. 2015. Estimating and explaining the effect of education and income on head and neck cancer risk:  INHANCE consortium pooled analysis of 31 case-​control studies from 27 countries. Int J Cancer, 136(5), 1125–​1139. PMID: 24996155. Conway DI, McKinney PA, McMahon AD, et al. 2010. Socioeconomic factors associated with risk of upper aerodigestive tract cancer in Europe. Eur J Cancer, 46(3), 588–​598. PMID: 19857956. Conway DI, Petticrew M, Marlborough H, Berthiller J, Hashibe M, and Macpherson LM. 2008. Socioeconomic inequalities and oral cancer risk: a systematic review and meta-​analysis of case-​control studies. Int J Cancer, 122(12), 2811–​2819. PMID: 18351646. Curtis RE, Freedman DM, Ron E, et al. 2006. New malignancies among cancer survivors: SEER Cancer Registries, 1973–​2000. Bethesda, MD: National Cancer Institute, NIH Publ. No. 05-​5302. D’Souza G, Kreimer AR, Viscidi R, et al. 2007. Case-​control study of human papillomavirus and oropharyngeal cancer. N Engl J Med, 356(19), 1944–​ 1956. PMID: 17494927. Demko CA, Sawyer D, Slivka M, Smith D, and Wotman S. 2009. Prevalence of oral lesions in the dental office. Gen Dent, 57(5), 504–​ 509. PMID: 19903642. Downer MC, and Petti S. 2005. Leukoplakia prevalence estimate lower than expected. Evid Based Dent, 6(1), 12; author reply 13–​ 14. PMID: 15789044. Edefonti V, Hashibe M, Ambrogi F, et al. 2012. Nutrient-​based dietary patterns and the risk of head and neck cancer: a pooled analysis in the International Head and Neck Cancer Epidemiology consortium. Ann Oncol, 23(7), 1869–​1880. PMCID: PMC3387823. Edefonti V, Hashibe M, Parpinel M, et  al. 2015a. Vitamin E intake from natural sources and head and neck cancer risk: a pooled analysis in the International Head and Neck Cancer Epidemiology consortium. Br J Cancer, 113(1), 182–​192. PMID: 25989276. Edefonti V, Hashibe M, Parpinel M, et al. 2015b. Natural vitamin C intake and the risk of head and neck cancer: a pooled analysis in the International Head and Neck Cancer Epidemiology Consortium. Int J Cancer, 137(2), 448–​462. PMCID: PMC4428957. El-​ Zaatari ZM, Chami HA, and Zaatari GS. 2015. Health effects associated with waterpipe smoking. Tob Control, 24(Suppl 1), i31–​ i43. PMCID: PMC4345795. Engholm G, Ferlay J, Christensen N, Kejs AMT, Johannesen TB, Khan S, Leinonen M, Milter MC, Ólafsdóttir E, Petersen T, Stenz F, Storm HH. NORDCAN:  Cancer Incidence, Mortality, Prevalence and Survival in the Nordic Countries, Version 7.1 (07.2015). Association of the Nordic Cancer Registries. Danish Cancer Society. Available from http://​www. ancr.nu, accessed October 2015. Ferlay J, Bray F, Pisani P, Parkin DM. GLOBOCAN 2002. Cancer Incidence, Mortality and Prevalence Worldwide. IARC Cancer Base No. 5 Version 2.0. Lyon, France: IARCPress; 2004. Ferlay J, Soerjomataram I, Ervik M, et  al. 2013. GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality Worldwide:  IARC CancerBase No. 11 [Internet]. Available from: http://​globocan.iarc.fr. Ferlay J, Bray F, Steliarova-​Foucher E and Forman D. (2014). Cancer Incidence in Five Continents, CI5plus. IARC CancerBase No. 9. Lyon: International Agency for Research on Cancer; 2014. Available from: http://​ci5.iarc.fr, accessed October 2014. Forman D, Bray F, Brewster DH, et al. 2013. Cancer incidence in five continents, Vol. X (IARC Scientific Publication No. 164). Lyon: International Agency for Research on Cancer. Forrest J, Campbell P, Kreiger N, and Sloan M. 2008. Salivary gland cancer: an exploratory analysis of dietary factors. Nutr Cancer, 60(4), 469–​473. PMID: 18584480. Freedman ND, Abnet CC, Leitzmann MF, Hollenbeck AR, Schatzkin A. 2007a. Prospective investigation of the cigarette smoking-​head and neck cancer association by sex. Cancer, 110(7), 1593–​1601. PMID: 17724671. Freedman ND, Schatzkin A, Leitzmann MF, Hollenbeck AR, and Abnet CC. 2007b. Alcohol and head and neck cancer risk in a prospective study. Br J Cancer, 96(9), 1469–​1474. PMCID: PMC2360181. Freedman ND, Abnet CC, Caporaso NE, et al. 2015. Impact of changing US cigarette smoking patterns on incident cancer:  risks of 20 smoking-​ related cancers among the women and men of the NIH-​AARP cohort. Int J Epidemiol, 45(3), 846–​856. PMID: 26411408. Friedman GD, Asgari MM, Warton EM, Chan J, and Habel LA. 2012. Antihypertensive drugs and lip cancer in non-​ Hispanic whites. Arch Intern Med, 172(16), 1246–​1251. PMCID: PMC3809130. Fritz A, Percy C, Jack A, et al. (2012). International classification of diseases for oncology (3rd ed.). Geneva: World Health Organization.

Fu JY, Gao J, Zhang ZY, et  al. 2013. Tea consumption and the risk of oral cancer incidence:  a case-​control study from China. Oral Oncol, 49(9), 918–​922. PMID: 23731795. Galeone C, Edefonti V, Parpinel M, et al. 2015a. Folate intake and the risk of oral cavity and pharyngeal cancer: a pooled analysis within the International Head and Neck Cancer Epidemiology Consortium. Int J Cancer, 136(4), 904–​914. PMCID: PMC4262536. Galeone C, Tavani A, Pelucchi C, et al. 2010. Coffee and tea intake and risk of head and neck cancer: pooled analysis in the international head and neck cancer epidemiology consortium. Cancer Epidemiol Biomarkers Prev, 19(7), 1723–​1736. PMCID: PMC3047460. Galeone C, Turati F, Zhang ZF, et  al. 2015b. Relation of allium vegetables intake with head and neck cancers: evidence from the INHANCE consortium. Mol Nutr Food Res, 59(9), 1641–​1650. PMID: 26018663. Gandini S, Negri E, Boffetta P, La Vecchia C, and Boyle P. 2012. Mouthwash and oral cancer risk quantitative meta-​analysis of epidemiologic studies. Ann Agric Environ Med, 19(2), 173–​180. PMID: 22742785. Gaudet MM, Kitahara CM, Newton CC, et al. 2015. Anthropometry and head and neck cancer:a pooled analysis of cohort data. Int J Epidemiol, 44(2), 673–​681. PMCID: PMC4481608. Gaudet MM, Olshan AF, Chuang SC, et al. 2010. Body mass index and risk of head and neck cancer in a pooled analysis of case-​control studies in the International Head and Neck Cancer Epidemiology (INHANCE) Consortium. Int J Epidemiol, 39(4), 1091–​1102. PMCID: PMC2929351. Goodman MT, Saraiya M, Thompson TD, et al. 2015. Human papillomavirus genotype and oropharynx cancer survival in the United States of America. Eur J Cancer, 51(18), 2759–​2767. PMCID: PMC4666760. Greenberg RS, Haber MJ, Clark WS, et  al. 1991. The relation of socioeconomic status to oral and pharyngeal cancer. Epidemiology, 2(3), 194–​200. PMID: 2054401. Grégoire V, Langendijk JA, and Nuyts S. 2015. Advances in radiotherapy for head and neck cancer. J Clin Oncol, 33(29), 3277–​3284. PMID: 26351354. Guha N, Warnakulasuriya S, Vlaanderen J, and Straif K. 2014. Betel quid chewing and the risk of oral and oropharyngeal cancers: a meta-​analysis with implications for cancer control. Int J Cancer, 135(6), 1433–​1443. PMID: 24302487. Gupta PC, and Warnakulasuriya S. 2002. Global epidemiology of areca nut usage. Addict Biol, 7(1), 77–​83. PMID: 11900626. Hamdi Cherif Mokhtar et al. Cancer in Setif 1986-​2010, Algeria. Incidences, trends, survival. Report available from http://​www.registre-​cancer-​setif. com, accessed September 2013. Hashibe M, Brennan P, Benhamou S, et  al. 2007. Alcohol drinking in never users of tobacco, cigarette smoking in never drinkers, and the risk of head and neck cancer:  pooled analysis in the International Head and Neck Cancer Epidemiology Consortium. J Natl Cancer Inst, 99(10), 777–​789. PMID: 17505073. Hashibe M, Brennan P, Chuang SC, et al. 2009. Interaction between tobacco and alcohol use and the risk of head and neck cancer: pooled analysis in the International Head and Neck Cancer Epidemiology Consortium. Cancer Epidemiol Biomarkers Prev, 18(2), 541–​550. PMCID: PMC3051410. Hashibe M, Galeone C, Buys SS, et al. 2015. Coffee, tea, caffeine intake, and the risk of cancer in the PLCO cohort. Br J Cancer, 113(5), 809–​816. PMID: 26291054. Hashibe M, Hunt J, Wei M, Buys S, Gren L, and Lee YC. 2013. Tobacco, alcohol, body mass index, physical activity, and the risk of head and neck cancer in the prostate, lung, colorectal, and ovarian (PLCO) cohort. Head Neck, 35(7), 914–​922. PMID: 22711227. Hashibe M, Jacob BJ, Thomas G, et al. 2003. Socioeconomic status, lifestyle factors and oral premalignant lesions. Oral Oncol, 39(7), 664–​ 671. PMID: 12907205. Hashibe M, McKay JD, Curado MP, et  al. 2008. Multiple ADH genes are associated with upper aerodigestive cancers. Nat Genet, 40(6), 707–​709. PMID: 18500343. Hashim D, Sartori S, Brennan P, et al. 2016. The role of oral hygiene in head and neck cancer:  results from International Head and Neck Cancer Epidemiology (INHANCE) Consortium. Ann Oncol, 27(8), 1619–​1625. PMID: 27234641. Hassona Y, Scully C, Almangush A, Baqain Z, and Sawair F. 2014. Oral potentially malignant disorders among dental patients: a pilot study in Jordan. Asian Pac J Cancer Prev, 15(23), 10427–​10431. PMID: 25556487. Herrero R, Castellsagué X, Pawlita M, et al. 2003. Human papillomavirus and oral cancer: the International Agency for Research on Cancer multicenter study. J Natl Cancer Inst, 95(23), 1772–​1783. PMID: 14652239. Hildebrand JS, Patel AV, McCullough ML, et al. 2013. Coffee, tea, and fatal oral/​pharyngeal cancer in a large prospective US cohort. Am J Epidemiol, 177(1), 50–​58. PMID: 23230042. Hobbs CG, Sterne JA, Bailey M, Heyderman RS, Birchall MA, and Thomas SJ. 2006. Human papillomavirus and head and neck cancer:  a

 57



Oral Cavity, Oropharynx, Lip, and Salivary Glands

systematic review and meta-​analysis. Clin Otolaryngol, 31(4), 259–​266. PMID: 16911640. Horn-​ Ross PL, Ljung BM, and Morrow M. 1997. Environmental factors and the risk of salivary gland cancer. Epidemiology, 8(4), 414–​ 419. PMID: 9209856. Horn-​Ross PL, Morrow M, and Ljung BM. 1999. Menstrual and reproductive factors for salivary gland cancer risk in women. Epidemiology, 10(5), 528–​530. PMID: 10468426. Howlader N, Noone AM, Krapcho M, et  al. 2016. SEER cancer statistics review, 1975–​2013, National Cancer Institute. Bethesda, MD, http://​seer. cancer.gov/​csr/​1975_​2013/​, based on November 2015 SEER data submission, posted to the SEER website April 2016. Huang YH, Zhang ZF, Tashkin DP, Feng B, Straif K, and Hashibe M. 2015. An epidemiologic review of marijuana and cancer:  an update. Cancer Epidemiol Biomarkers Prev, 24(1), 15–​31. PMCID: PMC4302404. Hübbers CU, and Akgül B. 2015. HPV and cancer of the oral cavity. Virulence, 6(3), 244–​248. PMCID: PMC4601238. IARC. 1988. Alcohol drinking, Volume 44. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Lyon: IARC Press, World Health Organization. IARC. 1998. Tobacco smoking, Volume 38. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Lyon: IARC Press, World Health Organization. IARC. 2007. Smokeless tobacco and some tobacco-​specific N-​nitrosamines, Volume 89. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Lyon: IARC Press, World Health Organization. IARC. 2010. Alcohol consumption and ethyl carbamate, Volume 96. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Lyon: IARC Press, World Health Organization. IARC. 2012a. Personal habits and indoor combustions, Volume 100E. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Lyon: IARC Press, World Health Organization. IARC. 2012b. Biological agents, Volume 100B. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Lyon: IARC Press, World Health Organization. IARC. 2012c. Radiation, Volume 100D. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Lyon:  IARC Press, World Health Organization. IARC. 2016. Some drugs and herbal products, Volume 108. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Lyon: IARC Press, World Health Organization. Ide R, Fujino Y, Hoshiyama Y, and M et al. 2007. A prospective study of green tea consumption and oral cancer incidence in Japan. Ann Epidemiol, 17(10), 821–​826. PMID: 17606381. Institute of Medicine (US) Committee on Asbestos. 2006. Asbestos: selected cancers. Washington, DC: The National Academies Press. Jayaprakash V, Natarajan KK, Moysich KB, et al. 2008. Wood dust exposure and the risk of upper aero-​digestive and respiratory cancers in males. Occup Environ Med, 65(10), 647–​654. PMID: 18182588. Johnson S, Corsten MJ, McDonald JT, and J C. 2010. Socio-​economic factors and stage at presentation of head and neck cancer patients in Ottawa, Canada:  a logistic regression analysis. Oral Oncol, 46(5), 366–​ 368. PMID: 20308010. Kaugars GE, Pillion T, Svirsky JA, Page DG, Burns JC, and Abbey LM. 1999. Actinic cheilitis: a review of 152 cases. Oral Surg Oral Med Oral Pathol Oral Radiol Endod, 88(2), 181–​186. PMID: 10468463. Keck MK, Zuo Z, Khattri A, et al. 2015. Integrative analysis of head and neck cancer identifies two biologically distinct HPV and three non-​HPV subtypes. Clin Cancer Res, 21(4), 870–​881. PMID: 25492084. Kreimer AR, Struyf F, Del Rosario-​Raymundo MR, et al; Costa Rica Vaccine Trial and PATRICIA study groups. 2015. Efficacy of fewer than three doses of an HPV-​16/​18 AS04-​adjuvanted vaccine: combined analysis of data from the Costa Rica Vaccine and PATRICIA trials. Lancet Oncol, 16(7):775–​786. PMCID: PMC4498478. Kumar S, Debnath N, Ismail MB, et al. 2015. Prevalence and risk factors for oral potentially malignant disorders in Indian population. Adv Prev Med, 2015, Article ID 208519, 7 pages. PMCID: PMC4548071. Kuo CL, Chen YT, Shiao AS, Lien CF, and Wang SJ. 2015. Acid reflux and head and neck cancer risk:  a nationwide registry over 13  years. Auris Nasus Larynx, 42(5), 401–​405. PMID: 25998850. Lachenmeier DW, Gumbel-​Mako S, Sohnius EM, et al. 2009. Salivary acetaldehyde increase due to alcohol-​containing mouthwash use: a risk factor for oral cancer. Int J Cancer, 125(3), 730–​735. PMID: 19444911. Lagiou P, Talamini R, Samoli E, et al. 2009. Diet and upper-​aerodigestive tract cancer in Europe: the ARCAGE study. Int J Cancer, 124(11), 2671–​2676. PMID: 19230023. Lam TK, Cross AJ, Freedman N, et  al. 2011. Dietary fiber and grain consumption in relation to head and neck cancer in the NIH-​ AARP

575

Diet and Health Study. Cancer Causes Control, 22(10):1405–​1414. PMCID: PMC3215506. La Vecchia C. 2009. Mouthwash and oral cancer risk: an update. Oral Oncol, 45(3), 198–​200. PMID: 18952488. Lee YC, Boffetta P, Sturgis EM, et  al. 2008. Involuntary smoking and head and neck cancer risk: pooled analysis in the International Head and Neck Cancer Epidemiology Consortium. Cancer Epidemiol Biomarkers Prev, 17(8), 1974–​1981. PMCID: PMC2561190. Lee YC, Hashibe M. 2014. Tobacco, alcohol, and cancer in low and high income countries. Ann Glob Health, 80(5):378–​383. PMID: 25512153. Lee YC, Marron M, Benhamou S, et al. 2009. Active and involuntary tobacco smoking and upper aerodigestive tract cancer risks in a multicenter case-​ control study. Cancer Epidemiol Biomarkers Prev, 18(12), 3353–​3361. PMCID: PMC4085143. Leitzmann MF, Koebnick C, Freedman ND, et al. 2008. Physical activity and head and neck cancer risk. Cancer Causes Control, 19(10), 1391–​1399. PMCID: PMC3507337. Leoncini E, Ricciardi W, Cadoni G, et  al. 2014. Adult height and head and neck cancer: a pooled analysis within the INHANCE Consortium. Eur J Epidemiol, 29(1), 35–​48. PMCID: PMC4122122. Leoncini E, Edefonti V, Hashibe M, et  al. 2015. Carotenoid intake and head and neck cancer: a pooled analysis in the International Head and Neck Cancer Epidemiology Consortium. Eur J Epidemiol, 31(4), 369–​383. PMID: 25930054 Li S, Lee YA, Li Q, et al. 2015. Oral lesions, chronic diseases and the risk of head and neck cancer. Oral Oncol, 51(12), 1082–​1087. PMID: 26526128. Li X, and Hemminki K. 2003. Familial upper aerodigestive tract cancers: incidence trends, familial clustering and subsequent cancers. Oral Oncol, 39(3), 232–​239. PMID: 12618195. Li WQ, Park Y, Wu JW, et al. 2014. Index-​based dietary patterns and risk of head and neck cancer in a large prospective study. Am J Clin Nutr. 99(3), 559–​566. PMCID: PMC3927689. López-​Pintor RM, Hernández G, de Arriba L, and de Andrés A. 2011. Lip cancer in renal transplant patients. Oral Oncol, 47(1), 68–​71. PMID: 21112239. Lubin JH, Gaudet MM, Olshan AF, et al. 2010. Body mass index, cigarette smoking, and alcohol consumption and cancers of the oral cavity, pharynx, and larynx: modeling odds ratios in pooled case-control data. Am J Epidemiol. 171(12), 1250-1261. doi: 10.1093/aje/kwq088.PubMed PMID: 20494999; PubMed Central PMCID: PMC2915496. Lubin JH, Purdue M, Kelsey K, et al. 2009. Total exposure and exposure rate effects for alcohol and smoking and risk of head and neck cancer:  a pooled analysis of case-​control studies. Am J Epidemiol, 170(8), 937–​ 947. PMC2800266 PMID: 19745021. Macfarlane TV, Macfarlane GJ, Thakker NS, et al. 2012. Role of medical history and medication use in the aetiology of upper aerodigestive tract cancers in Europe:  the ARCAGE study. Ann Oncol, 23(4), 1053–​1060. PMID: 21828376. Madani AH, Dikshit M, Bhaduri D, Jahromi AS, and Aghamolaei T. 2010. Relationship between selected socio-​ demographic factors and cancer of oral cavity:  a case control study. Cancer Inform, 9, 163–​168. PMCID: PMC2935817. Marks MA, Chaturvedi AK, Kelsey K, et al. 2014. Association of marijuana smoking with oropharyngeal and oral tongue cancers:  pooled analysis from the INHANCE consortium. Cancer Epidemiol Biomarkers Prev, 23(1), 160–​171. PMCID: PMC3947141. Marron M, Boffetta P, Moller H, et al. 2012. Risk of upper aerodigestive tract cancer and type of alcoholic beverage:  a European multicenter case-​ control study. Eur J Epidemiol, 27(7), 499–​517. PMID: 22692594. Marron M, Boffetta P, Zhang ZF, et  al. 2010. Cessation of alcohol drinking, tobacco smoking and the reversal of head and neck cancer risk. Int J Epidemiol, 39(1), 182–​196. PMCID: PMC2817090. Mayne ST, Cartmel B, Kirsh V, and Goodwin WJ Jr. 2009. Alcohol and tobacco use prediagnosis and postdiagnosis, and survival in a cohort of patients with early stage cancers of the oral cavity, pharynx, and larynx. Cancer Epidemiol Biomarkers Prev, 18(12), 3368–​ 3374. PMCID: PMC2789339. Mayne ST, Morse DE, and Winn DE. 2006. Cancers of the oral cavity and pharynx. In D Schottenfeld & JF Fraumeni (Eds.), Cancer epidemiology and prevention (3rd ed.). New York: Oxford University Press. McCredie M, Coates M, and Grulich A. 1994. Cancer incidence in migrants to New South Wales (Australia) from the Middle East, 1972–​1991. Cancer Causes Control, 5(5), 414–​421. PMID: 7999963. McKay JD, Truong T, Gaborieau V, et  al. 2011. A genome-​wide association study of upper aerodigestive tract cancers conducted within the INHANCE consortium. PLoS Genet, 7(3), e1001333. PMCID: PMC3060072. Mirghani H, Amen F, Moreau F, and Lacau St Guily J. 2015. Do high-​risk human papillomaviruses cause oral cavity squamous cell carcinoma. Oral Oncol, 51(3), 229–​236. PMID: 25488179.

576

576

Part IV:  Cancers by Tissue of Origin

Moles DR, Fedele S, Speight PM, Porter SR, and dos Santos Silva I. 2008. Oral and pharyngeal cancer in South Asians and non-​South Asians in relation to socioeconomic deprivation in South East England. Br J Cancer, 98(3), 633–​635. PMCID: PMC2243164. Mork J, Lie AK, Glattre E, et al. 2001. Human papillomavirus infection as a risk factor for squamous-​cell carcinoma of the head and neck. N Engl J Med, 344(15), 1125–​1131. PMID: 11297703. Murti PR, Bhonsle RB, Pindborg JJ, Daftary DK, Gupta PC, and Mehta FS. 1985. Malignant transformation rate in oral submucous fibrosis over a 17-​year period. Community Dent Oral Epidemiol, 13(6), 340–​341. PMID: 3866655. Ndiaye C, Mena M, Alemany L, et al. 2014. HPV DNA, E6/​E7 mRNA, and p16INK4a detection in head and neck cancers: a systematic review and meta-​analysis. Lancet Oncol, 15(12), 1319–​1331. PMID: 25439690. Nicolau SG, and Balus L. 1964. Chronic actinic cheilitis and cancer of the lower lip. Br J Dermatol, 76, 278–​289. PMID: 14154810. Nicolotti N, Chuang SC, Cadoni G, et al. 2011. Recreational physical activity and risk of head and neck cancer: a pooled analysis within the international head and neck cancer epidemiology (INHANCE) Consortium. Eur J Epidemiol, 26(8), 619–​628. PMID: 21842237. Öhman J, Rexius H, Mjörnstedt L, et al. 2015. Oral and lip cancer in solid organ transplant patients:  a cohort study from a Swedish Transplant Centre. Oral Oncol, 51(2), 146–​150. PMID: 25497104. Perea-​Milla López E, Miñarro-​Del Moral RM, Martínez-​García C, et al. 2003. Lifestyles, environmental and phenotypic factors associated with lip cancer: a case-​control study in southern Spain. Br J Cancer, 88(11), 1702–​ 1707. PMCID: PMC2377138. Petti S. 2003. Pooled estimate of world leukoplakia prevalence: a systematic review. Oral Oncol, 39(8), 770–​780. PMID: 13679200. Pogoda JM, and Preston-​Martin S. 1996. Solar radiation, lip protection, and lip cancer risk in Los Angeles County women (California, United States). Cancer Causes Control, 7(4), 458–​463. PMID: 8813434. Purdue MP, Hashibe M, Berthiller J, et al. 2009. Type of alcoholic beverage and risk of head and neck cancer: a pooled analysis within the INHANCE Consortium. Am J Epidemiol, 169(2), 132–​142. PMCID: PMC2727255. Radoï L, Paget-​Bailly S, Guida F, et al. 2013a. Family history of cancer, personal history of medical conditions and risk of oral cavity cancer in France: the ICARE study. BMC Cancer, 13, 560. PMCID: PMC4219459 Radoï L, Paget-​Bailly S, Menvielle G, et al. 2013b. Tea and coffee consumption and risk of oral cavity cancer:  results of a large population-​based case-​control study, the ICARE study. Cancer Epidemiol, 37(3), 284–​289. PMID: 23453554. Reagan-​Steiner S, Yankey D, Jeyarajah J, et al. 2015 National, regional, state, and selected local area vaccination coverage among adolescents aged 13–​ 17 years: United States, 2014. MMWR Morb Mortal Wkly Rep, 64(29), 784–​792. PMID: 26225476. Richiardi L, Corbin M, Marron M, et al. 2012. Occupation and risk of upper aerodigestive tract cancer:  the ARCAGE study. Int J Cancer, 130(10), 2397–​2406. PMID: 21671472. Sankaranarayanan R, Ramadas K, Thara S, et  al. 2013. Long term effect of visual screening on oral cancer incidence and mortality in a randomized trial in Kerala, India. Oral Oncol, 49(4), 314–​321. PMID: 23265945. Sapkota A, Zaridze D, Szeszenia-​Dabrowska N, et al. 2013. Indoor air pollution from solid fuels and risk of upper aerodigestive tract cancers in central and eastern Europe. Environ Res, 120, 90–​95. PMID: 23092716. Sato F, Oze I, Kawakita D, et al. 2011. Inverse association between toothbrushing and upper aerodigestive tract cancer risk in a Japanese population. Head Neck, 33(11), 1628–​1637. PMID: 21259377. SEER. 2005. Surveillance, epidemiology, and end results program (http://​ www.seer.cancer.gov) SEER*stat database:  incidence—​ racial ethnic mono, SEER 18 (excl AZ,AK,RG) Limited Use, Nov 2005 for detailed API ethnicities only (1998–​2002) , National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch, released April 2006, based on the November 2005 submission. SEER. 2016. Surveillance, Epidemiology, and End Results Program (www. seer.cancer.gov) SEER*Stat Database:  Incidence—​ SEER 18 Regs Research Data + Hurricane Katrina Impacted Louisiana Cases, Nov 2015 Sub (2000–​2013) Katrina/​Rita Population Adjustment—​Linked to County Attributes—​ Total U.S., 1969–​ 2013 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2016, based on the November 2015 submission. Shanmugham JR, Zavras AI, Rosner BA, and Giovannucci EL. 2010. Alcohol-​ folate interactions in the risk of oral cancer in women:  a prospective

cohort study. Cancer Epidemiol Biomarkers Prev, 19(10), 2516–​2524. PMCID: PMC2952069. Skálová A, Kašpírková J, Andrle P, Hostička L, and Vaneček T. 2013. Human papillomaviruses are not involved in the etiopathogenesis of salivary gland tumors. Cesk Patol, 49(2), 72–​75. PMID: 23641711. Spitz MR, Fueger JJ, Goepfert H, and Newell GR. 1990. Salivary gland cancer: a case-​control investigation of risk factors. Arch Otolaryngol Head Neck Surg, 116(10), 1163–​1166. PMID: 2206501. Spitz MR, Tilley BC, Batsakis JG, Gibeau JM, and Newell GR. 1984. Risk factors for major salivary gland carcinoma:  a case-​comparison study. Cancer, 54(9), 1854–​1859. PMID: 6478421. Stott-​Miller M, Chen C, Chuang SC, et al. 2012. History of diabetes and risk of head and neck cancer: a pooled analysis from the international head and neck cancer epidemiology consortium. Cancer Epidemiol Biomarkers Prev, 21(2), 294–​304. PMCID: PMC3275674. Swerdlow AJ, Marmot MG, Grulich AE, and Head J. 1995. Cancer mortality in Indian and British ethnic immigrants from the Indian subcontinent to England and Wales. Br J Cancer, 72(5), 1312–​ 1319. PMCID: PMC2033952. TCGA. 2015. Comprehensive genomic characterization of head and neck squamous cell carcinomas. Nature, 517(7536), 576–​ 582. PMCID: PMC4311405. Teerlink CC, Albright FS, Lins L, and Cannon-​Albright LA. 2012. A comprehensive survey of cancer risks in extended families. Genet Med, 14(1), 107–​114. PMID: 22237439. Toporcov TN, Znaor A, Zhang ZF, et al. 2015. Risk factors for head and neck cancer in young adults: a pooled analysis in the INHANCE consortium. Int J Epidemiol, 44(1), 169–​185. PMCID: PMC4339764. Turati F, Galeone C, La Vecchia C, Garavello W, and Tavani A. 2011. Coffee and cancers of the upper digestive and respiratory tracts: meta-​analyses of observational studies. Ann Oncol, 22(3), 536–​544. PMID: 20943597. USCS. 2016. US Cancer Statistics Working Group (USCS). United States cancer statistics: 1999–​2013 incidence and mortality web-​based report. Atlanta:  US Department of Health and Human Services, Centers for Disease Control and Prevention and National Cancer Institute. Available at: www.cdc.gov/​uscs. USPTF. 2015. Final update summary: oral cancer: screening. US Preventive Services Task Force. July 2015. Available at:  http://​www.uspreventiveservicestaskforce.org/ ​ P age/ ​ D ocument/ ​ U pdateSummaryFinal/​ oral-​cancer-​screening1. Walsh T, Liu JL, Brocklehurst P, et al. 2013. Clinical assessment to screen for the detection of oral cavity cancer and potentially malignant disorders in apparently healthy adults. Cochrane Database Syst Rev, 11, CD010173. PMID: 24258195. Wang LE, Li G, Sturgis EM, and Wei Q. 2010. Host susceptibility and molecular epidemiology. In AF Olshan (Ed.), Epidemiology, pathogenesis, and prevention of head and neck cancer (pp. 155–​186). New York, New York: Springer. Warnakulasuriya KA, Johnson NW, Linklater KM, and Bell J. 1999. Cancer of mouth, pharynx and nasopharynx in Asian and Chinese immigrants resident in Thames regions. Oral Oncol, 35(5), 471–​475. PMID: 10694946. Warnakulasuriya S. 2002. Areca nut use following migration and its consequences. Addict Biol, 7(1), 127–​132. PMID: 11900632. Warnakulasuriya S, Johnson NW, and van der Waal I. 2007. Nomenclature and classification of potentially malignant disorders of the oral mucosa. J Oral Pathol Med, 36(10), 575–​580. PMID: 17944749. Warnakulasuriya S, Kovacevic T, Madden P, et  al. 2011. Factors predicting malignant transformation in oral potentially malignant disorders among patients accrued over a 10-​year period in South East England. J Oral Pathol Med, 40(9), 677–​683. PMID: 21762430. Warnakulasuriya S. 2011. Waterpipe smoking, oral cancer and other oral health effects. Evid Based Dent, 12(2):44–​45. PMID: 21701545. Warnakulasuriya S, Ariyawardana A. 2016. Malignant transformation of oral leukoplakia: a systematic review of observational studies. J Oral Pathol Med, 45(3), 155–​166. PMID: 26189354. WCRF (World Cancer Research Fund)/​American Institute for Cancer Research. 2007. Food, nutrition, physical activity, and the prevention of cancer: a global perspective. Washington, DC: AICR. WHO (World Health Organization). 2014. Global status report on alcohol and health. Geneva, Switzerland: WHO Press. Williams RR, and Horm JW. 1977. Association of cancer sites with tobacco and alcohol consumption and socioeconomic status of patients: interview study from the Third National Cancer Survey. J Natl Cancer Inst, 58(3), 525–​547. PMID: 557114.

 57



Oral Cavity, Oropharynx, Lip, and Salivary Glands

Wyss A, Hashibe M, Chuang SC, et al. 2013. Cigarette, cigar, and pipe smoking and the risk of head and neck cancers: pooled analysis in the International Head and Neck Cancer Epidemiology Consortium. Am J Epidemiol, 178(5), 679–​690. PMCID: PMC3755640. Wyss A, Hashibe M, Chuang SC, et al. Smokeless tobacco use and the risk of head and neck cancer: pooled analysis of US studies in the INHANCE Consortium. Submitted. PMCID: PMC3755640. Xu L, Tang H, Chen DW, El-​Naggar AK, Wei P, and Sturgis EM. 2015. Genome-​wide association study identifies common genetic variants associated with salivary gland carcinoma and its subtypes. Cancer, 121(14), 2367–​2374. PMCID: PMC4564998.

577

Yanik EL, Katki HA, Silverberg MJ, Manos MM, Engels EA, and Chaturvedi AK. 2015. Leukoplakia, oral cavity cancer risk, and cancer survival in the U.S.  elderly. Cancer Prev Res (Phila), 8(9), 857–​ 863. PMCID: PMC4560597. Zeng XT, Leng WD, Zhang C, Liu J, Cao SY, and Huang W. 2015. Meta-​ analysis on the association between toothbrushing and head and neck cancer. Oral Oncol, 51(5), 446–​ 451. PMID: 25753558. Zheng W, Shu XO, Ji BT, and Gao YT. 1996. Diet and other risk factors for cancer of the salivary glands:a population-​based case-​control study. Int J Cancer, 67(2), 194–​198. PMID: 8760587.

578

 579

30 Esophageal Cancer WILLIAM J. BLOT AND ROBERT E. TARONE

OVERVIEW Cancer of the esophagus is the eighth most common malignancy worldwide in terms of incident cases, and the sixth most common for cancer deaths. The two main histopathologic subtypes, esophageal squamous cell carcinoma (ESCC) and esophageal adenocarcinoma (EADC), have strikingly different clinical and epidemiologic features. ESCC occurs throughout the esophagus and is the most common histologic subtype globally; over 90% of cases in the traditionally high-​risk regions of Eastern Asia and Eastern and Southern Africa are ESCC. The incidence of ESCC is decreasing worldwide; in some high-​risk areas in Asia the decrease was preceded by economic development and improvements in diet, whereas in high-​income countries the decrease followed reductions in cigarette smoking. In contrast, the incidence of EADC continues to increase in many high-​and middle-​income countries, especially among white men. EADC develops in the lower third of the esophagus, primarily because of gastroesophageal reflux disease (GERD) and obesity. This chapter will review the dynamic changes in the two histologic subtypes of esophageal cancer and will consider their natural history, molecular pathogenesis, etiology, and the potential for primary and secondary prevention.

INTRODUCTION The two main histopathologic subtypes of esophageal cancer are squamous cell carcinomas (ESCC) and adenocarcinomas (EADC). Striking differences exist in the geographic distribution, temporal trends, natural history, and etiology of these two histologic subtypes. This chapter will focus on the two most common forms of esophageal cancer and will emphasize research advances that have occurred since the previous edition of Cancer Epidemiology and Prevention (Blot et al., 2006). We will not discuss other rare malignancies of the esophagus, such as small cell carcinoma, melanoma, and sarcoma. Collectively, the latter subtypes account for less than 1%–​2% of all esophageal cancers (National Cancer Institute, 2015). They are discussed in Chapters 28, 43, and 57 of this volume.

CLASSIFICATION BY ANATOMIC SITE The International Classification of Diseases for Oncology, version ICD-​O-​2/​3, classifies esophageal cancers as C15.0–​15.9, based on the anatomic location of the tumor. Squamous cell carcinomas occur throughout the esophagus, although somewhat more often in the middle third in high-​income countries. For example, among cases of ESCC reported to the US Surveillance, Epidemiology, and End Results (SEER) registries between 2000 and 2012, 21% were in the upper third, 46% in the middle third, and 33% in the lower third (National Cancer Institute, 2015). In contrast, the great majority (89%) of EADC in SEER during this period involved the lower third of the esophagus, often just above the gastric-​esophageal (GE) junction; only 1.5% arose in the upper third, with 9.5% in the middle third.

DISEASE BURDEN According to the International Agency for Research on Cancer (IARC), an estimated 456,000 new cases and 400,000 deaths from esophageal

cancer occurred in 2012 (Ferlay et al., 2013). These estimates combine the two major subtypes, which together represent the eighth most common malignancy in terms of incident cases and the sixth most common for cancer deaths. ESCC predominates globally, accounting for over 90% of cases in the extremely high-​risk regions of Eastern Asia and Eastern and Southern Africa. The incidence rate of ESCC varies widely within and among countries, as discussed later in this chapter in the section on descriptive epidemiology. In certain high-​risk regions in China and Iran, ESCC is reported to be the most common form of cancer (Blot et al., 2006).

NATURAL HISTORY Although both histologic subtypes of esophageal cancer usually develop in areas of chronic inflammation, their precursor lesions and natural history differ. ESCCs arise from the multilayered epithelium comprising the innermost lining of the esophagus, and dysplasia is an established precursor of ESCC (Blot et  al., 2006; Taylor et  al., 2013). With the exception of high-​risk geographic regions, relatively few dysplasia cases undergo malignant transformation into in situ or invasive cancer, but ESCC risk rises rapidly with increasing severity of dysplasia (Taylor et al., 2013). In a high-​risk area of China, only 8% of ESCC arose in people without a prior diagnosis of esophageal dysplasia (Taylor et al., 2013). In contrast, adenocarcinomas of the esophagus arise from glandular columnar tissue that has displaced squamous cells in the lining of the lower esophagus. Metaplastic columnar tissue known as Barrett’s esophagus is a well-​established precursor lesion for EADC (Wild and Hardie, 2003). Nearly all EADCs are thought to arise from Barrett’s esophagus (Chen and Yang, 2001). This abnormal columnar tissue develops during the healing phase of severe, recurrent gastroesophageal reflux and supplants damaged squamous epithelium (Rustgi and El-​Serag, 2014; Verbeek et al., 2014; Yousef et al., 2008). Both Barrett’s esophagus and EADC occur in the same areas of the lower esophagus and GE junction. Likewise, their epidemiologic characteristics are similar. Both have increased in prevalence during the last 30  years in Western countries, are more common in men than women and in Caucasians than African Americans or Asians, and are associated with obesity and tobacco smoking (Blot et al., 2006; Runge et al., 2015; Schneider and Corley, 2015; Spechler, 2013). For squamous cell carcinomas, the strongest evidence connecting the severity of the premalignant lesions to cancer risk was seen in Linxian, China, where in a 13-​year follow-​up, 50% of individuals with moderate and 74% of those with severe esophageal dysplasia developed ESCC (Taylor et al., 2013). Not surprisingly, it has been easier to study the risk of progression from Barrett’s esophagus to EADC in high-​income countries than to monitor progression to ESCC in remote areas of Asia or Africa. Older studies estimated the annual risk of progression to EADC to be as high as 1%; more recent data indicate an overall progression rate of less than 0.5% (Falk, 2015; Gatenby et al., 2014; Spechler, 2013). Patients with Barrett’s esophagus but with no dysplasia have an annual probability of developing EADC of approximately 0.25% per year; this increases to about 6% per year for patients with Barrett’s esophagus and high-​grade dysplasia (Spechler, 2013).

579

580

580

PART IV:  Cancers by Tissue of Origin

TREATMENT, PROGNOSIS, AND SURVIVAL Treatment successes have been limited for both histologic subtypes of esophageal cancer, partly because most tumors are diagnosed at an advanced stage. The percentage of esophageal cancers diagnosed at a local stage in the United States has remained stubbornly under 30% for four decades (Hur et  al., 2013). Five-​year relative survival was less than 10% prior to 1980 (Blot et al., 2006; Hur et al., 2013), and relatively small improvements have occurred since then. For cases of esophageal cancer diagnosed from 2000 through 2012 in the United States, the 5-​year relative survival for ESCC was 15%; that for EADC was 19% (National Cancer Institute, 2015). For both cell types, relative survival was slightly better among whites than blacks. For ESCC, survival was slightly better among men than women; no sex difference was observed for EADC (National Cancer Institute, 2015).

DESCRIPTIVE EPIDEMIOLOGY International Variation Table 30–1 shows esophageal cancer incidence rates by sex for various geographic regions of the world in 2012, along with the estimated percentage contributed by squamous cell cancers. Esophageal cancer exhibits great international variation, with more than 20-​fold differences in rates between world regions in both sexes (Ferlay et  al., 2015). Highest overall rates in both sexes are observed in Eastern Asia, Eastern and Southern Africa, and South-​Central Asia, and the lowest rates are observed in Western Africa and Central America (Table 30–1). Mortality data on esophageal cancer usually cannot be partitioned by subsite (i.e., into ESCC or EADC). The geographic variation is even greater if histologic subtype is considered in addition to the overall incidence rate. Table 30–1 shows that in regions with the highest overall esophageal cancer incidence rates, ESCC predominates (> 90%), whereas in high-​income Western countries, which have relatively low overall incidence rates, the percentage of ESCC is low, especially among men. Approximately 96% of cases among men in Eastern Asia are ESCC, corresponding to an annual incidence rate for this subtype of approximately 16.2 per 100,000. In contrast, only 33% of cases among men in North America are ESCC, corresponding to an incidence rate of about 1.8 per 100,000. Low-​and middle-​income regions have a high percentage of ESCC, regardless of

their total incidence rates of esophageal cancer, whereas high-​income countries in North America, Northern Europe, Australia, and New Zealand have a low percentage of ESCC and relatively low overall incidence rates of esophageal cancer (Arnold et al., 2015). Countries with the highest overall esophageal cancer incidence rates (per 100,000 annually) for men and women combined include China (10.9), Mongolia (15.5), Malawi (22.9), Kenya (16.5), Uganda (15.9), Burundi (12.2), Lesotho (13.9), Turkmenistan (18.5), Tajikistan (13.6), and Bangladesh (11.7) (Ferlay et al., 2013). Because the incidence rates are highly variable within countries, however, the actual geographic variation is much larger than is reflected by national averages. For example, the age-​standardized incidence rate in some parts of North-​Central China exceeds 100 per 100,000 annually (Blot et al., 2006). While squamous cell carcinomas are the predominant histologic subtype throughout most of the world, adenocarcinomas are more common among men in high-​income countries (Table 30–1; Arnold et al., 2015). The seven countries with the highest incidence rates (per 100,000) of EADC among men include the United Kingdom (7.2), The Netherlands (7.1), Ireland (5.4), New Zealand (4.0), Iceland (3.9), the United States (3.6), and Belgium (3.5) (Arnold et al., 2015).

Temporal Trends By the late 1980s and early 1990s, it had become apparent that there was a major shift in esophageal cancer occurrence in the United States and certain other high-​income Western nations. Incidence rates for ESCC, which had previously been rising as a result of cigarette smoking patterns, began to plateau, or even fall, while incidence rates of EADC began to increase (Blot et  al., 1991, 2006; Bosetti et  al., 2008). The trends in incidence by histologic subtype, race, and sex are shown for the United States in Figure 30–1. The incidence rate of ESCC has decreased by over 40% since the early 1980s in male and female whites and African Americans. In contrast, the incidence rate of adenocarcinoma has increased over the past three decades in all subgroups, but especially among white men. The proportion of esophageal cancers that are adenocarcinomas rose steadily in white men from less than 15% around 1970, to 34% by the mid-​1980s, to 60% by 1992–​1994, and currently to 80% (Blot et al., 2006; National Cancer Institute, 2015). The increase in the incidence rate of EADC appears to have slowed, and possibly ended, in the United States (Figure 30–1; Hur et al., 2013), although mathematical models suggest that EADC rates may continue to rise slowly until 2030 (Kong et al., 2014).

Table 30–1. International Esophageal Cancer Incidence Rates by Sex Incidence Rate* Country North America Central America South America Western Europe Central and Eastern Europe Northern Europe Southern Europe Australia and New Zealand Eastern Asia Southeast Asia South-​Central Asia Western Asia Eastern Africa Western Africa Southern Africa Northern Africa

Squamous Cell Cancer (%)

Male

Female

Male

Female

5.4 1.7 7.0 6.8 5.6

1.1 0.6 2.0 1.6 0.8

33 76 78 62 85

62 77 78 72 81

8.1 3.2 5.4

2.7 0.6 1.7

31 72 34

61 77 70

16.9 3.6 6.5 2.9 11.9 0.8 13.7 2.4

5.4 1.0 3.9 2.1 7.8 0.4 6.7 1.5

96 92 91 71 94 96 93 75

96 91 94 90 95 95 90 84

* Rates per 100,000 person-​years, age-​adjusted using World Standard. Source: Globocan 2012 (Ferlay et al., 2013); Arnold et al. (2015).

Gender The incidence of esophageal cancer is consistently higher in men than women for both ESCC and EADC in virtually all populations studied. Table 30–2 shows incidence rates in the United States for cases diagnosed in 2003–​2012 by gender, race/​ethnicity, and histology. The male:female ratio is higher for adenocarcinoma than for squamous cell carcinoma. Barrett’s esophagus is more common among men than women, but differences in known risk factors do not appear to explain the male excess of EADC, with limited evidence that hormonal factors may play a role (Xie and Lagergren, 2016). The male:female ratios vary widely worldwide for both EADC and ESCC, with male rates almost always exceeding female rates (Arnold et al., 2015). In Northern Africa and Western Asia, however, regions where rates are more similar by sex (Table 30–1), some countries have reported higher ESCC rates among women than among men (Al-​Samawi and Aulaqi, 2014; Arnold et al., 2015; Mohammed et al., 2012).

Race/​Ethnicity The reasons underlying the variation in the incidence rates of esophageal cancer subtypes by race or ethnicity are as yet incompletely understood. The incidence of ESCC is about four times higher in black than in white men in the United States (Table 30–2). This is in marked contrast to the low incidence of esophageal cancer in Western Africa

 581

581

Esophageal Cancer Squamous Cell Carcinoma

100

10

1

1

0.1

0.1

0.01

0.01

19 7

19 7

White Female

White Male

Adenocarcinoma

3– 19 19 77 76 – 19 19 81 80 – 19 19 85 84 – 19 19 89 88 – 19 19 93 92 – 19 19 97 96 – 20 20 01 00 – 20 20 05 04 – 20 20 09 08 –2 01 2

10

3– 19 197 77 6 – 19 198 81 0 – 19 198 85 4 – 19 198 89 8 – 19 199 93 2 – 19 199 97 6 – 20 200 0 01 – 20 200 05 4 – 20 200 09 8 –2 01 2

Rate per 100,000 person-years

100

Black Male

Black Female

Figure 30–1.  Trends for esophageal cancer incidence rates by histology in the United States for blacks and whites by sex.

(Table 30–1), the historical region of origin for most American slaves. For EADC, the incidence rate is over four times higher in white than black men. The ESCC rates are similar for Hispanic and non-​Hispanic whites, but Hispanic whites have EADC rates one-​half those for non-​ Hispanic whites. This is in contrast to gastric adenocarcinoma, for which Hispanic whites have incidence rates twice as high as those for non-​Hispanic whites (National Cancer Institute, 2015). The percentages of esophageal cancers with adenocarcinoma histology are higher among African Americans and Asian Americans than the 10% or less observed in most Asian and African countries.

Socioeconomic Status

Table 30–2. Incidence Rates* for Squamous Cell Carcinomas and Adenocarcinomas of the Esophagus, by Racial/​Ethnic Group and Sex, SEER Program (2003–​2012)

Tobacco Use

Males Number Squamous Cell Carcinoma Whites 5,020 Blacks 2,319 Asians 789 Native Americans 45 White (non-​Hispanic) 4,322 White (Hispanic) 698 Adenocarcinoma Whites 17,760 Blacks 434 Asians 331 Native Americans 83 White (non-​Hispanic) 16,583 White (Hispanic) 1,177

Females Rate

Number

Rate

1.65 6.64 2.51 1.32 1.61 1.99

3,152 1,021 295 28 2,919 233

0.84 2.18 0.75 0.68 0.89 0.53

5.73 1.23 1.05 2.05 6.13 2.99

2,839 145 74 27 2,640 199

0.74 0.32 0.19 0.62 0.79 0.43

* Rates per 100,000 person-​years, age-​adjusted using 2000 US Standard. Source: SEER (National Cancer Institute, 2015).

ESCC has long been associated with low socioeconomic status. This association persists, even after adjustment for tobacco and alcohol, although residual confounding may still exist (Blot et  al., 2006). In contrast, EADC has not been consistently linked to socioeconomic status (Blot et al., 2006).

ETIOLOGIC FACTORS

Tobacco smoking has long been established as a major cause of esophageal cancer (IARC, 1986), with early studies documenting the association in populations in which ESCC predominated. Among smoking-​associated cancers, few have higher relative risks than ESCC (IARC, 2004; Vineis et al., 2004). Early cohort and case-​ control studies consistently showed that risk of esophageal cancers increased nearly 5-​fold among current cigarette smokers compared to nonsmokers, with the excess reaching nearly 10-​fold among heaviest smokers (Blot et  al., 2006). Alcohol consumption and cigarette smoking are correlated, and alcohol is also a risk factor for ESCC, but numerous studies have demonstrated that smoking increases ESCC risk independent of alcohol consumption (Table 30–3; Blot et  al., 2006; Freedman et  al., 2007). Studies assessing ESCC risk among smokers who did not consume alcohol also reported elevated risk (Blot et al., 2006; Prabhu et al., 2014). ESCC risk declines following cessation of smoking (Blot et al., 2006; Freedman et al., 2007), and the attributes required for establishing a cause-​and-​effect relationship between smoking and ESCC hold (IARC, 1986). It is highly likely that much of the decline in ESCC incidence rates since the early 1980s in the United States (Figure 30–1) is due to reductions in cigarette consumption. Smoking prevalence in adult men peaked in the 1960s, when about 60% were smokers, and steadily decreased to 20% in 2012 (Agaku et al., 2014).

582

582

PART IV:  Cancers by Tissue of Origin

Cigarette smoking is the only exposure known to increase risk of both ESCC and EADC. Although cigarette smoking affects adenocarcinoma risk to a lesser extent than squamous cell carcinoma risk in the esophagus, it is still an important risk factor for this histology (Table 30–3; Cook et al., 2010; Drahos et al., 2016; Tramacere et al., 2011). The pattern of lower smoking-​associated risk for EADC compared to ESCC has been consistently observed (Blot et al., 2006; Freedman et al. 2007; Vineis et al., 2004). Cigarette smoking also increases the risk of the adenocarcinoma precursor, Barrett’s esophagus (Andrici et al., 2013; Cook et al., 2012). Although the risk of adenocarcinoma declines after smoking cessation, there is evidence that the rate of decline for adenocarcinoma is much slower than the relatively rapid rate of decline for ESCC (Blot et al., 2006; Cook et al., 2010; Freedman et al., 2007; Schneider and Corley, 2015). This may partly explain why the sharp decline in smoking prevalence in the United States over the past five decades has not had an apparent dampening effect on the increase in adenocarcinoma rates over the past two decades (Figure 30–1). The impact of smoking on the esophageal cancer rate in a given population will depend heavily on the mix of histologic subtypes, but a recent pooled analysis of deaths in five US cohorts during the period 2000–​2011 reported that the relative risk for esophageal cancer was markedly elevated, even though EADC was the predominant histology during those years (Carter et al., 2015). Data on cigar and pipe smoking are more limited than that for cigarettes, but the available evidence indicates that the effects on esophageal cancer (predominantly ESCC) resemble those associated with cigarettes (Chang et  al., 2015; IARC, 2012b; National Cancer Institute, 1998). Evaluation of esophageal cancer risk from use of smokeless tobacco products is complicated by extreme geographic variation in the magnitude of relative risk estimates. Asian studies consistently show evidence of greater than 2.5-​fold elevations in esophageal cancer risk among smokeless tobacco users, but relative risk estimates from Western countries are much smaller, and the slight to moderate increases in Western risk estimates may be affected by residual confounding from socioeconomic status and other personal habits (Lagergren et al., 2000; Lee and Hamling, 2009; Siddiqi et al., 2015; Sinha et al., 2016). The geographic variation in risk estimates is likely explained in part by the different composition of some smokeless tobacco products in Asia (e.g., some contain betel leaves, betel nuts, or areca nuts). IARC has concluded that smokeless tobacco use is a cause of esophageal cancer in humans (2012b).

Alcohol Consumption Alcohol consumption, like smoking, was identified early as a major risk factor for esophageal cancer (IARC, 1968). Unlike smoking, alcohol consumption only increases the risk of ESCC (IARC, 2012b). Only oral cavity and pharynx cancers have estimated alcohol-​associated relative risks as high as those for ESCC (Bagnardi et al., 2015). Multiple studies have shown a strong dose–​ response relationship with the amount of alcohol consumed and ESCC risk (Bagnardi et al., 2015; Blot et al., 2006; Freedman et al., 2011; Islami et al., 2011). Relative risks of 5 or higher for heavy alcohol consumption have been consistently reported (Table 30–4). Large studies have had sufficient numbers to demonstrate a rising risk of ESCC with increasing alcohol intake in nonsmokers, and ESCC risk declines following quitting drinking (Blot et al., 2006; IARC, 1988; Islami et al., 2011). Table 30–3. Relative Risks* of Esophageal Cancer by Histology, According to Cigarette Smoking History Pack Years of Smoking 0 1–​29 30–​39 40–​49 50–​59 60+

Squamous Cell Carcinoma

Table 30–4. Relative Risks* of Esophageal Cancer by Histology, According to Alcohol Intake Alcohol Intake (Drinks/​Day) 0 > 0–​< 0.5 0.5–​< 1.0 1.0–​< 3.0 3.0–​< 5.0 5.0–​< 7.0 ≥7

Squamous Cell Carcinoma 1.0 0.80 (0.56–​1.14)† 1.23 (0.55–​2.74) 2.56 (1.10–​5.96) 4.56 (2.32–​8.96) 7.17 (2.98–​17.3) 9.62 (4.26–​21.7)

* Relative risks adjusted for age, sex, education, alcohol, and BMI. † 95% confidence interval. Source: BEACON consortium pooled analysis (Lubin et al., 2012).

1.0 1.66 (1.1–​2.4) 1.45 (0.8–​2.5) 2.22 (1.2–​4.0) 1.92 (1.0–​3.6) 2.77 (1.4–​5.6)

1.0 0.86 (0.65–​1.13) 0.63 (0.41–​0.99) 0.81 (0.60–​1.09) 0.86 (0.59–​1.24) 0.93 (0.66–​1.31) 0.97 (0.68–​1.36)

* Relative risks adjusted for age, sex, BMI, education, cigarette smoking, and if available, GERD. † 95% confidence interval. Source: BEACON consortium pooled analysis (Freedman et al., 2011).

Alcohol consumption is responsible for the clustering of ESCC in several hot-​spot areas of the world. These include clusters of elevated rates in northern France associated with Calvados apple brandies, in the South African Transkei associated with maize beer, in Puerto Rico and South America associated with rum and sugar-​cane distilled beverages, and in coastal South Carolina associated with moonshine whiskies (Blot, 1994). The common ingredient is alcohol (ethanol), although a role for other compounds in the alcoholic beverages is possible. These diverse findings also suggest that all forms of alcoholic beverage, convey risk of ESCC, and studies comparing spirit, wine, and beer intake tend to report increased risks regardless of the type of alcoholic beverage drunk (Blot, 1992). In certain very high risk areas of the world, including Linxian China and along the Caspian littoral of Iran, alcohol consumption does not appear to be involved in the excess ESCC risk (Muñoz and Day, 1996). The lack of association between alcohol consumption and EADC is striking, especially among those who report high levels of consumption. A  meta analysis taking into account data from 20 case-​ control and four cohort studies showed a summary relative risk for EADC of 0.87 (95% CI = 0.74–​1.01) for drinkers versus nondrinkers (Tramacere et  al., 2012). The risk of EADC did not increase even for those who reported consumption of seven or more drinks per day, as shown in Table 30–4. Two studies that assessed risk of both Barrett’s esophagus and EADC found no association with alcohol consumption for either endpoint (Freedman et al., 2011; Thrift et al., 2014a).

Interaction Between Smoking and Drinking The combination of smoking and alcohol consumption produces a much greater increase in ESCC risk than either exposure alone or than the sum of their independent effects. Table 30–5 presents risk estimates from a case-​control study in a high-​risk area of France showing one of the strongest reported links between drinking and smoking and esophageal cancer (in early studies, cell type was generally not specified, but nearly all tumors were of squamous cell histology). Both smoking and drinking independently increased risk. The combination Table 30–5. Relative Risks of Esophageal Cancer Associated with Alcohol Drinking and Tobacco Smoking in Brittany, France Tobacco (g/​day)

Adenocarcinoma

1.0 2.63 (1.8–​4.0)† 2.69 (1.6–​4.6) 3.93 (2.2–​7.1) 4.62 (2.5–​8.5) 5.63 (2.7–​11.7)

Adenocarcinoma

Alcohol (g/​day) 0–​40 41–​80 81–​120 ≥121

0–​9 1.0* 7.3 11.8 49.6

10–​19 3.4 8.4 13.6 65.9

20–​29 3.2 8.8 12.6 137.6

* Reference category. Source: Muñoz and Day (1996). Adapted from Tuyns et al. (1977).

≥ 30 7.8 35.0 83.0 155.6

 583

Esophageal Cancer of heavy smoking and drinking was associated with a 100-​ fold increase in ESCC risk compared to that of nonsmoking nondrinkers. Combined exposure to alcohol and tobacco smoking accounts for a very high percentage of ESCCs in many Western regions of the world, but tobacco accounts for a small fraction of EADCs, with no contribution from alcohol. Tobacco use is not as strongly related to risk of ESCC in Asia. However, a recent meta-​analysis of five studies showed increased risk from smoking, particularly among those who both smoked cigarettes and drank alcoholic beverages (Prabhu et al., 2014).

Betel Quid In India and Southeast Asia, elevated ESCC risk was linked to betel nut chewing (Znaor et al., 2003), with a recent meta-​analysis reporting a 3-​fold increased risk associated with areca nut chewing, independent of tobacco smoking (Akhtar, 2013). Betel quid has been classified as an esophageal carcinogen in humans (IARC, 2012b).

Obesity Obesity is firmly established as a major risk factor for EADC, and the association between obesity and EADC may be the strongest observed among obesity-​associated cancers (Lagergren, 2011). The most commonly used measure of obesity, particularly in early studies, is body mass index (BMI), and elevated BMI has consistently been shown to be associated with increased EADC risk (Hoyo et al., 2012; Kubo and Corley, 2006; O’Doherty et al., 2012; Ryan et al., 2011; Steffen et al., 2015; Turati et al., 2013). The positive association between BMI and EADC risk has been corroborated in a Mendelian randomization study (Thrift et al., 2014c). In sharp contrast to the strong association between BMI and EADC risk, there is no evidence of increased ESCC risk among those with elevated BMI (Blot et al., 2006; Corley et al., 2008). Table 30–6 presents data from the US multicenter study showing rising risk of EADC, but no increase in ESCC risk, with increasing BMI. The pattern of risk for ESCCs, with no evidence of increased risk among overweight and obese individuals, and high risk often observed among those with lowest weight, may be due to poor nutritional status as well as residual confounding from tobacco and alcohol consumption among those with low BMI. In Table 30–6 the increases in EADC risk are substantial, reaching 3-​fold among those in the highest BMI quartile (with a lower bound in this study of 27 kg/​m2). Even higher relative risks have been reported elsewhere for BMI values exceeding 30 or 40 kg/​m2 (Hoyo et al., 2012; Lagergren et al., 1999). Height appears to be inversely associated with risk of both Barrett’s esophagus and EADC, in contrast to the positive associations observed between height and many other types of cancers (Thrift et al., 2014b). This suggests a high-​risk phenotype for EADC characterized by short stature and overweight. Recent studies investigating obesity have reported that central (or abdominal) adiposity (measured as waist circumference, abdominal diameter, or waist-​to-​hip ratio) is a consistent risk factor for EADC, with some of the studies showing that abdominal adiposity is a stronger risk factor than BMI (Corley et al., 2008; Lagergren, 2011; O’Doherty et al., 2012; Singh et al., 2013b; Steffen et al., 2015). The

Table 30–6. Relative Risks* of Esophageal Cancer, by Cell Type, Associated with Obesity BMI Quartile I (low) II III IV

Squamous Cell Carcinoma 1.0 0.5 (0.3–​0.9)† 0.8 (0.5–​1.3) 0.6 (0.3–​1.0)

* Relative risks adjusted for age, sex, race, smoking. † 95% confidence interval. Source: Chow et al. (1998).

Adenocarcinoma 1.0 1.3 (0.8–​2.2) 2.0 (1.3–​3.3) 2.9 (1.8–​4.7)

583

prevalence of abdominal obesity is higher in men than women, and thus the positive association between central adiposity and EADC risk likely contributes to the striking male predominance of EADC (Lagergren, 2011; Singh et  al., 2013b). Two reviews concluded that BMI and measures of central adiposity produced similar relative risk estimates for EADC (Runge et al., 2015; Singh et al., 2013b), but in a more recent follow-​up study of nearly 400,000 European adults, risk of EADC was strongly associated with waist circumference, and the association between BMI and EADC risk was nearly eliminated after adjustment for waist circumference (Steffen et al., 2015). The conclusion that central adiposity is a risk factor for EADC independent of BMI led Singh et al. (2013b) to recommend that future studies aimed at understanding the mechanistic effect of obesity on EADC should focus on measures of visceral fat rather than overall obesity. The rise in the incidence of EADC in the United States and other Western countries seems at least in part due to the concomitant rise in obesity. The prevalence of obesity in the United States has steadily increased over the past three decades, with a leveling off only recently beginning (Kroep et al., 2014). With a 2-​fold higher risk among the obese versus non-​obese individuals, and an increase in obesity prevalence of 20% (say from 10%–​30% in the US after 1980), the expected increase in EADC can be estimated to be just under 20%. In the United States, age-​adjusted EADC incidence rates among white males increased nearly 100% from 1990 to 2010 (Figure 30–1), outpacing the rise attributable solely to the obesity epidemic. It is also the case that differences in obesity prevalence and trends do not explain differences among countries in rates or trends of EADC (Kroep et al., 2014). It is paradoxical that the prevalence of obesity in the United States is highest among black women (Ogden et al., 2014), the group with the lowest incidence of EADC. It is difficult to assemble adequately powered studies of EADC among black women to explore this paradox because of their low EADC incidence rate (Table 30–2).

Gastroesophageal Reflux Disease (GERD)

Barrett’s esophagus develops as a consequence of GERD, a condition caused by reflux into the lower esophagus of gastric acid and/​or bile salts and alkaline duodenal contents (Spechler, 2014; Wild and Hardie, 2003). GERD is among the strongest risk factors for both Barrett’s esophagus and EADC (Cook et al., 2014; Drahos et al., 2016; Lada et al., 2013; Lagergren and Lagergren, 2013; Rubenstein et al., 2011; Runge et  al., 2015), and obesity is a major risk factor for GERD (Friedenberg et al., 2008; Lagergren, 2011). Increasing GERD prevalence over time may be contributing to the increase in EADC rates in developed countries (Lagergren and Lagergren, 2013). The strong association between central adiposity and EADC has led to suggestions that obesity may influence adenocarcinoma risk via a mechanical mechanism, that is, an increase in intra-​abdominal pressure leading to promotion of GERD with transition to Barrett’s esophagus (Chandar and Iyer, 2015). Consistent with this hypothesized mechanism, obesity is a strong risk factor for both GERD and Barrett’s esophagus (Chandar and Iyer, 2015; Corley et al., 2007; Lagergren, 2011; Runge et al., 2015). The observation that associations between central adiposity and Barrett’s esophagus and EADC risk are independent of GERD has led to suggestions that visceral fat may also increase EADC risk through non-​mechanical mechanisms (Lagergren, 2011; Singh et al., 2013b). Thus it has been suggested that visceral fat plays a metabolic role in the etiology of EADC, perhaps through the secretion of adipokines and cytokines (Chang and Friedenberg, 2014; Chandar and Iyer, 2015; Lagergren, 2011; Lagergren and Lagergren, 2013; Singh et al., 2013b).

Hiatal Hernia

Obesity increases the risk of hiatal hernias, and hiatal hernias are associated with increased risk of GERD (Friedenberg et al., 2008; Chang and Friedenberg, 2014). Although not widely investigated as a risk factor for EADC in epidemiologic studies, hiatal hernia appears to be a risk factor for Barrett’s esophagus (Nguyen et al., 2014; Spechler, 2013). A recent study indicates that both a family history of hiatal hernia and a personal history of hiatal hernia may increase the risk of EADC (Jiang et al., 2014).

584

584

PART IV:  Cancers by Tissue of Origin

Nutritional Deficiency Earlier research identified several diseases that predispose to nutritional deficiency, such as Plummer-​Vinson syndrome, celiac disease, and pernicious anemia, as risk factors for ESCC. These conditions are uncommon in most populations, however, and there have been few subsequent reports of associations (Blot et al., 2006). Nutritional deficiencies were suspected as major factors in the clustering of the very high incidence of squamous cell cancers of the esophagus in Central Asia, partly because tobacco and alcohol seemed not to be major determinants in these areas (Blot et  al., 2006). Indeed, the Linxian nutritional intervention trials were launched in the mid-​1980s to evaluate whether supplementation of a high-​risk rural population, with limited variety in food and nutrient intake, could help reduce the substantial impact of these cancers in this region of North-​Central China. Two trials were conducted, one randomizing nearly 30,000 residents into four groups of vitamin/​mineral supplement combinations (retinol and zinc; riboflavin and niacin; ascorbic acid and molybdenum; and beta carotene, alpha tocopherol and selenium). The other trial randomized nearly 3,000 persons with esophageal dysplasia to a multivitamin/​mineral pill or placebo (Blot et al., 2006). After 5 years of supplementation, cancer mortality rates were significantly lower among those who received a combination of beta-​carotene, vitamin E, and selenium, although the reduction was more pronounced for stomach than for esophageal cancer. In subgroup analyses, those with high baseline levels of selenium and alpha tocopherol, but not beta-​carotene or vitamin C, had lower esophageal cancer mortality (Blot et al., 2006; Lam et al., 2013). Supplementation ended after 5 years, but an additional 10-​year follow-​up of the trial participants indicated that the benefit in the beta carotene/​alpha tocopherol/​ selenium group persisted into the 2000s (Qiao et al., 2009). No significantly reduced cancer occurrence was seen for the other three vitamin/​ mineral combinations either at the end of supplementation or in the extended follow-​up. The Linxian trials were initially heralded as providing evidence that simple nutrient supplementation with beta carotene, vitamin E, or selenium might reduce the burden of cancer in humans. However, hopes for a generalized reduction in cancer incidence were soon dashed when randomized clinical trials of beta carotene and/​or retinol supplementation were reported to increase rather than decrease risk of lung cancer in the US Carotene and Retinol Efficacy Trial (Omenn et al., 1996) and Finnish Alpha-​Tocopherol, Beta-​Carotene Cancer Prevention Study (ATBC Study Group, 1994). In the two decades since the Linxian trials, strong evidence of a protective role for vitamin or mineral supplementation on cancer in general, and esophageal cancer in particular, has yet to emerge (Myung and Yang, 2013). The benefits of supplementation may be limited to populations with extreme nutritional deficiency, such as in Linxian.

Other Dietary Factors Dietary factors long have been thought to be important in the development of esophageal cancer (Blot et  al., 2006), but quantifying risks associated with food and nutrient consumption has been difficult, and it remains unclear to what extent diet (other than alcohol drinking) contributes to risk of either ESCC or EADC. The most comprehensive review of dietary factors and cancer lists no specific foods or nutritional factors as having “convincing” evidence of a causal relationship with esophageal cancer (World Cancer Research Fund/​ American Institute for Cancer Research [WCRF/​AICR], 2007). Fruits, vegetables, and foods containing beta carotene or vitamin C are designated “probable” protective factors, and red and processed meats as “possible” risk factors. Reduced risks associated with vegetable and fruit intake have been observed among multiple case-​control studies, as well as the few cohort studies that have reported on esophageal cancer (Blot et al., 2006). Most early studies applied largely to squamous cell carcinomas, but similar findings have since been reported for adenocarcinomas (Abnet et  al., 2015; Kubo et  al., 2010; Petrick et al., 2015; Steevens et al., 2011). Increased risks associated with red and processed meat consumption have been observed in most observational studies, with generally similar findings for both ESCC and

EADC (Abnet et al., 2015; Cross et al., 2011; O’Doherty et al., 2011; Petrick et al., 2015; Zhu et al., 2014). Consumption of exceptionally hot beverages, including maté and very hot teas in South America and Asia, has been associated with increased risk of ESCC (Andrici and Eslick, 2013; Chen et al., 2015; Islami et al., 2009; Tang et al., 2013). Two meta-​analyses have reported increased risk of ESCC, but not EADC, associated with consumption of very hot beverages and foods (Andrici and Eslick, 2015; Chen et al., 2015). Tea drinking itself, however, has been associated with reduced risk of these cancers in China (Wu et al., 2009; Zheng et al., 2013), mainly for green and oolong teas with higher concentrations of flavonoids and other agents that have been shown to inhibit esophageal cancer in experimental animals (Lambert and Yang, 2003). Studies of overall dietary patterns have been more promising than studies of individual foods. Those who reported following a healthy diet had significantly lower risk of ESCC in a meta-​analysis of nine case-​control studies (Liu et al., 2014), and consumption of a healthy diet was associated with reduced risk of EADC in a case-​control study (Chen et  al., 2002). In the National Institutes of Health–​American Association of Retired Persons (NIH-​AARP) cohort tracking nearly 500,000 adults, those in the highest quintile of a composite Healthy Eating Index-​2005 score had significantly reduced risks of both ESCC (HR = 0.51) and EADC (HR = 0.75) (Li et al., 2013c). Analyses of a Mediterranean diet score in the same cohort demonstrated lower risk of ESCC (HR = 0.44), but not EADC (HR = 0.91) (Li et al., 2013c). Collectively, these studies suggest that consumption of a healthy, balanced diet may reduce the risk of esophageal cancers, although the data on individual food items remain inconclusive.

Occupation Occasional reports have suggested increased risks of esophageal cancer, primarily ESCC, among various occupational groups. These associations tend to be neither consistent nor strong, however, and have not always been adjusted for smoking, alcohol, and social class (Blot et al., 2006). Many of these associations were derived from record-​linkage or other exploratory studies that examined multiple occupations, and are thus subject to the problem of false positives arising from multiple statistical comparisons. Early research among insulation workers heavily exposed to asbestos initially suggested an increased risk of ESCC, but two reviews and meta-​analyses have concluded that there is no evidence of increased risk. One meta-​analysis reported a summary risk estimate of 1.02 (Goodman et al., 1999). A review by the US Institute of Medicine reported a summary risk estimate of 0.99 (National Academies of Science [NAS], 2006). Fewer studies have assessed occupational risk factors for EADC, but again, the occasional associations reported in individual studies do not provide compelling evidence of carcinogenicity. A  few studies suggested that chlorinated solvent exposure might be associated with increased EADC risk, but an IARC working group review found no consistent or reliable evidence to support this finding (IARC, 2014). Based on current information, esophageal cancer appears to have little or no association with occupational exposures.

Radiation Studies of populations exposed to ionizing radiation have established that both short-​term high-​level exposure and chronic low-​level exposure increases risk of esophageal cancer. The diverse study populations include Japanese atomic bomb survivors, patients who received X-​ray therapy for ankylosing spondylitis, patients receiving radiotherapy for breast cancer and Hodgkin lymphoma, medical diagnostic X-​ray workers, and people exposed to ionizing radiation from environmental pollutants released from a plutonium production plant (Blot et al., 2006; Davis et al., 2015; Kamiya et al., 2015; Morton et al., 2012, 2014; Wang et al., 2015). The dose of radiation to the esophagus was estimated in a study of long-​term breast cancer survivors. There was little evidence of increased risk of esophageal cancer in women who received less than 20 Gy, but a monotonic increase in ESCC risk in those receiving higher doses. Women with over 35 Gy exposure to the

 58

Esophageal Cancer esophagus had an 8-​fold increase in ESCC risk (Morton et al., 2012). In contrast, EADC risk did not increase significantly with increasing radiation dose. ESCC comprised nearly all (97%) esophageal cancers in women who received more than 20 Gy, but 79% among those who received less than this amount (Morton et al., 2012). Another review of second cancer risk after radiotherapy for breast cancer reported a summary relative risk estimate of 2.2 (95% CI = 1.1–​4.2) for women who survived 15 or more years after breast cancer diagnosis (Grantzau and Overgaard, 2015). A study of esophageal cancer following radiotherapy for Hodgkin lymphoma also found significantly increased risk in those with high estimated doses to the esophagus (Morton et  al., 2014). A nested case-​control study of X-​ray technicians occupationally exposed to radiation reported significantly increased risk of both esophageal and breast cancer (Wang et  al., 2015). A study of solid cancer incidence associated with environmental radiation exposure in the Techna River cohort in Russia reported significantly increased risk for esophageal and uterine cancer (Davis et al., 2015). There is now convincing evidence that ESCC is a radiogenic malignancy, but whether EADCs are caused by ionizing radiation remains uncertain.

Medications When the epidemic rise of EADC first became apparent, one of the hypothesized causes was the rising use of medications used to treat GERD. These drugs included H2 receptor antagonists, proton-​pump inhibitors, and over-​the-​counter antacids. Drugs that relax the gastroesophageal sphincter and thus could increase gastric reflux, including tricyclic antidepressants, calcium channel blockers, asthma drugs containing theophylline or ß agonists, and anticholinergics, were also hypothesized to increase the risk of EADC. However, studies in the 1990s tended to dispel this notion, despite the difficulty in sorting out effects of drugs from the effects of the underlying conditions for which the drugs were applied (Blot et al., 2006). Notably, a nested case-​control study in a US prepaid health plan showed that the 4-​fold increase in EADC risk among those using H2 blockers was nearly eliminated after control for GERD (Chow et al., 1995). Since that time, no further conclusive evidence has implicated these medications, including proton pump inhibitors, in EADC risk (Spechler, 2013; Thrift, 2015). Indeed, recent research has focused on the possible protective effects of medications on esophageal cancer (Thrift, 2015). Observational studies suggest that proton pump inhibitors can prevent EADC, particularly among patients with Barrett’s esophagus (Dunbar et al., 2015; Krishnamoorthi et al., 2016; Singh et al., 2014; Thrift, 2015). This benefit has not been confirmed in clinical trials (Dunbar et al., 2015). Early observational studies suggested that non-​steroidal anti-​inflammatory drugs (NSAIDs) might protect against esophageal cancer (Blot et al., 2006), and evidence from recent studies now strongly supports a preventive benefit from NSAIDs for both EADCs and ESCCs (Beales et al., 2013; Drahos et al., 2016; Lagergren and Lagergren, 2013; Liao et al., 2012; Sun and Yu, 2011; Wang et al., 2011). NSAID use also appears to protect against Barrett’s esophagus (Schneider et al., 2015). In long-​ term follow-​up of participants in three low-​dose aspirin trials in the United Kingdom conducted primarily for the evaluation of cardiovascular events, significantly lower esophageal cancer (type not specified) mortality was observed among those randomized to the aspirin arms (Rothwell et al., 2011). Statin use may decrease the risk of EADC, particularly among patients with Barrett’s esophagus, although evidence of a protective effect is more limited than for NSAIDs (Beales et al., 2013; Krishnamoorthi et  al., 2016; Lagergren and Lagergren, 2013; Singh et al., 2013a). Statin use may also protect against development of Barrett’s esophagus (Beales et al., 2016). Although the observational evidence for pharmacologic prevention of esophageal cancer appears strong, more definitive evidence is needed to establish the efficacy and safety of these agents for the prevention of either EADC or ESCC.

Infectious Agents Infectious agents have been suspected of affecting risk of both ESCCs and EADCs, with most attention focused on human papillomaviruses

585

(HPV) for ESCC and Helicobacter pylori (H. pylori) for EADC. Numerous studies, especially in Asia, have reported associations between HPV infection and increased risk of ESCC (Blot et al., 2006). Meta-​analyses have reported relative risk estimates of approximately 3 for the association between HPV and ESCC (Hardefeldt et al., 2014; Liyanage et al., 2013). Caution is advised in interpreting these results, however, because the prevalence of HPV in tumor tissue varied by almost 2-​fold among the individual studies included in the meta-​analyses (Li et al., 2014). The prevalence of HPV detected in tumor tissue from ESCC averaged 28% across samples, but varied by almost 2-​fold depending on the detection method, with the highest prevalence occurring in China (Petrick et al., 2014). Data from six case-​control studies that assessed serologic markers of 14 HPV types in several parts of the world reported positive associations between the risk of ESCC and two subtypes of HPV (16 and 6), suggesting a limited causal role (Sitas et  al., 2012). Moreover, not all studies have found an association between HPV and ESCC, including some negative studies in China (Koshiol et al., 2010; Peixoto et al., 2001; Teng et al., 2014). Accordingly, an evaluation of the evidence for the association between HPV and ESCC in 2011 noted that the overall evidence for a causal role for HPV was inconclusive (IARC, 2012a), a conclusion shared by other recent reviews (Ludmir et  al., 2015; Rustgi and El-​Serag, 2014; Zandberg et al., 2013). It would be difficult to reconcile a strong causal association between HPV and ESCC with the downward secular trend observed for ESCC (Figure 30–1) since the late 1990s among white men in the United States, especially given the large increase in the HPV-​related squamous cell carcinomas of the oropharynx, tonsil, and base of the tongue (Simard et al., 2012). HPV infection has been reported to be unrelated to risk of EADC or Barrett’s esophagus (El-​Serag et al., 2013). While bacterial H. pylori infection is a major risk factor for gastric cancer, and may contribute to increased risk of colorectal and perhaps other cancers, it has now been consistently associated with lower risk of EADC (IARC, 2012a; Nie et  al., 2014; Runge et  al., 2015; Xie et al., 2013). A number of meta-​analyses have reported evidence that H.  pylori infection is associated with a reduced risk of EADC, particularly when infection involves CagA-​positive strains (Islami and Kamangar, 2008; Rokkas et  al., 2007; Xie et  al., 2013). In contrast, H.  pylori infection seems generally to be unrelated to ESCC risk (IARC, 2012a; Nie et al., 2014; Rokkas et al., 2007; Xie et al., 2013), although a finding of significantly reduced risk among Asian populations warrants further investigation (Xie et al., 2013). The mechanism(s) by which H. pylori might protect against EADC have been hypothesized to involve a reduction of gastric acid secretion resulting from gastric atrophy due to chronic infection with H. pylori. H.  pylori infection has been consistently associated with reduced risk of Barrett’s esophagus (Blot et  al., 2006; Fishbach et  al., 2012; Rubenstein et al., 2014). The association between H. pylori infection and GERD is uncertain, however, with recent studies indicating that H.  pylori may be associated with a reduced risk of GERD in Asian populations, but not Western populations (Rubenstein et  al., 2014). H. pylori prevalence has been declining in Western countries due to improvements in sanitation and widespread use of antibiotics. This decline coincides with the rise in EADC incidence. The trends may be linked (Blaser, 1999, 2008), as might be the higher H. pylori infection rates and lower EADC rates among blacks compared to whites in the United States. Analysis of the esophageal microbiome is in the early stages, but based on a comprehensive review of existing studies it has been hypothesized that an emerging Campylobacter species may be involved in the etiology of EADC (Kaakoush et al., 2015; Neto et al., 2016).

HOST FACTORS Predisposing Medical Conditions Several medical conditions predispose to the development of ESCC or EADC. Plummer-​Vinson syndrome, celiac disease, and pernicious anemia increase the risk of ESCC (Blot et al., 2006). Tylosis, a rare

586

586

PART IV:  Cancers by Tissue of Origin

genetic syndrome characterized by autosomal dominant inheritance and palmar and plantar hyperkeratosis, is associated with increased risk of ESCC (Blaydon et al., 2012; Robertson et al., 2008). Mutations in the RHBDF2 gene within chromosomal region 17q25 have been identified as the cause of tylosis (Blaydon et al., 2012). The relative contribution of RHBDF2 mutations to sporadic esophageal cancer risk is currently unknown. Another rare condition linked to squamous cell carcinomas is achalasia, in which the gastro-​esophageal sphincter fails to relax due to disturbance of the autonomic nervous system. This allows food to be retained in the esophagus for extended periods (Blot et al., 2006; Eckardt and Eckardt, 2010). Achalasia appears to increase the risk of EADC as well as ESCC (Leeuwenburgh, et  al., 2013; Zendehdel et al., 2011).

Family History Familial associations are less strong for esophageal cancer than for a number of other cancers, especially in Western countries (Blot et al., 2006; Gao et al., 2009; Jiang et al., 2014). It is difficult to distinguish the effects of inherited risk from those caused by shared environmental exposures. This challenge is compounded in high-​risk areas of the world where ESCC predominates. In a case-​control study of ESCC in China, for example, the relative risk estimate for having a first-​degree relative with esophageal cancer was 2.3, but the relative risk estimate was nearly the same for having a non-​blood relative with esophageal cancer (Gao et al., 2009).

Inherited Susceptibility Genes Candidate Gene Studies

Because of the strong associations of tobacco smoking and alcohol consumption with ESCC risk, genetic studies in the era preceding genome-​wide association studies (GWAS) concentrated on genes that might be likely to modify the risk associated with these known risk factors. The CYP1A1 gene was studied extensively because of the ability of the enzyme CYP1A1 to convert polycyclic aromatic hydrocarbons to reactive carcinogenic metabolites (Wang et  al., 2012). A large meta-​analysis reported a strong association between CYP1A1 polymorphisms and ESCC risk in people in Asia and in Western countries (Shen et al., 2013). No association with CYP1A1 polymorphisms was observed for EADC in this meta-​analysis. The CYP2E1 gene may also be associated with ESCC risk; CYP2E1 can activate a number of potential carcinogens (Wang et al., 2012). Alcohol is metabolized to acetaldehyde, which when associated with consumption of alcoholic beverages is classified as a human carcinogen (IARC, 2012b). Alcohol metabolism is under genetic control, with polymorphic variation in alcohol dehydrogenase (ADH) and aldehyde dehydrogenase (ALDH) genes and substantial differences in the prevalence of slow versus fast metabolizers across broad population groups. Early candidate gene studies extensively investigated polymorphisms in ADH and ALDH, and identified three genes, ALDH2, ADH1B, and ADH2, as likely genes associated with ESCC risk (Yokoyama et al., 2002; Zhang et al., 2010). A meta-​analysis reported significant associations between ADH1B and ALDH2 polymorphisms and ESCC risk (Zhang et al., 2010). As was the case for head and neck cancers (Brennan et al., 2004; Chang et  al., 2012), increased risk of esophageal cancer was observed among slow metabolizers based on the ADH1B genotype. A recent case-​control study provided compelling evidence that ADH4 polymorphisms may be associated with squamous cell carcinoma risk (Xu et  al., 2015). A small candidate gene study implicated germline mutations in MSR1 as possibly being associated with Barrett’s esophagus and esophageal adenocarcinoma (Orloff et al., 2011).

Genome-​Wide Association Studies

Although the findings of candidate gene studies are often not replicated in GWAS, the first two GWAS of esophageal carcinoma,

conducted in China and Japan, confirmed that ALDH2 and ADH1B polymorphisms were associated with risk of ESCC, with strong interactions for alcohol consumption and smoking (Cui et al., 2009; Tanaka et al., 2010). Subsequent large GWAS in Chinese populations identified several genes or gene regions associated with ESCC (Abnet et al., 2010; Wang et al., 2010; Wu et al., 2011, 2012, 2014). Genes identified in multiple GWAS included ADH1B, ALDH2, PLCE1, and CHEK2. In a GWAS conducted in a region where alcohol consumption and smoking are not major risk factors, the alcohol dehydrogenase genes were not significantly associated with cancer risk (Abnet et al., 2010). In another study, the alcohol dehydrogenase genes were only associated with cancer risk in analyses restricted to smokers and drinkers (Wang et al., 2010). One small GWAS found no association between the PLCE1 gene and ESCC in Caucasians, although this gene has been consistently found to be associated with ESCC risk in Asians (Dura et al., 2013b). However, the null study was based on only 83 patients with ESCC and 258 with EADC. A Chinese GWAS found no evidence that common variants in the SLC39A6 gene were associated with survival among ESCC patients (Wu et al., 2013). Because almost all GWAS have been conducted in Eastern Asian populations, the extent to which these results apply to Western populations is uncertain. Two recent studies have evaluated genes in specific pathways for evidence of an association with ESCC and gastric adenocarcinomas in a Chinese population. Alterations in DNA repair pathway genes were associated with ESCC (but not gastric cancer) risk; CHEK2 was the most significant single gene (Li et al., 2013a). The epidermal growth factor receptor pathway was significant for gastric cancer but not for ESCC (Li et al., 2013b). Far fewer GWAS have been conducted of EADC. Many studies are based on populations that combined EADC and Barrett’s esophagus patients. It is noteworthy that analysis of germline mutations indicates that EADC shares an underlying genetic basis with Barrett’s esophagus, but not with GERD (Ek et al., 2013). GWAS of Barrett’s esophagus and esophageal adenocarcinoma have identified polymorphisms at FOXF1, CRTC1, BARX1, FOXP1, and within the major histocompatibility complex (MHC) as being associated with risk (Levine et al., 2013; Palles et al., 2015; Su et al., 2012). A  case-​control study confirmed associations between esophageal adenocarcinoma and polymorphisms at FOXF1 and within the MHC (Dura et al., 2013a). A large GWAS evaluating miRNA-​related single nucleotide polymorphisms (associated with 389 genes) found no evidence of an association with risk of esophageal adenocarcinoma or Barrett’s esophagus (Buas et al., 2015). An integrative post-​ GWAS analysis suggested that germline CDKN2A mutations may be associated with progression from Barrett’s esophagus to esophageal adenocarcinoma (Buas et al., 2014).

Somatic Mutations Sequencing of the genome in cancer tissue has identified several genes that are mutated at high frequencies in ESCC. TP53 is by far the most frequently mutated gene; other genes frequently mutated include CDKN2A, PIK3CA, NFE2L2, NOTCH1, MLL2, FAT1, and ZNF750 (Abedi-​Ardekani and Hainaut, 2014; Gao et al., 2014; Lin et al., 2014; Sawada et al., 2016; Song et al., 2014; Zhang et al., 2015). Based on a detailed analysis of mutational signatures, the APOEC family of cytidine deaminases has been implicated in the mutation process for as many as 50% of the ESCCs examined (Sawada et  al., 2016; Zhang et al., 2015). Sequencing of the genomes in disease tissue has indicated a high frequency of mutations in several genes in EADC and Barrett’s esophagus. TP53 is the most frequently mutated gene, perhaps more so for EADC than for ESCC. Other frequently mutated genes include CDKN2A, SMAD4, ARID1A, PIK3CA, and MYO18B (Abedi-​ Ardekani and Hainaut, 2015; Dulak et  al., 2013; Galipeau et  al., 2007; Ross-​Innes et al., 2015; Stachler et al., 2015; Weaver et al., 2014). ARID1A has been identified as a suppressor gene in Barrett’s esophagus (Streppel et  al., 2014). TP53 mutations appear to be

 587

Esophageal Cancer associated with Barrett’s esophagus, particularly Barrett’s esophagus with high-​grade dysplasia (Galipeau et  al., 2007; Ross-​Innes et al., 2015; Stachler et al., 2015; Weaver et al., 2014). Progression to adenocarcinoma is frequently associated with chromosomal gain (increases in aneuploidy and copy number) or loss (deletions/​loss of heterozygosity) of genomic material (Galipeau et  al., 2007; Ross-​ Innes et al., 2015; Stachler et al., 2015).

Molecular Pathogenesis The molecular pathogenesis of esophageal cancer is poorly understood. Commonly mutated genes associated with ESCC risk are involved in cell cycle control, apoptosis, and histone modification. Cellular pathways affected by mutated genes in ESCC are the RTK-​MAPK-​PI3K, Wnt, Hippo, and Notch pathways. Based on evidence from mouse studies, it has been hypothesized that an unfolded protein response to stress in the endoplasmic reticulum, which may be linked to the Notch pathway, might play a role in the etiology of ESCC (Rosekrans et al., 2015b). Imputation analysis of an ESCC GWAS identified XBP1, a gene involved in the unfolded protein response, as possibly being associated with ESCC risk (Wu et al., 2012). The possible roles played by the various implicated pathways in the etiology of ESCC remain to be elucidated. Although the stepwise development of EADC is relatively well characterized, the roles of various pathways in cancer development are poorly understood. Frequently mutated genes associated with EADC risk are involved in cell signaling, apoptosis, and genomic stability. Cellular pathways affected by mutated genes in EADC are the Wnt pathway and the Bmp signaling pathway (Rosekrans et  al., 2015a). Several genes identified as being associated with Barrett’s esophagus or EADC are involved in the Bmp pathway, which may play a role in the development of Barrett’s esophagus (Rosekrans et al., 2015a). Recent studies that have evaluated the development and carcinogenic progression of Barrett’s esophagus have suggested that the Notch and Sonic Hedgehog pathways may also be involved in early steps in the development of EADC (Saraggi et al., 2016; Streppel et al., 2014). The relative contributions of the various implicated pathways to the etiology of EADC remain uncertain.

PREVENTIVE MEASURES The poor survival following a diagnosis of esophageal cancer highlights the need for measures aimed at cancer prevention.

Primary Prevention Eliminating or reducing the exposures that cause ESCC and EADC offers the greatest potential to reduce the incidence and mortality of both cancer subtypes. Thus for ESCC, curtailment of tobacco smoking (by helping smokers quit and ensuring that nonsmokers never take up the habit) and limiting alcohol consumption to moderate or low (including abstention) amounts may provide the best means currently available to reduce the incidence of this cancer in Western societies. The advantages of quitting smoking will be seen in the near term, since risk of ESCC declines rapidly following smoking cessation. The impact of declining smoking prevalence has already been reflected in the declining rates of this cancer subtype in the United States. For EADC, cigarette smoking also increases risk; quitting smoking will result in reduced risk of these esophageal cancers, although the reduction is slower than for ESCC and may not be evident for over a decade following cessation. In areas of the world with clusters of ESCC related to high alcohol intake, and in Western societies generally, reduction in alcohol consumption is critical to prevention, since excess risk occurs mainly among heavy drinkers. Alcohol consumption seems unrelated to EADC risk, so reducing consumption will be of benefit only for ESCC. Dietary factors may be involved in risk of both types of esophageal cancer, but if and until more specific components are identified,

587

the most prudent approach to risk reduction will be the adoption of a balanced diet. For EADC, obesity and GERD are clear risk factors. Large-​scale clinical trials to investigate reductions in adenocarcinoma incidence following weight loss or control of GERD are not practical given the relative rarity of the cancers, but the strong observational epidemiologic evidence indicates that measures aimed at obesity control should be helpful in lowering adenocarcinoma risk. Whether treatment of GERD will lower risk of esophageal adenocarcinoma is also unclear but seems plausible. However, the complexity of the gastric-​esophageal microbiome hinders evaluation of how treatment of GERD, or H. pylori or other microbes, may influence subsequent risk of cancers near the gastroesophageal junction.

Chemoprevention There is promise that aspirin and other NSAIDs may reduce the risk of both forms of esophageal cancer, since the epidemiologic evidence of lowered risk among users of these drugs has become more convincing over the past decade. In a meta-​analysis combining data across 51 randomized controlled trials designed primarily to assess cardiovascular outcomes, significantly reduced incidence of GI tract cancers was found among those taking daily aspirin, but the numbers were too small to sort out an effect for esophageal cancers (Rothwell et al., 2012). However, in an up to 20-​year follow-​up of participants in three large randomized prevention trials in the United Kingdom, with 62 deaths from esophageal cancer observed, mortality was less than half as high among those prescribed aspirin (HR = 0.42; 95% CI = 0.25–​0.71) (Rothwell et al., 2011). In the Linxian intervention trials, reduced incidence of the primary ESCC/​stomach cancer endpoint was found among those randomized to receive a combination of beta carotene, selenium, and vitamin E supplementation (Blot et al., 1993), although applicability beyond the nutritionally limited local setting is questionable. While no new strong candidates appear on the immediate horizon, prospects for chemoprevention may emerge as more is learned about the potential mechanisms of esophageal carcinogenesis and the human microbiome (Chung et al., 2015; Neto et al., 2016).

Secondary and Tertiary Prevention One approach to prevent esophageal cancer onset is to block the formation of its precursor lesions and/​or their transformation to malignancy. Both ESCC and EADC develop through a multistage sequence of events with detectable early lesions (esophagitis and dysplasia for the former, and Barrett’s metaplasia and dysplasia for the latter), providing targets for inhibition of the carcinogenic process. Methods to prevent the occurrence of these early lesions are not yet available, however. Targeted secondary prevention will require further characterization of molecular markers that predict the progression of dysplasia and Barrett’s esophagus, which may in turn depend upon host factors related to the metabolism and detoxification of esophageal carcinogens. The multistage process also provides opportunities for surveillance of those at high risk and early detection of cancers at a stage when they may be more amenable to surgical or other treatment. Adults with Barrett’s esophagus can be followed more intensely, although screening the general population for Barrett’s and subsequently following those in whom the lesions are found has not yet been shown to lower mortality from esophageal cancer and may not be efficient due to the low rate of transition from Barrett’s to cancer (Conio et al., 2003; Falk, 2015; Spechler, 2013).

FUTURE RESEARCH The epidemiologic evidence accumulated to date indicates that most ESCC in Western countries are preventable by reductions in tobacco and alcohol consumption and suggests that pathways to reduce risk of EADC will involve the prevention or treatment of obesity and reflux disease. Further characterization of the roles of central obesity, sedentary activity, and diet in esophageal cancer risk may yield better

58

588

PART IV:  Cancers by Tissue of Origin

strategies for risk reduction, especially for EADC. In other countries with pockets of excess ESCC incidence, preventive options seem less clear. Continued research in genomic, metabolomic, microbiome, and other molecular approaches toward personalized precision medicine may provide new insights and clues into the mechanisms of the development of esophageal cancers and clues toward their prevention and control. References Abedi-​Ardekani B, Hainaut P. 2014. Cancers of the upper gastro-​intestinal tract: a review of somatic mutation distributions. Arch Iran Med 17:286–​ 292. PMID: 24724606. Abnet CC, Corley DA, Freedman ND, Kamangar F. 2015. Diet and upper gastrointestinal malignancies. Gastroenterol 148:1234–​1243. PMCID: PMC4414068. Abnet CC, Freedman ND, Hu N, Wang Z, Yu K, Shu XO, et al. 2010. A shared susceptibility locus in PLCE1 at 10q23 for gastric adenocarcinoma and esophageal squamous cell carcinoma. Nature Genet 9:764–​767. PMCID: PMC2947317. Agaku IT, King BA, Dube SR. 2014. Current cigarette smoking among adults: United States, 2005–​2012. MMWR 63:29–​34. PMID: 24430098. Akhtar S. 2013. Areca nut chewing and esophageal squamous-​cell carcinoma risk in Asians:  a meta-​analysis of case-​control studies. Cancer Causes Control 24:257–​265. PMID: 23224324. Al-​ Samawi AS, Aulaqi SM. 2014. Esophageal cancer in Yemen. J Coll Physicians Surg Pak 24:182–​185. PMID: 24613114. Andrici J, Cox MR, Eslick GD. 2013. Cigarette smoking and the risk of Barrett’s esophagus:  a systematic review and meta-​ analysis. J Gastroenterol Hepatol 28:1258–​1273. PMID: 23611750. Andrici J, Eslick GD. 2013. Maté consumption and the risk of esophageal squamous cell carcinoma: a meta-​analysis. Dis Esophagus 26:807–​816. PMID: 22891687. Andrici J, Eslick GD. 2015. Hot food and beverage consumption and the risk of esophageal cancer:  a meta-​analysis. Am J Prev Med 49:952–​960. PMID: 26590941. Arnold M, Soerjomataram I, Ferlay J, Forman D. 2015. Global incidence of oesophageal cancer by histological subtype in 2012. Gut 64:381–​387. PMID: 25320104. ATBC Study Group. 1994. The effect of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers. N Engl J Med 330:1029–​1035. PMID: 8127329. Bagnardi V, Rota M, Botteri E, Tramacere I, Islami F, Fedirko V, et  al. 2015. Alcohol consumption and site-​ specific cancer risk:  a comprehensive dose-​response meta-​analysis. Br J Cancer 112:580–​593. PMCID: PMC4453639. Beales ILP, Vardi I, Dearman L, Broughton T. 2013. Statin use is associated with a reduction in the incidence of esophageal adenocarcinoma: a case control study. Dis Esophagus 26:838–​846. PMID: 22989236. Beales ILP, Dearman L, Vardi I, Loke Y. 2016. Reduced risk of Barrett’s esophagus in statin users:  case-​control study and meta-​analysis. Dig Dis Sci 61:238–​246. PMID: 26386857. Blaser MJ. 1999. Hypothesis: the changing relationships of Helicobacter pylori and humans: implications for health and disease. J Infect Dis 179:1523–​ 1530. PMID: 10228075. Blaser MJ. 2008. Disappearing microbiota:  Helicobacter pylori protection against esophageal adenocarcinoma. Cancer Prev Res 1:308–​311. PMID: 19138974. Blaydon DC, Etheridge SL, Risk JM, Hennies HC, Gay LJ, Carroll R, et al. 2012. RHBDF2 mutations are associated with tylosis, a familial esophageal cancer syndrome. Am J Hum Genet 90:340–​346. PMCID: PMC3276661. Blot WJ. 1992. Alcohol and cancer. Cancer Res 52:2119s–​2123s. PMID: 1544150. Blot WJ, Li JY, Taylor PR, Guo W, Dawsey S, Wang GQ, et al. 1993. Nutrition intervention trials in Linxian, China:  supplementation with specific vitamin/​ mineral combinations, cancer incidence, and disease-​ specific mortality in the general population. J Natl Cancer Inst 85:1483–​1492. PMID: 8360931. Blot WJ. 1994. Esophageal cancer trends and risk factors. Semin Oncol 21:403–​ 410. PMID: 8042037. Blot WJ, Devesa SS, Kneller RW, Fraumeni JFF Jr. 1991. Rising incidence of adenocarcinoma of the esophagus and gastric cardia. JAMA 265:1287–​ 1289. PMID: 1995976. Blot WJ, McLaughlin JK, Fraumeni JF Jr. 2006. Esophageal cancer. In: Schottenfeld D, Fraumeni JFJr (eds.), Cancer epidemiology and prevention, 3rd ed. New York: Oxford University Press, pp. 697–​706.

Bosetti C, Levi F, Ferlay J, Garavello W, Lucchini F, Bertuccio P, et al. 2008. Trends in oesophageal cancer incidence and mortality in Europe. Int J Cancer 122:1118–​1129. PMID: 17990321. Brennan P, Lewis S, Hashibe M, Bell DA, Boffetta P, Bouchardy C, et al. 2004. Pooled analysis of alcohol dehydrogenase genotypes and head and neck cancer: a HuGE review. Am J Epidemiol 159:1–​16. PMID: 14693654. Buas MF, Levine DM, Makar KW, Utsugi H, Onstad L, Li X, et  al. 2014. Integrative post-​genome-​wide association analysis of CDKN2A and TP53 SNPs and risk of esophageal adenocarcinoma. Carcinogenesis 35:2740–​ 2747. PMCID: PMC4247522. Buas MF, Onstad L, Levine DM, Risch HA, Chow WH, Liu G, et  al. 2015. MiRNA-​ related SNPs and risk of esophageal adenocarcinoma and Barrett’s esophagus:  post genome-​ wide association analysis in the BEACON consortium. PLoS One 10(6):e0128617. doi: 10.1371/​journal. pone.0128617. PMCID: PMC4454432. Carter BD, Abnet CC, Feskanich D, Freedman ND, Hartge P, Lewis CE, et al. 2015. Smoking and mortality: beyond established causes. N Engl J Med 372:631–​640. PMID: 25671255. Chandar AK, Iyer PG. 2015. Role of obesity in the pathogenesis and progression of Barrett’s esophagus. Gastroenterol Clin N Am 44:249–​264. PMID: 26021193. Chang CM, Corey CG, Rostron BL, Apelberg BJ. 2015. Systematic review of cigar smoking and all cause and smoking related mortality. BMC Public Health Apr 24;15:390. doi:  10.1186/​s12889-​015-​1617-​5. PMCID: PMC4408600. Chang JS, Straif K, Guha N. 2012. The role of alcohol dehydrogenase genes in head and neck cancers: a systematic review and meta-​analysis of ADH1B and ADH1C. Mutagenesis 27:275–​286. PMID: 22042713. Chang P, Friedenberg F. 2014. Obesity and GERD. Gastroenterol Clin N Am 43:161–​173. PMCID: PMC3920303. Chen H, Ward MH, Graubard BJ, Heineman EF, Markin RM, Potischman NA, et al. 2002. Dietary patterns and adenocarcinoma of the esophagus and distal stomach. Am J Clin Nutr 75:137–​144. PMID: 11756071. Chen X, Yang C. 2001. Esophageal adenocarcinoma:  a review and perspectives on the mechanism of carcinogenesis and chemoprevention. Carcinogenesis 22:1119–​1129. PMID: 11470739. Chen Y, Tong Y, Yang C, Gan Y, Sun H, Bi H, et  al. 2015. Consumption of hot beverages and foods and risk of esophageal cancer: meta-​analysis of observational studies. BMC Cancer 15:449. PMCID: PMC4457273. Chow WH, Blot WJ, Vaughn TL, Risch HA, Gammon MD, Stanford JL, et al. 1998. Body mass index and risk of adenocarcinomas of the esophagus and gastric cardia. J Natl Cancer Inst 90:150–​155. PMID: 9450576. Chow WH, Finkle WD, McLaughlin JK, Frankl H, Ziel HK, Fraumeni JF Jr. 1995. The relation of gastroesophageal reflux disease and its treatment to adenocarcinomas of the esophagus and gastric cardia. JAMA 274:474–​ 477. PMID: 7629956. Chung CC, Lee YC, Wu MS. 2015. Prevention strategies for esophageal cáncer:  perspective of the East vs. West. Best Practices Res Clin Gastroenterol 29:869–​883. PMID: 26651249. Conio M, Blanchi S, Lapertosa G, Ferraris R, Sablich R, Marchi S, et al. 2003. Long-​term endoscopic surveillance of patients with Barrett’s esophagus. Incidence of dysplasia and adenocarcinoma:  a prospective study. Am J Gastroenterol 98:1931-​1939. PMID:14499768. Cook MB, Corley DA, Murray LJ, Liao LM, Kamangar F, Ye W, et al. 2014. Gastroesophageal reflux in relation to adenocarcinomas of the esophagus: a pooled analysis from the Barrett’s and Esophageal Adenocarcinoma consortium. PLoS One 9:e103508. PMCID: PMC4116205. Cook MB, Kamangar F, Whiteman DC, Freedman ND, Gammon M, Bernstein L, et  al. 2010. Cigarette smoking and adenocarcinomas of the esophagus and esophagogastric junction:  a pooled analysis from the international BEACON consortium. J Natl Cancer Inst 102:1344–​1353. PMCID: PMC2935475. Cook MB, Shaheen NJ, Anderson LA, Giffen C, Chow WH, Vaughan TL, et al. 2012. Cigarette smoking increases risk of Barrett’s esophagus: an analysis of the Barrett’s and Esophageal Adenocarcinoma Consortium. Gastroenterol 142:744–​753. PMCID: PMC3321098. Corley DA, Kubo A, Zhao W. 2007. Abdominal obesity, ethnicity and gastro-​ oesophageal reflux symptoms. Gut 56:756–​762. PMCID: PMC1954862. Corley DA, Kubo A, Zhao W. 2008. Abdominal obesity and the risk of esophageal and gastric cardia carcinomas. Cancer Epidemiol Biomarkers Prev 17:352–​358. PMCID: PMC2670999. Cross AJ, Freedman ND, Ren J, Ward MH, Hollenbeck AR, Schatzkin A, et  al. 2011. Meat consumption and risk of esophageal and gastric cancer in a large prospective study. Am J Gastroenterol 106:432–​442. PMCID: PMC3039705. Cui R, Kamatani Y, Takahashi A, Usami M, Hosono N, Kawaguchi T, et  al. 2009. Functional variants in ADH1B and ALDH2 coupled with alcohol and

 589

Esophageal Cancer smoking synergistically enhance esophageal cancer risk. Gastroenterol 137:1768–​1775. PMID: 19698717. Davis FG, Krestinina LY, Preston D, Epifanova S, Degteva M, Akleyev AV. 2015. Solid cancer incidence in the Techna River Incidence Cohort. Radiation Res 184:56–​65. PMID: 26121228. Drahos J, Xiao Q, Risch HA, Freedman ND, Abnet CC, Anderson LA, et al. 2016. Age-​specific risk factor profiles of adenocarcinomas of the esophagus: a pooled analysis from the international BEACON consortium. Int J Cancer 138:55–​64. PMCID: PMC4607633. Dulak AM, Stojanov P, Peng S, Lawrence MS, Fox C, Stewart C, et al. 2013. Exome and whole-​genome sequencing of esophageal adenocarcinoma identifies recurrent driver events and mutational complexity. Nature Genet 45:478–​486. PMCID: PMC3678719. Dunbar KB, Souza RF, Spechler SJ. 2015. The effect of proton pump inhibitors on Barrett’s esophagus. Gastroenterol Clin N Am 44:415–​424. PMID: 26021202. Dura P, van Veen EM, Salomon J, te Morsche RHM, Roelofs HMJ, Kristinsson JO, et  al. 2013a. Barrett associated MHC and FOXF1 variants also increase esophageal carcinoma risk. Int J Cancer 133:1751–​1756. PMID: 23504527. Dura P, Bregitha CVV, te Morsche RHM, Roelofs HMJ, Kristinsson JO, Wobbes T, et  al. 2013b. GWAS-​uncovered SNPs in PLCE1 and RFT2 genes are not implicated in Dutch esophageal adenocarcinoma and squamous cell carcinoma etiology. Eur J Cancer Prev 22:417–​419. PMID: 23222411. Eckardt AJ, Eckardt VF. 2010. Cancer surveillance in achalasia: better late than never? Am J Gastroenterol 105:2150–​2152. PMID: 20927062. Ek WE, Levine DM, D’Amato M, Pedersen NL, Magnusson PKE, Bresso F, et  al. 2013. Germline genetic contributions to risk for esophageal adenocarcinoma, Barrett’s esophagus, and gastroesophageal reflux. J Natl Cancer Inst 105:1711–​1718. PMCID: PMC3833931. El-​Serag HB, Hollier JM, Gravitt P, Alsarraj A, Younes M. 2013. Human papillomavirus and the risk of Barrett’s esophagus. Dis Esophagus 26:517–​ 521. PMCID: PMC4412355. Falk GW. 2015. Barrett’s oesophagus:  frequency and prediction of dysplasia and cancer. Best Pract Res Clin Gastroenterol 29:125–​138. PMCID: PMC4352690. Fishbach LA, Nordenstedt H, Kramer JR, Gandhi S, Dick-​Onuoha S, Lewis A, El-​Serag HB. 2012. The association between Barrett’s esophagus and Helicobacter pylori infection: a meta-​analysis. Helicobacter 17:163–​175. PMCID: PMC3335759. Ferlay J, Soerjomataram I, Ervik M, Dikshit R, Eser S, Mathers C, et  al. 2013. GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality Worldwide:  IARC CancerBase No. 11. Lyon, France:  International Agency for Research on Cancer. Available from: http://​globocan.iarc.fr, accessed August 27, 2015. Ferlay J, Soerjomataram I, Dikshit R, Eser S, Mathers C, Rebelo M, et  al. 2015. Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012. Int J Cancer 136:E359–​E386. PMID: 25220842. Freedman ND, Abnet CC, Leitzmann MF, Mouw T, Subar AF, Hollenbeck AR, Schatzkin A. 2007. A prospective study of tobacco, alcohol, and the risk of esophageal and gastric cancer subtypes. Am J Epidemiol 165:1424–​ 1433. PMID: 17420181. Freedman ND, Murray LJ, Kamangar F, Abnet CC, Cook MB, Nyrén O, et  al. 2011. Alcohol intake and risk of oesophageal adenocarcinoma:  a pooled analysis from the BEACON consortium. Gut 60:1029–​1037. PMCID: PMC3439838. Friedenberg FK, Xanthopoulos M, Foster GD, Richter JE. 2008. The association between gastroesophageal reflux disease and obesity. Am J Gastroenterol 103:2111–​2122. PMID: 18796104. Galipeau PC, Li X, Blount PL, Maley CC, Sanchez CA, Odze RD, et al. 2007. NSAIDs modulate CDKN2A, TP53, and DNA content risk for progression to esophageal adenocarcinoma. PLoS Med 4:e67. PMCID: PMC1808095. Gao Y, Hu N, Han XY, Giffen C, Ding T, Goldstein A, Taylor P. 2009. Family history of cancer and risk for esophageal and gastric cancer in Shanxi, China. BMC Cancer 9:269. PMCID: PMC2729777. Gao YB, Chen ZL, Li JG, Hu XD, Shi XJ, Sun ZM, et al. 2014. Genetic landscape of esophageal squamous cell carcinoma. Nature Genet 46:1097–​ 1102. PMID: 25151357. Gatenby P, Caygill C, Wall C, Bhatacharjee S, Ramus J, Watson A, Winslet M. 2014. Lifetime risk of esophageal adenocarcinoma in patients with Barrett’s esophagus. World J Gastroenterol 20:9611–​9617. PMCID: PMC4110596. Goodman M, Morgan RW, Ray R, Malloy CD, Zhao K. 1999. Cancer in asbestos-​exposed occupational cohorts: a meta-​analysis. Cancer Causes Control 10:453–​465. PMID: 10530617.

589

Grantzau T, Overgaard J. 2015. Risk of second non-​breast cancer after radiotherapy for breast cancer:  a systematic review and meta-​ analysis of 762,468 patients. Radiother Oncol 114:56–​65. PMID: 25454172. Hardefeldt HA, Cox MR, Eslick GD. 2014. Association between human papillomavirus (HPV) and oesophageal squamous cell carcinoma:  a meta-​ analysis. Epidemiol Infect 142:1119–​1137. PMID: 24721187. Hoyo C, Cook MB, Kamangar F, Freedman ND, Whiteman DC, Bernstein L, et  al. 2012. Body mass index in relation to oesophageal and oesophagogastric junction adenocarcinomas:  a pooled analysis from the international BEACON consortium. Int J Epidemiol 41:1706–​1718. PMCID: PMC3535758. Hur C, Miller M, Kong CY, Dowling EC, Nattinger KJ, Dunn M, Feuer EJ. 2013. Trends in esophageal adenocarcinoma incidence and mortality. Cancer 119:1149–​1158. PMCID: PMC3744155. International Agency for Research on Cancer (IARC). 1986. Tobacco smoking. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Vol. 38. Lyon, France: IARC. PMID: 3460963. International Agency for Research on Cancer (IARC). 1988. Alcohol drinking. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Vol. 44. Lyon, France: IARC. PMID: 3236394. International Agency for Research on Cancer (IARC). 2004. Tobacco smoke and involuntary smoking. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Vol. 83. Lyon, France:  IARC. PMCID: PMC4781536. International Agency for Research on Cancer (IARC). 2012a. Biologic agents. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Vol. 100B. Lyon, France: IARC. PMCID: PMC4781184. International Agency for Research on Cancer (IARC). 2012b. Personal habits and indoor combustions. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Vol. 100E. Lyon, France:  IARC. PMCID: PMC4781577. International Agency for Research on Cancer (IARC). 2014. Trichloroethylene, tetrachloroethylene, and some other chlorinated agents. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Vol. 106. Lyon, France: IARC. PMCID: PMC4781308. Islami F, Boffetta P, Ren JS, Pedoeim L, Khatib D, Kamangar F. 2009. High-​ temperature beverages and foods and esophageal cancer risk: a systematic review. Int J Cancer 125:491–​524. PMCID: PMC2773211. Islami F, Fedirko V, Tramacere I, Bagnardi V, Jenab M, Scotti L, et al. 2011. Alcohol drinking and esophageal squamous cell carcinoma with focus on light-​drinkers and never-​smokers: a systematic review and meta-​analysis. Int J Cancer 129:2473–​2484. PMID: 21190191. Islami F, Kamangar F. 2008. Helicobacter pylori and esophageal cancer risk: a meta-​analysis. Cancer Prev Res 1:329–​338. PMCID: PMC3501739. Jiang X, Tseng CC, Bernstein L, Wu AH. 2014. Family history of cancer and gastroesophageal disorders and risk of esophageal and gastric adenocarcinomas:  a case-​ control study. BMC Cancer 14:60. PMCID: PMC3915076. Kaakoush NO, Castaño-​ Rodríguez N, Man SM, Mitchell HM. 2015. Is Campylobacter to esophageal adenocarcinoma as Helicobacter is to gastric adenocarcinoma? Trends Microbiol 23:455–​462. PMID: 25937501. Kamiya K, Ozasa K, Akiba S, Niwa O, Kodama K, Takamura N, et al. 2015. From Hiroshima and Nagasaki to Fukushima 1: long-​term effects of radiation exposure on health. Lancet 386:469–​478. PMID: 26251392. Kong CY, Kroep S, Curtius K, Hazleton WD, Jeon J, Meza R, et  al. 2014. Exploring the recent trend in esophageal adenocarcinoma incidence and mortality using comparative simulation modeling. Cancer Epidemiol Biomarkers Prev 23:997–​1006. PMCID: PMC4048738. Koshiol J, Wei WQ, Kreimer AR, Chen W, Gravitt P, Ren JS. 2010. No role for human papillomavirus in esophageal squamous cell carcinoma in China. Int J Cancer 127:93–​100. PMCID: PMC3069961. Krishnamoorthi R, Borah B, Heien H, Das A, Chak A, Iyer P. 2016. Rates and predictors of progression to esophageal carcinoma in a large population-​ based esophagus cohort. Gastrointest Endosc 84:40–​46. PMID: 26772891. Kroep S, Lansdorp-​Vogelaar I, Rubenstein JH, Lemmens VEPP, van Heijningen EB, Aragonés N, et al. 2014. Comparing trends in esophageal adenocarcinoma incidence and lifestyle factors between the United States, Spain, and The Netherlands. Am J Gastroenterol 109:336–​343. PMID: 24343546. Kubo A, Corley DA, Jensen CD, Kaur R. 2010. Dietary factors and the risks of oesophageal adenocarcinoma and Barrett’s esophagus. Nutr Res Rev 23:230–​246. PMCID: PMC3062915. Kubo A, Corley DA. 2006. Body mass index and adenocarcinomas of the esophagus or gastric cardia:  a systematic review and meta-​ analysis. Cancer Epidemiol Biomarkers Prev 15:872–​878. PMID: 16702363. Lada MJ, Nieman DR, Han M, Timratana P, Alsalahi O, Peyre CG, et al. 2013. Gastroesophageal reflux disease, proton-​pump inhibitor use and Barrett’s

590

590

PART IV:  Cancers by Tissue of Origin

esophagus in oesophageal adenocarcinoma:  trends revisited. Surgery 154:856–​866. PMID: 24074425. Lagergren J. Influence of obesity on the risk of esophageal disorders. 2011. Nat Rev Gastroenterol Hepatol 8:340–​347. PMID: 21643038. Lagergren J, Bergström R, Lindgren A, Nyrén O. 2000. The role of tobacco, snuff and alcohol use in the aetiology of cancer of the oesophagus and gastric cardia. Int J Cancer 85:340–​346. PMID: 10652424. Lagergren J, Bergström R, Nyrén O. 1999. Association between body mass index and adenocarcinoma of the esophagus and gastric cardia. Ann Intern Med 130:883–​890. PMID: 10375336. Lagergren J, Lagergren P. 2013. Recent developments in esophageal adenocarcinoma. CA Cancer J Clin 63:232–​248. PMID: 23818335. Lam TK, Freedman ND, Fan JH, Qiao YL, Dawsey SM, Taylor PR, Abnet CC. 2013. Prediagnostic plasma vitamin C and risk of gastric adenocarcinoma and esophageal squamous cell carcinoma in a Chinese population. Am J Clin Nutr 98:1289–​1297. PMCID: PMC3798080. Lambert JD, Yang CS. 2003. Mechanisms of cancer prevention by tea constituents. J Nutr 133:3262s–​3267s. PMID: 14519824. Lee PN, Hamling J. 2009. Systematic review of the relation between smokeless tobacco and cancer in Europe and North America. BMC Medicine 7:36. PMCID: PMC2744672. Leeuwenburgh I, Scholten P, Caljé TJ, Vaessen RJ, Tilanus HW, Hansen BE, Kuipers EJ. 2013. Barrett’s esophagus and esophageal adenocarcinoma are common after treatment for achalasia. Dig Dis Sci 58:244–​252. PMID: 23179142. Levine DM, Ek WE, Zhang R, Liu X, Onstad L, Sather C, et al. 2013. A genome-​ wide association study identifies new susceptibility loci for esophageal adenocarcinoma and Barrett’s esophagus. Nature Genet 45:1487–​1493. PMCID: PMC3840115. Li X, Gao C, Yang Y, Zhou F, Li M, Jin Q, Gao L. 2014. Systematic review with meta-​analysis: the association between human papillomavirus infection and oesophageal cancer. Aliment Pharmacol Ther 39:270–​281. PMID: 24308856. Li WQ, Hu N, Hyland PL, Gao Y, Wang ZM, Yu K, et al. 2013a. Genetic variations in DNA repair pathway genes and risk of esophageal squamous cell carcinoma and gastric adenocarcinoma in a Chinese population. Carcinogenesis 34:1536–​1542. PMCID: PMC3697889. Li WQ, Hu N, Wang Z, Yu K, Su H, Wang L, et al. 2013b. Genetic variants in epidermal growth factor receptor pathway genes and risk of esophageal squamous cell carcinoma and gastric cancer in a Chinese population. PLoS One 8:e68999. PMCID: PMC3715462. Li WQ, Park Y, Wu JW, Ren JS, Goldstein AM, Taylor PR, et  al. 2013c. Index-​based dietary patterns and risk of esophageal and gastric cancer in a large cohort study. Clin Gastroenterol Hepatol 11:1130–​1136. PMCID: PMC3758458. Liao LM, Vaughan TL, Corley DA, Cook MB, Casson AG, Kamangar F, et al. 2012. Nonsteroidal anti-​inflammatory drug use reduces the risk of adenocarcinomas of the esophagus and esophagogastric junction in a pooled analysis. Gastroenterol 142:442–​452. PMCID: PMC3488768. Lin DC, Hao JJ, Nagata Y, Xu L, Shang L, Meng X, et  al. 2014. Genomic and molecular characterization of esophageal squamous cell carcinoma. Nature Genet 46:467–​473. PMCID: PMC4070589. Liu X, Wang X, Lin S, Yuan J, Yu ITS. 2014. Dietary patterns and oesophageal squamous cell carcinoma:  a systematic review and meta-​analysis. Br J Cancer 110:2785–​2795. PMCID: PMC4037820. Liyanage SS, Rahman B, Ridda I, Newall AT, Tabrizi SN, Garland SM, et  al. 2013. The aetiological role of human papillomavirus in oesophageal squamous cell carcinoma:  a meta-​analysis. PLoS One 8:e69238. PMCID: PMC3722293. Lubin JH, Cook MB, Pandeya N, Vaughan TL, Abnet CC, Giffen C, et  al. 2012. The importance of exposure rate on odds ratios by cigarette smoking and alcohol consumption for esophageal adenocarcinoma and squamous cell carcinoma in the Barrett’s Esophagus and Esophageal Adenocarcinoma Consortium. Cancer Epidemiol 36:306–​316. PMCID: PMC3489030. Ludmir EB, Stephens SJ, Palta M, Willett CG, Czito BG. 2015. Human papillomavirus tumor infection in esophageal squamous cell carcinoma. J Gastrointest Oncol 6:287–​295. PMCID: PMC4397257. Mohammed ME, Abuidris DO, Elgaili EM, Gasmelseed N. 2012. Predominance of females with oesophageal cancer in Gezira, Central Sudan. Arab J Gastroenterol 13:174–​177. PMID: 23432985. Morton LM, Gilbert ES, Hall P, Andersson M, Joensuu H, Vaalavirta L, et al. 2012. Risk of treatment-​related esophageal cancer among breast cancer survivors. Ann Oncol 23:3081–​3091. PMCID: PMC3501231. Morton LM, Gilbert ES, Stovall M, van Leeuwen FE, Dores GM, Lynch CF, et al. 2014. Risk of esophageal cancer following radiotherapy for Hodgkin lymphoma. Haematologica 99:e193. PMCID: PMC4181268.

Muñoz N, Day NE. 1996. Esophageal cancer. In: Schottenfeld D, Fraumeni JFJr (eds.), Cancer epidemiology and prevention, 2nd ed. New York: Oxford University Press, pp. 681–​706. Myung SK, Yang HJ. 2013. Efficacy of vitamin and antioxidant supplements in prevention of esophageal cancer: meta-​analysis of randomized controlled trials. J Cancer Prev 18:135–​143. PMCID: PMC4189455. National Academies of Science. 2006. Asbestos: selected cancers. Washington, DC: The National Academies Press. National Cancer Institute (NCI). 1998. Cigars:  health effects and trends. Smoking and Tobacco Control Mongraph 9.  NIH Publication no.  98–​ 4302. Bethesda, MD: NCI. National Cancer Institute (NCI). 2015. DCCPS, Surveillance Research Program, Cancer Statistics Branch. SEER Program Public Use Data Tapes 1973–​2012, released April 2015, based on the November 2014 submission. Neto AG, Whitaker A, Pei Z. 2016. Microbiome and potential targets for chemoprevention of esophageal adenocarcinoma. Semin Oncol 43:86–​96. PMC4789168. Nguyen TH, Thrift AP, Ramsey D, Green L, Shaib YH, Graham DY, El-​Serag HB. 2014. Risk factors for Barrett’s esophagus compared between African Americans and non-​ Hispanic whites. Am J Gastroenterol 109:1870–​ 1880. PMID: 25420546. Nie S, Chen T, Yang X, Huai P, Lu M. 2014. Association of Helicobacter pylori infection with esophageal adenocarcinoma and squamous cell carcinoma: a meta-​analysis. Dis Esophagus 27:645–​653. PMID: 24635571. O’Doherty MG, Cantwell MM, Murray LJ, Anderson LA, Abnet CC; on behalf of the FINBAR study group. 2011. Dietary fat and meat intakes and risk of reflux esophagitis, Barrett’s esophagus and esophageal adenocarcinoma. Int J Cancer 129:1493–​1502. PMCID: PMC3144995. O’Doherty MG, Freedman ND, Hollenbeck AR, Schatzkin A, Abnet CC. 2012. A prospective cohort study of obesity and risk of oesophageal and gastric adenocarcinoma in the NIH–​AARP Diet and Health Study. Gut 61:1261–​ 1268. PMCID: PMC3504700. Ogden CL, Carroll MD, Kit BK, Flegal KM. 2014. Prevalence of childhood and adult obesity in the United States, 2011–​2012. JAMA 311:806–​814. PMCID: PMC4770258. Omenn GS, Goodman GE, Thornquist MD, Balmes J, Cullen MR, Glass A, et al. 1996. Effects of a combination of beta carotene and vitamin A on lung cancer and cardiovascular disease. N Engl J Med 334:1150–​1155. PMID: 8602180. Orloff M, Peterson C, He X, Ganapathi S, Heald B, Yang YR, et  al. 2011. Germline mutations in MSR1, ASCC1, and CTHRC1 in patients with Barrett esophagus and esophageal adenocarcinoma. JAMA 306:410–​419. PMCID: PMC3574553. Palles C, Chegwidden L, Li X, Findlay JM, Farnham G, Giner FC, et  al. 2015. Polymorphisms near TBX5 and GDF7 are associated with increased risk for Barrett’s esophagus. Gastroenterol 148:367–​378. PMCID: PMC4315134. Peixoto GD, Hsin LS, Snijders P, Wilmotte R, Herrero R, Lenoir G, et al. 2001. Absence of association between HPV DNA, TP53 codon 72 polymorphism, and risk of oesophageal cancer in high-​risk area of China. Cancer Lett 162:231–​235. PMID: 11146230. Petrick JL, Li N, McClain KM, Steck SE, Gammon MD. 2015. Dietary risk reduction factors for Barrett’s esophagus-​ esophageal adenocarcinoma continuum:  a review or the recent literature. Curr Nutr Rep 4:47–​65. PMCID: PMC4349493. Petrick JL, Wyss AB, Butler AM, Cummings C, Sun X, Poole C, et al. 2014. Prevalence of human papillomavirus among oesophageal squamous cell carcinoma cases:  systematic review and meta-​ analysis. Br J Cancer 110:2369–​2377. PMCID: PMC4007246. Prabhu A, Obi KO, Rubenstein JH. 2014. The synergistic effects of alcohol and tobacco consumption on the risk of esophageal squamous cell carcinoma: a meta-​analysis. Am J Gastroenterol 109:822–​827. PMID: 24751582. Qiao YL, Dawsey SM, Kamangar F, Fan JH, Abnet CC, Sun XD, et al. 2009. Total and cancer mortality after supplementation with vitamins and minerals: follow-​up of the Linxian General Population Nutrition Intervention Trial. J Natl Cancer Inst 101:507–​518. PMCID: PMC2664089. Robertson EV, Jankowski JA. 2008. Genetics of gastroesophageal cancer: paradigms, paradoxes, and prognostic utility. Am J Gastroenterol 103:443–​ 449. PMID: 17925001. Rokkas T, Pistiolas D, Sechopoulos P, Robotis I, Margantinis G. 2007. Relationship between Helicobacter pylori infection and esophageal neoplasia:  a meta-​analysis. Clin Gastroenterol Hepatol 5:1413–​1417. PMID: 17997357. Rosekrans SL, Baan B, Muncan V, van den Brink GR. 2015a. Esophageal development and epithelial homeostasis. Am J Physiol Gastronintest Liver Physiol 309:G216–​G228. PMID: 26138464.

 591

Esophageal Cancer Rosekrans SL, Heijmans J, Büller NVJA, Westerlund J, Lee AS, Muncan V, van den Brink GR. 2015b. ER stress induces epithelial differentiation in the mouse esophagus. Gut 64:195–​202. PMID: 24789843. Ross-​Innes CS, Becq J, Warren A, Cheetham RK, Northern H, O’Donovan M, et  al. 2015. Whole-​genome sequencing provides new insights into the clonal architecture of Barrett’s esophagus and esophageal adenocarcinoma. Nature Genet 47:1038–​1046. PMCID: PMC4556068. Rothwell PM, Fowkes FGR, Belch JFF, Ogawa H, Warlow CP, Meade TW. 2011. Effect of daily aspirin on long-​term risk of death due to cancer:  analysis of individual patient data from randomized trials. Lancet 377:31–​41. PMID: 21144578. Rothwell PM, Price JF, Fowkes FGR, Zanchetti A, Roncaglioni MC, Tognoni G, et  al. 2012. Short-​term effects of daily aspirin on cancer incidence, mortality and non-​vascular death:  analysis of the time course of risks and benefits in 51 randomized controlled trials. Lancet 379:1602–​1612. PMID: 22440946. Rubenstein JH, Inadomi JM, Scheiman J, Schoenfeld P, Appelman H, Zhang M, et  al. 2014. Association between Helicobacter pylori and Barrett’s esophagus, erosive esophagitis, and gastroesophageal reflux symptoms. Clin Gastroenterol Hepatol 12:239–​245. PMCID: PMC3947027. Rubenstein JH, Scheiman JM, Sadeghi S, Whiteman D, Inadomi JM. 2011. Esophageal adenocarcinoma incidence in individuals with gastroesophageal reflux:  synthesis and estimates from population studies. Am J Gastroenterol 106:254–​260. PMCID: PMC3901355. Runge TM, Abrams JA, Shaheen NJ. 2015. Epidemiology of Barrett’s esophagus and esophageal adenocarcinoma. Gastroenterol Clin N Am 44:203–​ 231. PMCID: PMC4449458. Rustgi AK, El-​Serag HB. 2014. Esophageal carcinoma. N Eng J Med 371:2499–​ 2509. PMID: 25539106. Ryan AM, Duoung M, Healy L, Ryan SA, Parekh N, Reynolds JV, Power DG. 2011. Obesity, metabolic syndrome and esophageal adenocarcinoma: epidemiology, etiology and new targets. Cancer Epidemiol 35:309–​319. PMID: 21470937. Saraggi D, Fassan M, Bornschein J, Farinati F, Realdon S, Valeri N, Rugge M. 2016. From Barrett metaplasia to esophageal adenocarcinoma:  the molecular background. Histol Histopathol 31:25–​32. PMID: 26334343. Sawada G, Niida A, Uchi R, Hirata H, Shimamura T, Suzuki Y, et  al. 2016. Genomic landscape of esophageal squamous cell carcinoma in a Japanese population. Gastroenterol 150:1171–​1182. PMID: 26873401. Schneider JL, Corley DA. 2015. A review of the epidemiology of Barrett’s oesophagus and oesophageal adenocarcinoma. Best Pract Res Clin Gastroenterol 29:29–​39. PMID: 25743454. Schneider JL, Zhao WK, Corley DA. 2015. Aspirin and Nonsteroidal anti-​ inflmmatory drug use and the risk of Barrett’s esophagus. Dig Dis Sci 60:436–​443. PMCID: PMC4304909. Shen FF, Zhou FY, Xue QS, Pan Y, Zheng L, Zhang H, et al. 2013. Association between CYP1A1 polymorphisms and esophageal cancer:  a meta-​ analysis. Mol Biol Rep 40:6035–​6042. PMID: 24065535. Siddiqi K, Shah S, Abbas SM, Vidyasagaran A, Jawad M, Dogar O, Sheikh A. 2015. Global burden of disease due to smokeless tobacco consumption in adults:  analysis of data from 113 countries. BMC Med 13:194. PMCID: PMC4538761. Simard EP, Ward EM, Siegel R, Jemal A. 2012. Cancers with increasing incidence trends in the United States: 1999 through 2008. CA Cancer J Clin 62:118–​128. PMID: 22281605. Singh S, Garg SK, Singh PP, Iyer PG, El-​Serag HB. 2014. Acid-​suppressive medications and risk of oesophageal adenocarcinoma in patients with Barrett’s oesophagus:  a systematic review and meta-​ analysis. Gut 63:1229–​1237. PMCID: PMC4199831. Singh S, Goyal A, Singh PP, Murad MH, Iyer PG. 2013a. Statins are associated with reduced risk of esophageal cancer, particularly in patients with Barrett’s esophagus:  a systematic review and meta-​analysis. Clin Gastroenterol Hepatol 11:620–​629. PMCID: PMC3660516. Singh S, Sharma AN, Murad MH, Buttar NS, El-​Serag HB, Katzka DA, Iyer PG. 2013b. Central adiposity is associated with increased risk of esophageal inflammation, metaplasia, and adenocarcinoma: a systematic review and meta-​analysis. Clin Gastroenterol Hepatol 11:1399–​1412. PMCID: PMC3873801. Sinha DN, Abdulkader RS, Gupta PC. 2016. Smokeless tobacco-​associated cancers:  a systematic review and meta-​analysis of Indian studies. Int J Cancer 138:1368–​1379. PMID: 26443187. Sitas F, Egger S, Urban MI, Taylor PR, Abnet CC, Boffetta P, et  al. 2012. InterSCOPE study: associations between esophageal squamous cell carcinoma and human papillomavirus serologic markers. J Natl Cancer Inst 104:147–​158. PMCID: PMC3260131.

591

Song Y, Li L, Ou Y, Gao Z, Li E, Li X, et al. 2014. Identification of genomic alterations in oesophageal squamous cell cancer. Nature 509:91–​95. PMID: 24670651. Spechler SJ. 2013. Barrett esophagus and risk of esophageal cancer: a clinical review. JAMA 310:627–​636. PMID: 23942681. Spechler SJ. 2014. Does Barrett’s esophagus regress after surgery (or proton pump inhibitors)? Dig Dis 32:156–​163. PMID: 24603402. Stachler MD, Taylor-​Weiner A, Peng S, McKenna A, Agoston AT, Odze RD, et al. 2015. Paired exome analysis of Barrett’s esophagus and adenocarcinoma. Nature Genet 47:1047–​1055. PMCID: PMC4552571. Steevens J, Schouten LJ, Goldbohm RA, van den Brandt PA. 2011. Vegetables and fruits consumption and risk of esophageal and gastric cancer subtypes in the Netherlands Cohort Study. Int J Cancer 129:2681–​2693. PMID: 21960262. Steffen A, Huerta JM, Weiderpass E, Bueno-​ de-​ Mesquita HB, May AM, Siersema PD, et  al. 2015. General and abdominal obesity and risk of esophageal and gastric adenocarcinoma in the European Prospective Investigation into Cancer and Nutrition. Int J Cancer 137:646–​657. PMID: 25598323. Streppel MM, Lata S, DelaBastide M, Montgomery EA, Wang JS, Canto MI, et al. 2014. Next-​generation sequencing of endoscopic biopsies identifies ARID1A as a tumor-​suppressor gene in Barrett’s esophagus. Oncogene 33:347–​357. PMCID: PMC3805724. Streppel MM, Montgomery EA, Maitra A. 2014. New advances in the pathogenesis and progression of Barrett’s esophagus. Curr Mol Med 14:58–​68. PMID: 23865431. Su Z, Gay LJ, Strange A, Palles C, Band G, Whiteman DC, et  al. 2012. Common variants at the MHC locus and at chromosome 16q24.1 predispose to Barrett’s esophagus. Nature Genet 44:1131–​1136. PMCID: PMC3459818. Sun L, Yu S. 2011. Meta-​analysis:  non-​steroidal anti-​inflammatory drug use and the risk of esophageal squamous cell carcinoma. Dis Esophagus 24:544–​549. PMID: 21539676. Tanaka F, Yamamoto K, Suzuki S, Inoue H, Tsurumaru M, Kajiyama Y, et al. 2010. Strong interaction between the effects of alcohol consumption and smoking on oesophageal squamous cell carcinoma among individuals with ADH1B and/​or ALDH2 risk alleles. Gut 59:1457–​1464. PMID: 20833657. Tang L, Xu, F, Zhang T, Lei J, Binns CW, Lee AHW. 2013. High temperature food and beverage intake increases the risk of oesophageal cancer in Xinjiang, China. Asian Pac J Cancer Prev 14:5085–​5088. PMID: 24175780. Taylor PR, Abnet CC, Dawsey SM. 2013. Squamous dysplasia:  the precursor lesion for esophageal squamous cell carcinoma. Cancer Epidemiol Biomarkers Prev 22:540–​552. PMCID: PMC3681095. Teng H, Li X, Liu X, Wu J, Zhang J. 2014. The absence of human papillomavirus in esophageal squamous cell carcinoma in East China. Int J Exp Pathol 7:4184–​4193. PMCID: PMC4129033. Thrift AP. 2015. Esophageal adenocarcinoma:  the influence of medications used to treat comorbidities on cancer prognosis. Clin Gastroenterol Hepatol 13:2225–​2232. PMID: 25835331. Thrift AP, Cook MB, Vaughan TL, Anderson LA, Murray LJ, Whiteman DC, et al. 2014a. Alcohol and the risk of Barrett’s esophagus: a pooled analysis from the international BEACON consortium. Am J Gastroenterol 109:1586–​1594. PMCID: PMC4189971. Thrift AP, Risch HA, Onstad L, Shaheen NJ, Casson AG, Bernstein L, et al. 2014b. Risk of esophageal adenocarcinoma decreases with height, based on consortium analysis and confirmed by Mendelian randomization. Clin Gastroenterol Hepatol 12:1667–​1676. PMCID: PMC4130803. Thrift AP, Shaheen NJ, Gammon MD, Bernstein L, Reid BJ, Onstat L, et al. 2014c. Obesity and risk of esophageal adenocarcinoma and Barrett’s esophagus:  a Mendelian randomization study. J Natl Cancer Inst 106:dju252. doi: 10.1093/​jnci/​dju252. PMCID: PMC4200028. Tramacere I, La Vecchia C, Negri E. 2011. Tobacco smoking and esophageal and gastric cardia adenocarcinoma: a meta-​analysis. Epidemiology 22:344–​349. PMID: 21330928. Tramacere I, Pelucchi C, Bagnardi V, Rota M, Scotti L, Islami F, et al. 2012. A meta-​analysis on alcohol drinking and esophageal and gastric cardia adenocarcinoma risk. Ann Oncol 23:287–​297. PMID: 21551004. Turati F, Tramacere, I, La Vecchia C, Negri E. 2013. A meta-​analysis of body mass index and esophageal and gastric cardia adenocarcinoma. Ann Oncol 24:609–​617. PMID: 22898040. Tuyns AJ, Pequignot G, Jensen OM. 1977. [Esophageal cancer in Ille-​et-​ Vilaine in relation to levels of alcohol and tobacco consumption. Risks are multiplying]. Bull Cancer 64:45–​60. PMID: 861389. Verbeek RE, Leenders M, ten Kate FJW, van Hillegersberg R, Vleggaar FP, van Baal JWPM, et  al. 2014. Surveillance of Barrett’s esophagus and

592

592

PART IV:  Cancers by Tissue of Origin

mortality from esophageal adenocarcinoma:  a population-​based cohort study. Am J Gastroenterol 109:1215–​1222. PMID: 24980881. Vineis P, Alavanja M, Buffler P, Fontham E, Franceschi S, Gao YT, et al. 2004. Tobacco and cancer:  recent epidemiologic findings. J Natl Cancer Inst 96:99–​106. PMID: 14734699. Wang D, Su M, Tian D, Liang S, Zhang J. 2012. Associations between CYP1A1 and CYP2E1 polymorphisms and susceptibility to esophageal cancer in Chaoshan and Taihang areas of China. Cancer Epidemiol 36:276–​282. PMID: 22088806. Wang F, Lv ZS, Fu YK. 2011. Nonsteroidal anti-​inflammatory drugs and esophageal inflammation–​ Barrett’s esophagus–​ adenocarcinoma sequence:  a meta-​analysis. Dis Esophagus 24:318–​324. PMID: 21166737. Wang FR, Fang QQ, Tang WM, Xu XS, Mahapatra T, Liu YF, et al. 2015. Nested case-​ control study of occupational radiation exposure and breast and esophagus cancer risk among medical diagnostic X ray workers in Jiangsu of China. Asian Pac J Cancer Prev 16:4699–​4704. PMID: 26107226. Wang LD, Zhou FY, Li XM, Sun LD, Song X, Jin Y, et al. 2010. Genome-​wide association study of esophageal squamous cell carcinoma in Chinese subjects identifies a susceptibility locus at PLCE1. Nature Genet 9:759–​763. PMID: 20729853. WCRF/​AICR. 2007. Food, nutrition, physical activity, and prevention of cancer: a global perspective. Washington, DC: AICR. http://​www.aicr.org/​ assets/​docs/​pdf/​reports/​Second_​Expert_​Report.pdf. Weaver JM, Ross-​Innes CS, Shannon N, Lynch AG, Forshew T, Barbera M, et al. 2014. Ordering of mutations in preinvasive disease stages of esophageal carcinogenesis. Nature Genet 46:837–​843. PMCID: PMC4116294. Wild CP, Hardie LJ. 2003. Reflux, Barrett’s oesophagus and adenocarcinoma: burning questions. Nat Rev Cancer 3:676–​684. PMID: 12951586. Wu C, Hu Z, He Z, Jia W, Wang F, Zhou Y, et al. 2011. Genome-​wide association study identifies three new susceptibility loci for esophageal squamous cell carcinoma in Chinese populations. Nature Genet 43: 679–​684. PMID: 21642993. Wu C, Kraft P, Zhai K, Chang J, Wang Z, Li Y, et  al. 2012. Genome-​wide association analyses of esophageal squamous cell carcinoma in Chinese identify multiple susceptibility loci and gene-​environment interactions. Nature Genet 44:1090–​1096. PMCID: PMC3513832. Wu C, Li D, Jia W, Hu Z, Zhou Y, Yu D, et al. 2013. Genome-​wide association study identifies common variants in SLC39A6 associated with length of survival in esophageal squamous-​cell carcinoma. Nature Genet 45:632–​ 638. PMID: 23644492. Wu C, Wang Z, Song X, Feng XS, Abnet CC, He J, et  al. 2014. Joint analysis of three genome-​wide association studies of esophageal squamous cell carcinoma in Chinese populations. Nature Genet 46:1001–​1006. PMCID: PMC4212832. Wu M, Liu AM, Kampman E, Zhang ZF, van’t Veer P, Wu DL, et  al. 2009. Green tea drinking, high tea temperature and esophageal cancer in

high-​and low-​risk areas of Jiangsu Province, China: a population-​based case-​control study. Int J Cancer 124:1907–​1913. PMID: 19123468. Xie FJ, ZhangYP, Zheng QQ, Jin HC, Wang FL, Chen M, et al. 2013. Helicobacter pylori infection and esophageal cancer risk: and updated meta-​analysis. World J Gastroenterol 19:6098–​6107. PMCID: PMC3785633. Xie SH, Lagergren J. 2016. The male predominance in esophageal adenocarcinoma. Clin Gastroenterol Hepatol 14:338–​347. PMID: 26484704. Xu X, Wang J, Zhu SM, Yang M, Fang Y, Zhao A, et al. 2015. Impact of alcohol dehydrogenase gene 4 polymorphisms on esophageal squamous cell carcinoma risk in a Chinese population. PLoS One 10:e0127304. doi: 10.1371/​journal.pone.0127304. PMCID: PMC4454665. Yokoyama A, Kato H, Yokoyama T, Tsujinaka T, Muto M, Omori T, et al. 2002. Genetic polymorphisms of alcohol and aldehyde dehydrogenases and glutathione S-​transferase M1 and drinking, smoking, and diet in Japanese men with esophageal squamous cell carcinoma. Carcinogenesis 23:1851–​ 1859. PMID: 12419833. Yousef F, Cardwell C, Cantwell MM, Galway K, Johnston BT, Murray L. 2008. The incidence of esophageal cancer and high-​grade dysplasia in Barrett’s esophagus:  a systematic review and meta-​ analysis. Am J Epidemiol 168:237–​249. PMID: 18550563. Zandberg DP, Bhargava R, Badin S, Cullen KJ. 2013. The role of human papillomavirus in nongenital cancers. CA Cancer J Clin 63:57–​81. PMID: 23258613. Zendehdel K, Nyrén O, Edberg A, Ye W. 2011. Risk of esophageal adenocarcinoma in achalasia patients, a retrospective cohort study in Sweden. Am J Gastroenterol 106:57–​61. PMID: 21212754. Zhang GH, Mai RQ, Huang B. 2010. Meta-​ analysis of ADH1B and ALDH2 polymorphisms and esophageal cancer risk in China. World J Gastroenterol 16:6020–​6025. PMCID: PMC3007115. Zhang L, Zhou Y, Cheng C, Cui H, Cheng L, Kong P, et al. 2015. Genomic analyses reveal mutational signatures and frequently altered genes in esophageal squamous cell carcinoma. Am J Hum Genet 96:597–​611. PMCID: PMC4385186. Zheng JS, Yang J, Fu YQ, Huang T, Li D. 2013. Effects of green tea, black tea, and coffee consumption on the risk of esophageal cancer:  a systematic review and meta-​analysis of observational studies. Nutr Cancer 65:1–​16. PMID: 23368908. Zhu HC, Yang X, Xu LP, Zhao LJ, Tao GZ, Zhang C, et al. 2014. Meat consumption is associated with esophageal cancer risk in a meat-​and cancer-​histological-​type dependent manner. Dig Dis Sci 59:664–​673. PMID: 24395380. Znaor A, Brennan P, Gajalakshmi V, Mathew A, Shanta V, Varghese C, Boffetta P. 2003. Independent and combined effects of tobacco smoking, chewing and alcohol drinking on the risk of oral, pharyngeal and esophageal cancers in Indian men. Int J Cancer 105:681–​686. PMID: 12740918.

 593

31 Stomach Cancer CATHERINE DE MARTEL AND JULIE PARSONNET

OVERVIEW Stomach cancer is the fifth most common incident cancer worldwide and the third leading cause of cancer death. Almost half of the world’s cases occur in Asia, with 42% in China alone. Although the incidence and mortality from stomach cancer are decreasing, global disease burden remains high. Moreover, the absolute number of cases continues to rise because of population aging. Adenocarcinomas comprise over 90% of gastric malignancies. The adenocarcinomas are further classified according to anatomic location (cardia vs. non-​cardia), histology (e.g., intestinal or diffuse, signet ring or non-​signet ring) and most recently by molecular classification. Adenocarcinomas in the stomach’s body and antrum are usually caused by chronic infection with Helicobacter pylori (H. pylori); the incidence of these tumors is decreasing worldwide. Cardia tumors have epidemiological characteristics more similar to those of esophageal adenocarcinoma; the incidence of these tumors is increasing, particularly in high-​income, Western countries. New molecular classification systems have been proposed based on investigations of tumors in high-​income countries. The Cancer Genome Atlas Program has identified four molecular subtypes: (1) tumors positive for Epstein-​Barr virus; (2) those marked by microsatellite instability; (3) genomically stable tumors; and (4) tumors with chromosomal instability and extensive somatic copy-​ number aberrations. These subtypes have not yet been integrated into etiologic and descriptive studies. The most promising public health strategy for preventing gastric cancer is the eradication of H. pylori with antibiotics. This approach is currently being tested in randomized clinical trials.

INTRODUCTION At the dawn of the twentieth century, stomach cancer represented an astonishing one-​third of all cancers, approximately 1% of hospital admissions, and 2% of all deaths investigated by necropsy (Fenwick and Fenwick, 1903). Although stomach cancer remained the leading cause of cancer death in the world until the mid-​twentieth century, overall, the rapid decline in stomach cancer throughout the last 100 years has been touted as an “unplanned triumph” (Howson et al., 1986). The term “unplanned” highlights the lack of directed intervention in preventing this disease; rather, stomach cancer’s decline parallels the improvements in nutrition, sanitation, and hygiene in the twentieth century. This steady decrease in incidence over time provides insights into stomach cancer etiology, as well as directions for the ultimate elimination of stomach cancer as an important health problem worldwide. Over 90% of stomach cancer cases are adenocarcinomas arising from the gastric glands (Coleman et  al., 1993; World Health Organization [WHO], 2010). Other histologic types of epithelial stomach cancer include adenosquamous carcinoma, carcinoma with lymphoid stroma (i.e., medullary carcinoma), hepatoid carcinoma, squamous cell carcinoma, and the small group of neuroendocrine neoplasms (WHO, 2010). Gastric cancers of non-​epithelial origin include lymphomas and mesenchymal tumors, such as leiomyoma, schwannoma, and Kaposi sarcoma in immunosuppressed patients (WHO, 2010). Secondary tumors are rare, the stomach being one of the five least common metastatic sites (Disibio and French, 2008).

Because they represent the vast majority of gastric tumors, this chapter focuses on adenocarcinomas of the stomach, including cancers of both the gastric cardia (ICD-​O code 16.0) and non-​cardia (ICD-​O codes C16.1–​C16.6) (WHO, 2013).

DISEASE BURDEN According to GLOBOCAN, an estimated 952,000 new cases of gastric cancer occurred worldwide in 2012 (Ferlay et al., 2013), making it the fifth most common cause of cancer worldwide (6.8% of all cancers). More than 70% of cases (677,000) occurred in less developed countries, with 553,000 cases occurring in Eastern Asia, and nearly half of the total number (405,000 cases or 42.5% of the total) in China alone. Global incidence and death rates are shown in Figures 31–​1a and 31–​1b. Europe contributed nearly 15% of the global burden (140,000 cases), and Latin America contributed a further 6% (61,000 cases) (Ferlay et al., 2013). Eastern Europe and the Andes are areas with particularly high risk. In the United States, the American Cancer Society predicts that 26,370 new cases of stomach cancer (16,480 in men and 9890 in women) will be diagnosed in 2016 with 10,730 deaths (6540 men and 4190 women) (American Cancer Society, 2016). This places stomach cancer as the 15th most common malignancy, in terms of both incidence and mortality, in the United States. Survival rates for stomach cancer are generally poor. A  recent international comparison of survival in 279 population-​based cancer registries in 67 countries (Allemani et  al., 2015)  shows that the 5-​ year age standardized net survival from stomach cancer (i.e., the proportion of cancer patients who survive 5  years, after eliminating the background mortality due to other causes) ranges from 15% to 35% for adults. Survival is considerably higher in Japan and South Korea (50%–​60%), where systematic screening allows the early detection and surgical treatment of tumors. It is not known how much of the improvement in survival in these countries represents lead-​time bias, however. Some studies suggest a 30%–​ 60% reduction in mortality with screening (Hamashima et al., 2013; Hamashima et al., 2015). The slope of improvement in mortality has remained relatively constant since the early 1970s, prior to the onset of widespread endoscopic and radiographic screening programs (Whitlock, 2012). Because of its high case-​fatality, gastric cancer accounts for a larger fraction of cancer deaths than incident cases globally (8.8% versus 6.8%). Despite its declining incidence, gastric cancer remains the third leading cause of cancer death worldwide (723,000 deaths estimated per annum), after lung and liver cancers (American Cancer Society, 2016; Ferlay et  al., 2013). The preceding estimates represent the figures for adenocarcinoma, the most common histological type of gastric cancer. The worldwide total number of lymphomas of gastric origin in 2012 was estimated to be 18,000, or less than 2% the number of adenocarcinomas (Plummer et al., 2016). Other gastric histologic types are even less common.

CLASSIFICATION Epidemiologic studies over the last 50  years have classified gastric cancers according to several systems, beginning with histopathology

593

594

594

PART IV:  Cancers by Tissue of Origin

> 15.4 9.7–15.4 6.6–9.7 4–6.6 13 7.2–13 5.5–7.2 3.6–5.5 < 3.6 No Data

Not applicable

Figure 31–​1b.  Estimated worldwide stomach cancer mortality rates per 100,000 in men for the year 2012. Source: GLOBOCAN 2012, IARC, WHO.

in the mid-​1960s, followed by anatomic location in the early 1990s. Molecular classification systems have recently been proposed but have not yet been integrated into epidemiologic studies, nor are molecular markers available at this point from population-​ based tumor registries.

Gastric Anatomy and Function Grossly, the stomach has four anatomical regions: the cardia, the fundus, the body, and the pylorus, as shown in Figure 31–​2. The gastric cardia (also known as the gastroesophageal [GE] or esophagogastric junction [WHO,  2010]) is a narrow circular band, 1.5–​3  cm wide,

located at the junction where the tubular esophagus joins the stomach. The cardia’s proximal border can usually be seen on endoscopy, but the distal boundary is poorly defined (WHO, 2010). The fundus and body, the sections of the stomach that secrete acid, comprise the majority of the stomach. The pylorus is the section of the stomach that transitions from stomach to the small bowel; the proximal portion of the pylorus located within the stomach and before the pyloric sphincter is known as the antrum. All areas of the stomach are covered with mucus-​secreting foveolar and columnar epithelial cells lining the luminal surfaces and invaginations called gastric pits. Midway down the pits are the gastric stem cells; at the base of the pits are the glands. Secretory glands in the

 59

595

Stomach Cancer

CIN

• Intestinal histology • TP53 mutation • RTK-RAS activation

Cardia GE Junction Fundus EBV

Body

Pylorus

• PIK3CA mutation • PD-L1/2 overexpression • EBV-CIMP • CDKN2A silencing • Immune cell signaling

Antrum MSI

• Hypermutation • Gastric-CIMP • MLH1 silencing • Mitotic pathways

GS • Diffuse histology • CDH1, RHOA mutations • CLDN18–ARHGAP fusion • Cell adhesion

Figure 31–​2.  Cancer Genome Atlas: key features of the four tumor types by gastric site. MSI = microsatellite instability; EBV = Epstein Barr virus; CIN = chromosomal instability; GS = genomically stable. Source: The Cancer Genome Atlas Research Network, Comprehensive molecular characterization of gastric adenocarcinoma. Adapted from Nature 2014;513(7517):202–​209.

body and fundus of the stomach produce hydrochloric acid and intrinsic factor (parietal cells), serotonin (enteroendocrine or Kulchitsky cells), and pepsinogen, leptin, and lipase (chief cells). Enteroendocrine cells in the gastric antrum secrete gastrin (G cells) and somatostatin (D cells). Cardiac and pyloric glands have neither parietal nor chief cells and largely secrete mucus. The length of the cardiac mucosa increases with age and with central obesity, often developing histopathological signs of moderate to severe inflammation (Derakhshan et  al., 2015; Miao et al., 2014). When H. pylori infection occurs, it typically starts in the pylorus, extending to the body and fundus over time, causing infiltration of the mucosa with inflammatory cells—​a condition known as chronic, active gastritis (Correa and Piazuelo, 2008).

Anatomic Subtypes In the United States, overall rates of stomach cancer have continued to decline over the last three decades from an age-​standardized rate of 11.7 per 100,000 in 1975 to 6.7 per 100,000 in 2013. However, in the 1990s, investigators noted strikingly different epidemiologic trends in the incidence rate of cardia versus non-​cardia tumors: gastric cardia cancer increased continuously beginning in the mid-​1970s, whereas the incidence of non-​cardia cancers sharply declined (Blot et al., 1993; Henson et al., 2004; Levi et al., 1990; Powell and McConkey, 1992; Wu et al., 2009). Since 1990, however, the age-​standardized incidence rate of cardia cancer—​unlike adenocarcinoma of the esophagus—​seems to have stabilized in the United States at approximately 2.1 per 100,000 (Figure 31–​3; Wu et al., 2009). In contrast, the incidence rate of non-​cardia cancer in the overall population has decreased progressively to 4.1 per 100,000, although incidence remains higher among Asians, blacks, and Hispanic whites (Devesa et al., 1998). Similar trends have been reported worldwide. In some countries of Europe and the Americas, the incidence of cardia cancers now approximates or exceeds that of non-​cardia cancer among men (Derakhshan et al., 2016; WHO, 2010).

Although the differentiation of anatomical subsite of gastric cancers has improved over time (Camargo et  al., 2011a), it can be difficult to determine the origin of large tumors that bridge the lower esophagus and upper stomach (Misumi et  al., 1989)  or that arise in the poorly defined distal boundary of the gastric cardia. As of 2016, tumors registered with overlapping (ICD-​O C16.8) or unspecified (C16.9) anatomical subsites constitute nearly half (40%–​50%) of all stomach cancers in Japan, Korea, and the United States; 50%–​60% in the United Kingdom, Italy, and Australia; and up to 90% in Brazil (Machii and Saika, 2016).

Histologic Subtypes As noted earlier, over 90% of stomach cancer cases are adenocarcinomas arising from the gastric glands (Coleman et al., 1993; WHO, 2010). Several histologic classification systems have been proposed for gastric adenocarcinoma. Among these, the Laurén system has been the most influential.

Laurén Classification Described by Pekka Laurén in 1965, the Laurén classification divides gastric cancer into two histological types:  intestinal and diffuse (Laurén, 1965; Nevalainen, 2013). Intestinal-​type cancers form recognizable glands, whereas diffuse adenocarcinomas consist of poorly cohesive cells that infiltrate diffusely into the gastric wall with little or no gland formation (WHO, 2010). These two types are usually easy to distinguish under the microscope. Moreover, since intestinal and diffuse tumors differ in their clinical, epidemiologic, and etiologic characteristics, the Laurén classification provides a useful framework for understanding pathology, epidemiology, and clinical treatment. Because many of the seminal epidemiologic and clinical investigations of gastric adenocarcinoma have employed the Laurén classification

596

596

PART IV:  Cancers by Tissue of Origin

Rate per 100,000 person-years

10

Diffuse

10

Intestinal

10

1

1

1

0.1

0.1

0.1

Other Types

Anatomic site Cardia Specified Non-Cardia Overlapping and Non-Specified 0.01

0.01

0.01 1980

1990

2000

1980 1990 2000 Year of diagnosis

1980

1990

2000

Figure 31–​3.  Trends in stomach carcinoma incidence by histologic type and anatomic site (SEER 1978–​1983 to 2001–​2005). Source: Wu H et al., Cancer Epidemiol Biomarkers Prev 2009;18:1945–​1952.

system, the terms “intestinal” and “diffuse” will be used frequently throughout this chapter. The intestinal subtype historically represented the majority of gastric adenocarcinomas worldwide, and the decrease in this subtype has been the major contributor to the overall decline in stomach cancer (Figure 31–​3). In contrast to the decreasing incidence rate of the intestinal subtype, the incidence of diffuse gastric cancers has increased by over 400% (Henson et  al., 2004). Consequently, the proportion of gastric cancers contributed by the two subtypes has changed dramatically over time. In the early 1970s, intestinal-​ type cancers represented over 90% of gastric adenocarcinomas in the United States; by 2000, this proportion had decreased by 50%. Similar trends have been reported in Europe (Laurén and Nevalainen, 1993), Latin America (Rampazzo et  al., 2012), and Asia (Hajmanoochehri et al., 2013; Kaneko and Yoshimura, 2001), although one study from Sweden showed decreases in both tumor types (Ekstrom et al., 2000a). The Laurén classification has also proven useful in evaluating the natural history of gastric cancer and temporal trends in cancer incidence and precursor lesions, as discussed in the following (Lewin and Appelman, 1995; WHO, 2010).

Perhaps because of its increased complexity, the WHO classification has been used less frequently in epidemiologic studies than the Laurén system. Among the intestinal-​type tumors, tubular carcinomas are the most common, accounting for 56% of stomach tumors in some studies (Terada, 2016). Papillary carcinomas (13% of tumors) tend to occur in the proximal part of the stomach and to be more aggressive than other histologic types, with higher likelihood of metastasis and mortality. Table 31–​1.  Comparison of Laurén and WHO Gastric Adenocarcinoma Classifications Laurén Classification Intestinal type

Papillary adenocarcinoma Tubular adenocarcinoma Mucinous adenocarcinoma

Diffuse type

Signet ring cell carcinoma Other poorly cohesive carcinoma

Not classified

Rare adenocarcinomas including: Hepatoid adenocarcinoma Oncocytic adenocarcinoma Paneth cell carcinoma Gastric carcinoma with lymphoid stroma and others

WHO Histologic Classification In 2010, the WHO devised a more granular classification for gastric cancer (Hu et al., 2012; WHO, 2010) (Table 31–​1). The intent of this new classification was to harmonize the histological typing of gastric cancers with that used for cancers of the small and large intestine.

WHO Classification

(Hu et al., 2012)

 597

Stomach Cancer Mucinous adenocarcinomas comprise approximately 10% of gastric adenocarcinomas, and can have features of both intestinal and diffuse-​ type malignancies. Signet ring cell and other poorly cohesive carcinomas comprise the majority of diffuse types and tend to be highly invasive. In their early stage, hereditary gastric malignancies usually manifest themselves as patchy intramucosal distributions of signet ring cells (Hu et al., 2012).

Molecular Classifications In 2014, the Cancer Genome Atlas program (TCGA) tested 295 gastric cancer samples with array-​based analyses of somatic copy number, whole-​ exome sequencing, array-​ based DNA methylation profiling, messenger RNA sequencing, microRNA sequencing, reverse-​phase protein array, microsatellite instability testing, and, on a subset, low-​ pass whole genome sequencing (The Cancer Genome Atlas [TCGA] Research Network, 2014). These methods identified four molecular subtypes of gastric cancer: (1) tumors positive for Epstein-​Barr virus (EBV); (2) tumors marked by microsatellite instability (MSI); (3) genomically stable tumors (GS); and (4) tumors with chromosomal instability (CIN), characterized by the presence of extensive somatic copy-​ number aberrations (The Cancer Genome Atlas Research Network, 2014) (Figure 31–​2). The last of these subtypes comprised approximately half of the tested tumors, whereas the EBV tumors represented less than 10%. Although TCGA found no prognostic significance associated with these categories in their small data set, they did identify statistical associations with gender (men were more likely to have EBV-​related tumors), anatomic subsite (CIN more likely in the cardia and EBV more likely in the body), Laurén classification (75% of GS were diffuse-​type), and age (GS tumors occurred at younger ages and MSI, older). Four subtypes with different nomenclature were also identified by the Asian Cancer Research Group (ACRG), based on gene expression profiles, and later were confirmed using the three other cohorts, including TCGA (Cristescu et al., 2015). These subtypes were labeled microsatellite instability (MSI), microsatellite stable-​epithelial to mesenchymal transition (MSS/​EMT), MSS/​tumor protein 53 (TP53) positive, and MSS/​TP53 negative. Unlike the TCGA subtypes, the ACRG subtypes were clearly linked to prognosis. Recurrence and survival were worst for the MSS/​EMT tumors, which typically displayed loss of e-​cadherin (CDH1) expression and diffuse, signet ring histopathology. Prognosis was best for MSI tumors. Other studies have categorized tumors based on mutations of single genes. For example, germline mutations in CDH1 account for most familial gastric cancers. More recently, tumors of the cardia and the gastroesophageal junction have been classified based on overexpression of the human epidermal growth factor 2 (HER-​2) receptor (Gravalos and Jimeno, 2008). Because these more aggressive tumors may respond favorably to monoclonal antibody treatments, expression of HER-​2 may prove to be useful as a clinical marker (Bang et al., 2010).

TUMOR GRADE AND STAGE The grading of gastric adenocarcinoma is based on the degree of differentiation (WHO, 2010). Well-​differentiated tumors have well-​ formed glands that often resemble metaplastic intestinal epithelium, whereas poorly differentiated tumors are composed of highly irregular glands that are recognized with difficulty, or single cells that remain isolated or are arranged in small or large clusters. Moderately differentiated tumors display patterns intermediate between well and poorly differentiated.

TNM Staging The system most often used to stage stomach cancers in the United States is the American Joint Commission on Cancer (AJCC) TNM system. The TNM staging considers the size and the extent of the primary

597

tumor (T), the involvement of regional lymph nodes (N), and presence or absence of distant metastasis (M) (Washington, 2010). The staging system used by the SEER program classifies cancer cases into three categories: localized, regional, and distant. In the United States, 27% of stomach cancer cases reported to SEER registries between 2006 and 2012 were diagnosed when localized, 28% were regional, and 35% were distant. The remaining 10% were classified as unknown stage at diagnosis (Howlader et al., 2015). Other classification systems for stomach cancer make the distinction between early disease (i.e., invasive cancer not extending beyond the submucosa) and advanced disease (i.e., invasion of the muscularis or beyond). This approach is commonly used in countries that employ routine endoscopic screening for gastric cancer.

PRECANCEROUS OR PRECURSOR LESIONS For nearly a century, it has been known that gastric cancers with the intestinal-​type histology typically originate from mucosa affected by chronic gastritis (Hartfall, 1936). It was not until the advent of modern endoscopy, however, that serial inspection of unautolyzed specimens clarified the temporal sequence of preneoplastic histopathology. For intestinal-​type tumors, the neoplastic cascade begins with chronic gastritis, an inflammatory condition in which both acute and chronic inflammatory cells invade the mucosa, and mucosal cell turnover is accelerated. In a subset of people, the chronic gastritis progresses to chronic atrophic gastritis with loss of gastric glands and diminished ability to produce gastric acid. As the atrophic gastritis progresses to involve large areas of the stomach, the mucosa may develop islands of intestinal metaplasia in which the gastric mucosa recapitulates the morphologic appearance of intestinal mucosa. These metaplastic zones can have different histologic morphologies termed “complete” (type I) or “incomplete” (types II and III), as discussed later, and can eventually extend to involve much of the stomach. After many years, the mucosa in a subset of individuals progresses to low-​and then high-​ grade gastric epithelial dysplasia and ultimately to early and then invasive intestinal-​type cancer.

Preneoplasia Correa and colleagues (Correa, 2002; Correa et al., 1976; Correa and Yardley, 1992) have proposed a multistep premalignant process for the intestinal-​type gastric carcinoma, involving a sequence of histopathological changes in the mucosa from normal to non-​atrophic gastritis, multifocal atrophic gastritis, intestinal metaplasia, and dysplasia. In high-​risk populations, chronic gastritis predominantly affects the antrum and is associated with multifocal atrophic gastritis and gland loss (Correa, 2002). The metaplastic process tends to begin at the antrum-​corpus junction, then enlarge and extend to the antrum and/​ or the corpus. A patchy distribution of dysplastic foci may eventually appear within the area of intestinal metaplasia. As mentioned, the two main types of intestinal metaplasia are “complete” (also called small intestinal type or type I) and “incomplete” (also called colonic type or types II and III). The epithelium in the complete type resembles the small intestinal phenotype, small intestine digestive enzymes are present, and only sialomucins are expressed. In incomplete metaplasia, small intestine digestive enzymes are absent or only partially expressed, the epithelium resembles the colonic phenotype, and sulfomucins are expressed, either in combination with (type II) or without (type III) sialomucins (Camargo et  al., 2011a). Incomplete metaplasia is frequently associated with frank dysplasia and early carcinoma. Controversies still exist regarding the utility of subtyping intestinal metaplasia as a marker of stomach cancer risk. However, the results from prospective cohorts suggest that complete intestinal metaplasia occurs first and may transform over time into incomplete metaplasia, which has a much higher risk of malignant transformation (Camargo et al., 2014). Furthermore, the great majority of individuals with intestinal-​type gastric cancer have some evidence of metaplasia in surrounding tissue (Lauwers and Srivastava, 2007).

598

598

PART IV:  Cancers by Tissue of Origin

Dysplasia (also called intraepithelial neoplasia), which arises in either native gastric or intestinalized gastric epithelia, is characterized by partial or complete loss of differentiation (IARC, 2010). The magnitude of risk of cancer for each of these precursor stages is difficult to measure, due to their patchy distribution, inconsistent terminology among pathologists, and the challenge of performing repeated endoscopy in large cohort studies. To rectify some of these problems, several groups (e.g., Padova International Classification, Vienna classification, Paris endoscopic classification) (Rugge et al., 2000; Schlemper et al., 2000; The Paris Classification, 2003) have tried to eliminate inconsistencies and semantic misunderstandings across countries—​especially between Japan and Europe/​North America—​regarding the terminology for preneoplastic lesions and early invasive cancers (IARC, 2010; Stolte, 2003). Despite differences in nomenclature, however, it is clear that each step increases risk, even though only a minority of cases ultimately progress from preneoplasia to invasive cancer. In a nationwide cohort study in the Netherlands, the annual incidence of gastric cancer was 0.10% for patients whose most severe premalignant lesion was chronic atrophic gastritis, 0.25% for those with intestinal metaplasia, 0.60% for those with mild-​to-​moderate dysplasia, and 6.0% for those with severe dysplasia (de Vries et al., 2008). A large study from China demonstrated a similar magnitude of increasing risk with advancing precancerous conditions: < 0.1% patients with atrophy gastritis, 2.7% with deep intestinal metaplasia, and 7% with moderate or severe dysplasia developed cancer over 5 years (Camargo et al., 2011a). More recently, data from a low-​risk Swedish cohort of over 400,000 people indicated that, over 20 years, one of 256 individuals with normal gastric mucosa would develop cancer (standardized incidence ratio [SIR] = 1.0) compared to 1 in 85 for gastritis (SIR = 1.8), 1 in 50 for atrophic gastritis (SIR = 2.8), 1 in 39 for intestinal metaplasia (SIR = 3.4), and 1 in 19 for dysplasia (SIR = 6.5) (Song et al., 2015). Precursor lesions can be identified via endoscopy or by using serum markers for chronic atrophic gastritis. Because gastric glands, and the chief cells in the glands, can be destroyed in atrophic gastritis, biomarkers for proteins produced by chief cells have been used to detect this damage. The most studied of these proteins are pepsinogens I and II. Both low pepsinogen I and a low ratio of pepsinogen I to pepsinogen II have been used clinically (Charvat et al., 2016; Yamaguchi et al., 2016), although the sensitivity and specificity for the detection of atrophic gastritis are modest (approximately 65% and 85%, respectively) (Huang et al., 2015). Given the low incidence of gastric cancer and its precursors, pepsinogen screening is not recommended in the United States or other low-​incidence countries due to the low predictive power of positive tests (PDQ® Screening and Prevention Editorial Board, 2016). No precursor lesions have yet been identified for diffuse cancer or the rarer subtypes of adenocarcinoma.

2. Chemotactism 3. BMDC recruitment NFκB TNFα/SDF1

In 1975, Pelayo Correa proposed a model for intestinal-​type gastric carcinogenesis positing that gastric cancer arose from chronic gastritis and that subsequent steps in tumor progression were caused by other exposures or cofactors (Correa et  al., 1975). With the discovery of H. pylori in the 1980s and subsequent research confirming its central role in causing chronic gastritis, Correa revised his model to incorporate H. pylori infection as the initiating step in carcinogenesis (Correa, 2004). In this model, dietary factors—​including salt-​preserved foods, high fats, high nitrates, and decreased fruits and vegetables—​fostered subsequent tumor progression. Bacterial overgrowth in the hypochlorhydric stomach was also thought to contribute to mutation and mucosal changes by inducing the formation of carcinogenic N-​nitroso compounds and free radicals. Some of the stages of the Correa model have been confirmed experimentally, notably the role for salts in magnifying damage in H. pylori–​infected stomachs (D’Elia et al., 2014; Gaddy et  al., 2013). Molecular aspects of this model of tumor progression are also being elucidated. For example, metaplasia associated with gastric overproduction of a trefoil protein normally found in the intestine (spasmolytic polypeptide) is a particularly strong marker for cancer risk. A different model of tumor progression was proposed by Houghton et al. (2004). These researchers showed experimentally that H. pylori–​ related tumors of the stomach derive from bone marrow–​derived stem cells, rather than resident peripheral stem cells or mucosal cells. The bone marrow–​derived stem cells are recruited to the stomach by the chronic inflammatory process, fuse with gastric epithelial cells, and then replicate, regenerating the mucosal surface and replacing the original stem cells that were destroyed by chronic gastritis (Bessede et al., 2015) (Figure 31–​4). As with the Correa model, chronic inflammation due to H. pylori infection is the dominant factor in the carcinogenic process. Some studies suggest that precursor lesions can be reversed or their progression aborted by eradication of Helicobacter pylori infection (Fukase et  al., 2008). Thus, several international guidelines recommend H.  pylori eradication for those with precancerous conditions and even early gastric cancer. The latter are thought to be true malignancies that have not yet invaded the muscularis. They can be cured by the combination of surgical excision and H.  pylori eradication (Malfertheiner et al., 2012; Rollan et al., 2014). No analogous tumor progression model has yet been identified for sporadic diffuse gastric cancers. However, genome-​wide analysis studies (GWAS) have identified several germline mutations associated with hereditary diffuse gastric cancer (HDGC). Here, the initial stages of carcinogenesis are postulated to begin with germline

GFP+BMDC

Chronic infection with H. pylori

1. Inflammation and epithelial damages

TUMOR PROGRESSION MODELS

4. Homing and differentiation cell/cell fusion

CD44

EMT affecting EMT confering differentiation CSC properties 5. Altered 6. Dysplasia during differentiation regenerative Composed at 22% of BMDC /metaplasia hyperplasia Composed of CD44+ cells with CSC properties

7. Emergence of CD44high CSC and carcinoma

Figure 31–​4.  Stem cell model of gastric carcinogenesis. EMT = epithelial-​mesenchymal transition; CSC = cancer stem cells; BMDC = bone marrow-​derived cells. Source: Bessede E et al., Helicobacter pylori infection and stem cells at the origin of gastric cancer. Oncogene 2015;34(20):2547–​2555.

 59

Stomach Cancer mutations—​frameshifts, point mutations, or deletions—​that typically affect one of two genes:  CDH1 or, less commonly, CTNNA1 (alpha e-​catenin) (Caldas et  al., 1999; Majewski et  al., 2013). The loss of function of the second allele in HDGC patients—​either due to loss of heterozygosity or promoter hypermethylation—​may then lead to gastric cancer (Grady et al., 2000; Majewski et al., 2013).

SURVIVAL Survival among patients with stomach cancer is still among the lowest of all cancer sites in most regions of the world (Allemani et al., 2015). Only cancers of the lung, liver, and pancreas have worse survival. Five-​year net survival from stomach cancer diagnosed between 2005 and 2009 is generally in the range of 25% to 30%. Five-​year relative survival is slightly better in some Western countries (i.e., up to 33% in Belgium and Austria) and in China (31%) and substantially better in Japan and Korea (58% and 54%, respectively). The success in these latter countries is due to the larger proportion of early-​stage, curable cancers diagnosed through the intensive screening for stomach cancer in these countries. Between the periods 1995–​1999 and 2005–​2009, survival statistics have shown very large increases in South Korea (33% to 58%) and China (15% to 31%), but relative survival rose by less than 10% in most other locations (Allemani et al., 2015). Diagnosis at an early stage is critical for survival. In stomach cancer patients in the United States diagnosed between 1992 and 1998, the overall relative survival at 5 years for all disease stages combined was 22% (Jemal et al., 2003). Stage-​specific survival was the highest for localized disease (59%), followed by regional disease (22%), and lowest for distant disease (2%) (Jemal et al., 2003). Between 2006 and 2012, the overall relative survival improved slightly to 30%. Stage-​ specific survival had also improved since 1992–​1998, with 67%, 31%, and 5% observed for localized, regional, and distant disease, respectively (American Cancer Society, 2016). Survival is lower for patients with cancer of the cardia than for non-​cardia cancers, even in early stage disease (Amini et al., 2015).

DESCRIPTIVE EPIDEMIOLOGY Traditionally, descriptive analyses of stomach cancer have classified gastric tumors as either a single entity or as cardia versus non-​cardia cancers. Mortality data do not capture either the anatomic location or the histologic characteristics of the tumor. Even when tumors are subclassified as cardia or non-​cardia, tumor registration does not currently capture the heterogeneity that particularly affects cardia cancers. These are thought to represent a mix of at least two etiologies: the first involves severe atrophic gastritis due to H. pylori, similar to the pathway for non-​cardia cancers; the second is thought to result from a transformation of squamous to columnar metaplasia due to the reflux of gastric contents, similar to the development of esophageal adenocarcinoma (Derakhshan et  al., 2008; Miao et  al., 2014). The former predominates in high-​risk populations where chronic infection is still common, such as Northwest Iran and central China. The latter predominates in Western countries where central obesity and esophageal adenocarcinoma are common (Derakhshan et al., 2015). Cardia and non-​cardia cancers differ in their demographic, geographical, and temporal distributions. The extent of these differences depends on the mix of the two etiologies of cardia cancer (H. pylori–​ related versus reflux-​related) in a given area. We describe in the following the epidemiological features of the most frequent non-​cardia subsite, and highlight the differences with the cardia subsite at the end of each paragraph, when necessary.

Demographic Characteristics Stomach cancer is extremely rare before the age of 30. After this age, the age-​specific incidence rate of all subsites combined increases slowly until the age of 50 and then increases more sharply, with the highest incidence seen in the oldest age groups. In the United

599

States, the median age at diagnosis is 69 for both sexes (Derakhshan et al., 2009; Sipponen and Correa, 2002). At all ages, men are twice as likely as women to develop non-​cardia stomach cancer and four times as likely to develop cardia cancer. Interestingly, the male-​to-​ female ratio varies across countries, and is reported to be only 1.5 in Sub-​Saharan Africa. This variation probably reflects differences in the prevalence of the main etiologic factors for cardia cancer, as mentioned earlier (Colquhoun et  al., 2015). Within each gender, African Americans and Asian Pacific Islanders have twice the incidence rate of overall stomach cancer than non-​Hispanic whites (Colquhoun et al., 2015). Within any country or population, non-​cardia gastric cancer is most often seen in lower socioeconomic groups and has been associated with risk factors that correlate with lower socioeconomic status (SES; i.e., lower income, educational level, and occupational standing, greater number of siblings, and crowding). These factors also correlate with H. pylori infection. A large European multicenter study reported that adjustment for H. pylori infection eliminated the association between lower SES and non-​cardia gastric cancer (Nagel et al., 2007). Other factors, such as fruit and vegetable consumption, cigarette smoking, and physical activity, may also confound any observed association with SES. Notably, although higher SES is inversely associated with non-​cardia gastric cancer, it is strongly associated with cardia gastric cancer and esophageal adenocarcinoma, especially in the middle-​and high-​income countries.

Geographic Variation A global map of male age-​ standardized incidence rates from GLOBOCAN 2012 shows that the highest rates of stomach cancers occur in Eastern and Southeastern Asia, Eastern Europe, and parts of Central and South America. The map of female incidence rates is nearly identical except that, in any given country, the rates in women are approximately half those in men. Internationally, the incidence of stomach cancer varies approximately 10-​fold among men, from more than 60 per 100,000 in Korea to less than 6 per 100,000 in the United States, Sweden, or Kuwait (Forman et al., 2013). The extremely high incidence rates among men in Korea (62 per 100,000) and Japan (46 per 100,000) may partly reflect the intensive endoscopic surveillance conducted in these countries. Screening detects even very small lesions that might not otherwise progress or be diagnosed. Even excluding Korea and Japan, however, a four-​fold variation in male incidence persists, with rates over 20 per 100,000 in Belarus, Lithuania, the Russian Federation, and Costa Rica. Within Europe, there is considerable variation between the countries at highest risk (generally in Eastern Europe) and those with the lowest risk (in Scandinavia, Switzerland, and the United Kingdom). In the United States, the National Cancer Institute’s Surveillance, Epidemiology, and End Results Program (SEER), estimates a national standardized incidence rate for stomach cancers using nine population-​ based cancer registries (San Francisco–​Oakland, Connecticut, Detroit, Hawaii, Iowa, New Mexico, Seattle–​Puget Sound, Utah, and Atlanta). For the period 2008–​2012, annual incidence rates per 100,000 were 10.1 for men and 5.3 for women, for all races combined. Age-​ standardized incidence rates per 100,000 were higher in blacks than in whites for both men (14.6 vs. 9.2) and women (8.4 vs. 4.5) (Howlader et al., 2015). In individual registries, age-​standardized incidence rates for white men range from 5.4 per 100,000 in Utah to 11.7 in Los Angeles, and those for white women range from 2.4 per 100,000 in Hawaii to 6.9 in Los Angeles. Incidence rates of stomach cancer for black men range from 11.5 per 100,000 in Greater California to 19.1 in Louisiana, and for black women from 6.2 per 100,000 in Kentucky to 10.5 in Connecticut. Male rates are approximately two-​fold higher than female rates across the registries for both blacks and whites. On average, among the 18 SEER registries during the period 2008–​2012, black males had the highest incidence (14.6 per 100,000) followed by Asian or Pacific Islanders (14.5), Hispanic (14.2), American Indian/​ Alaska Native (12.3), and white men (9.2). Female rates showed a similar pattern, with the highest incidence among Asian or Pacific

60

600

PART IV:  Cancers by Tissue of Origin

Islanders (8.8), followed by blacks and Hispanics (both 8.4), American Indians/​Alaska Natives (7.5), and whites (4.5) (Howlader et al., 2015).

Migrant Studies Comparisons of cancer risk among people who migrate from a low-​ risk to a high-​risk population, or vice versa, can provide insights into the relative contributions of inherited versus acquired risk factors. In a landmark study of Japanese migrants to Hawaii and their descendants (Kolonel et al., 1985), age-​adjusted incidence rates of stomach cancer in the 1970s were lower in Japanese-​born migrants to Hawaii (“Issei” or first generation) than among the Japanese in Japan. The rates were even lower in the Hawaiian-​born Japanese (“Nisei” or second generation), although still higher than the Caucasian rates. Migrants to Australia from seven European countries with higher stomach cancer rates than Australia (England, Scotland, Ireland, Poland, Yugoslavia, Greece, and Italy) showed a risk reduction with increased duration of residence in Australia (McMichael et  al., 1980). A  study of migrant populations within Italy (Fascioli et  al., 1995)  suggested that place of birth was a stronger predictor of stomach cancer risk than current place of residence. Mortality data from several European countries also showed a closer relation of stomach cancer risk to county of birth than county of death (Coggon et al., 1990; Spallek et al., 2012), indicating the persistent significance of environmental factors in earlier life. Collectively, these data suggest that environmental factors acting early in life have a crucial role in gastric carcinogenesis.

Temporal Trends Over the last half century, the overall incidence and death rates from stomach cancer have decreased steadily almost everywhere. The database Cancer Incidence in Five Continents (CI5plus) contains updated annual incidence rates for 118 selected populations from 102 cancer registries through 2007 (Ferlay et al., 2013). A comparison over a 30-​ year period (1977–​2007) shows that, apart from year-​to-​year sporadic fluctuations, the trends of the decline of stomach cancer incidence are remarkably similar in men and in women, and in high-​risk countries such as Japan and low-​risk countries such as Denmark. In the United States, incidence rates have decreased by more than 80% since 1950 (Siegel et al., 2016). The magnitude and consistency of this downward trend worldwide parallels the decrease in the prevalence of H. pylori. Analysis of cancer registry data over time has provided an average estimated annual percentage change (EAPC) in gastric cancer incidence of  –​2.5% per annum (Bray et  al., 2012). The decrease in the incidence rate is outweighed, however, by the increase in the world’s population. Thus, the absolute number of gastric cancer cases is expected to increase from 952,000 new cases in 2012 to 1,520,000 new cases by 2030. Mortality trends from GLOBOCAN (Ferlay et al., 2013) show that the decline in mortality is slightly greater and shows more variability across countries than incidence, suggesting that some improvement in survival has accompanied the reduction in the incidence rate. A recent global overview of gastric mortality for the period 1980–​2010 demonstrated an average EAPC of –​3.7% in men in the European Union, –​ 2.7% in the United States, and  –​4.1% in Korea (Ferro et  al., 2014). The rate of decrease has slowed in the United States, France, and some other European countries. In the United States, the EAPC decreased to  –​1.6 during the interval 2006–​2010. A  European study of birth cohort trends shows that the rate of decrease has diminished among cohorts born after the 1940s, particularly among women (Malvezzi et al., 2010). Incidence trends also vary by histological subtype and anatomic site, with widespread decreases in non-​cardia and intestinal cancers but increases in cardia and, less commonly, diffuse cancers (Kaneko and Yoshimura, 2001; Wu et  al., 2009). The SEER Program in the United States identified a strong decrease of the intestinal type in both sexes and all age groups, whereas the diffuse type—​particularly the signet ring cell type—​progressively increased (Henson et al., 2004). The increase in signet ring cell type has not been found in other

populations, however (Chapelle et al., 2016; Kaneko and Yoshimura, 2001). A SEER study by Wu et al. showed a 23% increase in cardia cancer (ICD code 16.0) from 1978–​1983 to 1996–​2000, whereas non-​ cardia cancer (ICD code 16.1–​6) and overlapping/​unspecified cancer of the stomach (ICD code 16.8–​9) decreased (Wu et al., 2009) (Figure 31–​ 2). Although the increase in cardia cancer may partly reflect improvements in gastric cancer subsite recording during the period, a true increase is likely and has been reported in other studies in the United States and Northern Europe, including the United Kingdom (Steevens et  al., 2010). Data from the nine oldest US SEER cancer registries (covering 10% of the population) showed that the incidence rates for cardia cancer significantly increased among whites from 1976 to 2007, but did not change among blacks or other races (Camargo et al., 2011a). This is consistent with white males having higher rates of cardia cancer than blacks, in contrast to the opposite pattern for gastric cancer overall.

ETIOLOGIC FACTORS The strongest risk factor for stomach cancer identified to date is chronic infection with the bacterium Helicobacter pylori (H. pylori). Because H. pylori infection was established as a risk factor for stomach cancer relatively recently (IARC, 1994), studies conducted earlier did not have information on H. pylori infection. Some risk factors and protective factors, such as smoking and certain dietary components, may be correlated with H.  pylori infection, leading to confounding. Alternatively, non–​H.  pylori risk factors and protective factors may modify the risk due to H. pylori infection, which is likely since only a small minority of people infected with H. pylori ever develop stomach cancer. Hence, the most informative studies are those that collect information on both H. pylori and other putative risk factors.

Helicobacter pylori Historical Aspects

Long before the discovery of H. pylori, gastric adenocarcinoma was known to arise within areas of gastritis. The type of gastritis associated with cancer—​termed “chronic active gastritis” or “chronic type B gastritis” because of the presence of both lymphocytes and neutrophils—​ was extremely common in elderly populations and was thought to be a natural consequence of aging. In detailed studies from Northern Europe and Latin America, the majority of individuals over the age of 50 years were found to have chronic type B gastritis (Correa et al., 1976; Siurala et al., 1985). H. pylori had been described almost a century (Kreinitz, 1906) before Marshall and Warren’s groundbreaking work (Warren and Marshall, 1983). However, with the rediscovery of H. pylori in the early 1980s, the idea that gastric inflammation preceded cancer took on new meaning. Experimental ingestions and clinical trials of H. pylori eradication all demonstrated that H. pylori was the dominant cause of type B gastritis (Dixon et al., 1996), prompting reconsideration of traditional dietary and genetic theories of gastric carcinogenesis. If H. pylori caused gastritis and gastritis was a precursor to malignancy in the majority of cases, H. pylori was likely to be a critical factor in carcinogenesis. In 1994, an expert working group convened by the International Agency for Research on Cancer (IARC) classified infection with H. pylori as carcinogenic to humans, based on its association with gastric adenocarcinoma and MALT lymphoma (IARC, 1994). This conclusion was upheld in 2009 by a second IARC working group (IARC, 2012a), with the added precision that H. pylori was designated a cause of non-​cardia gastric carcinoma, the most common subtype globally. It was recently estimated that 89% of non-​cardia gastric cancers worldwide, or 730,000 incident cases in 2012, were attributable to H. pylori (Plummer et al., 2015).

Evidence for Carcinogenicity

Helicobacter pylori is a spiral Gram-​negative bacterium that colonizes the stomach. Although most infections are asymptomatic, H.  pylori

 601

Stomach Cancer is associated with chronic gastritis, peptic ulcer disease, gastric B-​ cell mucosa-​ associated lymphoid tissue (MALT) lymphoma, and gastric adenocarcinoma. It is believed that H. pylori was once ubiquitous in humans, but that its prevalence has declined in successive birth cohorts, especially in Western Europe, North America, Oceania, and Japan, so that infection is now rare among children (Herrera and Parsonnet, 2009). The risk of H. pylori infection is associated with low socioeconomic status, particularly with overcrowding and poor sanitation (Palli et al., 1994; Rothenbacher et al., 1999). Thus the gradual disappearance of H. pylori in these regions may be largely a byproduct of economic development. The widespread use of antibiotics and improvements in diet may also have played a role. It is noteworthy that the reduction in H.  pylori prevalence matches the decline in gastric cancer incidence and mortality. Almost all of the epidemiological evidence on the relationship between H. pylori and gastric cancer comes from serological assessment of anti–​H.  pylori immunoglobin G (IgG) antibodies. It is now widely accepted that serological assessment of H. pylori infection has poor sensitivity in retrospective studies, so that case control studies systematically underestimate the strength of the association. This is because atrophic gastritis, a precancerous lesion, reduces the burden of H.  pylori infection. This in turn decreases titers of IgG antibody, potentially causing the H.  pylori infection to become serologically undetectable. For this reason, the best evidence of H. pylori prevalence in relation to cancer comes from prospective studies. The most comprehensive relative risk estimates for H. pylori and gastric cancer come from a pooled analysis of 12 prospective studies, which included 762 cases of non-​cardia gastric cancer and 2250 controls. The pooled odds ratio was 2.97 (95% CI = 2.34–​3.77) for H. pylori infection (Forman and Helicobacter and Cancer Collaborative Group, 2001). The same study included 274 cases of cardia gastric cancer and 827 controls with an odds ratio of 0.99 (95% CI = 0.40–​ 1.77) for H. pylori infection. When the pooled analysis was restricted to cases occurring at least 10 years after the blood draw used for H. pylori diagnosis, the odds ratio increased to 5.93 (95% CI = 3.41–​ 10.3) for non-​cardia cancer but decreased to 0.46 (95% CI = 0.23–​ 0.90) for cardia cancer. This subgroup analysis underscores the much stronger relationship with non-cardia than cardia cancer and the need to account for the effect of premalignant disease on the detection of H. pylori infection, even in prospective studies. Further follow-​up of the individual studies contributing to this pooled analysis was reviewed in the IARC Monographs volume 100 part B, but did not substantively change the conclusions (IARC, 2012a). Although H.  pylori infection clearly increases risk for cancers of the gastric body and antrum, its relationship to cardia cancers has not been firmly established. Some data suggest that these topographical subtypes may represent a mixture of two different tumors, resembling either non-​cardia adenocarcinoma (H. pylori induced) or esophageal adenocarcinoma (reflux induced) in their etiology. In Western countries, tumors of the cardia occur more frequently in white males (Yang and Davis, 1988) and a large proportion occur in the setting of gastroesophageal reflux disease (GERD). Conversely, in high-​risk areas for H.  pylori and non-​cardia gastric cancers, most cancers of the cardia occur in H. pylori–​infected people, and may conceivably be related to H. pylori infection (Shakeri et al., 2015).

Virulence Factors

H.  pylori readily loses and acquires DNA fragments and undergoes various structural genetic changes such as point mutations and chromosomal rearrangements (Blaser and Berg, 2001). As a consequence, H.  pylori isolates have an extraordinary degree of genetic variability between and even within infected hosts, and this diversity may contribute to the clinical outcome of the infection (Aras et  al., 2002; Patra et al., 2012). A number of genetic factors associated with H.  pylori colonization (babA, sabA, alphAB, hopZ, and OipA) and virulence (cagA, vacA) have been identified (Keilberg and Ottemann, 2016)  (Figure 31–​ 5). The genetic marker that has attracted most attention in epidemiological studies is the presence of the cag pathogenicity island (PAI), a DNA sequence of 40 kbp present in 70% of H. pylori strains in Europe and North America, but ubiquitous in

601

Asia and most of Africa (Peek and Crabtree, 2006). PAI-​containing organisms cause greater inflammation and are more closely associated with intestinal-​type dysplasia and malignancy than are strains without the PAI (Parsonnet et al., 1997). In contrast, diffuse-​type cancers have similar associations with both PAI-​positive and PAI-​negative isolates. Other polymorphic genes—​ such as the vacuolating cytotoxin and babA-​2 adherence gene—​have been linked to pathogenicity, but none as strongly as the PAI. The PAI encodes for a type four secretion system that injects the cagA protein into the host cell, where it is phosphorylated and binds to the Shp-​2 tyrosine phosphatase (Stein et al., 2002; Yamazaki et al., 2003). The host cell consequently elongates and acquires a growth-​ factor-​like phenotype (Segal et al., 1997). The cagA protein is highly immunogenic, which allows serological detection of infection with CagA-​positive H.  pylori by the detection of anti-​cagA antibodies. CagA-​positive strains are associated with higher risk of gastric cancer than cagA-​negative strains. A  meta-​analysis of 16 cohort and case-​control studies including 778 cases of non-​cardia gastric cancer and 1409 matched controls found an elevated risk of cagA-​positive H. pylori infections, with an odds ratio of 2.01 (95% CI = 1.21–​3.32) for cagA-​positivity among all H.  pylori–​infected individuals (Shiota et al., 2010). The cag pathogenicity island is also associated with precancerous gastric lesions. Plummer et al (2007) analyzed a cross-​sectional endoscopic survey of 2145 individuals from Venezuela, in which both the presence of H. pylori DNA and presence of the cagA gene were determined by polymerase chain reaction (PCR) on gastric biopsies. Infection with cagA-​positive H. pylori strains but not cagA-​negative strains was associated with the severity of precancerous lesions. Using individuals with normal gastric mucosa or superficial gastritis as controls, the OR for dysplasia was 15.5 (95% CI = 6.4–​37.2) for cagA-​positive H.  pylori compared with 0.90 (95% CI  =  0.37–​2.17) for cagA-​negative H. pylori. González et al. (2011) analyzed a follow-​ up study of 312 individuals from Spain with an average of 12.8 years of follow-​up between two endoscopies, also using PCR detection and genotyping of H. pylori. The relative risk for progression of precancerous lesions was 2.28 (95% CI = 1.13–​4.58) for cagA-​positive strains compared with cagA-​negative strains.

Host Response and Other Interacting Factors

The host also plays an important role in H. pylori outcome. El-​Omar et al. (2000) reported that H. pylori–​infected subjects who developed gastric cancer were more likely to have specific genotypes of interleukin (IL)-​1β or the IL-​1β receptor antagonist. Interestingly, the higher risk genotypes of IL-​1β not only induce more inflammation than lower risk genotypes but also increase suppression of gastric acid secretion, supporting the pathogenic model devised by Correa. Moreover, the IL-​1β genotype is not associated with cancer risk in the absence of infection. A  subsequent study identified similar, though less strong, interactions between H.  pylori and TNF-​α (Machado et  al., 2003). Other putative host factors that are being explored include p53 polymorphisms and variants of the HLA genotype. Environmental factors such as tobacco smoking, diet, and medications may also enhance or diminish H.  pylori’s deleterious effects, although relatively few of these studies have measured the effects of combined exposures. Studies of dietary factors are particularly sparse. A  prospective Scandinavian cohort demonstrated a protective association with ascorbic acid (vitamin C) and beta-​carotene in H. pylori–​infected subjects but not in uninfected subjects (Ekstrom et al., 2000b). A study by Correa and colleagues on gastric dysplasia and other preneoplastic conditions supported this finding. They observed that both H. pylori eradication therapy and dietary antioxidants (ascorbic acid and beta-​carotene) prevented preneoplastic progression. Combining antioxidants with H. pylori eradication therapy did not provide added benefit, however, suggesting that the benefit of antioxidants may be limited to infected hosts. In animals, dietary salt magnifies H. pylori–​associated gastric carcinogenesis (Fox et al., 2003); this finding has not been substantiated in humans. Non-​ dietary exposures that may alter H.  pylori outcome include aspirin and non-​steroidal anti-​inflammatory drugs (NSAIDs), which appear

602

602

HopZ/OipA

AlpA/B

VacA

SabA

BabA

PART IV:  Cancers by Tissue of Origin

α5

α5 β1 CagA β1

? Lex Laminin Le b

Abi Src

Vacuolation

P-CagA Shp2

IL8

ATP

IL2

Apoptosis

TNF

Cell integrity Proliferation cell polarity

Inflammation

Figure 31–​5.  Adherence and virulence factors used by H. pylori to promote direct interactions with epithelial cells. H. pylori possesses multiple adherence factors to attach to epithelial cells, including BabA, SabA, AlphA/​B, HopZ, and OipA. Adherence is important for CagA delivery via the T4SS. CagA is phosphorylated inside the host cell, and alters signaling pathways, leading to loss in cell integrity and alteration of cell proliferation and cell polarity, and induces inflammation. Independent of attachment, VacA is secreted by H. pylori and can enter the cells via T5SS. VacA leads to vacuolation and apoptosis of its host cells. Source: Keilberg D, Ottemann KM. How Helicobacter pylori senses, targets and interacts with the gastric epithelium. Environ Microbiol 2016;18(3):791–​806.

to protect against gastric cancer only in infected subjects (Zaridze et al., 1999). Smoking was associated with increased risk of stomach cancer in H. pylori–​infected subjects (OR = 2.2; 95% CI = 1.2–​4.2) in a nested case control study from Sweden, where cases and controls were tested by enzyme-​linked immunosorbent assay (ELISA) (Siman et al., 2001). Compared to nonsmokers, active smokers had significantly higher risk of colonization by cagA-​positive virulent strains and a non-​significant increase in bacterial load (Santibanez et al., 2015). In a population-​based case-​control study in Germany, the relative risk of gastric cancer was 2.6 (95% CI  =  1.2–​5.7) for nonsmoking subjects with CagA-​positive H.  pylori infections and 7.2 (95% CI  =  2.2–​23.6) for smoking subjects with CagA-​positive H. pylori infections, compared with subjects without these risk factors (Brenner et al., 2002). Hence, smoking could act indirectly on cancer risk by promoting the emergence of the more virulent strains in the gastric mucosa. The age at which H. pylori infection is acquired may also modulate risk, since infection in high-​risk areas is usually acquired in childhood. Only indirect data actually support this hypothesis, however (Blaser et al., 1995). Nevertheless, the theory remains popular, both because of its biological plausibility (increased duration of chronic inflammation increases the risk for genetic mutations to accumulate in gastric epithelium) and the epidemiological patterns of disease (cancer occurs more frequently in regions where childhood infection is common). Even among infected children, however, differences in gastric response to infection exist between high-​risk and low-​risk regions (Bedoya et al., 2003), suggesting that age at acquisition cannot explain important differences in clinical progression.

Epstein-​Barr Virus In 2014, The Cancer Genome Atlas Research Network recognized EBV-​associated gastric cancer (EBVaGC) as one of the four subtypes of a new molecular classification of gastric adenocarcinoma (The Cancer Genome Atlas Research Network, 2014). EBVaCG tumors are distinguished from the other subtypes by higher levels of DNA methylation in CpG islands of promoter regions and by distinctive genetic alterations. The latter include a high frequency of mutations in PIK3CA and ARID1A, a mutation in BCOR, and amplification of PD-​ L1 and PD-​L2 (The Cancer Genome Atlas Research Network, 2014).

In epidemiological studies, the established way to identify EBV in gastric tumors has been through detection of EBV-​encoded small RNAs (EBERs) or EBV DNA using in situ hybridization (ISH) (Fukayama and Ushiku, 2011; Hamilton-​Dutoit and Pallesen, 1994). Two meta-​analyses have now been published of gastric cancer studies that used ISH detection methods (Lee et al., 2009; Murphy et al., 2009). One of these was followed by a pooled analysis of over 5000 cancer cases from 15 populations (Camargo et al., 2011b). The pooled analysis detected EBV in malignant epithelial cells in 9% of all gastric cancers, although with high heterogeneity among the studies. EBVaGC displays distinctive epidemiological and clinical features. The proportion of EBV-​associated gastric cancer is higher at younger than at older cases, higher in men than in women (11% versus 5%, respectively), and slightly higher in American or Caucasian patients than in Asians (10% versus 8%, respectively) (Lee et al., 2009; Murphy et al., 2009). EBV-​related tumors also typically occur in the gastric body or cardia of the stomach, rather than in the antrum (Takada, 2000), and are common in gastric stumps following gastric resection. A large multicenter case series examined the association of EBV status with survival after gastric cancer diagnosis, accounting for tumor stage and other prognostic factors, and found a lower mortality rate (Hazard Ratio [HR] = 0.72; 95% CI = 0.61–​0.86) (Camargo et al., 2014). A small percentage (probably less than 1%) of all gastric cancers are categorized histopathologically as lymphoepithelioma-​like carcinomas (LELC). These are epithelial tumors with intense lymphoid infiltration in the stroma, similar in appearance to nasopharyngeal carcinomas. Between 80% and 100% of gastric LELC contain monoclonally integrated Epstein-​Barr virus (Burke et al., 1990; Herrmann and Niedobitek, 2003; Wu et al., 2000). Apart from these histopathological features and the consistent association with EBV, LELC resembles the conventional form of EBVaGC (Cheng et al., 2015). How and when EBV acts to induce malignant transformation remains largely unknown. In the inflammatory mucosa, EBV is probably transferred from EBV-​infected B lymphocytes to gastric epithelial cells. The viral genome is not integrated into the host genome but becomes circular and episomal within the cytoplasm of infected cells. EBV uses the cellular machinery of the host cell to propagate its monoclonal viral genome, epigenetically silence viral and host genes, and control the behavior and microenvironment of the infected cell (Abe et al., 2015). EBV does not replicate in gastric tumors but does express latent genes according to specific latency patterns. EBVaGC

 603

Stomach Cancer belongs to latency type I or II, and typically expresses EBERs, EBNA-​ 1, BARTs, and BART miRNAs. Approximately half of the cases also express LMP-​2A (Shinozaki-​Ushiku et  al., 2015). The clonality of the viral genome in tumor cells suggests that infection with EBV is an early event of gastric carcinogenesis (Fukayama et al., 1994; Imai et al., 1994). The relationship between EBV and H. pylori is still unclear. In vitro studies have shown that the H. pylori–​associated oxidant NH2Cl may convert the latent epithelial EBV infection into lytic infection with activation of the early gene (Minoura-​Etoh et al., 2006). Recently, seroprevalence of antibodies to 15 specific H. pylori proteins using a multiplex assay showed that the serological profile of H. pylori is equivalent in EBV-​ positive and in EBV-​ negative gastric biopsies obtained from 169 non-​cardia gastric cancer patients in five countries. These results suggest that H. pylori is the main etiological factor in EBV-​positive gastric cancers, and that EBV might act as a cofactor by increasing the likelihood of malignant transformation. In 2011, EBV was re-​evaluated by an IARC Working Group as part of Monograph 100B. The group concluded that there was as yet insufficient epidemiological evidence to implicate EBV as a cause of gastric cancer. The expert group added the following sentence to their evaluation: “However, the fact that the EBV genome is present in the tumor cell in a monoclonal form, and that transforming EBV proteins are expressed in the tumor cell provides a mechanistic explanation for how EBV might cause a proportion of gastric cancer” (IARC, 2012a).

Smoking IARC first evaluated the relationship between tobacco smoking and gastric cancer in 1985 (IARC, 1986). At that time, the available data were considered insufficient to conclude that the association between tobacco smoking and stomach cancer was causal. Based on additional data that had accumulated by 2002, IARC concluded that confounding by other factors, such as alcohol consumption, H. pylori infection, and dietary factors, could be reasonably ruled out (IARC, 2004). Supporting evidence included dose–​response relationships with the duration of smoking and number of cigarettes smoked daily and a progressive decrease in the association following cessation. The findings from the American Cancer Society’s Cancer Prevention Study II (Chao et al., 2002), published after the 2002 IARC review, further supported the conclusion of the IARC Working Group. In a 14-​year follow-​up of 467,788 men and 588,053 women observed from 1982 through 1996, cigarette smoking and use of other tobacco products were significantly associated with stomach cancer mortality. The relative risk (RR) of dying from stomach cancer among male smokers versus never smokers, after adjusting for age, race, education, family history of stomach cancer, dietary habits, and aspirin intake—​but not H. pylori infection—​was 2.16 (95% CI = 1.75–​2.67) for cigarette smokers and 2.29 (95% CI = 1.49–​3.51) for cigar smokers, with larger risks observed with increasing smoking duration. The magnitude of association between cigarette smoking and stomach cancer mortality was smaller but still statistically significant in women. A meta-​analysis of prospective studies (cohorts, case cohorts, and nested case-​control studies) reported a summary estimate of relative risk of 1.62 (1.50, 1.75) in male smokers and 1.20 (1.01, 1.43) in female smokers, compared to never smokers (Ladeiras-​Lopes et  al., 2008). The same meta-​analysis suggested that the association may vary by subsite, although this is not a consistent finding in the literature. A  systematic review and meta-​analysis of nine cohort studies reported a summary RR of 1.87 (95% CI = 1.31–​2.67) for cardia cancers and 1.60 (95% CI = 1.41–​1.80) for non-​cardia cancers (Ladeiras-​ Lopes et al., 2008). As noted earlier, smoking was associated with increased risk of stomach cancer in H. pylori–​infected subjects in a nested case-​control study in Sweden (OR = 2.2; 95% CI = 1.2–​4.2) (Siman et al., 2001) and higher risk of colonization by cagA-​positive virulent strains (Santibanez et al., 2015). A population-​based case-​control study in Germany reported a relative risk of gastric cancer of 2.6 (95% CI = 1.2–​ 5.7) for nonsmoking subjects with cagA-​positive H. pylori infections and 7.2 (95% CI = 2.2–​23.6) for smoking subjects with cagA-​positive

603

H. pylori infections, compared with subjects without these risk factors (Brenner et al., 2002). Smoking and H. pylori strains, and especially the cagA-​positive, appear to act synergistically in increasing the risk of non-​cardia cancer. In 2004 (IARC, 2004) and again in 2009 (IARC, 2012b), IARC concluded that active tobacco smoking increases the risk of gastric cancer but that the evidence for involuntary exposure to tobacco smoke was inconclusive.

Food and Nutrition In June 2015, an expert panel was convened by the World Cancer Research Fund (WCRF) and the American Institute for Cancer Research (AICR) to update a previous report from 2007 and to grade nutritional risk factors for stomach cancer on a five-​level scale, according to the strength of the evidence (1  =  convincing; 2  =  probable; 3 = suggestive; 4 = inconclusive; and 5 = unlikely) (Wiseman, 2008; World Cancer Research Fund International/​ American Institute for Cancer Research [WCRF/​AICR], 2016). In this report, only prospective studies (cohort, nested case control, and randomized controlled trials) were considered, and the two anatomic subsites of the stomach were analyzed separately when possible. The authors reported on 89 new studies conducted around the world, comprising 77,000 cases of stomach cancers; they noted, however, that whereas most studies adjusted for tobacco smoking, few did so for H. pylori infection. According to the WCRF/​AICR panel, no dietary risk factor for gastric cancer met the criteria for level one (convincing evidence), but several qualified for level two. The evidence was considered probable that consuming more than three drinks (45 g) of alcohol per day and eating processed meat or foods preserved by salting increased the risk of stomach cancer. A high body mass was considered to be a probable risk factor for cardia cancer. Eating grilled/​barbecued meat or fish and low consumption of fruit were considered suggestive risk factors (level 3) for both cardia and non-​cardia cancers. Citrus fruit consumption was designated a suggestive protective factor for cardia cancer (WCRF/​AICR, 2016).

Fruits and Vegetables

In a meta-​analysis based on 22 cohort studies looking at all stomach cancers irrespective of anatomic location, Wang et al. (2014) reported a small but statistically significant 10% decreased risk of stomach cancer in the high versus low consumers of fruits (RR = 0.90; 96% CI = 0.83–​0.98) (Wang et al., 2014). Similarly, a dose response analysis by the WCRF panel reported a 5% reduction in cancer per 100 g of fruit consumed daily (RR = 0.95; 95% CI = 0.91–​0.99). The association was nonlinear, with most of the risk reduction observed in the low to middle categories of intake. The association was strongest when considering only European studies (RR = 0.81; 95% CI = 0.68–​0.96) or subgroups of high-​quality studies conducted elsewhere. No inverse association was found between vegetable intake and gastric cancer risk. The WCRF expert panel concluded that these data and others published up to 2014 were reasonably consistent to classify the evidence for low fruit consumption and increased risk of both cardia and non-​cardia cancers as level 3 (suggestive). Three studies conducted in Europe and in the United States reported that citrus fruit consumption was associated with decreased risk of cardia gastric cancer when comparing the highest to lowest categories of intake (Freedman et al., 2008; Gonzalez et al., 2012; Steevens et al., 2011). A meta-​analysis of the dose response in these three studies showed a significant decrease of 24% per 100 g of citrus fruit consumed per day (RR = 0.76; 95% CI = 0.58–​0.99) (WCRF/​AICR, 2016). Given the paucity of published studies, however, the evidence regarding citrus consumption and cardia gastric cancer was considered only suggestive.

Salt and Salted Preserved Food

According to WCRF/​AICR, a relationship between gastric cancer incidence and consumption of salt-​preserved foods is strongly supported by the literature (WCRF/​AICR, 2016). Two meta-​analyses showed that high intake of salt-​preserved vegetables (“pickled

604

604

PART IV:  Cancers by Tissue of Origin

food”) was associated with summary relative risks for gastric cancer of 1.27 (95% CI = 1.09–​1.49) and 1.32 (95% CI = 1.10–​1.59), respectively (D’Elia et al., 2012; Ren et al., 2012). A subsequent meta-​analysis of nine studies by the WCRF/​AICR expert panel identified a dose response with no heterogeneity. Risk increased by 9% per 20 g increase in daily consumption of salt-​preserved vegetables (RR = 1.09; 95% CI = 1.05–​1.13) (WCRF/​AICR, 2016). Salt consumption was a particularly strong predictor of gastric cancer in studies in Japan. In contrast, a population study in Norway found no association between high consumption of dietary salt and risk of gastric cancer (Sjodahl et al., 2008b). In the few studies that controlled for H. pylori infection, the effect of salt appeared to be independent of cagA status but was stronger in the presence of H. pylori–​associated atrophic gastritis (Peleteiro et al., 2011; Shikata et al., 2006). Animal models have not shown that increased dietary salt increases the risk of H. pylori–​associated cancer (Rogers et al., 2005). Cardia and non-​cardia tumors were not investigated separately in any of these studies. Salt itself is not carcinogenic, but salted foods contain lower amounts of micronutrients and may undergo fermentation during preservation. Salt causes mucosal damage in the stomach, induces inflammation, and increases DNA synthesis and cell proliferation, all of which can facilitate carcinogenesis (Ames and Gold, 1990; Charnley and Tannenbaum, 1985).

Other Dietary Factors

Studies of processed meat consumption (i.e., sausages, bacon, meatballs, ham, or pieces of meat having undergone smoking, fermentation, or salt preservation) have been recently reviewed by both WCRF/​AICR and IARC (Bouvard et al., 2015; WCRF/​AICR, 2016). Processed meat was classified as carcinogenic to humans (Group  1) by IARC, based on sufficient evidence in humans regarding colorectal cancer. For stomach cancer, however, the evidence was considered limited. Based on the results of three cohort studies, the WCRF designated consumption of processed meat as a probable cause of non-​ cardia cancer (Cross et al., 2011; Gonzalez et al., 2006; Keszei et al., 2012)  with the mechanism possibly relating to nitrites and nitrate (Takahashi et  al., 1994). A  dose–​response meta-​analysis showed an 18% increased risk for non-​cardia cancer per 50 g of processed meat consumed per day (RR = 1.18; 95% CI = 1.01–​1.38). No association was found with cardia cancer. Diets high in grilled or barbecued meat or fish were also deemed to possibly increase stomach cancer risk. Despite a plausible biological mechanism (i.e., the formation of carcinogenic heterocyclic amines when meat and fish are cooked at high temperature), data specifically linking heterocyclic amines to gastric cancer are lacking. Similarly, although laboratory studies support a causal link between N-​nitroso compounds and stomach cancer, the WCRF/​ AICR panel designated the evidence for carcinogenicity of N-​ nitroso compounds in humans to be limited/​suggestive (WCRF/​ AICR, 2016). Few studies have been able to accurately evaluate the effect of vitamin C on the risk of gastric cancer. Within the EPIC cohort, higher plasma vitamin C levels were associated with a lower risk of cancer, and did not appear to be limited to a particular anatomical subsite or histological subtype (Jenab et al., 2006). In contrast, dietary vitamin C showed no significant association with cancer risk. The WCRF expert panel was unable to draw any conclusion about vitamin C based on available evidence (Wiseman, 2008; WCRF/​ AICR, 2016).

Alcohol A possible relationship between alcoholic beverage consumption and risk for stomach cancer has long been hypothesized, but in most prospective studies, no overall association has been observed (IARC, 2010, 2012c). A 2010 meta-​analysis by anatomic subtype found no statistically significant association between heavy drinking and either non-​cardia cancer (pooled RR =1.17; 95% CI = 0.78–​ 1.75) or cardia cancer (RR = 0.99; 95% CI = 0.67–​1.47) (Tramacere

et al., 2012). The strongest evidence for a causal association comes from a meta-​analysis of 30 studies (12,000 cases) conducted by WCRF/​AICR in 2015 (WCRF/​AICR, 2016). A dose–​response trend in risk was observed with increasing alcohol consumption above 45 g of ethanol daily. The relative risk estimates were 1.06 (95% CI = 1.01–​1.11), 1.15 (95% CI = 1.06–​1.26), and 1.28 (95% CI = 1.08–​ 1.52) for alcohol intake of 45 g, 80 g, and 120 g per day. However, alcohol consumption is difficult to evaluate as an independent variable in observational studies due to underreporting, and its association with smoking and poor nutrition.

Ionizing Radiation Ionizing radiation increases risk of gastric carcinoma (IARC, 2012c). The best evidence comes from the longitudinal study of 38,576 atomic-​bomb survivors in Hiroshima and Nagasaki, Japan, followed between 1980 and 1999 (Sauvaget et al., 2005). For a person-​years weighted mean dose of 1.6 Gy, the RR was 1.71 (95% CI = 1.27–​2.30) compared to the lowest dose, with a significant dose response trend (p = 0.009).

Body Mass Index and Physical Activity Overweight and obesity have been associated with increased risk of many cancers (Lukanova et  al., 2006). A  meta-​analysis has shown an elevated risk for cardia gastric cancer with a summary relative risk estimate of 1.4 (95% CI  =  1.16–​1.68) for overweight (body mass index [BMI] 25–​30), and 2.06 (95% CI = 1.63–​2.61) for obesity (BMI ≥ 30) (Yang et al., 2009). A dose response meta-​analysis by WCRF stratified by geographic regions confirmed the positive association in Europe and North America, but not in Asia (WCRF/​ AICR, 2016). In Western countries, cardia cancer occurs predominantly in obese white men, presumably with similar pathogenesis as Barrett’s esophagus and esophageal adenocarcinoma. It is notable that in healthy volunteers without reflux symptoms, the cardiac mucosa lengthens with the age of the subjects and increasing abdominal obesity and histologically resembles Barrett’s esophagus. The mechanism by which BMI is associated with cardia cancer could therefore involve not only chronic inflammatory states and hormonal disruptions associated with obesity, but also acidic damage at or around the esophagogastric junction, caused by subclinical acidic reflux (Derakhshan et al., 2015). No association has been observed between BMI and non-​cardia gastric cancer (Corley et  al., 2008; Kuriyama et  al., 2005; Sjodahl et al., 2008a). An inverse relationship has been reported between regular physical activity and non-​cardia gastric cancer. Among many studies that have looked at physical activity in cancer patients, two recent prospective studies specifically examined gastric cancer, and both found a protective association (Leitzmann et al., 2009; Sjodahl et al., 2008a)

HOST FACTORS Familial Risk Increased risk of stomach cancer has long been observed in those with a family history of stomach cancer (Terry et al., 2002). This moderate to strong increase in risk does not necessarily indicate heritable susceptibility, however, since many of the established and suspected risk factors for stomach cancer tend to aggregate in families. These include H. pylori infection, smoking and dietary habits. Hereditary cancer syndromes involving familial clustering of stomach cancer have been recognized for more than 50 years and account for up to 3% of cases (McLean and El-​Omar, 2014). In a study of Maori kindred in New Zealand, the familial pattern was consistent with dominant inheritance of a susceptibility gene with incomplete penetrance (Guilford et al., 1998). A linkage analysis of this and two other Maori pedigrees with early-​onset, diffuse-​type stomach cancer and

 605

605

Stomach Cancer subsequent molecular genetic studies led to a discovery that these pedigrees had germline mutations in the E-​cadherin/​CDH1 gene (Guilford et  al., 1998). Germline mutations in the E-​cadherin gene have also been found in stomach cancer pedigree studies of families of European (Gayther et al., 1998; Guilford et al., 1999), Japanese (Shinmura et al., 1999; Yabuta et  al., 2002), and Korean (Yoon et  al., 1999)  descent. These familial cases led to the recognition of a new syndrome defined by Guilford et al. (1998) as hereditary diffuse gastric cancer (HDGC). Male and female carriers of germline E-​cadherin gene mutations included in studies of the International Gastric Cancer Linkage Consortium have a greater than 80% cumulative risk of stomach cancer by age 80 (Fitzgerald et al., 2010). Somatic mutations in the E-​cadherin gene are frequently detected in the diffuse but not the intestinal type of stomach cancer (Becker et al., 1994). The critical role of E-​cadherin in cell-​cell adhesion is notable in this regard. Inactivating mutations in CDH1 are also common in sporadic gastric tumors as well as HDGC (Pinheiro et al., 2014). Two mutations induce complete inactivation of the CDH1 gene, which initiates the carcinogenic process of diffuse gastric cancer. Until very recently, CDH1 was the only identified gene related to HDGC. However, approximately 60% of HDGC tumors do not harbor a defective CDH1 allele. In 2013, a CTNNA1 germline truncating mutation was found in a large HDGC pedigree, prompting an active search for other susceptibility genes (Majewski et al., 2013; Pinheiro et al., 2014). Stomach cancer has also been observed as part of other hereditary cancer predisposition syndromes, as shown in Table 31–​2. These include hereditary breast/​ovarian cancer due to germline BRCA1 and BRCA2 mutations (Brose et al., 2002); Lynch syndrome (Chen et al., 2006; Watson et  al., 2008); Li-​Fraumeni syndrome due to germline p53 mutations (Gonzalez et  al., 2009); and the much rarer familial adenomatous polyposis (Garrean et  al., 2008), Peutz-​Jeghers (Howe et  al., 2004), and juvenile polyposis syndromes (Giardiello et  al., 2000) associated with germline mutations in the APC gene, SMAD4 and BMPR1A genes, and STK11 gene, respectively (reviewed in Chun and Ford, 2012). Lynch syndrome carries a lifetime risk of around 10% for gastric cancer, considerably lower than the risk of colorectal or endometrial cancer associated with this syndrome (Chen et al., 2006; Watson et al., 2008). Approximately 90% of cases have intestinal type histology. A  significantly increased risk of stomach cancer (SIR  =  2.78; 95% CI = 1.59–​4.52) was observed in Swedish hereditary prostate cancer families (Gronberg et  al., 2000), but germline E-​cadherin mutations do not seem to contribute to the elevated stomach cancer risk in these families (Jonsson et al., 2002).

Other Genetic Factors The role of host susceptibility genes in stomach cancer has also been investigated in populations without strong familial aggregation. For instance, a higher incidence of stomach cancer in blood type A  individuals than in those with blood type O was noticed as

early as the 1950s (Aird et al., 1953). Prevalence of chronic atrophic gastritis, intestinal metaplasia, and dysplasia is also higher in subjects with blood type A than type O (Haenszel et al., 1976; Kneller et al., 1992). Numerous studies including a large Swedish and Danish cohort of blood donors (Edgren et  al., 2010)  and a meta-​analysis published in 2012 (Wang et al., 2012) confirmed these observations. Individuals with blood type A  demonstrated a higher risk of gastric cancer (OR = 1.20; 95% CI = 1.02–​1.42; and OR = 1.11; 95% CI = 1.05–​1.15) than donors with other blood types in the Swedish and Danish cohorts, respectively. Further studies found that adherence of H. pylori to human gastric epithelium can be mediated by the blood-​group antigen-​binding adhesin (BabA) produced by H. pylori that targets human fucosylated blood group antigens H type I (type O substance) and Lewis b (Leb) (Prinz et al., 2001). The presence of the babA2 gene, encoding for BabA, in the H. pylori genome is crucial for H. pylori–​related pathogenesis and correlates with the activity of gastritis in the infected stomach. The establishment of H. pylori infection as a risk factor, along with advances in molecular genetic techniques, has facilitated studies of interaction between H.  pylori and host genetic factors with regard to stomach cancer risk. Despite the consistent association between H. pylori infection and stomach cancer risk, only a small fraction of the infected individuals develop stomach cancer (Parsonnet, 1999). It is plausible to hypothesize that some individuals are more susceptible to acquiring persistent infection when exposed to H.  pylori and to develop preneoplastic lesions and eventually cancer once infection persists. Such variation in susceptibility may be due to interindividual variability regarding response to and interaction with H.  pylori. As mentioned earlier, El-​Omar et al. (2000) reported that IL-​1 gene cluster polymorphisms were associated with gastric cancer. Polymorphisms in these clusters are suspected of enhancing IL-​1β, an important pro-​ inflammatory cytokine and a powerful inhibitor of gastric acid secretion. Many studies have investigated a possible association between selected candidate genes and gastric cancer risk (see comprehensive review by McLean and El-​Omar, 2014). Genes coding for the inflammatory proteins IL-​1β, IL-​1ra, IL-​8, IL-​10, and TNF-​α were recently examined in a systematic review and meta-​analysis by the Human Genome Epidemiology (HuGE) project (Persson et  al., 2011). The analyses revealed risk differences among histologic types, anatomic sites, geographic locations, and H. pylori infection status. An increased risk for gastric cancer was clearly observed for IL-​1RN2; however, this risk appeared to be confined to non-​Asian populations and was particularly observed for non-​cardia cancer, of both intestinal and diffuse types. Conversely, in Asian populations, IL-​1B-​31C carrier status was associated with a reduced risk of gastric cancer (Persson et al., 2011). In recent years, IL-​17 has been identified as another candidate gene. The IL-​17 family of pro-​ inflammatory cytokines has been implicated in many inflammatory-​ driven diseases, including autoimmune diseases and colorectum or breast cancers. Several studies and meta-​analyses have evaluated the 187G > A  polymorphism in the IL-​17A gene (rs2275913), which seems to be associated with an increased risk of gastric cancer, especially in Asian populations (Dai et al., 2016).

Table 31–​2.  Inherited Predisposition Syndromes for Gastric Cancer Cancer Syndrome Hereditary diffuse gastric cancer (HDGC) Hereditary breast/​ovarian cancer Lynch syndrome Li-​Fraumeni syndrome Familial adenomatous polyposis Juvenile polyposis Peutz-​Jeghers syndrome

Gene

Frequency

Gastric Cancer Risk

References

CGH1

Very rare

> 80%

Fitzgerald et al., 2010

BRCA1/​2 MLH1, MSH2, MSH6, PMS2, Epcam p53 APC SMAD4, BMPR1A STK11

1/​40 to 1/​400 1/​440

2.6%–​5.5% 6%–​13%

1/​5000 1/​10,000 to 1/​15,000 1/​16,000 to 1/​100,000 1/​25,000 to 1/​250,000

2.8% 0.5%–​2% 21% 29%

Brose et al., 2002 Chen et al., 2006; Watson et al., 2008 Gonzalez et al., 2009 Garrean et al., 2008 Howe et al., 2004 Giardiello et al., 2000

Source: Chun N, Ford JM, Genetic Testing by Cancer Site: Stomach. The Cancer Journal 2012;18(4):355–​363.

60

606

PART IV:  Cancers by Tissue of Origin

Genome-​ wide association studies (GWAS) have also identified novel susceptibility loci that may offer new insights about gastric carcinogenesis. For example, gene variant rs2294008, a polymorphic variation in the gene encoding for the prostate stem cell antigen (PSCA), was identified in 2008 as statistically significantly associated with diffuse gastric cancer in a Japanese population (Sakamoto et al., 2008). This association was later confirmed in other populations in China and in Europe, with a similar strength of association with diffuse and intestinal cancer types (Lu et al., 2010; Sala et al., 2012). Recently the rs2294008T gene variant has been associated with a worse overall survival in patients with diffuse type cancers (HR = 1.85; 95% CI = 1.12–​3.06) (Garcia-​Gonzalez et al., 2015).

Other Predisposing Conditions Pernicious anemia, an autoimmune disorder characterized by atrophic damage restricted to the gastric body mucosa (gastric atrophy type A), predisposes to gastric adenocarcinoma, with an incidence rate similar to that seen in gastric atrophy due to H.  pylori (gastric atrophy type B) (Annibale et  al., 2000). This excess risk seems to be independent of H. pylori infection, although the potential interaction between infection and pernicious anemia has not yet been thoroughly studied. Prior gastric surgery for benign disorders (mainly gastric ulcer) was linked to increased risk of adenocarcinoma long before the discovery of H. pylori. However, it is not clear if prior gastric surgery is merely a surrogate for long-​term H. pylori infection with more aggressive cagA strains or if it causes increased risk in the remnant stomach, perhaps acting synergistically with H. pylori through adverse consequences of the surgery, such as bile reflux) (Leivonen et al., 1997; Mezhir et al., 2011). To date, bariatric surgery has not been linked to gastric cancer risk (Orlando et al., 2014), but as obesity prevalence increases and the procedures correspondingly become more common, further research will be needed.

FUTURE RESEARCH The history of gastric cancer is one of impressive, although inadvertent, success. Despite the decreasing incidence of this disease, gastric cancer remains common in many areas of the world, and the absolute number of cases is increasing. Several areas of epidemiologic investigation need attention. First, we need improved surveillance for gastric cancer, particularly in low-​income countries where this malignancy is frequently missed or misdiagnosed. Hand in hand with this surveillance, research is needed to dissect the epidemiologic significance of the various molecular subtypes of gastric cancer that have recently been identified. This research will, in turn, stimulate development of new screening methodologies as well as development of targeted treatments. Studies indicate that endoscopic screening for gastric cancer in high-​risk areas with resection of early tumors results in longer survival, but the extent to which this favorable trend reflects lead-​time bias has not been established. What also remains to be established is the best public health strategy for managing H.  pylori infection, both in high and low gastric cancer risk countries; although H. pylori treatment appears to reduce gastric cancer incidence, the broader consequences of large-​scale antibiotic use, the optimal therapy in an age of increasing antibiotic resistance, and the best implementation strategies need to be defined on local levels. Finally, much more work needs to be focused on understanding the epidemiology and prevention of gastric cardia cancer that, in the future, could very well surpass non-​cardia cancer in incidence worldwide. References Abe H, Kaneda A, and Fukayama M. 2015. Epstein-​Barr virus-​associated gastric carcinoma: use of host cell machineries and somatic gene mutations. Pathobiology, 82(5), 212–​223. PMID: 26337667. Aird I, Bentall HH, and Roberts JA. 1953. A relationship between cancer of stomach and the ABO blood groups. Br Med J, 1(4814), 799–​801. PMCID: PMC2015995.

Allemani C, Weir HK, Carreira H, et al. 2015. Global surveillance of cancer survival 1995–​2009: analysis of individual data for 25,676,887 patients from 279 population-​based registries in 67 countries (CONCORD-​2). Lancet, 385(9972), 977–​1010. PMID: 25467588. American Cancer Society. 2016. Cancers facts & figures:  stomach cancer. Atlanta: American Cancer Society. Ames BN, and Gold LS. 1990. Too many rodent carcinogens:  mitogenesis increases mutagenesis. Science, 249(4972), 970–​971. PMID: 2136249. Amini N, Spolverato G, Kim Y, et al. 2015. Clinicopathological features and prognosis of gastric cardia adenocarcinoma:  a multi-​ institutional US study. J Surg Oncol, 111(3), 285–​292. PMID: 25308915. Annibale B, Lahner E, Bordi C, et al. 2000. Role of Helicobacter pylori infection in pernicious anaemia. Dig Liver Dis, 32(9), 756–​762. Aras RA, Small AJ, Ando T, and Blaser MJ. 2002. Helicobacter pylori interstrain restriction-​modification diversity prevents genome subversion by chromosomal DNA from competing strains. Nucleic Acids Res, 30(24), 5391–​5397. PMCID: PMC140068. Bang YJ, Van Cutsem E, Feyereislova A et al. 2010. Trastuzumab in combination with chemotherapy versus chemotherapy alone for treatment of HER2-​ positive advanced gastric or gastro-​ oesophageal junction cancer (ToGA): a phase 3, open-​label, randomised controlled trial. Lancet, 376(9742), 687–​697. PMID: 20728210. Becker KF, Atkinson MJ, Reich U, et al. 1994. E-​cadherin gene mutations provide clues to diffuse type gastric carcinomas. Cancer Res, 54(14), 3845–​ 3852. PMID: 8033105. Bedoya A, Garay J, Sanzon F, et  al. 2003. Histopathology of gastritis in Helicobacter pylori-​infected children from populations at high and low gastric cancer risk. Hum Pathol, 34(3), 206–​213. PMID: 12673553. Bessede E, Dubus P, Megraud F, and Varon C. 2015. Helicobacter pylori infection and stem cells at the origin of gastric cancer. Oncogene, 34(20), 2547–​2555. PMID: 25043305. Blaser MJ, and Berg DE. 2001. Helicobacter pylori genetic diversity and risk of human disease. J Clin Invest, 107(7), 767–​773. PMCID: PMC199587. Blaser MJ, Chyou PH, and Nomura A. 1995. Age at establishment of Helicobacter pylori infection and gastric carcinoma, gastric ulcer, and duodenal ulcer risk. Cancer Res, 55(3), 562–​565. Blot WJ, Devesa SS, and Fraumeni JF, Jr. 1993. Continuing climb in rates of esophageal adenocarcinoma:  an update. JAMA, 270(11), 1320. PMID: 8360967. Bouvard V, Loomis D, Guyton KZ, et al. 2015. Carcinogenicity of consumption of red and processed meat. Lancet Oncol, 16(16), 1599–​ 1600. PMID: 26514947. Bray F, Jemal A, Grey N, Ferlay J, and Forman D. 2012. Global cancer transitions according to the Human Development Index (2008–​ 2030):  a population-​based study. Lancet Oncol, 13(8), 790–​801. PMID: 22658655. Brenner H, Arndt V, Bode G, et al. 2002. Risk of gastric cancer among smokers infected with Helicobacter pylori. Int J Cancer, 98(3), 446–​449. PMID: 11920598. Brose MS, Rebbeck TR, Calzone KA, et  al. 2002. Cancer risk estimates for BRCA1 mutation carriers identified in a risk evaluation program. J Natl Cancer Inst, 94(18), 1365–​1372. PMID: 12237282. Burke AP, Yen TS, Shekitka KM, and Sobin LH. 1990. Lymphoepithelial carcinoma of the stomach with Epstein-​Barr virus demonstrated by polymerase chain reaction. Mod Pathol, 3(3), 377–​380. PMID: 2163534. Caldas C, Carneiro F, Lynch HT, et  al. 1999. Familial gastric cancer:  overview and guidelines for management. J Med Genet, 36(12), 873–​880. PMID: 10593993. Camargo MC, Anderson WF, King JB, et al. 2011a. Divergent trends for gastric cancer incidence by anatomical subsite in US adults. Gut, 60(12), 1644–​ 1649. PMCID: PMC3202077. Camargo MC, Kim WH, Chiaravalli AM, et  al. 2014. Improved survival of gastric cancer with tumour Epstein-​Barr virus positivity: an international pooled analysis. Gut, 63(2), 236–​243. PMCID: PMC4384434. Camargo MC, Murphy G, Koriyama C, et al. 2011b. Determinants of Epstein-​ Barr virus-​positive gastric cancer: an international pooled analysis. Br J Cancer, 105(1), 38–​43. PMCID: PMC3137422. Chao A, Thun MJ, Henley SJ, et  al. 2002. Cigarette smoking, use of other tobacco products and stomach cancer mortality in US adults: The Cancer Prevention Study II. Int J Cancer, 101(4), 380–​389. PMID: 12209964. Chapelle N, Manfredi S, Lepage C, et al. 2016. Trends in gastric cancer incidence: a period and birth cohort analysis in a well-​defined French population. Gastric Cancer, 19(2), 508–​514. PMID: 26118904. Charnley G, and Tannenbaum SR. 1985. Flow cytometric analysis of the effect of sodium chloride on gastric cancer risk in the rat. Cancer Res, 45(11 Pt 2), 5608–​5616. PMID: 4053034. Charvat H, Sasazuki S, Inoue M et al. 2016. Prediction of the 10-​year probability of gastric cancer occurrence in the Japanese population: the JPHC study cohort II. Int J Cancer, 138(2), 320–​331. PMID: WOS:000369161200007.

 607

Stomach Cancer Chen S, Wang W, Lee S, et  al. 2006. Prediction of germline mutations and cancer risk in the Lynch syndrome. JAMA, 296(12), 1479–​ 1487. PMCID: PMC2538673. Cheng N, Hui DY, Liu Y, et  al. 2015. Is gastric lymphoepithelioma-​like carcinoma a special subtype of EBV-​associated gastric carcinoma? New insight based on clinicopathological features and EBV genome polymorphisms. Gastric Cancer, 18(2), 246–​255. PMID: 24771002. Chun N, and Ford JM. 2012. Genetic testing by cancer site: stomach. Cancer J, 18(4), 355–​363. PMID: 22846738. Coggon D, Osmond C, and Barker DJ. 1990. Stomach cancer and migration within England and Wales. Br J Cancer, 61(4), 573–​ 574. PMCID: PMC1971373. Coleman MP, Esteve J, Damiecki P, Arslan A, and Renard H. 1993. Trends in cancer incidence and mortality. IARC Sci Publ, 121, 1–​ 806. PMID: 8258476. Colquhoun A, Arnold M, Ferlay J, et al. 2015. Global patterns of cardia and non-​cardia gastric cancer incidence in 2012. Gut, 64(12), 1881–​1888. PMID: 25748648. Corley DA, Kubo A, and Zhao W. 2008. Abdominal obesity and the risk of esophageal and gastric cardia carcinomas. Cancer Epidemiol Biomarkers Prev, 17(2), 352–​358. PMCID: PMC2670999. Correa P. 2002. Gastric neoplasia. Curr Gastroenterol Rep, 4(6), 463–​470. PMID: 12441036. Correa P. 2004. The biological model of gastric carcinogenesis. In: Buffler P, Rice J, Baan R, Bird M, and Boffetta P (Eds.), Mechanisms of carcinogenesis:  contributions of molecular epidemiology (Vol. 157, pp. 301–​ 310). Lyon: International Agency for Research on Cancer. Correa P, Cuello C, Duque E, et  al. 1976. Gastric cancer in Colombia. III. Natural history of precursor lesions. J Natl Cancer Inst, 57(5), 1027–​ 1035. PMID: 1003539. Correa P, Haenszel W, Cuello C, Tannenbaum S, and Archer M. 1975. A model for gastric cancer epidemiology. Lancet, 2, 58. PMID: 49653. Correa P, and Piazuelo MB. 2008. Natural history of Helicobacter pylori infection. Dig Liver Dis, 40(7), 490–​496. PMCID: PMC3142999. Correa P, and Yardley JH. 1992. Grading and classification of chronic gastritis:  one American response to the Sydney system. Gastroenterology, 102(1), 355–​359. PMID: 1727769. Cristescu R, Lee J, Nebozhyn M, et  al. 2015. Molecular analysis of gastric cancer identifies subtypes associated with distinct clinical outcomes. Nat Med, 21(5), 449–​456. PMID: 25894828. Cross AJ, Freedman ND, Ren J, et  al. 2011. Meat consumption and risk of esophageal and gastric cancer in a large prospective study. Am J Gastroenterol, 106(3), 432–​442. PMCID: PMC3039705. D’Elia L, Galletti F, and Strazzullo P. 2014. Dietary salt intake and risk of gastric cancer. Cancer Treat Res, 159, 83–​95. PMID: 24114476. D’Elia L, Rossi G, Ippolito R, Cappuccio FP, and Strazzullo P. 2012. Habitual salt intake and risk of gastric cancer: a meta-​analysis of prospective studies. Clin Nutr, 31(4), 489–​498. Dai ZM, Zhang TS, Lin S, et  al. 2016. Role of IL-​17A rs2275913 and IL-​ 17F rs763780 polymorphisms in risk of cancer development: an updated meta-​analysis. Sci Rep, 6, 20439. PMCID: PMC4740815. de Vries AC, van Grieken NC, Looman CW, et al. 2008. Gastric cancer risk in patients with premalignant gastric lesions: a nationwide cohort study in the Netherlands. Gastroenterology, 134(4), 945–​952. PMID: 18395075. Derakhshan MH, Arnold M, Brewster DH, et  al. 2016. Worldwide inverse association between gastric cancer and esophageal adenocarcinoma suggesting a common environmental factor exerting opposing effects. Am J Gastroenterol, 111(2), 228–​239. PMID: 26753891. Derakhshan MH, Liptrot S, Paul J, et  al. 2009. Oesophageal and gastric intestinal-​ type adenocarcinomas show the same male predominance due to a 17  year delayed development in females. Gut, 58(1), 16–​23. PMID: 18838486. Derakhshan MH, Malekzadeh R, Watabe H, et  al. 2008. Combination of gastric atrophy, reflux symptoms and histological subtype indicates two distinct aetiologies of gastric cardia cancer. Gut, 57(3), 298–​305. PMID: 17965056. Derakhshan MH, Robertson EV, Lee YY, et  al. 2015. In healthy volunteers, immunohistochemistry supports squamous to columnar metaplasia as mechanism of expansion of cardia, aggravated by central obesity. Gut, 64(11), 1705–​1714. PMID: 25753030. Devesa SS, Blot WJ, and Fraumeni JF, Jr. 1998. Changing patterns in the incidence of esophageal and gastric carcinoma in the United States. Cancer, 83(10), 2049–​2053. PMID: 9827707. Disibio G, and French SW. 2008. Metastatic patterns of cancers:  results from a large autopsy study. Arch Pathol Lab Med, 132(6), 931–​939. PMID: 18517275. Dixon MF, Genta RM, Yardley JH, and Correa P. 1996. Classification and grading of gastritis: the updated Sydney System. International Workshop on

607

the Histopathology of Gastritis, Houston 1994. Am J Surg Pathol, 20(10), 1161–​1181. PMID: 8827022. Edgren G, Hjalgrim H, Rostgaard K, et al. 2010. Risk of gastric cancer and peptic ulcers in relation to ABO blood type: a cohort study. Am J Epidemiol, 172(11), 1280–​1285. PMID: 20937632. Ekstrom AM, Hansson LE, Signorello LB, et al. 2000a. Decreasing incidence of both major histologic subtypes of gastric adenocarcinoma: a population-​ based study in Sweden. Br J Cancer, 83(3), 391–​396. PMCID: 2374560. Ekstrom AM, Serafini M, Nyren O, et al. 2000b. Dietary antioxidant intake and the risk of cardia cancer and noncardia cancer of the intestinal and diffuse types:  a population-​based case-​control study in Sweden. Int J Cancer, 87(1), 133–​140. PMID: 10861464. El-​ Omar EM, Carrington M, Chow WH, et  al. 2000. Interleukin-​ 1 polymorphisms associated with increased risk of gastric cancer. Nature, 404(6776), 398–​402. PMID: 10746728. Fascioli S, Capocaccia R, and Mariotti S. 1995. Cancer mortality in migrant populations within Italy. Int J Epidemiol, 24(1), 8–​18. PMID: 7797360. Fenwick S, and Fenwick WS. 1903. Cancer and other tumours of the stomach. Philadelphia: P. Blakiston’s Son. Ferlay J, Soerjomataram I, Ervik M, et  al. 2013. GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality Worldwide:  IARC CancerBase No. 11 [Internet]. Available from: http://​globocan.iarc.fr. Accessed May 2015. Ferro A, Peleteiro B, Malvezzi M, et al. 2014. Worldwide trends in gastric cancer mortality (1980–​2011), with predictions to 2015, and incidence by subtype. Eur J Cancer, 50(7), 1330–​1344. PMID: 24650579. Fitzgerald RC, Hardwick R, Huntsman D, et  al. 2010. Hereditary diffuse gastric cancer:  updated consensus guidelines for clinical management and directions for future research. J Med Genet, 47(7), 436–​444. PMCID: PMC2991043. Forman D, Bray F, Brewster DH, et al. 2013. Cancer incidence in five continents, Vol. X. Available from: http://​ci5.iarc.fr. Accessed May 2015. Forman D, and Helicobacter and Cancer Collaborative Group. 2001. Gastric cancer and Helicobacter pylori:  a combined analysis of 12 case control studies nested within prospective cohorts. Gut, 49(3), 347–​ 353. PMCID: PMC1728434. Fox JG, Rogers AB, Ihrig M, et al. 2003. Helicobacter pylori-​associated gastric cancer in INS-​GAS mice is gender specific. Cancer Res, 63(5), 942–​950. PMID: 12615707. Freedman ND, Subar AF, Hollenbeck AR, et  al. 2008. Fruit and vegetable intake and gastric cancer risk in a large United States prospective cohort study. Cancer Causes Control, 19(5), 459–​467. PMID: 18166992. Fukase K, Kato M, Kikuchi S, et al. 2008. Effect of eradication of Helicobacter pylori on incidence of metachronous gastric carcinoma after endoscopic resection of early gastric cancer:  an open-​label, randomised controlled trial. Lancet, 372(9636), 392–​397. PMID: 18675689. Fukayama M, Hayashi Y, Iwasaki Y, et al. 1994. Epstein-​Barr virus-​associated gastric carcinoma and Epstein-​Barr virus infection of the stomach. Lab Invest, 71(1), 73–​81. PMID: 8041121. Fukayama M, and Ushiku T. 2011. Epstein-​Barr virus-​associated gastric carcinoma. Pathol Res Pract, 207(9), 529–​537. PMID: 21944426. Gaddy JA, Radin JN, Loh JT, et al. 2013. High dietary salt intake exacerbates Helicobacter pylori-​induced gastric carcinogenesis. Infect Immun, 81(6), 2258–​2267. PMCID: 3676043. Garcia-​Gonzalez MA, Bujanda L, Quintero E, et al. 2015. Association of PSCA rs2294008 gene variants with poor prognosis and increased susceptibility to gastric cancer and decreased risk of duodenal ulcer disease. Int J Cancer, 137(6), 1362–​1373. PMID: 25721731. Garrean S, Hering J, Saied A, Jani J, and Espat NJ. 2008. Gastric adenocarcinoma arising from fundic gland polyps in a patient with familial adenomatous polyposis syndrome. Am Surg, 74(1), 79–​83. PMID: 18274437. Gayther SA, Gorringe KL, Ramus SJ, et al. 1998. Identification of germ-​line E-​ cadherin mutations in gastric cancer families of European origin. Cancer Res, 58(18), 4086–​4089. PMID: 9751616. Giardiello FM, Brensinger JD, Tersmette AC, et  al. 2000. Very high risk of cancer in familial Peutz-​Jeghers syndrome. Gastroenterology, 119(6), 1447–​1453. PMID: 11113065. Gonzalez CA, Figueiredo C, Lic CB, et  al. 2011. Helicobacter pylori cagA and vacA genotypes as predictors of progression of gastric preneoplastic lesions:  a long-​term follow-​up in a high-​risk area in Spain. Am J Gastroenterol, 106(5), 867–​874. Gonzalez CA, Jakszyn P, Pera G, et al. 2006. Meat intake and risk of stomach and esophageal adenocarcinoma within the European Prospective Investigation Into Cancer and Nutrition (EPIC). J Natl Cancer Inst, 98(5), 345–​354. Gonzalez CA, Lujan-​Barroso L, Bueno-​de-​Mesquita HB, et al. 2012. Fruit and vegetable intake and the risk of gastric adenocarcinoma: a reanalysis of the European Prospective Investigation into Cancer and Nutrition (EPIC-​ EURGAST) study after a longer follow-​up. Int J Cancer, 131(12), 2910–​ 2919. PMID: 22473701.

608

608

PART IV:  Cancers by Tissue of Origin

Gonzalez KD, Buzin CH, Noltner KA, et al. 2009. High frequency of de novo mutations in Li-​Fraumeni syndrome. J Med Genet, 46(10), 689–​693. PMID: 19556618. Grady WM, Willis J, Guilford PJ, et al. 2000. Methylation of the CDH1 promoter as the second genetic hit in hereditary diffuse gastric cancer. Nat Genet, 26(1), 16–​17. PMID: 10973239. Gravalos C, and Jimeno A. 2008. HER2 in gastric cancer:  a new prognostic factor and a novel therapeutic target. Ann Oncol, 19(9), 1523–​1529. PMID: 18441328. Gronberg H, Bergh A, Damber JE, and Emanuelsson M. 2000. Cancer risk in families with hereditary prostate carcinoma. Cancer, 89(6), 1315–​1321. PMID: 11002228. Guilford P, Hopkins J, Harraway J, et al. 1998. E-​cadherin germline mutations in familial gastric cancer. Nature, 392(6674), 402–​405. PMID: 9537325. Guilford PJ, Hopkins JB, Grady WM, et al. 1999. E-​cadherin germline mutations define an inherited cancer syndrome dominated by diffuse gastric cancer. Hum Mutat, 14(3), 249–​255. PMID: 10477433. Haenszel W, Correa P, Cuello C, et al. 1976. Gastric cancer in Colombia. II. Case-​control epidemiologic study of precursor lesions. J Natl Cancer Inst, 57(5), 1021–​1026. PMID: 1003538. Hajmanoochehri F, Mohammadi N, Nasirian N, and Hosseinkhani M. 2013. Patho-​epidemiological features of esophageal and gastric cancers in an endemic region: a 20-​year retrospective study. Asian Pac J Cancer Prev, 14(6), 3491–​3497. PMID: 23886134. Hamashima C, Ogoshi K, Okamoto M, et al. 2013. A community-​based, case-​ control study evaluating mortality reduction from gastric cancer by endoscopic screening in Japan. PLoS One, 8(11), e79088. PMCID: 3827316. Hamashima C, Shabana M, Okada K, Okamoto M, and Osaki Y. 2015. Mortality reduction from gastric cancer by endoscopic and radiographic screening. Cancer Sci, 106(12), 1744–​1749. PMCID: 4714659. Hamilton-​Dutoit SJ, and Pallesen G. 1994. Detection of Epstein-​Barr virus small RNAs in routine paraffin sections using non-​isotopic RNA/​RNA in situ hybridization. Histopathology, 25(2), 101–​111. PMID: 7982672. Hartfall SJ. 1936. Gastritis in theory and practice. Br Med J, 1(3936), 1200–​ 1204. PMCID: PMC2458877. Henson DE, Dittus C, Younes M, Nguyen H, and Albores-​Saavedra J. 2004. Differential trends in the intestinal and diffuse types of gastric carcinoma in the United States, 1973–​2000: increase in the signet ring cell type. Arch Pathol Lab Med, 128(7), 765–​770. PMID: 15214826. Herrera V, and Parsonnet J. 2009. Helicobacter pylori and gastric adenocarcinoma. Clin Microbiol Infect, 15(11), 971–​976. Herrmann K, and Niedobitek G. 2003. Epstein-​Barr virus-​associated carcinomas: facts and fiction. J Pathol, 199(2), 140–​145. PMID: 12533825. Houghton J, Stoicov C, Nomura S, et  al. 2004. Gastric cancer originating from bone marrow-​derived cells. Science, 306(5701), 1568–​1571. PMID: 15567866. Howe JR, Sayed MG, Ahmed AF, et  al. 2004. The prevalence of MADH4 and BMPR1A mutations in juvenile polyposis and absence of BMPR2, BMPR1B, and ACVR1 mutations. J Med Genet, 41(7), 484–​ 491. PMCID: PMC1735829. Howlader N, Noone AM, Krapcho M, et  al. 2015. SEER Cancer Statistics Review, 1975–​2012, National Cancer Institute. Bethesda, MD, http://​ seer.cancer.gov/​csr/​1975_​2012/​, based on November 2014 SEER data submission. Howson CP, Hiyama T, and Wynder EL. 1986. The decline in gastric cancer:  epidemiology of an unplanned triumph. Epidemiol Rev, 8, 1–​27. PMID: 3533579. Hu B, El Hajj N, Sittler S, et al. 2012. Gastric cancer: classification, histology and application of molecular pathology. J Gastrointest Oncol, 3(3), 251–​ 261. PMCID: PMC3418539. Huang YK, Yu JC, Kang WM, et  al. 2015. Significance of serum pepsinogens as a biomarker for gastric cancer and atrophic gastritis screening:  a systematic review and meta-​ analysis. PLoS One, 10(11), 23. PMID: WOS:000364430700050. IARC. 1986. Tobacco smoking. Vol. 38. Lyon, France. February 12–​20, 1985. IARC Monogr Eval Carcinog Risks Hum, 38. IARC. 1994. Schistosomes, liver flukes and Helicobacter pylori. Vol. 61. Lyon, France, June 7–​ 14, 1994. IARC Monogr Eval Carcinog Risks Hum, 61, 1–​241. IARC. 2004. Tobacco smoke and involuntary smoking. Vol. 83. Lyon, France, June 11–​18, 2002. IARC Monogr Eval Carcinog Risks Hum, 83, 1–​1438. PMID: 15285078. IARC. 2010. Alcohol consumption and ethyl carbamate. Vol. 96. Lyon, France, February 6–​ 13, 2007. IARC Monogr Eval Carcinog Risks Hum, 96, 3–​1383. IARC. 2012a. Biological agents. Vol. 100B. A review of human carcinogens. Lyon, France, February 24–​March 3, 2009. IARC Monogr Eval Carcinog Risks Hum, 100B.

IARC. 2012b. Personal habits and indoor combustions. Vol. 100E. A review of human carcinogens. IARC Monogr Eval Carcinog Risks Hum, 100(Pt E), 1–​538. IARC. 2012c. Radiation. Vol. 100D. A review of human carcinogens. Lyon, France, June 2–​9, 2009. IARC Monogr Eval Carcinog Risks Hum, 100(Pt D), 7–​303. Imai S, Koizumi S, Sugiura M, et al. 1994. Gastric carcinoma: monoclonal epithelial malignant cells expressing Epstein-​Barr virus latent infection protein. Proc Natl Acad Sci USA, 91(19), 9131–​9135. PMCID: PMC44761. Jemal A, Murray T, Samuels A, et al. 2003. Cancer statistics, 2003. CA Cancer J Clin, 53(1), 5–​26. PMID: 12568441. Jenab M, Riboli E, Ferrari P, et al. 2006. Plasma and dietary vitamin C levels and risk of gastric cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC-​EURGAST). Carcinogenesis, 27(11), 2250–​ 2257. PMID: 16774936. Jonsson BA, Bergh A, Stattin P, Emmanuelsson M, and Gronberg H. 2002. Germline mutations in E-​cadherin do not explain association of hereditary prostate cancer, gastric cancer and breast cancer. Int J Cancer, 98(6), 838–​843. PMID: 11948460. Kaneko S, and Yoshimura T. 2001. Time trend analysis of gastric cancer incidence in Japan by histological types, 1975–​1989. Br J Cancer, 84(3), 400–​405. PMCID: 2363747. Keilberg D, and Ottemann KM. 2016. How Helicobacter pylori senses, targets and interacts with the gastric epithelium. Environ Microbiol, 18(3), 791–​ 806. PMID: 26768806. Keszei AP, Schouten LJ, Goldbohm RA, and van den Brandt PA. 2012. Red and processed meat consumption and the risk of esophageal and gastric cancer subtypes in The Netherlands Cohort Study. Ann Oncol, 23(9), 2319–​ 2326. PMID: 22351741. Kneller RW, You WC, Chang YS, et al. 1992. Cigarette smoking and other risk factors for progression of precancerous stomach lesions. J Natl Cancer Inst, 84(16), 1261–​1266. PMID: 1640486. Kolonel LN, Hankin JH, and Nomura AM. 1985. Multiethnic studies of diet, nutrition, and cancer in Hawaii. Princess Takamatsu Symp, 16, 29–​40. PMID: 3916200. Kreinitz W. 1906. Ueber das Auftreten von Spirochaeten verschiedener Form im Mageninhalt bei Carcinoma ventriculi. Deut Med Wochenschr, 32, 872. Kuriyama S, Tsubono Y, Hozawa A, et al. 2005. Obesity and risk of cancer in Japan. Int J Cancer, 113(1), 148–​157. Ladeiras-​Lopes R, Pereira AK, Nogueira A, et al. 2008. Smoking and gastric cancer:  systematic review and meta-​analysis of cohort studies. Cancer Causes Control, 19(7), 689–​701. Laurén P. 1965. The two histological main types of gastric cancer: diffuse and so-​called intestinal type carcinoma. Acta Pathol Microbiol Scand, 64, 31. PMID: 173. Laurén PA, and Nevalainen TJ. 1993. Epidemiology of intestinal and diffuse types of gastric carcinoma: a time-​trend study in Finland with comparison between studies from high-​and low-​risk areas. Cancer, 71(10), 2926–​ 2933. PMID: 8490820. Lauwers GY, and Srivastava A. 2007. Gastric preneoplastic lesions and epithelial dysplasia. Gastroenterol Clin North Am, 36(4), 813–​ 829, vi. PMID: 17996792. Lee JH, Kim SH, Han SH, et al. 2009. Clinicopathological and molecular characteristics of Epstein-​Barr virus-​associated gastric carcinoma:  a meta-​ analysis. J Gastroenterol Hepatol, 24(3), 354–​365. PMID: 19335785. Leitzmann MF, Koebnick C, Freedman ND, et al. 2009. Physical activity and esophageal and gastric carcinoma in a large prospective study. Am J Prev Med, 36(2), 112–​119. Leivonen M, Nordling S, and Haglund C. 1997. Does Helicobacter pylori in the gastric stump increase the cancer risk after certain reconstruction types? Anticancer Res, 17(5B), 3893–​3896. Levi F, La Vecchia C, and Te VC. 1990. Descriptive epidemiology of adenocarcinomas of the cardia and distal stomach in the Swiss Canton of Vaud. Tumori, 76(2), 167–​171. PMID: 2330608. Lewin KJ, and Appelman HD. 1995. Stomach cancer. In: Atlas of tumor pathology. Washington, DC: Armed Forces Institue of Pathology. Lu Y, Chen J, Ding Y, et al. 2010. Genetic variation of PSCA gene is associated with the risk of both diffuse-​and intestinal-​type gastric cancer in a Chinese population. Int J Cancer, 127(9), 2183–​2189. PMID: 20131315. Lukanova A, Bjor O, Kaaks R, et al. 2006. Body mass index and cancer: results from the Northern Sweden Health and Disease Cohort. Int J Cancer, 118(2), 458–​466. Machado JC, Figueiredo C, Canedo P, et al. 2003. A proinflammatory genetic profile increases the risk for chronic atrophic gastritis and gastric carcinoma. Gastroenterology, 125(2), 364–​371. PMID: 12891537. Machii R, and Saika K. 2016. Subsite distribution of stomach cancer from Cancer Incidence in Five Continents Vol. X. Jpn J Clin Oncol, 46(1), 98. PMID: 26709380.

 609

Stomach Cancer Majewski IJ, Kluijt I, Cats A, et  al. 2013. An alpha-​E-​catenin (CTNNA1) mutation in hereditary diffuse gastric cancer. J Pathol, 229(4), 621–​629. PMID: 23208944. Malfertheiner P, Megraud F, O’Morain CA et  al. 2012. Management of Helicobacter pylori infection:  the Maastricht IV/​Florence Consensus Report. Gut, 61(5), 646–​664. PMID: 22491499. Malvezzi M, Bonifazi M, Bertuccio P, et  al. 2010. An age-​ period-​ cohort analysis of gastric cancer mortality from 1950 to 2007 in Europe. Ann Epidemiol, 20(12), 898–​905. PMID: 21074104. McLean MH, and El-​Omar EM. 2014. Genetics of gastric cancer. Nat Rev Gastroenterol Hepatol, 11(11), 664–​674. PMID: 25134511. McMichael AJ, McCall MG, Hartshorne JM, and Woodings TL. 1980. Patterns of gastro-​intestinal cancer in European migrants to Australia: the role of dietary change. Int J Cancer, 25(4), 431–​437. PMID: 7372370. Mezhir JJ, Gonen M, Ammori JB, et  al. 2011. Treatment and outcome of patients with gastric remnant cancer after resection for peptic ulcer disease. Ann Surg Oncol, 18(3), 670–​676. Miao Q, Ma YZ, Cai GH, and Chen XY. 2014. Etiology and morphology of carditis: experiences from a single center in China. J Dig Dis, 15(2), 71–​77. PMID: 24151905. Minoura-​Etoh J, Gotoh K, Sato R, et al. 2006. Helicobacter pylori-​associated oxidant monochloramine induces reactivation of Epstein-​ Barr virus (EBV) in gastric epithelial cells latently infected with EBV. J Med Microbiol, 55(Pt 7), 905–​911. PMID: 16772418. Misumi A, Murakami A, Harada K, Baba K, and Akagi M. 1989. Definition of carcinoma of the gastric cardia. Langenbecks Arch Chir, 374(4), 221–​ 226. PMID: 2761324. Murphy G, Pfeiffer R, Camargo MC, and Rabkin CS. 2009. Meta-​analysis shows that prevalence of Epstein-​Barr virus-​positive gastric cancer differs based on sex and anatomic location. Gastroenterology, 137(3), 824–​833. PMCID: PMC3513767. Nagel G, Linseisen J, Boshuizen HC, et al. 2007. Socioeconomic position and the risk of gastric and oesophageal cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC-​ EURGAST). Int J Epidemiol, 36(1), 66–​76. Nevalainen TJ. 2013. Pekka Laurén and histological classification of gastric carcinoma. Gut, 62(8), 1230–​1231. Orlando G, Pilone V, Vitiello A, et al. 2014. Gastric cancer following bariatric surgery: a review. Surg Laparosc Endosc Percutan Tech, 24(5), 400–​405. PMID: 25238176. Palli D, Galli M, Caporaso NE, et al. 1994. Family history and risk of stomach cancer in Italy. Cancer Epidemiol Biomarkers Prev, 3(1), 15–​18. Parsonnet J. 1999. Helicobacter pylori and gastric adenocarcinoma. In:  Parsonnet J (Ed.), Microbes and malignancies (pp. 372–​ 408). New York: Oxford University Press. Parsonnet J, Friedman GD, Orentreich N, and Vogelman H. 1997. Risk for gastric cancer in people with CagA positive or CagA negative Helicobacter pylori infection. Gut, 40(3), 297–​301. PMCID: PMC1027076. Patra R, Chattopadhyay S, De R, et al. 2012. Multiple infection and microdiversity among Helicobacter pylori isolates in a single host in India. PLoS One, 7(8), e43370. PMCID: PMC3428359. PDQ® Screening and Prevention Editorial Board. 2016. PDQ Stomach (Gastric) Cancer Screening. Bethesda, MD: National Cancer Institute. Available at: http:// www.cancer.gov/types/stomach/hp/ stomach- screening- pdq. Accessed April 25, 2016. [PMID: 26389174]. Peek RM, Jr, and Crabtree JE. 2006. Helicobacter infection and gastric neoplasia. J Pathol, 208(2), 233–​248. PMID: 16362989. Peleteiro B, Lopes C, Figueiredo C, and Lunet N. 2011. Salt intake and gastric cancer risk according to Helicobacter pylori infection, smoking, tumour site and histological type. Br J Cancer, 104(1), 198–​207. PMCID: PMC3039805. Persson C, Canedo P, Machado JC, El-​ Omar EM, and Forman D. 2011. Polymorphisms in inflammatory response genes and their association with gastric cancer: a HuGE systematic review and meta-​analyses. Am J Epidemiol, 173(3), 259–​270. PMCID: PMC3105271. Pinheiro H, Oliveira C, Seruca R, and Carneiro F. 2014. Hereditary diffuse gastric cancer: pathophysiology and clinical management. Best Pract Res Clin Gastroenterol, 28(6), 1055–​1068. PMID: 25439071. Plummer M, De Martel C, Vignat J, et al. 2016. Global burden of cancer attributable to infections in 2012. Lancet Public Health, 4(9), e609–​e616. doi:  10.1016/​ S2214-​ 109X(16)30143-​ 7. Epub 2016 Jul 25. PubMed PMID: 27470177. Plummer M, Franceschi S, Vignat J, Forman D, and de Martel C. 2015. Global burden of gastric cancer attributable to Helicobacter pylori. Int J Cancer, 136(2), 487–​490. PMID: 24889903. Plummer M, van Doorn LJ, Franceschi S, et  al. 2007. Helicobacter pylori cytotoxin-​associated genotype and gastric precancerous lesions. J Natl Cancer Inst, 99(17), 1328–​1334.

609

Powell J, and McConkey CC. 1992. The rising trend in oesophageal adenocarcinoma and gastric cardia. Eur J Cancer Prev, 1(3), 265–​ 269. PMID: 1467772. Prinz C, Schoniger M, Rad R, et al. 2001. Key importance of the Helicobacter pylori adherence factor blood group antigen binding adhesin during chronic gastric inflammation. Cancer Res, 61(5), 1903–​1909. PMID: 11280745. Rampazzo A, Mott GL, Fontana K, and Fagundes RB. 2012. Gastric adenocarcinoma trends in the central region of Rio Grande do Sul (Southern Brazil):  what has changed in 25  years? Arq Gastroenterol, 49(3), 178–​ 183. PMID: 23011238. Ren JS, Kamangar F, Forman D, and Islami F. 2012. Pickled food and risk of gastric cancer: a systematic review and meta-​analysis of English and Chinese literature. Cancer Epidemiol Biomarkers Prev, 21(6), 905–​915. PMID: 22499775. Rogers AB, Taylor NS, Whary MT, et al. 2005. Helicobacter pylori but not high salt induces gastric intraepithelial neoplasia in B6129 mice. Cancer Res, 65(23), 10709–​10715. PMID: 16322215. Rollan A, Arab JP, Camargo MC, et  al. 2014. Management of Helicobacter pylori infection in Latin America: a Delphi technique-​based consensus. World J Gastroenterol, 20(31), 10969–​10983. PMCID: PMC4138478. Rothenbacher D, Bode G, Berg G, et al. 1999. Helicobacter pylori among preschool children and their parents: evidence of parent-​child transmission. J Infect Dis, 179(2), 398–​402. Rugge M, Correa P, Dixon MF, et  al. 2000. Gastric dysplasia:  the Padova international classification. Am J Surg Pathol, 24(2), 167–​176. PMID: 10680883. Sakamoto H, Yoshimura K, Saeki N, et al. 2008. Genetic variation in PSCA is associated with susceptibility to diffuse-​type gastric cancer. Nat Genet, 40(6), 730–​740. PMID: 18488030. Sala N, Munoz X, Travier N, et  al. 2012. Prostate stem-​cell antigen gene is associated with diffuse and intestinal gastric cancer in Caucasians: results from the EPIC-​EURGAST study. Int J Cancer, 130(10), 2417–​2427. PMID: 21681742. Santibanez M, Aguirre E, Belda S, et al. 2015. Relationship between tobacco, cagA and vacA i1 virulence factors and bacterial load in patients infected by Helicobacter pylori. PLoS One, 10(3), e0120444. PMCID: PMC4368826. Sauvaget C, Lagarde F, Nagano J, et al. 2005. Lifestyle factors, radiation and gastric cancer in atomic-​bomb survivors (Japan). Cancer Causes Control, 16(7), 773–​780. Schlemper RJ, Riddell RH, Kato Y, et  al. 2000. The Vienna classification of gastrointestinal epithelial neoplasia. Gut, 47(2), 251–​ 255. PMCID: PMC1728018. Segal ED, Lange C, Covacci A, Tompkins LS, and Falkow S. 1997. Induction of host signal transduction pathways by Helicobacter pylori. Proc Natl Acad Sci U S A, 94(14), 7595–​7599. PMCID: PMC23867. Shakeri R, Malekzadeh R, Nasrollahzadeh D, et al. 2015. Multiplex H. pylori serology and risk of gastric cardia and noncardia adenocarcinomas. Cancer Res, 75(22), 4876–​4883. PMCID: PMC4792189. Shikata K, Kiyohara Y, Kubo M, et al. 2006. A prospective study of dietary salt intake and gastric cancer incidence in a defined Japanese population: the Hisayama study. Int J Cancer, 119(1), 196–​201. Shinmura K, Kohno T, Takahashi M, et  al. 1999. Familial gastric cancer:  clinicopathological characteristics, RER phenotype and germline p53 and E-​ cadherin mutations. Carcinogenesis, 20(6), 1127–​ 1131. PMID: 10357799. Shinozaki-​Ushiku A, Kunita A, and Fukayama M. 2015. Update on Epstein-​ Barr virus and gastric cancer (review). Int J Oncol, 46(4), 1421–​1434. PMID: 25633561. Shiota S, Matsunari O, Watada M, and Yamaoka Y. 2010. Serum Helicobacter pylori CagA antibody as a biomarker for gastric cancer in east-​Asian countries. Future Microbiol, 5(12), 1885–​1893. PMCID: PMC3044821. Siegel RL, Miller KD, and Jemal A. 2016. Cancer statistics, 2016. CA Cancer J Clin, 66(1), 7–​30. PMID: 26742998. Siman JH, Forsgren A, Berglund G, and Floren CH. 2001. Tobacco smoking increases the risk for gastric adenocarcinoma among Helicobacter pylori-​infected individuals. Scand J Gastroenterol, 36(2), 208–​ 213. PMID: 11252415. Sipponen P, and Correa P. 2002. Delayed rise in incidence of gastric cancer in females results in unique sex ratio (M/​F) pattern:  etiologic hypothesis. Gastric Cancer, 5(4), 213–​219. Siurala M, Sipponen P, and Kekki M. 1985. Chronic gastritis: dynamic and clinical aspects. Scand J Gastroenterol Suppl, 109, 69–​76. PMID: 3895391. Sjodahl K, Jia C, Vatten L, et al. 2008a. Body mass and physical activity and risk of gastric cancer in a population-​based cohort study in Norway. Cancer Epidemiol Biomarkers Prev, 17(1), 135–​140. Sjodahl K, Jia C, Vatten L, et  al. 2008b. Salt and gastric adenocarcinoma:  a population-​based cohort study in Norway. Cancer Epidemiol Biomarkers Prev, 17(8), 1997–​2001.

610

610

PART IV:  Cancers by Tissue of Origin

Song H, Ekheden IG, Zheng Z, et al. 2015. Incidence of gastric cancer among patients with gastric precancerous lesions: observational cohort study in a low risk Western population. Brit Med J, 351, h3867. PMCID: PMC4516137. Spallek J, Arnold M, Razum O, et al. 2012. Cancer mortality patterns among Turkish immigrants in four European countries and in Turkey. Eur J Epidemiol, 27(12), 915–​921. PMID: 23179631. Steevens J, Botterweck AA, Dirx MJ, van den Brandt PA, and Schouten LJ. 2010. Trends in incidence of oesophageal and stomach cancer subtypes in Europe. Eur J Gastroenterol Hepatol, 22(6), 669–​678. PMID: 19474750. Steevens J, Schouten LJ, Goldbohm RA, and van den Brandt PA. 2011. Vegetables and fruits consumption and risk of esophageal and gastric cancer subtypes in the Netherlands Cohort Study. Int J Cancer, 129(11), 2681–​2693. PMID: 21960262. Stein M, Bagnoli F, Halenbeck R et  al. 2002. c-​Src/​ Lyn kinases activate Helicobacter pylori CagA through tyrosine phosphorylation of the EPIYA motifs. Mol Microbiol, 43(4), 971–​980. PMID: 11929545. Stolte M. 2003. The new Vienna classification of epithelial neoplasia of the gastrointestinal tract: advantages and disadvantages. Virchows Arch, 442(2), 99–​106. PMID: 12596058. Takada K. 2000. Epstein-​Barr virus and gastric carcinoma. Mol Pathol, 53(5), 255–​261. PMCID: PMC1186978. Takahashi M, Nishikawa A, Furukawa F, et al. 1994. Dose-​dependent promoting effects of sodium chloride (NaCl) on rat glandular stomach carcinogenesis initiated with N-​methyl-​N’-​nitro-​N-​nitrosoguanidine. Carcinogenesis, 15(7), 1429–​1432. PMID: 8033321. Terada T. 2016. Histopathological study using computer database of 10 000 consecutive gastric specimens: (2) malignant lesions. Gastroenterol Rep (Oxf), 4(1), 54–​58. PMCID: 4760062. Terry MB, Gaudet MM, and Gammon MD. 2002. The epidemiology of gastric cancer. Semin Radiat Oncol, 12(2), 111–​127. PMID: 11979413. The Cancer Genome Atlas Research Network. 2014. The Cancer Genome Atlas Research Network: comprehensive molecular characterization of gastric adenocarcinoma. Nature, 513(7517), 202–​209. PMCID: PMC4170219. The Paris Classification. 2003. The Paris endoscopic classification of superficial neoplastic lesions: esophagus, stomach, and colon: November 30 to December 1, 2002. Gastrointest Endosc, 58(6 Suppl), S3–​43. PMID: 14652541. Tramacere I, Negri E, Pelucchi C, et al. 2012. A meta-​analysis on alcohol drinking and gastric cancer risk. Ann Oncol, 23(1), 28–​36. Wang Q, Chen Y, Wang X, et al. 2014. Consumption of fruit, but not vegetables, may reduce risk of gastric cancer: results from a meta-​analysis of cohort studies. Eur J Cancer, 50(8), 1498–​1509. PMID: 24613128. Wang Z, Liu L, Ji J, et  al. 2012. ABO blood group system and gastric cancer: a case-​control study and meta-​analysis. Int J Mol Sci, 13(10), 13308–​ 13321. PMCID: PMC3497328. Warren JR, and Marshall BJ. 1983. Unidentified curved bacilli on gastric epithelium in active chronic gastritis. Lancet, 1(8336), 1273–​1275. Washington K. 2010. 7th edition of the AJCC cancer staging manual: stomach. Ann Surg Oncol, 17(12), 3077–​3079. PMID: 20882416.

Watson P, Vasen HF, Mecklin JP, et al. 2008. The risk of extra-​colonic, extra-​ endometrial cancer in the Lynch syndrome. Int J Cancer, 123(2), 444–​ 449. PMCID: PMC2627772. Whitlock G. 2012. Japanese stomach cancer mortality at age 35–​ 69:  1952–​2006. From www.mortality-​trends.org, using WHO and UN Population Division data. Available from:  http://​www.mortality-​ trends.org/​specify.php?place=jap&cod=nst&age=35–​69&submit=Go. Accessed 2016. WHO. 2010. WHO Classifications of tumours of the digestive system. Lyon, France: IARC Press. WHO. 2013. International classification of diseases for oncology, 3rd ed. Geneva: World Health Organization. Wiseman M. 2008. The second World Cancer Research Fund/​American Institute for Cancer Research expert report. Food, nutrition, physical activity, and the prevention of cancer:  a global perspective. Proc Nutr Soc, 67(3), 253–​256. World Cancer Research Fund International/​ American Institute for Cancer Research (WCRF/​AICR). 2016. Continuous update project report: diet, nutrition, physical activity and stomach cancer. Available at:  wcrf.org/​ stomach-​cancer-​2016. Wu H, Rusiecki JA, Zhu K, Potter J, and Devesa SS. 2009. Stomach carcinoma incidence patterns in the United States by histologic type and anatomic site. Cancer Epidemiol Biomarkers Prev, 18(7), 1945–​1952. PMCID: PMC2786772. Wu MS, Shun CT, Wu CC, et al. 2000. Epstein-​Barr virus-​associated gastric carcinomas: relation to H. pylori infection and genetic alterations. Gastroenterology, 118(6), 1031–​1038. PMID: 10833477. Yabuta T, Shinmura K, Tani M, et al. 2002. E-​cadherin gene variants in gastric cancer families whose probands are diagnosed with diffuse gastric cancer. Int J Cancer, 101(5), 434–​441. PMID: 12216071. Yamaguchi Y, Nagata Y, Hiratsuka R, et al. 2016. Gastric cancer screening by combined assay for serum anti-​Helicobacter pylori IgG anbitody and serum pepsinogen levels: the ABC method. Digestion, 93, 13–​18. Yamazaki S, Yamakawa A, Ito Y, et al. 2003. The CagA protein of Helicobacter pylori is translocated into epithelial cells and binds to SHP-​2 in human gastric mucosa. J Infect Dis, 187(2), 334–​337. PMID: 12552462. Yang P, Zhou Y, Chen B, et al. 2009. Overweight, obesity and gastric cancer risk: results from a meta-​analysis of cohort studies. Eur J Cancer, 45(16), 2867–​2873. Yang PC, and Davis S. 1988. Epidemiological characteristics of adenocarcinoma of the gastric cardia and distal stomach in the United States, 1973–​ 1982. Int J Epidemiol, 17(2), 293–​297. PMID: 3403124. Yoon KA, Ku JL, Yang HK, et al. 1999. Germline mutations of E-​cadherin gene in Korean familial gastric cancer patients. J Hum Genet, 44(3), 177–​180. PMID: 10319582. Zaridze D, Borisova E, Maximovitch D, and Chkhikvadze V. 1999. Aspirin protects against gastric cancer: results of a case-​control study from Moscow, Russia. Int J Cancer, 82(4), 473–​476. PMID: 10404057.

 61

32

Cancer of the Pancreas SAMUEL O. ANTWI, RICK J. JANSEN, AND GLORIA M. PETERSEN

OVERVIEW Pancreatic cancer (PC) is an uncommon but often rapidly lethal malignancy. Worldwide, PC is the twelfth most commonly diagnosed cancer and the seventh most common cause of cancer-​related death. Globally, the estimated incidence and death rates from PC were almost identical in 2012, resulting in 338,000 new cases and 330,400 deaths. In high-​resource countries, the recorded incidence rates are higher than in developing countries, at least partly reflecting more available means of diagnosis, but the prognosis is only slightly better. In the United States, PC ranks twelfth in cancer incidence among men and ninth among women; it ranks fourth in cancer mortality in both sexes, and third when men and women are combined. Etiologic research on PC is complicated by the relatively inaccessible location of the pancreas, the obstacles to early diagnosis, the aggressiveness and resistance to therapy of these malignancies, and the tendency of PC to progress rapidly from diagnosis to death. Until recently, the only etiologic factors considered definitive causes of PC were tobacco use, chronic pancreatitis, and several rare high-​penetrance genetic disorders. In the past 10–​15 years, the evidence for other causal relationships has strengthened, especially for metabolic risk factors (obesity, type 2 diabetes mellitus, insulin, and insulin-​like growth factor), chronic local inflammation, heavy alcohol consumption, dietary consumption of grilled meat, and non-​O ABO blood type. Among US whites, the incidence and mortality rates from PC, which were decreasing in the late twentieth century following large reductions in smoking, are now increasing, paralleling large increases in the prevalence of obesity and type 2 diabetes. Looking at SEER 13 area trends from 2005-​2014, increasing incidence and mortality were observed for all racial groups, except Blacks and American Indians/​ Alaska Natives who show a decreasing trend in both incidence and mortality. Research on PC is focused on factors along the entire chain connecting structural changes in DNA (germline and somatic), through intermediate epigenetic and other influences on transcription and translation, to the phenotypic abnormalities of neoplasia.

INTRODUCTION Research on PC has long been impeded by the relatively inaccessible location of the pancreas, the lack of access to tissue during the preclinical development of tumors, the difficulties of early diagnosis, aggressiveness of the disease, resistance to therapy, and the typically rapid progression from diagnosis to death (Anderson et  al., 2006). Partly because of these barriers, etiologic studies in the twentieth century identified few factors that were considered definite causes of PC. Multiple epidemiologic studies established that tobacco use was an important and potentially modifiable cause. Clinical studies confirmed strong relationships between PC and chronic pancreatitis, and smaller increases in risk associated with several rare high-​penetrance genetic disorders and a history of gastric surgery. Despite numerous studies of other exposures, the evidence remained inconsistent and/​ or difficult to interpret into the early twenty-​first century, as reviewed by Anderson et  al. in the third edition of this text (Anderson et  al., 2006). This chapter is divided into sections on the two major types of pancreatic carcinoma (ductal adenocarcinoma and neuroendocrine/​ islet cell tumors), and discusses classification, diagnosis, demographic

patterns, environmental and host risk factors, molecular pathogenesis, and preventive measures.

CLINICAL AND PATHOLOGICAL FEATURES Anatomy, Presenting Symptoms, and Diagnosis The pancreas is a glandular organ of the digestive system, located behind the stomach and lying against and connected to the duodenum by a main duct that joins to the common bile duct. The head of the pancreas is closest to the duodenum, and distally from it are the body and tail. The pancreas has two functions. Its exocrine function is to produce digestive enzymes (proteases, lipase, and amylase) that are secreted through a network of smaller pancreatic ducts that join the main pancreatic duct. Its endocrine function is to produce hormones (glucagon and insulin) in the islet cells located throughout the pancreas, which help maintain glucose homeostasis. Cancer can occur either in the cells lining the ducts or in the islet cells. Most symptoms of PC do not appear until the tumor is at a late stage. Approximately 80% of patients have unresectable disease at the time of diagnosis due to metastatic spread or locally advanced disease. Typically, individuals diagnosed with PC report non-​specific abdominal pain, jaundice, and/​or unintended weight loss. Jaundice is usually due to ductal tumors located in the pancreatic head, obstructing the bile duct (Ryan et al., 2014; Yamada et al., 2009). Approximately 70% of ductal tumors are located in the head of the pancreas, 5%–​ 10% in the body, and 10%–​15% in the tail. Diagnosis of pancreatic cancer is usually confirmed by imaging that includes a combination of computed tomography (CT) scanning, magnetic resonance imaging (MRI), and positron emission tomography (PET), and/​or endoscopic ultrasound (EUS) and endoscopic retrograde cholangiopancreatography (ERCP), pancreatic biopsy, and serum levels of CA 19-​9.

Tumor Subtypes The most common form of pancreatic cancer is ductal adenocarcinoma (in the exocrine portion of the gland), accounting for > 95% of all pancreatic cancers. Ductal cells constitute 10%–​15% of the volume but give rise to 90% of all tumors, as discussed in the following section of the chapter. Acinar cell tumors are rare (1% of exocrine pancreatic cancer), but these cells comprise 80% of the volume of the gland (Antonello et al., 2009). The other main group of pancreatic tumors is neuroendocrine in origin (islet cells), as discussed in the subsequent section. It is proposed that ductal adenocarcinoma progresses in a stepwise fashion through acquisition of molecular alterations and histologically well-​defined non-​invasive precursor lesions (Maitra and Hruban, 2008). Precursor lesions progress from a benign intraductal tumor through increasing grades of dysplasia to invasive adenocarcinoma, providing models of neoplastic pancreatic progression (Grutzmann et al., 2010; Mettu and Abbruzzese, 2016). The classic morphologic progression is suggested to occur through pancreatic intraepithelial neoplasias (PanINs). PanINs are microscopic lesions in small (less than 5 mm) pancreatic ducts and are classified into two grades (low-​ grade PanIN-​ 1 and PanIN-​ 2), and high-​ grade (PanIN-​ 3, formerly known as carcinoma in situ) (Basturk et al., 2015; Hruban et al., 2001; Ying et al., 2016). Another precursor of invasive pancreatic carcinomas

611

612

612

PART IV:  Cancers by Tissue of Origin

is pancreatic intraductal papillary mucinous neoplasia (IPMN), which are cystic lesions. With advances in endoscopic imaging, more patients are being diagnosed with IPMN and other cystic tumors, although the vast majority are asymptomatic and do not progress to malignancy. An estimated 2% of adults, and up to 10% of individuals aged 70 and older, have IPMNs (Ryan et al., 2014).

Molecular Pathogenesis/​Somatic Mutations Following a tumor progression model delineated by genetic events in pancreatic tumorigenesis (Bardeesy and DePinho, 2002), there has been consistency and confirmation in the body of research over many studies since 2002. The categories of molecular events include the following: (a) Activation by mutations in the KRAS oncogene, among the earliest events; 95% of pancreatic ductal adenocarcinomas possess a KRAS mutation, most commonly in codon 12. (b) Inactivation of tumor suppressor genes, TP53 in 60% of tumors, SMAD4 in half of tumors, and CDKN2A in 95% of tumors, occur later in the progression pathway. (c) Mutations of mismatch repair genes, such as MLH1 and MSH2, have been found in 4% of pancreatic tumors. It is thought that these molecular events, whether gradual or rapid, collectively have an impact in several core pathways (Notch, Hedgehog, beta-​ catenin, axon guidance, chromatin remodeling, and DNA repair) that have been proposed by Jones et al. (2008) and further refined and expanded by Biankin and colleagues (2012). Understanding these molecular events will shed light on future treatment and early detection strategies.

DUCTAL ADENOCARCINOMA Descriptive Epidemiology The diagnosis of PC is challenging, even in high-​resource countries, and tends to be underdiagnosed, particularly in countries that lack resources for sophisticated diagnostic testing (Wolfgang et al., 2013). Thus, the interpretation of incidence and mortality data is more tractable when comparisons involve countries and populations at a similar level of economic development.

Demographic Characteristics Age

PC occurs almost exclusively in adults, ages 20 years or older (American Cancer Society 2016). It is rarely diagnosed in persons younger than age 40; disease incidence rises sharply by age 50 and continues to rise until about age 80 (http://​seer.cancer.gov/​). The median age at diagnosis of PC in the United States is 71 years (American Cancer Society 2016). The majority (~80%) of PC patients are diagnosed between ages 60 and 80 years (Ahlgren, 1996; Gold and Goldin, 1998). Persons diagnosed before age 50 may be more likely to have a positive family history or inherited genetic predisposition (Lowenfels and Maisonneuve, 2004; Raimondi et  al., 2007; Wolfgang et  al., 2013). The age distribution at diagnosis in a high-​resource country is illustrated by data from the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER-​18) Registries between 2008 and 2012. The percentage of incident cases under age 20 years was approximately 0%. This increased to 3% at ages 20–​44  years, 9% at ages 45–​54  years, 22% at ages 55–​64 years, 27% at ages 55–​64 years, 26% at ages 75–​ 84 years, and 13% among those older than age 85 (The Surveillance Epidemiology and End Results [SEER] Program Fact Sheets: Pancreas Cancer [http://​seer.cancer.gov/​canques/​incidence.html]).

Gender

The incidence rate of PC is about 36% higher in men than women (American Cancer Society, 2013). Men are usually diagnosed at earlier

ages than women (American Cancer Society 2013; Siegel et al., 2015), and have a higher death rate from PC (American Cancer Society 2016; Lowenfels and Maisonneuve, 2006; Torre et al., 2015). The age-​adjusted incidence rate in the United States in 2012 was also higher in men than in women (14.5 vs. 11.7 per 100,000), with correspondingly higher mortality rates (12.6 vs. 9.6 per 100,000) (http://​seer.cancer.gov/​). In contrast to the age-​standardized rates, the lifetime risk of developing PC is similar among men and women (1.5% for both), due to the greater longevity of women and the decreasing gap in exposure to tobacco use between men and women (American Cancer Society, 2013).

Race/​Ethnicity

Racial disparities in PC incidence and mortality rates exist both within and among countries (American Cancer Society, 2016; Arnold et al., 2009; Center et  al., 2011; Khawja et  al., 2015; Parkin et  al., 2002; Silverman et  al., 2003). In the United States, African Americans have the highest incidence and mortality rates from PC, while Asian Americans and Pacific Islanders have the lowest rates (Table 32–​1). The recorded incidence and mortality are actually higher among African Americans than among Native Africans (Curado et al., 2007; Kovi and Heshmat, 1972; Parkin et  al., 2002; Walker et  al., 1993), although this at least partly reflects differences in detection. Between 2000 and 2012, African Americans were diagnosed with PC at a rate of 15.7 per 100,000 versus 11.9 per 100,000 among whites (Table 32–1). Whereas Hispanic women and white women in the United States have similar incidence rates, white men have about a 15% higher incidence rate than Hispanic men (Table 32–​1). PC incidence rates among American Indian/​Alaska Natives are the second lowest of the racial and ethnic groups in the United States. While reasons for the increased incidence of PC among African Americans are not entirely clear, it has been suggested that a higher prevalence of risk factors (such as cigarette smoking, diabetes, family history of PC, and high body mass index [BMI]) among African Americans may explain the excess risk in this population (Khawja et al., 2015; Parkin et  al., 2002). However, studies that attempt to explain the racial disparity in PC between African Americans and whites have reached varying conclusions (Arnold et  al., 2009; Silverman et  al., 2003). In a population-​based, case-​control study of African Americans and whites recruited from metropolitan regions of Atlanta, Georgia; Detroit, Michigan; and 10 counties in New Jersey between 1986 and 1989, Silverman et al. (2003) examined racial differences in PC risk among 526 individuals with PC and 2153 cancer-​free controls. They observed that the determinants of higher incidence of PC among African Americans compared to whites varied by sex. Among men, the excess risk of PC in African Americans was explained largely by cigarette smoking, long-​standing history of diabetes, and a positive family history of PC. Among women, the higher risk in African Americans was explained mainly by heavy alcohol use and elevated BMI (Silverman et al., 2003). Arnold et al. (2009) performed a similar but much larger study that included 6243 cases and 1,054,146 controls using data from the American Cancer Society’s Cancer Prevention Study II (CPS II). These investigators used PC mortality as a surrogate for incidence, and observed that variation in smoking habits, diabetes status, BMI, family history of PC, and cholecystectomy did not explain the higher incidence of PC in African Americans compared to whites, even in sex-​specific analyses (Arnold et al., 2009). These studies suggest that other unexplored factors or attributes of exposure not captured by the available metrics may contribute to the observed racial differences in PC. Even though rates of PC incidence and mortality are higher in economically developed than in developing regions of the world, the male: female ratios are identical (1.4:1) for incidence and similar (1.5:1 and 1.4:1) for mortality in developed and developing regions, respectively (Torre et al., 2015).

Geographic Variation The recorded incidence and mortality rates of PC vary substantially across the globe due, in part, to differences in diagnosis and surveillance patterns. In general, higher rates are reported in countries with

 613



(Ferlay et  al., 2013). The highest age-​standardized incidence rates in 2012 were recorded in the Czech Republic, Slovakia, Armenia, and Hungary (with 9.7, 9.4, 9.3, and 9.3 per 100,000, respectively) (Ferlay et  al., 2013). Countries with a high incidence of PC also include Canada (6.4 per 100,000), Australia (6.6 per 100,000), and many European countries (Figure 32–​2). In South America, PC incidence rates are high in French Guyana, Uruguay, and Argentina (with 8.1, 7.6, and 6.7 per 100,000, respectively), moderately high in Peru, Brazil, and Paraguay (with 4.6 per 100,000 in each), intermediate in Colombia, Venezuela, and Ecuador (with 3.8, 3.6, and 3.1 per 100,000, respectively), and lowest in Bolivia (2.3 per 100,000). In Asia, incidence rates are high in Japan, Kazakhstan, and South Korea (8.5, 6.8, and 6.7 per 100,000, respectively), while countries such as India, Pakistan, Nepal, Bangladesh, Laos, and Vietnam have some of the world’s lowest rates of PC (ranging from ~0.5 to 1.2 per 100,000) (Ferlay et al., 2013). In Africa, rates are moderately high in South African and Libya (4.7 per 100,000 for both) (Figure 32–​2). The lowest recorded rates in Africa are observed mainly in Central African countries such as Cameroon, the Central African Republic, the two Republics of Congo, Gabon, and Angola, with rates that range from 0.7 to 1.1 per 100,000, although these estimates are based on very limited data (Ferlay et  al., 2013). As mentioned, the low incidence in these countries may also reflect the limited availability

Table 32–​1.  Incidence of Pancreatic Cancer in the United States per 100,000 Population, 2000–2012 African American Overall Men Women

613

Cancer of the Pancreas

15.7 17.1 14.4

White 11.9 13.6 10.4

Hispanic 11.1 11.8 10.4

American Indian/​Alaska Native 10.2 11.0 9.6

Asian or Pacific Islander 9.7 10.6 8.9

Source:  Modified from the Surveillance, Epidemiology, and End Results (SEER) Program SEER*Stat Database: Incidence - SEER 18 Registries Research Data. National Cancer Institute. Bethesda, MD, http://seer.cancer.gov/canques/incidence.html. All reported materials are publicly available.

better diagnostic imaging (high-​income countries) and lower rates in countries lacking high-​level diagnostic imaging (low-​income countries). The reported PC incidence and mortality rates are high in North America, Western Europe, and Central and Eastern Europe and low in South-​Central Asia and Central Africa (Figure 32–​1). Worldwide, the United States ranked 20th in PC incidence and 23rd in PC-​ related deaths (with 7.5 and 7.0 per 100,000, respectively) in 2012 Male North America Western Europe More developed regions Central and Eastern Europe Northern Europe Australia/New Zealand Southern Europe

Female 6.4 5.9 6.3 5.8 5.9 5.5

8.5

8 8.3 8 8.6 8.3 8.9 9 7.3 7.1 7.5 6.3 7.6 7.4

South America Eastern Asia Southern Africa World Western Asia

5 4.9

5 5.1 5.5 5.2 5.3 5.2 4.9 4.7 4.7 4.7

Central America 4.4 4.4

1.9 1.9 2.4 2.3 1.8 1.8 2 1.9 1.1 1.2 1.3 1.2 1.7 1.6 0.8 0.3 1 0.9 0.8 0.7

3.3 3.2 3.3 3.2 2.5 2.5 2.4 2.3

Less developed regions Northern Africa Southeast Asia Melanesia Western Africa Eastern Africa Polynesia

3.6

2 1.9 1.5 1.5 2.2 1.3 1.2 1.4 1.3

South-Central Asia Middle Africa 10

3.6 3.4 3.5 3.4 3.6 3.4 3.1 3 3.6 3.6 3.7 3.5

4.2 4.1 3.7 3.6

Caribbean

Micronesia

5.9 5.6 5.4 4.9 5.3 4.9 4.4 4.4

5

0 Incidence

5

10

Mortality

Figure 32–​1.  Age-​standardized incidence and mortality rates (per 100,000) for pancreatic cancer among men and women, by geographic region, 2012. Source: International Agency for Research on Cancer (IARC) (2015).

614

614

PART IV:  Cancers by Tissue of Origin

Pancreatic cancer 6.3+ 4.1–6.3 2.4–4.1 1.2–2.4 30 gram/​day or > 45 grams/​day or > 9 drinks/​day or > 3 drinks/​day). All studies showed a significantly increased risk for PC among heavy drinkers. Among three recent case-​ control studies, one in Japan estimated that those who drank > 80 grams/​day were diagnosed with PC on average 3.1 years earlier than those who drank < 50 grams/​day. The other two studies, one in China (Zheng et al., 2016) and one in Canada (Rahman et al., 2015), showed no significant increased risk when comparing any alcohol consumption to no consumption or heavy to low categories, and type consumed. In individual epidemiological studies, this association is difficult to detect since they typically are limited by sample size, potential recall bias, or possible selection bias. Additionally, power issues arise when alcohol is split based on the type of alcohol consumed (i.e., beer, wine, or liquor).

Medical Conditions Diabetes

Type 2 diabetes mellitus is well established as both a cause and consequence of PC. Epidemiologic studies have consistently shown a higher risk of PC among individuals with preexisting diabetes, with relative risk estimates ranging from 1.5 to 2.0 (Li, 2012). In studies where medical records of PC patients were reviewed for diagnosis of diabetes or where PC patients were screened for diabetes using fasting glucose values or glucose tolerance tests, it has been observed that ~80% of PC patients have a diagnosis of diabetes or glucose intolerance that meets the American Diabetes Association’s diagnostic criteria for diabetes (Pannala et al., 2009). However, this includes both preexisting diabetes and situations where the diagnosis of PC preceded the diagnosis of diabetes (referred to as “PC-​associated diabetes”). This reverse causation is based on the observation that diabetes generally appears within 3 years before the diagnosis of PC and is thus an early manifestation of PC (Chari et al., 2008; Pannala et al., 2009). Studies have reported a wide range of frequencies of new-​onset diabetes attributable to PC, from as high as 52% to as low as 16% (Chari et al., 2008; Gullo et al., 1994). In patients with PC-​associated diabetes, it is believed that the tumor produces “diabetogenic” substances that impair glucose metabolism (Li, 2012; Pannala et al., 2009). This is supported by observations that PC-​associated diabetes generally resolves after pancreatic tumor resection, whereas removal of a pancreatic tumor does not resolve preexisting diabetes (Chari, 2007). Thus, while long-​standing diabetes is a risk factor for PC, new-​onset diabetes in later adulthood could be an early sign of undiagnosed PC. At least five overlapping meta-​analyses (Batabyal et al., 2014; Ben et  al., 2011; Everhart and Wright, 1995; Huxley et  al., 2005; Song et al., 2015) and three pooled analyses (Bosetti et al., 2014; Elena et al., 2013; Li et al., 2011) attempted to clarify the bidirectional relationship between diabetes and PC by examining the duration of diabetes relative to PC diagnosis (Batabyal et al., 2014; Ben et al., 2011; Everhart and Wright, 1995; Huxley et al., 2005; Song et al., 2015). Although all of the studies reported ≥ 50% increase in risk of PC among individuals with preexisting diabetes, there were suggestions of decreasing PC risk as the duration of diabetes increases. A meta-​analysis by Huxley and colleagues (2005) that included 19 prospective studies and 17 case-​control studies reported smaller magnitudes of association as the duration of diabetes increased, with RRs for 1–​4 years, 5–​9 years, and ≥ 10 years duration of diabetes of 2.05 (95% CI: 1.87, 2.25), 1.54 (95% CI:  1.31, 1.81), and 1.51 (95% CI:  1.16, 1.96), respectively. The majority of diabetic cases in the 1–​4  year group are most certainly PC-​associated diabetes. In another meta-​analysis involving 30

 619



Cancer of the Pancreas

cohort studies, Ben et al. observed a significant decline in the association between preexisting diabetes and PC as the duration of diabetes increased (Ben et al., 2011). The RRs for the subgroups of 1–​4 years, 5–​ 9  years, and ≥ 10  years duration of diabetes were 1.95 (95% CI: 1.65, 2.31), 1.49 (95% CI: 1.05, 2.12), and 1.47 (95% CI: 0.94, 2.31), respectively (Ben et al., 2011). In a pooled analysis of 15 case-​ control studies involving 8305 PC cases and 13,987 controls, Bosetti and colleagues (2014) also observed decreasing PC risk with increasing duration of diabetes, although the risk persisted in those with more than 20 years duration of diabetes. The RRs were 2.92 (95% CI: 2.44, 3.50), 1.84 (1.54, 2.20), 1.69 (1.36, 2.09), 1.54 (1.17, 2.03) and 1.30 (1.03, 1.63) for diabetes duration of 2–​4 years, 5–​9 years, 10–​14 years, 15–​19 years, and ≥ 20 years, respectively (Bosetti et al., 2014). Elena et al. examined the association between diabetes duration and PC by pooling primary data from 12 cohort studies that included 1621 incident PC cases and 1719 controls. They observed a strong association among individuals with diabetes duration of 2–​8  years (OR  =  1.79; 95 % CI: 1.25, 2.55) and no association among those who had been diagnosed with diabetes for ≥ 9 years (OR = 1.02; 95 % CI: 0.68, 1.52) (Elena et al., 2013). It is possible that the decreasing PC risk associated with a longer duration of diabetes could be related to lifestyle modifications (e.g., healthy diet, physical activity) following the diabetes diagnosis or the use of some diabetic medications, such as metformin, which has been associated with reduced risk of PC in some studies (Bosetti et al., 2014; Li et al., 2009). Very few studies have investigated the association between type 1 diabetes and PC. The large majority of these have been null, but individual studies have included only small numbers of both type 1 diabetes and PC cases. One cohort study of 109,581 participants identified 11 individuals with type 1 diabetes, among whom the risk of PC was greater than that of non-​diabetics; it reported a standardized PC incidence ratio of 3.50 (95% CI: 1.80, 6.30) (Wideroff et al., 1997). In a meta-​analysis of three cohort studies and six case-​control studies, individuals with type 1 diabetes or onset before age 40 had a 2-​fold increase in PC risk (RR = 2.0; 95% CI: 1.37, 3.01). This observation was based on 39 PC cases with diabetes (Stevens et al., 2007). It has been suggested that the use of exogenous insulin in type 1 diabetes may partly account for the increased PC risk (Li, 2012). Given the small number of diabetic cases included in the existing studies, further research with larger numbers of both conditions is needed to verify the findings. Reasons for the association between preexisting type 2 diabetes and PC include the fact that they share many risk factors, such as obesity, physical inactivity, cigarette smoking, and excessive alcohol consumption (Herman, 2007; Lowenfels and Maisonneuve, 2006). Proposed mechanisms for the link between diabetes and PC include insulin resistance, chronic hyperinsulinemia, increased circulating levels of insulin-​like growth factors (IGFs), and chronic inflammation (Bao et al., 2011; Li, 2012). In the setting of insulin resistance, excessive amounts of insulin are secreted by the pancreatic β-cells as a compensatory mechanism to regulate blood glucose levels. This exposes the exocrine pancreas to high concentrations of endogenous insulin for many years (Bao et al., 2011). Experimental evidence shows that chronic hyperinsulinemia can promote cell proliferation and decrease cellular apoptosis, which are conditions that favor carcinogenesis (Li, 2012; Magruder et al., 2011). Other studies have investigated the hypothesis that impaired glycemic control, insulin resistance, and its compensatory hyperinsulinemia may promote pancreatic tumorigenesis. In 2005, a study by Stolzenberg-​Solomon et  al. examined the association between pre-​ diagnostic fasting glucose level, blood insulin concentration, the extent of insulin resistance, and risk of PC in a case-​cohort study of 169 incident PC cases and 400 non-​cancer controls (Stolzenberg-​ Solomon et al., 2005). The study participants were randomly selected from the Alpha-​Tocopherol Beta-​Carotene Cancer Prevention (ATBC) trial of 29,133 Finnish male smokers of age 50–​65 years, who were followed for up to 17  years. The study found that increased blood glucose, insulin levels, and insulin resistance were positively associated with PC risk. The associations were particularly strong for PC cases that were diagnosed after 10 years of follow-​up (HRs and 95%

619

CIs for highest vs. lowest quartiles: glucose 2.16 [1.05, 4.42], insulin concentration 2.90 [1.22, 6.92], insulin resistance 2.71 [1.19, 6.18]) (Stolzenberg-​Solomon et al., 2005). Maisonneuve et al. also reported on the use of insulin injection versus oral hypoglycemic medication for control of diabetes and risk of PC in a population-​based study of 823 newly diagnosed PC cases and 1679 controls (Maisonneuve et al., 2010). The study showed over 2-​fold increase in PC risk among diabetes patients compared to non-​diabetics (OR = 2.16; 95% CI: 1.60, 2.91). The management of diabetes with insulin injection was associated with a greater risk (OR = 3.54; 95% CI: 1.64, 7.61) than the use of oral hypoglycemic medication (OR  =  1.78; 95% CI:  1.16, 2.75) (Maisonneuve et al., 2010). These epidemiological reports are corroborated by an earlier experimental study that demonstrated that physiologic levels of insulin stimulate pancreatic tumor cell proliferation and glucose utilization through the activation of mitogen-​activated protein (MAP) kinase and phosphatidylinositol 3-​kinase (PI3-​kinase), as well as increased expression of glucose transporter 1 (GLUT1) (Ding et al., 2000). Insulin’s effect on pancreatic tumor development and progression has been shown in animal models to occur also through promoting insulin-​like growth factor 1 (IGF-​1) activity (Li, 2012). High levels of insulin can stimulate synthesis of IGF-​1 and suppress IGF-​1 binding activity in the liver at the same time, which increases circulating levels of IGF-​1 (Bao et al., 2011; Li, 2012). Increased levels of free circulating IGF-​1 have been shown to promote cell proliferation, inhibit apoptosis, and enhance angiogenesis in the tumor microenvironment (Bao et al., 2011). Moreover, high IGF-​1 levels have been associated directly with increased risk of PC (McCarty, 2001). Thus, preexisting diabetes may contribute to PC risk through diabetes-​associated insulin resistance, its impact on insulin secretion from the pancreas or administration of exogenous insulin, and the subsequent effect of insulin on IGF-​1 activity. Diabetes is also associated with chronic inflammation (Garcia et al., 2010), and chronic inflammation is known to promote pancreas tumorigenesis, possibly by high levels of inflammatory cytokines in the setting of diabetes (Li, 2012).

Chronic Pancreatitis

The exocrine portion of the pancreas is susceptible to two main diseases: pancreatitis and PC (Raimondi et al., 2010). Pancreatitis is generally classified as acute or chronic, based on presenting symptoms, past medical history, and clinical course. While acute pancreatitis, which is typically caused by gallstones or heavy alcohol consumption, is characterized by inflammation that resolves after removal of the offending agent, chronic pancreatitis is a continuing inflammatory condition marked by irreversible morphological changes to the pancreas that include calcification, fibrosis, and pancreatic ductal inflammation (Forsmark, 2007; Raimondi et  al., 2010). Although chronic pancreatitis has been linked to habitual alcohol abuse (~70% of cases) and cigarette smoking (< 10% of cases), in a significant proportion of patients (~20%), the etiology is unknown (Forsmark, 2007). There is strong epidemiological and clinical evidence linking long-​standing history of chronic pancreatitis with increased risk of PC (Duell et al., 2012; Howes and Neoptolemos, 2002; Lowenfels et al., 1993; Raimondi et al., 2010). However, because pancreatic tumors can obstruct the flow of enzymes from the pancreatic duct into the bloodstream, pancreatitis can result from tumor-​associated ductal obstruction (Forsmark, 2007). Therefore, as with diabetes and PC, there is a two-​way relationship between chronic pancreatitis and PC. In studies that excluded from analyses cases of PC diagnosed temporally close to diagnosis of pancreatitis in order to reduce confounding by reverse causation, results showed a strong association between antecedent chronic pancreatitis and PC (Duell et  al., 2012; Karlson et  al., 1997; Lowenfels et  al., 1993; Olson, 2012). For example, in 1993, Lowenfels et al. conducted a cohort study of 2015 chronic pancreatitis patients who were followed at major hospitals in Europe and in the United States for an average of 7.4 years (Lowenfels et al., 1993). After excluding PC cases that occurred in the first two years of follow-​up, they observed a 16 times higher incidence of PC in the chronic pancreatitis patients than expected in the general population (standardized incidence ratio [SIR] = 16.5;

620

620

PART IV:  Cancers by Tissue of Origin

95% CI: 11.1, 23.7). The excess risk of PC remained when analyses were performed separately for patients who had lived with chronic pancreatitis for 10  years or more, suggesting an etiological link between preexisting chronic pancreatitis and PC (Lowenfels et al., 1993). In 1999, Talamini et  al. also conducted a cohort study of 715 chronic pancreatitis patients followed for a median of 10 years (Talamini et al., 1999). After eliminating PCs that occurred within 4 years of diagnosis of pancreatitis, they observed a 13 times greater risk of PC in the chronic pancreatitis patients than expected in the general population (SIR = 13.3; 95% CI: 6.4, 24.5). However, after excluding the early-​onset PCs, no cases of PC had occurred among non-​smokers. The investigators, therefore, concluded that cigarette smoking and perhaps other factors associated with chronic inflammation (e.g., alcohol abuse) may play a key role in the association between chronic pancreatitis and PC (Talamini et  al., 1999). In a Swedish hospital-​based cohort study of 4546 chronic pancreatitis patients conducted in 1997, Karlson et al. reported a nearly 8-​fold excess risk of PC in chronic pancreatitis patients than expected (SIR  =  7.6; 95% CI:  6.0, 9.7) (Karlson et  al., 1997). They also observed a generally declining, but persistently higher magnitude of PC risk for patients living with chronic pancreatitis for 1–​4 years, 5–​9 years, and 10–​24 years, with SIRs of 22.2 (95% CI: 16.2, 29.6), 4.6 (95% CI: 2.6, 7.5), and 2.2 (95% CI: 0.9, 4.4), respectively. The study further showed that the risk of PC remained significantly high among patients living with alcohol-​related chronic pancreatitis for 10 years or more (SIR = 3.8; 95% CI: 1.5, 7.9), suggesting that alcohol abuse may contribute to reported associations between chronic pancreatitis and PC risk (Karlson et al., 1997). In more recent studies where both alcohol intake and cigarette smoking were accounted for in the analyses, having a history of chronic pancreatitis remained significantly associated with an increase in risk of PC (Duell et al., 2012; Olson, 2012; Raimondi et al., 2010). Duell et al. reported a higher risk of PC among individuals with more than 2  years duration of chronic pancreatitis (OR  =  2.71; 95% CI:  1.96, 3.74), after adjusting for known risk factors of PC, including alcohol intake and cigarette smoking in a pooled analysis of data from 10 case-​ control studies consisting of 5048 cases and 10,947 controls (Duell et  al., 2012). They also reported a stronger association for chronic pancreatitis cases diagnosed less than 2 years before diagnosis of PC (OR = 13.56; 95% CI: 8.72, 21.90), which likely reflects reverse causation or initial misdiagnosis of PC as pancreatitis (Duell et al., 2012). In addition, they observed that the association remained statistically significant for individuals with 10 years or more duration of chronic pancreatitis. Raimondi et al. also reported a 5.8-​fold higher risk of PC among individuals with chronic pancreatitis from a meta-​analysis of 22 studies after excluding PC cases diagnosed within 2 years of the diagnosis of pancreatitis (RR  =  5.8; 95% CI:  2.1, 15.9) (Raimondi et al., 2010). This pair of reports is consistent with findings from an earlier review of 25 epidemiological studies that included eight studies that examined the time interval between diagnosis of pancreatitis and PC (Olson, 2012). The review showed that although the association generally declined with increasing duration of pancreatitis, all eight studies reported elevated risk of PC among individuals with pancreatitis, even when pancreatitis was diagnosed many years before the diagnosis of PC (RR ranging from 1.8 to 5.1) (Olson, 2012). Moreover, PC has been associated with tropical chronic pancreatitis (Chari et al., 1994) and hereditary pancreatitis (a germline autosomal-​dominant disorder) (Lowenfels et al., 2000). Thus, it appears that patients with any form of chronic pancreatitis have a high risk of PC. However, chronic pancreatitis is a relatively rare disease (annual incidence is ~5–​10 per 100,000), and it is estimated to account for only 1% of the overall burden of PC (Duell et al., 2012; Raimondi et al., 2010).

Drugs Non-​Steroidal Anti-​Inflammatory Drugs (NSAIDs)

Experimental studies suggest that NSAIDs can inhibit pancreatic tumor growth (Kokawa et al., 2001; Molina et al., 1999; Perugini et al.,

2000); however, the majority of epidemiologic studies do not show an association between NSAID use and PC risk (Anderson et al., 2002; Capurso et  al., 2007; Coogan et  al., 2000; Larsson et  al., 2006; Tan et  al., 2011). Several studies suggest that regular use of aspirin, the most frequently studied NSAID, is associated with reduced risk of PC, but the overall evidence is inconsistent. In a prospective cohort study of 28,283 postmenopausal women in the Iowa Women’s Health Study, 80 of whom had PC, Anderson et al. observed that current use of aspirin was associated with lower risk of PC (OR = 0.57; 95% CI: 0.36, 0.90, users vs. non-​users) (Anderson et  al., 2002). They found also that higher frequency of aspirin use was associated with lower risk of PC (OR = 0.40; 95% CI: 0.20, 0.82; ≥ 6 times per week vs. never users, P trend = 0.005) (Anderson et al., 2002). In a clinic-​based, case-​ control study of 904 rapidly ascertained incident PC cases and 1224 controls, Tan et al. observed a lower risk of PC among aspirin users (OR = 0.74; 95% CI: 0.60,0.91; ≥ 1 days/​month vs. < 1 day/​month) and among those with greater frequency of aspirin use (OR  =  0.63; 95% CI: 0.47, 0.85; ≥ 6 day/​week vs. < 1 day/​month, P trend = 0.007) (Tan et  al., 2011). A  similar finding was reported by Streicher and colleagues (2014). In contrast, Schernhanner et  al. observed a suggestive increased PC risk among aspirin users in the Nurses’ Health Study among 88,378 women, of whom 161 had PC (OR = 1.20; 95% CI:  0.87, 1.65; current users vs. never users) (Schernhammer et  al., 2004). The study further showed a pattern of increasing risk of PC with increasing duration of aspirin use (OR  =  1.58; 95% CI:  1.03, 2.43; ≥ 20 years of regular use vs. non-​use, P trend = 0.01) (Schernhammer et al., 2004). Results from meta-​analyses do not support the suggestion that aspirin or NSAID use reduce PC risk (Bosetti et al., 2012b; Capurso et al., 2007; Larsson et al., 2006).

Metformin

Epidemiologic reports indicate that the use of metformin, a commonly prescribed oral antidiabetic medication, is associated with reduced risk of PC (Li, 2012). A hospital-​based case-​control study of 973 cases and 863 controls observed a 62% lower risk of PC among diabetic patients who had taken metformin when compared to non-​diabetic control patients and after controlling for duration of diabetes, use of insulin, and other risk factors of PC (OR = 0.38; 95% CI: 0.21, 0.67) (Li et al., 2009). A further slight reduction in risk was observed among diabetics with longer duration of metformin use (OR = 0.30; 95% CI: 0.13, 0.69; ≥ 5 years vs. never use) (Li et al., 2009). This finding is corroborated by a retrospective cohort study of 62,809 diabetics, of whom 89 had PC (HR = 0.20; 95% CI: 0.011, 0.36; metformin users vs. non-​users) (Currie et al., 2009), and by a recent meta-​analysis involving 10 cohort studies and three case-​control studies (HR = 0.63; 95% CI: 0.46, 0.86; users vs. non-​users) (Wang et al., 2014). Potential mechanisms for the effect of metformin on PC include lowering circulating levels of insulin, improving peripheral insulin resistance, and blocking the mitogenic effects of IGF-​1 (Li et al., 2009; Li, 2012).

Statins

In vitro (Gbelcova et  al., 2008; Kusama et  al., 2002; Wong et  al., 2001) and in vivo (Gbelcova et al., 2008; Kusama et al., 2002; Sumi et  al., 1992; Wong et  al., 2001)  studies suggest that statins, a class of cholesterol-​ lowering drugs, have anti-​ pancreatic tumor properties. Postulated mechanisms of these 3-​ hydroxy-​ 3-​ methylglutaryl-​ coenzyme A  (HMG-​ CoA) reductase inhibiting agents include pro-​ apoptotic activity and blockade of the mevalonate metabolic pathway (Kusama et  al., 2002; Wong et  al., 2001). However, in a 2008 meta-​ analysis that included three randomized clinical trials, four cohort studies, and five case-​control studies, no association was observed between statin use and PC risk (Bonovas et al., 2008). It is, however, possible that the antitumor effectiveness of statins depends on their combination with other anticancer agents (Bocci et al., 2005; Yao et al., 2006).

Insulin/​Insulin Secretagogues

While there is mounting evidence linking the use of insulin or insulin secretagogues (drugs that increase insulin secretion by pancreatic β-​cells, e.g., sulfonylureas) with increased risk of PC (Bodmer et al.,

 621



Cancer of the Pancreas

2012; Currie et al., 2009; Li et al., 2009; Stolzenberg-​Solomon et al., 2005), it remains unclear whether long-​term insulin use is associated with an increase in PC risk. In a pooled analysis of three case-​control studies involving 2192 cases and 5113 controls, insulin use for 3 years or less was associated with higher risk of PC, whereas use of insulin for more than 10 years was associated with lower risk (Li et al., 2011). The study reported ORs for ≤ 3  years, >3–​≤ 10  years, and > 10 years of 2.4 (95% CI: 1.6, 3.7), 1.2 (95% CI: 0.7, 1.9), and 0.50 (95% CI: 0.30, 0.90), respectively, for insulin-​using diabetes patients compared to non-​insulin-​using diabetics (Li et al., 2011). Other studies that reported an overall increase in PC risk among insulin users also reported null associations between PC and insulin use for more than 5 years (Hassan et al., 2007; Wang et al., 2006) or more than 10 years (Silverman et al., 1999). Thus, although considerable evidence links insulin use to increased PC risk, the association may be confounded by duration of insulin use and, by extension, duration of diabetes because of its complex relationship with PC, especially in studies where analyses were not restricted to diabetic patients or where duration of diabetes was not considered in the analyses.

Occupation and Physical Environment PC incidence has been associated with various occupational exposures, including chlorinated hydrocarbons (CHC), polycyclic aromatic hydrocarbons (PAHs, mainly derived from diesel exhaust and aluminum production), organochlorine pesticides, and heavy metals such as nickel, chromium, and cadmium (Andreotti and Silverman, 2012; Ojajarvi et al., 2000). Results from a meta-​analysis of 92 occupational studies published between 1969 and 1998 showed excess risk of PC among individuals exposed to chlorinated hydrocarbons and nickel, with meta-​risk-​ratios of 1.4 (95% CI: 1.0, 1.8) and 1.9 (95% CI: 1.2, 3.2), respectively (Ojajarvi et  al., 2000). In 2012, a review of occupational studies published since this meta-​analysis concluded that work-​related exposure to CHC, pesticides, PAHs, nitrosamines, and radiation have been consistently associated with increased PC risk, with stronger evidence for CHC and PAHs (Andreotti and Silverman, 2012). Non-​occupational studies have also shown that increased exposure to environmental chemicals and heavy metals is associated with increased risk. A population-​based case-​control study in southeastern Michigan reported excess risk of PC among individuals with self-​ reported exposure to organochlorine pesticides, including ethylan, chloropropylate, and dichlorodiphenyltrichloroethane (DDT) (Fryzek et al., 1997). A hospital-​based case-​control study conducted in the East Nile Delta region of Egypt observed an elevated risk of PC among patients who self-​reported exposure to pesticides (OR  =  2.6; 95% CI: 0.97, 7.2) (Lo et al., 2007). In a clinic-​based, case-​control study in Rochester, Minnesota, excess risk of PC was observed among individuals with self-​reported exposure to pesticides, CHC, benzene, and asbestos (Antwi et al., 2015). However, these hospital or clinic-​based case-​control studies are limited by their retrospective nature, particularly the reliance on self-​reported exposure after cancer diagnosis. It has been suggested that environmental chemicals and heavy metals may reach the pancreas via the bloodstream or through bile reflux into the pancreatic duct and, in the process, these substances exert their carcinogenic effects, such as activation of oncogenes, formation of DNA adducts, and impairment of DNA damage repair mechanisms (Wogan et al., 2004). It remains to be demonstrated whether reducing exposures to these substances in occupational and non-​occupational settings would have a significant effect on the incidence of PC.

Genetic Susceptibility Family History and Familial PC

Family history of PC is a consistent and important risk factor of PC. Early familial clustering studies in which families with multiple siblings or over multiple generations were affected (Dat and Sontag, 1982; Ehrenthal et  al., 1987; Friedman and Fialkow, 1976; Ghadirian et al., 1987; MacDermott and Kramer, 1973) were followed

621

by familial aggregation studies and analysis of families using formal study designs, finding increased risk. A  comprehensive summary of these studies and estimated risks are listed in Table 32–​2. Seven case-​ control studies, two cohort studies, one population-​based genealogic analysis, and one case series that estimated incidence of PC in relatives have found that first-​degree relatives have at least a 2-​fold increased risk of developing PC. These findings are remarkably consistent, given that case ascertainment and data collection spanned 30 or more years, multiple countries and cultures, and different methods for estimating risk. A systematic review and meta-​analysis by Permuth-​Wey and Egan (2009) of a cohort study and seven case-​control studies totaling 6568 PC cases calculated an overall relative risk of 1.80 (95% CI: 1.48, 2.12). They also found that 1.3% of PC in the population is attributable to family history. The risk was consistent for both males and females, and did not differ by early or late age at diagnosis. With respect to risk for second-​degree relatives (aunts, uncles, grandparents, grandchildren), both Hassan et al. (2007) and Shirts et al. (2010) reported risks comparable to those of first-​degree relatives (relatives risks of 2.9 (95% CI: 1.3, 6.3) and 1.59 (95% CI: 1.31, 2.91), respectively. In addition, a large multicenter cohort study examined risk by number of affected individuals and showed even high risk associated with having two or more first-​degree relatives with PC with an odds ratio of 4.26 (95% CI: 0.48, 37) (Jacobs et al., 2010). A population-​based twin study of cancer in Sweden by Lichtenstein et al. (2000) estimated PC heritability to be 36%, similar to colorectal cancer (35%), higher than breast cancer (27%), and slightly lower than prostate cancer (42%). In the first published segregation analysis of PC, Klein et al. (2002) analyzed family histories in 287 PC patients seen during 1994–​1999 at the Johns Hopkins Hospital in Baltimore, Maryland. The analysis rejected non-​genetic transmission models. The data best fit a major gene model that was predicted to follow an autosomal dominant pattern of a rare allele; 0.7% of the population would carry a high risk of developing PC due to this putative gene. A smaller study of 70 families drew a similar conclusion (Banke et al., 2000). Familial PC (FPC) was defined to advance consistency in research, and was defined as kindreds containing at least a pair of individuals who were affected with pancreatic adenocarcinoma and who were first-​degree relatives (Hruban et  al., 1998). Numerous studies using this definition have focused on genetic epidemiology (Petersen et al., 2006) and susceptibility gene discovery (Amundadottir, 2016; Klein, 2012; Petersen, 2015). These studies have shown that FPC is genetically heterogeneous, and that increased risk of PC is seen in hereditary cancer syndromes. Using increasingly sophisticated genomic analysis technologies, susceptibility genes include most commonly BRCA1/​ 2, CDKN2A, PALB2, ATM, and mutations associated with hereditary colorectal cancer syndromes. These are summarized in Table 32–​3. It remains unclear whether or how lifestyle factors influence PC risk among FPC family members. Cigarette smoking is a well-​ established risk factor for sporadic PC. A  pooled case-​control study that included 6507 cases with PC found that the prevalence of never smokers was 36.5% and 60.5% for ever smokers, with 2.6% missing or other (Bosetti et al., 2012a). In the PACGENE study, 37% of affected members of FPC kindreds were never smokers, 47.1% were ever smokers, but smoking status was unknown in 14.9% (Petersen et al., 2006). In comparison, a regional hospital-​based Australian study found that 60.3% of 68 FPC patients were never smokers compared to 45.6% never smokers in 698 non-​FPC patients (p = .0315). However, there were no differences in alcohol intake in this study (Humphris et al., 2014). It has been suggested that smoking may potentiate PC risk among predisposition gene mutation carriers (McWilliams et al., 2011), but to date, this has not been possible to evaluate more comprehensively. Data on other PC risk factors, such as diet or obesity, are too sparse among FPC to draw meaningful inferences. Similarly, disease associations, such as diabetes or pancreatic cyst disease, have not been systematically studied in the context of FPC.

Genetic Predisposition Among Sporadic PC Patients Two distinct approaches have been taken to uncover genetic predisposition in patients unselected for family history. These include

62

Table 32–​2.  Family History and Estimated Risks of PC in Case-​Control and Cohort Studies Risk of PC in Family Members Location, Years of Study

Cases, N

Controls, N

Risk

Risk

Reference

Louisiana, 1979–​1983 Canada, 1984–​1988 Italy, 1983–​1992 United States, 1986–​1989 Japan, Cohort, 1988–​1999 United States, 1996–​1999 Texas, 2000–​2006 Italy, 1991–​2008 United States, 2005–​2009 China, 2011–​2013 International, PanScan Cohort Consortium (1 case-​control and 10 cohort studies), 1985–​2001 Utah, Genealogy database, 1966–​2010 Minnesota, Case series, 2000–​2004

362 174 363 484 200 247 888 326 654 323 1183

1408 136 1234 2099 2200 420 888 652 697 323 1205

5.25 5.0 2.8 3.2 2.09 2.49 3.3 1.23 2.79 3.67 1.76

2.1–​13.2 1.2–​24.5 1.3–​6.3 1.8–​5.6 1.01–​4.33 1.3–​4.7 1.8–​6.1 0.53–​2.85 1.44–​4.08 1.02–​13.14 1.19–​2.61

(Falk et al., 1988) (Ghadirian et al., 2002) (Fernandez et al., 1994) (Silverman et al., 1999) (Inoue et al., 2003) (Schenk et al., 2001) (Hassan et al., 2007) (Rosato et al., 2015) (Austin et al., 2013) (Zheng et al., 2016) (Jacobs et al., 2010)

1411 426

—​ —

RR=1.84 SIR=1.88

1.47–​2.29 1.27–​2.68

(Shirts et al., 2010) (McWilliams et al., 2005)

Case-​control study designs reported unless otherwise specified. Abbreviations: CI = confidence interval; RR = relative risk; SIR = standardized incidence ratio. Source: Updated from (Petersen, 2015).

Table 32–​3.  Associated Genes and Syndromes and Estimates of Risk of Developing Pancreatic Adenocarcinoma FPC Patients with Deleterious Mutations Predisposition Syndrome

Associated Malignancies

Gene

Chromosome

ATM

11q23

Familial breast cancer

Breast

Increased risk: not well defined

BRCA1

17q21.31

Hereditary breast and ovarian cancer

BRCA2

13q13.1

CDKN2A

9p21.3

Familial atypical mole and melanoma

Breast (particularly premenopausal), ovary, male breast, prostate Breast (particularly premenopausal), ovary, male breast, prostate, melanoma Melanoma

Mismatch repair: MLH1 MSH2 MSH6 PMS2

3p22.2 2p21 2p16.3 7p22.1

Hereditary non-​polyposis colorectal cancer (Lynch syndrome)

OR = 2.26 (95% CI: 1.26, 4.06); SIR = 2.55 (95% CI: 1.03, 5.31) OR = 3.5 (95% 19/​516 CI: 1.87, 6.58); SIR = 2.13 (95% CI: 0.36, 7.03) RR = 52 (95% 14/​519 CI: 13, 132); RR = 80.8 (95% CI: 44.7, 146) No effect up to 1/​186 (PMS2) SIR = 8.6 (95% CI: 4.7, 15.7)

PALB2

16p12.2

Familial breast cancer

PRSS1 SPINK1

7q34 5q32

Hereditary pancreatitis

19p13.3

Peutz Jeghers syndrome

STK11 (LKB1)

Colorectum, endometrial, ovary, stomach, small bowel, urinary tract (ureter, renal pelvis) biliary, glioblastoma, skin (sebaceous) Fanconi anemia, breast, esophagus, prostate, stomach —

Colorectum, small bowel, stomach, breast, gynecologic

Risk of PC

Proportion 2/​168 1/​39 6/​186 6/​516

%

References

1.2 2.6 3.2 1.2

(Chaffee et al., 2016; Grant et al., 2015; Roberts et al., 2012) (Iqbal et al., 2012; Zhen et al., 2015)

3.7

(Iqbal et al., 2012; Zhen et al., 2015)

2.7

(Goldstein et al., 2004; Potjer et al., 2015; Zhen et al., 2015)

0.5

(Chaffee et al., 2016; Kastrinos et al., 2009)

3.1 0.6

(Jones et al., 2009; Zhen et al., 2015)

Increased risk: not well defined

3/​96 3/​521

SIR = 53 (95% CI: 23, 105); SIR = 87 (95% CI: 42, 113) SIR = 132; RR = 76; RR = 139.7

—​

—​

(Lowenfels et al., 1997; Rebours et al., 2008)

—​

—​

(Giardiello et al., 2000; Korsse et al., 2013; Resta et al., 2013)

The probabilities of detecting a deleterious mutation in the predisposition genes shown were based upon studies that sequenced the entire gene in series of familial PC (FPC) patients. Abbreviations: FPC = familial PC; OR = odds ratio; CI = confidence interval; RR = relative risk; SIR = standardized incidence ratio. Source: Updated from (Petersen, 2015).

 623



Cancer of the Pancreas

genome-​ wide association studies (GWAS) and next-​ generation sequencing of patients.

Genome-​Wide Association Studies (GWAS).  In the search

for common variants with low penetrance, the GWAS approach has been the most-​used strategy. A series of these large case-​control studies performed by the PanScan Consortium and the PC Case-​Control Consortium (PanC4) have been completed, with ongoing meta-​ analysis. By applying agnostic genetic scans of single nucleotide polymorphisms (SNPs), highly stringent thresholds of significance (i.e., P < 1 × 10–​7), and independent validation, these consortia have identified a roster of at least 13 novel SNPs in genes, as well as confirming known associations. For example, the initial published GWAS (PanScan I  in 2009), using 2000 case-​control pairs, identified the most consistent SNP association on chromosome 9q34.2 (in the ABO blood group gene) (Amundadottir et al., 2009). In the following year, PanScan II reported from an analysis of 3850 pairs associations of SNPs on chromosome 1q32.1 (in NR5A2), 5p15.33 (in the CLPTM1L-​ TERT gene region) and 13q22.1 (in a non-​genic region between KLF5 and KLF12) (Petersen et al., 2010). PanScan III included 7683 cases and 14,397 controls and found more risk loci: on chromosome 5p15.33 (an association that independently identified the CLPTM1L-​TERT gene region), on 7q23.2 (LINC-​PINT), 16q23.1 (BCAR1), 13q12.2 (PDX1), 22q12.1 (ZNRF3), and a non-​genic SNP on 8q24.1 (Wolpin et  al., 2014). In 2015, the PanC4 studied 4164 cases and 3792 controls, combined with PanScan data on 9925 cases and 11,569 controls, with replication and reported four new loci on chromosomes 17q24.3 (LINC00673), 2p14 (ETAA1), 7p14.1 (SUGCT), and 3q28 (TP63) (Childs et al., 2015).

Genetic Analysis and Sequencing of  PC Patient Series  Among the first large unselected series reported was

sequencing for 39 known mutations in the cystic fibrosis transmembrane regulator (CFTR) gene of 949 PC patients by McWilliams et al. (2010), who reasoned from established data that the CFTR gene, known to be associated with chronic pancreatitis, may increase risk of PC. Compared with data on 13,340 controls from a clinical laboratory database, they found that 5.3% carried a common CFTR mutation versus 3.8% of controls, giving an OR = 1.40 (95% CI: 1.04, 1.89). Among patients who were younger when their disease was diagnosed (< 60  years), the carrier frequency was higher than in controls, and the resulting OR increased to 1.82 (95% CI: 1.14, 2.94). McWilliams et al. (2011) also estimated the prevalence of deleterious mutations in CDKN2A in a series of 1537 sporadic, unselected patients, and found a prevalence of 0.6%. Whole exome sequencing of patients with and without family history of PC led to the identification of PALB2 and ATM (Jones et al., 2009; Roberts et al., 2012), which had not been previously known to increase risk of PC. Moreover, the agnostic gene discovery efforts (such as whole exome or whole genome sequencing) are identifying novel genes (Roberts et  al., 2016), but the prevalence, gene by gene, is quite low, and in many cases is unique to specific individuals (Hu C et al., 2016). In summary, research to date has clearly revealed the extensive genetic heterogeneity of the FPC phenotype. In addition to expanding the catalog of genes, they provide an opportunity to study the potential effect of genetic mutations on age at diagnosis and the risk of developing other cancers. This promises to be an area for researchers to fill in the gaps of our knowledge about the epidemiology, functional, and clinical implications.

Interactive Effects Between Lifestyle and Genetic Variants on PC Risk Smoking and Genes

Genes from several pathways have been suggested to interact with smoking to influence PC risk. Hypothesized interactions include those with genes that affect carcinogen metabolism, DNA repair, nicotine dependence, oxidative stress, hormone metabolism, inflammation,

623

insulin secretion, and chromatin-​ remodeling (Ayaz et  al., 2008; Bartsch et al., 1998; Duell et al., 2002; Jiao et al., 2007; Liu et al., 2000; Vrana et  al., 2009). Laboratory evidence suggests that the major cigarette smoke carcinogen NNK can activate Cox2, EGFR, and Erk in pancreatic cancer cells and ductal cells (Askari et al., 2005; Blackford et al., 2009), thereby modifying proliferation and cell death (Edderkaoui and Thrower, 2013; Park et al., 2013). From an observational epidemiologic standpoint, 14 PC case-​control studies have investigated the potential interaction between smoking and polymorphisms using a candidate gene approach. In individual studies, an increased risk of PC has been reported between smoking status and those with a minor allele for SNPs in XRCC2 (p = 0.02; involved in DNA repair) (Jiao et al., 2008); XRCC4 (p = 0.02) (Shen et al., 2015); CAPN10 (involved in insulin uptake) (Fong et al., 2010); adiponectin gene (p = 0.022; involved in metabolic and hormone processes) (Yang et al., 2015); EPHX1 (p = 0.04; involved with metabolism); and NAT2 (p = 0.03; involved with detoxification of drugs and bioactivation of carcinogens) (Jang et al., 2012). Variants in the CYP1A2 (p = < 0.001; involved with metabolism) and NAT1 (p = 0.012; involved with detoxification of drugs and bioactivation of carcinogens) genes have also been observed to interact with heavy smoking among women (Suzuki et al., 2008b). Both genes are involved in detoxifying and bioactivation of aromatic amines, and NAT1 rapid acetylator genotypes have been identified (Hein et  al., 2000; Hirvonen, 1999). Genes in the IGF axis regulate cell differentiation, proliferation, and migration, and play an important part in initiating carcinogenesis (Gukovskaya et al., 2002; Haber et al., 2004; Verma, 2005; Zatonski et al., 1993). Two genes that encode components of the IGF-​axis, IGF2R and IRS1, interact with smoking, but the mechanisms are unknown (Dong et al., 2012). There is an observed interaction between XPD (involved in DNA repair) and smoking where having a polymorphism in XPD Asn312Asn and being an ever smoker (current and former) reduced the risk of PC (OR = 0.42 [0.21, .083]; p = 0.01) (Li J et al., 2007). Functionally, the Asn312Asn polymorphism may change the folding pattern of the resulting protein and corresponding function (Affatato et al., 2004). There is a significant interaction between smoking and cytotoxic T lymphocyte-​ associated protein (CTLA-​ 4; involved in immune response) and the risk of PC where smokers with at least one A allele have an increased risk of PC (p for interaction = 0.037) (Yang et al., 2012). Contrary to these individual study results, no significant smoking–​gene interactions were observed after multiple comparison correction in a discovery study using existing GWAS and smoking datasets and a combined sample of 2028 cases and 2109 controls (Tang et al., 2014a).

Obesity and Genes

Four studies have investigated the potential interaction for PC risk between obesity and genes responsible for regulating the balance of energy and tumor development and progression. Nakao et al. (2011) studied the interaction with the IGF-​1 gene in a Japanese hospital-​ based case-​control study. Weight was self-​reported at baseline and recalled for 20 years of age. Those with a minor allele for rs574214 and BMI ≥ 25 were at an increased risk of PC. In a previous study (Lin et al., 2011), this polymorphism was found to be associated with risk of PC and diabetes mellitus, but not BMI. Genetic variation in FTO has been associated with obesity (Hinney et al., 2007; Scott et al., 2007)  and is regulated by fasting and feeding status (Gerken et  al., 2007) and negatively regulates lipid metabolism (Klöting et al., 2008). Those with the FTO polymorphism and BMI < 25 have a reduced risk of PC, and those with BMI ≥ 25 have an increased risk (Tang et al., 2011). The mechanistic relationship with BMI is currently not known. ADIPOQ codes for adipocyte-​secreted hormone and has a low frequency of the homozygous variant in the study population. However, a significant interaction with BMI < 25 was observed (p = 0.005) (Tang et al., 2011). Among 172 cases and 181 controls in a Chinese population, 6 SNPs were analyzed in the adiponectin gene (Yang et al., 2015). One was significantly associated with a decreased risk (OR  =  0.66; 95% CI:  0.47, 0.93) of PC, while another was associated with an increased risk (OR = 1.42; 95% CI: 1.04, 1.94). Pooling information

624

624

PART IV:  Cancers by Tissue of Origin

from multiple sites into a discovery study using existing GWAS and smoking data sets for 2028 cases and 2109 controls, no significant obesity–​gene interactions were observed after multiple comparison correction (Tang et al., 2014b).

Diet and Genes

Dietary intake has been proposed to interact with genes involved with metabolism, antioxidant defense, and DNA repair (Cullen et al., 2003; Hauptmann and Cadenas, 1997; Jansen et al., 2013b; Ough et al., 2004; Suzuki et al., 2008b; Tang et al., 2010; Weydert et al., 2003; Yamauchi, 2007; Zhang et al., 2011). Alcohol and its major metabolite, acetaldehyde, are categorized as carcinogens (Baan et  al., 2009), which has been hypothesized to interact with genes that play a role in tumor development and progression (Appleman et al., 2000; Mohelnikova-​ Duchonova et al., 2010; Scheipers and Reiser, 1998; Yang et al., 2015, 2012). However, most of the reports based on potential interactions between diet or alcohol intake and candidate genes have not been replicated sufficiently to warrant conclusions about clear interaction.

Molecular Factors Telomere Length

Telomeres, the complex chromosomal end-​capping structures consisting of repetitive nucleotide sequences (TTAGGG) and the telomere binding-​protein complex called shelterin protect chromosomal DNA from degradation, and prevent end-​to-​end joining and aberrant recombination during cell division (Blackburn et al., 2015). Telomeres thereby preserve the integrity and stability of genomic material. Peripheral blood leukocyte telomere length (a marker of host telomere status) is approximately 10−16 kilobase pairs long at birth (Drury et al., 2015; Okuda et al., 2002), but progressively shortens by about 50–​100 base pairs with each cell division, due to incomplete replication (Collins and Mitchell, 2002). Because telomere length shortens with each cell division, telomere length correlates inversely with age and may serve as a marker of biological aging (Blackburn et al., 2015). Factors associated with chronic inflammation and oxidative stress, such as cigarette smoking, diabetes, and excess body weight are known contributors to the telomere length-​ shortening process (Blackburn et al., 2015; Harley, 1997; Jennings et al., 2000). Telomere attrition has been linked to increased risk of many cancer types. Five studies examined the association between telomere length and PC with mixed findings (Bao et  al., 2016; Campa et  al., 2014; Lynch et al., 2013; Skinner et al., 2012; Zhang et al., 2016). The first was a clinic-​based, case-​control study of 499 incident PC cases and 963 non-​cancer control patients (Skinner et al., 2012). This study reported a “U-​shaped” association between peripheral blood leukocyte telomere length and PC, such that individuals with extremely short (lower 1%) or extremely long (top 10%) telomere length had increased PC risk (Skinner et al., 2012). Among those within the 1st and 90th percentile distribution of telomere length, short telomere length was associated with an increased risk of PC (Skinner et  al., 2012). The second was a nested case-​control study among Finnish male smokers that consisted of 193 cases and 660 controls (Lynch et al., 2013). This prospective study reported that long leukocyte telomere length was associated with increased risk of PC in a dose–​response manner (OR = 1.57; 95% CI: 1.01, 2.43; longest vs. shortest telomere length quartile, P trend = 0.0007) (Lynch et al., 2013). The third report was from a nested case-​control study of 331 PC cases and 331 controls (Campa et al., 2014). It reported a modest increase in PC risk among individuals with long leukocyte telomere lengths (OR  =  1.13; 95% CI: 1.01, 1.27, continuous variable). Further analysis by cubic spline regression found that the association was non-​ linear (P for non-​ linearity  =  0.02) (Campa et  al., 2014). One of the two most recent studies was a pooled analysis of five prospective cohorts involving 386 cases and 896 controls (Bao et al., 2016). The results from this study showed that short pre-​diagnostic leukocyte telomere length was associated with increased PC in a dose-​dependent manner (OR = 1.72; 95% CI:  1.07, 2.78; shortest vs. longest telomere length quintile, P

trend = 0.048; P for non-​linearity > 0.05) (Bao et al., 2016). The final report was from a population-​based, nested case-​control study of 900 cases and 900 controls in Liaoning, China, and showed a “U-​shaped” association between leukocytes telomere length and PC risk (Zhang et al., 2016). Using the third quartile as the reference category, they found increased risk for those in the shortest telomere length quartile (OR  =  3.10; 95% CI:  1.84, 5.21) and those in the longest telomere length quartile (OR = 1.49; 95% CI: 1.11, 2.00) (Zhang et al., 2016). These studies differed in many ways, including differences in the studied populations, variations in the measurement of telomere length, and variations in the time between blood collection and diagnosis of PC, which may explain the inconsistent finding. Tumor-​based studies show that pancreatic tumors have shortened telomere lengths, suggesting that telomere shortening might be central to pancreatic tumorigenesis (Bardeesy and DePinho, 2002; Hashimoto et al., 2008).

Epigenetic Factors Our knowledge of the molecular structure of DNA beyond the gene sequence and how these mechanisms interact with the genetic structure is continuously expanding, but is currently limited. The genome is generally considered to represent inherited disease susceptibility and has been thought to be fairly stable over time with the help of efficient DNA repair mechanisms. We also know that additional mechanisms, collectively referred to as epigenetics, can be a means by which non-​inherited effects (i.e., due to environmental insults) can dynamically affect the expression (not coding) of the genetic sequence (Heard and Martienssen, 2014; Nagy and Turecki, 2015). Examples of epigenetic mechanisms include DNA methylation, histone modification, and alterations in microRNA and non-​ coding RNA. Epigenetic changes are thought to represent early influences in carcinogenesis (Verma et  al., 2014), with external factors such as diet, drugs, or infectious agents generating interactive effects to modify a person’s cancer risk (Costa, 2010; Hou et al., 2011; Verma, 2013). The potential ability to reverse or prevent harmful changes to these epigenetic mechanisms through behavior modification or other interventions is what makes epigenetic research attractive from a public health and disease prevention standpoint. Epigenetic markers and patterns are tissue-​specific, although many markers have been identified as playing an important role in different cancer types. To date, most research relevant to epidemiology of PC has been in DNA methylation of promoter sites in tumor suppressor genes (i.e., APC, BRCA1, and CDKN2A) identified as the most common regions in human pancreatic neoplasms (Guo et  al., 2014). Genome-​wide DNA methylation patterns in PDAC tissues suggest methylated genes involved in TGF-​beta, WNT, integrin signaling, cell adhesion, and stellate cell activation (Dutruel et al., 2014; Nones et al., 2014; Vincent et al., 2014).

Other Risk Factors Microbiome

Investigations into human microbiome and PC risk have highlighted how disruptions in host commensal bacterial population can promote PC risk (Zambirinis et al., 2014). The concept of microbiota perturbations as a potential modulator of PC risk assumes the importance of increasing the organ-​specific prevalence of emerging pathogenic bacteria in relationship to impaired host immune responses, a phenomenon referred to as dysbiosis or dysbacteriosis (Zambirinis et al., 2014). However, the role of shifts in the human microbiome composition as a causative factor of PC has yet to be thoroughly elucidated. It is well recognized that exposure to certain environmental agents can alter the composition of the gut microbiome and disrupt mucosal permeability. These bacteria or their products (e.g., lipopolysaccharides and bacterial DNA) can then evoke local and systemic inflammatory responses (Yan and Schnabl, 2012; Zambirinis et al., 2014). For example, studies have shown that excess consumption of alcohol can lead to disruption of the intestinal mucosal barrier, accompanied by increased

 625



Cancer of the Pancreas

proliferation of pathogenic bacteria in the gut (Purohit et  al., 2008; Yan and Schnabl, 2012; Zambirinis et al., 2014). Such gut microbes can migrate to the pancreas via reflux of intestinal contents into the main pancreatic duct. Studies also indicate that changes in oral microbiota because of poor hygiene may contribute to PC. In this context, periodontal disease and tooth loss in adults, caused by bacterial infection and periodontal inflammation, have been associated with increased PC risk (Michaud et al., 2007; Stolzenberg-​Solomon et al., 2003). Moreover, Farrell and colleagues examined microbiome profiles in human saliva and found specific oral microbiota profiles associated with increase of PC and pancreatitis (Farrell et al., 2012). The study found 31 bacterial species that were significantly higher in saliva pellets of PC patients compared to healthy controls, and 25 bacteria species that were significantly lower in saliva of PC patients than controls. The study further showed that the presence of two bacteria (Streptococcus mitis and Neisseria elongata), when combined, can distinguish PC patients from healthy controls with 96% sensitivity and 82% specificity. Although this was a small study consisting of 10 cases and 10 controls, the findings were validated in an independent sample of 28 PC cases and 28 controls (Farrell et al., 2012). The small samples in these studies limit generalizability, but provide intriguing clues regarding the role of microbiomes in PC. Larger studies are needed to refine and validate such biological markers. Investigations into the prognostic relevance of the oral cavity and gut microbiome may provide new insights for preventive interventions.

Inflammation

Experimental, clinical, and epidemiologic evidence show a clear link between inflammation and PC. The association appears to be mediated by a “field effect,” such that PC tends to be fostered in a pro-​ inflammatory microenvironment with high concentration of growth factors, cytokines, and chemokines, and activation of inflammatory pathways such as cyclooxygenase-​ 2 (COX-​ 2), and nuclear factor kappa B (NF-​kB) (Garcea et al., 2005; Greer and Whitcomb, 2009). The epidemiologic data show a strong association between PC and pro-​inflammatory stimuli, such as tobacco smoking, and inflammatory conditions, such as chronic pancreatitis, diabetes, and obesity (Lowenfels and Maisonneuve, 2006). A  chronic inflammatory state, particularly chronic inflammation of the pancreas (pancreatitis), favors pancreatic tumorigenesis in that it causes a cycle of repeated DNA damage and repair, leading to increased frequency of cell division, which, in turn, increases the chances of cellular aberrations, malignant cell formation, and ultimately, frank malignancy (Garcea et al., 2005; Greer and Whitcomb, 2009). Individuals with hereditary pancreatitis, a rare form of chronic pancreatitis that is caused by mutations in the cationic trypsinogen gene (PRSS1), have a 40% risk of developing PC by age 70 years (Lowenfels et al., 1997), while individuals with non-​genetic forms of chronic pancreatitis have a 13-​fold higher risk of PC compared to healthy populations (Yadav and Lowenfels, 2013). Cigarette smoking, diabetes, and obesity contribute to PC by promoting a persistent state of systemic inflammation, which has been associated with DNA adducts formation, altered gene expression, oncogene activation, and inhibition of apoptosis (Garcea et al., 2005; Greer and Whitcomb, 2009). Obesity accounts for ~12% of PCs (Parkin et al., 2011)  and is associated with increased circulating levels of the pro-​ inflammatory adipokines, resistin and leptin, and lower levels of the anti-​ inflammatory adipokine, adiponectin (Kowalska et  al., 2008). Diabetes, a well-​known risk factor of PC, has been associated with higher circulating levels of inflammatory cytokines such as C-​reactive protein (CRP) and interleukin 6 (IL-​6) (Pradhan et al., 2001), while cigarette smoking, which accounts for ~20%–​25% of PCs, is associated with elevated levels of many inflammatory cytokines including CRP, IL-​6, and intercellular adhesion molecule-​1 (ICAM-​1) (Levitzky et al., 2008). Thus, inflammation appears to be a common underlying mechanism through which these factors influence PC risk. Although inconclusive, emerging evidence suggest that diet-​derived inflammation, mainly from consumption of red meat, high-​fat and high-​ calorie foods, and sugar-​sweetened beverages, increases PC risk (Antwi et al., 2016). However, as with other environmental exposures such as

625

smoking, the effect of diet-​derived inflammation may be dependent on inter-​individual variation in metabolism of pro-​carcinogens, such that the effect might be less pronounced in persons with efficient metabolic detoxification of carcinogens than in those with inefficient detoxification of carcinogens. Studies also show that COX-​2 inhibitors can delay the development of the PC precursor lesion, pancreatic intraepithelial neoplasia (PanIN) (Hruban et al., 2008), while omega-​3 fatty acids can suppress the growth of pancreatic tumors (Hering et al., 2007), further supporting a role of inflammation in PC.

Infectious Agents

Pathogenic infections are estimated to account for ~10%–​20% of all cancers worldwide (Chang and Parsonnet, 2010). Although several studies suggest the possibility of an association with an infectious agent, no agent has been confirmed as causative for PC. Infection with hepatitis B and C has been associated with PC in some studies (Fiorino et al., 2013; Majumder et al., 2014). However, there is stronger evidence for a plausible role of Helicobacter pylori (H.  pylori), an infectious organism that can survive the acidic conditions of the gastric lumen and is associated with increased risk of gastric cancer (Lowenfels and Maisonneuve, 2006). A  1998 study by Raderer and colleagues was the first to report an epidemiologic association between H. pylori infection and PC (Raderer et al., 1998). This case-​ control study of 27 cases and 27 controls reported a 2-​fold increase in risk among H.  pylori positive individuals (OR  =  2.1; 95% CI:  1.09, 4.05) (Raderer et  al., 1998). In 2001, Stolzenberg-​Solomon et  al. measured seroprevalence of H. pylori among 121 cases and 226 controls in a nested case-​ control study within the Alpha-​ Tocopherol, Beta-​Carotene Cancer Prevention Study, a Prospective Cohort study of Finnish Male Smokers (Stolzenberg-​Solomon et al., 2001). These investigators observed increased risk among individuals with H. pylori infection (1.87; 95% CI: 1.05, 3.34), with a much stronger association among those with the cytotoxin-​associated gene-​A-​positive (CagA+) H.  pylori strains (2.01; 95% CI:  1.09, 3.70) (Stolzenberg-​Solomon et al., 2001). Since these studies, at least four additional studies have investigated this association, and none found a statistically significant association (de Martel et al., 2008b; Lindkvist et al., 2008; Risch et al., 2010; Schulte et al., 2015), although two reported a positive relationship (Lindkvist et al., 2008; Risch et al., 2010). A 2011 meta-​analysis of the six studies, including five of the studies cited earlier (de Martel et al., 2008a; Lindkvist et al., 2008; Raderer et al., 1998; Risch et al., 2010; Stolzenberg-​Solomon et  al., 2001)  reported a 38% increased risk among H. pylori seropositive individuals (OR 1.38; 95% CI: 1.08, 1.75) (Trikudanathan et al., 2011). In a more recent meta-​analysis of nine studies, increased PC risk was observed among H. pylori positive individuals (1.47; 95% CI: 1.22, 1.77) (Xiao et al., 2013). When analyses were stratified by geographic regions, an excess risk of 56% was observed in studies conducted in Europe, 2-​fold excess risk among studies conducted in East Asia, and a non-​significant 17% excess risk among studies conducted in North America (Xiao et  al., 2013). Therefore, while the evidence is not entirely consistent, it is plausible that H. pylori may be involved in pancreatic carcinogenesis, possibly through interaction with other host factors such as ABO blood type (Risch et al., 2010).

Allergies

Associations between self-​reported allergies and reduced risk of PC have been observed in many epidemiologic studies (Olson, 2012). In 2005, a meta-​analysis of 14 studies reported a significantly lower risk of PC among individuals with history of allergy in general (i.e., any allergy) (RR = 0.82; 95% CI: 0.68, 0.99), and particularly atopic allergies (RR  =  0.71; 95% CI:  0.64, 0.80), but not asthma (RR  =  1.01; 95% CI:  0.77, 1.31) (Gandini et  al., 2005). A  pooled analysis of individual data from 10 case-​control studies also reported a 26% and 38% lower PC risk among individuals with self-​reported history of hay fever and allergy to animals, respectively, which the investigators suggested indicates a potential role of immunoglobulin E (an allergic response antibody) in PC (Olson et  al., 2013). The results further showed a suggestion toward lower risk among those with a history of any allergy (OR = 0.79; 95% CI: 62, 1.00),

62

626

PART IV:  Cancers by Tissue of Origin

excluding asthma (Olson et al., 2013). A thorough review of 11 publications on allergy history and PC risk showed an overall inverse association, with ORs ranging between 0.39 (95% CI:  0.22, 0.71) and 0.77 (95% CI: 0.63, 0.95) (Olson, 2012). In general, the epidemiologic evidence indicates an inverse association between history of respiratory allergies (particularly hay fever) and PC, but does not support an association with asthma, possibly because of the complex and varied etiology of asthma (Olson, 2012).

Previous Surgery

Partial gastrectomy and cholecystectomy have been reported to be associated with PC risk. Among 15 studies that examined association between gastrectomy and PC risk, 11 reported elevated risk among post-​gastrectomy patients, with five showing statistically significant increased risk (Olson, 2012). In agreement with these results, a multicenter case-​control study of 4717 cases and 9374 controls reported increased PC risk among post-​gastrectomy patients (OR  =  1.53; 95% CI:  1.15, 2.03), but the association was limited to gastric resections occurring within 2  years before diagnosis of PC (OR = 6.18; 95% CI: 1.82, 20.96) (Bosetti et al., 2013). Other studies have suggested that cholecystectomy (surgical removal of gallbladder) may be associated with PC, but the evidence is inconclusive. While some studies have reported elevated PC risk after cholecystectomy (Silverman et al., 1999; Zhang et al., 2014), many show no association (Goldacre et  al., 2005; Schernhammer et  al., 2002; Ye et  al., 2001). The link between cholecystectomy and PC is based on the notion that removal of the gallbladder increases circulating levels of cholecystokinin (a digestive hormone thought to promotes PC growth), and enhances degradation of bile salt into bile acid and other metabolites that increases cell proliferation (Ura et al., 1986). However, this is not completely supported by the existing epidemiologic data.

NEUROENDOCRINE/​ISLET CELL TUMORS Pancreatic neuroendocrine tumors (PNETs) are rare tumors of neuroendocrine origin arising in the islet cells of the pancreas. PNETs comprise approximately 3% of PC. They are typically more indolent than pancreatic adenocarcinomas, although some PNETS can progress more rapidly to malignancy (Amin and Kim, 2016; Halfdanarson et  al., 2008b). However, the reasons for progression are unclear. Population-​based studies have reported wide estimates of 5-​year overall survival, ranging from 15% to 60%, based on disease stage and treatment center (Bilimoria et al., 2007; Strosberg et al., 2011). Based upon SEER data, the US crude annual incidence per 100,000 for all ages is 0.1 in females and 0.2 in males, and incidence increases with age (Halfdanarson et al., 2008a). Incidence of PNET has risen sharply in the United States over the past two decades; this may be related to improvements in imaging and increased screening of the pancreas (Yao et  al., 2008). In 1973, the incidence was 0.17/​100,000; it was 0.47/​100,000 by 2007 (Halfdanarson et al., 2008a; Hallet et al., 2015; Lawrence et al., 2011). PNETs are divided into two main types: functional (the tumors produce excess hormones of different varieties), often associated with clinical syndromes; and non-​ functional (non-​ secretory), which are often metastatic at diagnosis (Halfdanarson et  al., 2008b). Functional PNETs constitute 10%–​4 0% of all PNETs and are distinguished by the main hormones they secrete (insulin, gastrin, glucago-​vaso-​intestinal peptide, or associations with clinical conditions such as hyperinsulinemia, peptic ulcer disease). Inherited syndromes such as multiple endocrine neoplasia type 1 (MEN1) (Lakhani et  al., 2007) or von Hippel-​L indau disease (Lonser et al., 2003) can feature functional PNETs, although most PNETs are thought to be sporadic, not associated with any known hereditary or familial syndromes. The mean age at diagnosis of patients with PNETs is 58.5 + 14.9, with slightly older mean age for those with non-​functional tumors (58.8 + 14.7) compared to those with functional tumors (55.2 + 16.3) (Strosberg et al., 2011).

For unselected PNETs, a family history of other gastrointestinal neuroendocrine tumors and other cancers in first-​degree relatives was suggested (Hassan et al., 2008b; Hiripi et al., 2009). One study reported that patients with non-​functional PNETs were more likely to have a family history of pancreatic adenocarcinoma, but no formal comparison was done (Gullo et  al., 2003). In a study of 309 cases and 602 controls by Halfdanarson et al. (2014), PNET cases were more likely to report a family member with sarcoma, PNET, cancer of the gallbladder, stomach, and ovary, but in the absence of any association of PNET with a personal history of pancreatic adenocarcinoma. There are conflicting data on the association between smoking and alcohol use with risk of neuroendocrine tumors (Hassan et al., 2008a; Kaerlev et  al., 2002). In patients with five different types of neuroendocrine tumors, no association was observed between smoking or alcohol consumption and the development of neuroendocrine tumors (Hassan et al., 2008a). An association was suggested between smoking and the risk of neuroendocrine malignancies in two population-​based studies that included a limited sample of patients with small bowel neuroendocrine tumors (carcinoid tumors) (Hassan et  al., 2008a; Kaerlev et  al., 2002). In the study by Halfdanarson et  al., slightly more cases were likely to have a personal history of tobacco use than controls, but this difference was not statistically significant. However, alcohol use was less common among cases (54% vs. 67%, P < 0.001) (Halfdanarson et al., 2014). Several studies found associations with a personal history of diabetes (Feldman et al., 1975; Halfdanarson et al., 2014; Hassan et al., 2008a; Kaerlev et al., 2002). Interestingly, Hassan et al. reported that patients with PNETs were more likely to have been diagnosed with diabetes in the year leading up to the PNET diagnosis (Hassan et al., 2008a). Halfdanarson et al. found that diabetes was more commonly reported by cases than controls (19% vs. 11%, P < 0.001) (Halfdanarson et al., 2014)  and also suggested that the association between recent-​onset diabetes and pancreatic adenocarcinoma that they observed may represent the effect of proteins secreted by the tumor on beta-​cell function and insulin resistance. In summary, pancreatic neuroendocrine tumors (functional and non-​ functional), constitute a small fraction of PCs, but are increasing in incidence, possibly due to improved diagnostic procedures. Other than rare genetic syndromes that feature functional PNETs, studies of risk factors are limited and do not conclusively implicate any factors that are associated with pancreatic adenocarcinoma.

CURRENT STUDY LIMITATIONS All epidemiologic study designs are subject to limitations and biases that affect the interpretation and generalizability of reported results. For example, differential misclassification and recall of dietary patterns between cases and controls could contribute to biased risk estimates. Comorbidities associated with smoking, obesity, and alcohol intake affect selection of cases with these exposures. For many of the exposures discussed here, there are social stigmas associated with high levels of consumption that may influence how a participant completes survey questions. In retrospective population-​based studies of rapidly fatal disease, bias can occur due to the demise of eligible cases with a higher proportion of later-​stage disease, possibly resulting in non-​random non-​response. In prospective studies, the rarity of PC limits the number of potential cases seen during follow-​up. Both these situations lead to a reduced power to detect associations. Moreover, gene–​environment interaction studies are often criticized for being underpowered, and it has been suggested that the associations seen are often false positives and cannot be replicated.

OPPORTUNITIES FOR PREVENTION Smoking prevention and cessation, maintaining a healthy body weight, and consuming a well-​balanced diet present the best opportunities for

 627



Cancer of the Pancreas

the prevention of PC. The 2014 US Surgeon General’s Report estimates that ~10%–​14% of PC deaths in the United States are attributable to tobacco smoking (Reports of the Surgeon General, 2014), whereas European studies suggest that some 20%–​25% of PC cases are attributable to smoking (Lowenfels and Maisonneuve, 2006). While current smoking, smoking duration, and smoking dose are all strongly associated with increased PC risk, many studies have shown that the PC risk associated with smoking regresses to null after about 15  years of smoking cessation, a clear benefit of smoking cessation (Bosetti et  al., 2012a; Fuchs et  al., 1996; Iodice et  al., 2008; Lynch et al., 2009). The risk fraction attributable to excess adiposity is not completely clear; Calle and Kass (2004) attribute 27% of all PCs in the United States to overweight and obesity, while Parkin et al. (2011) attribute 12% of PCs in the United Kingdom to overweight and obesity. It is, nonetheless, plausible that at least 25% of PCs in the United States can be prevented through the elimination of smoking and long-​term maintenance of a healthy body weight. Although the role of diet in PC is not well characterized, there is evidence indicating that excessive alcohol consumption and increased intake of saturated fat, red meat, and processed meat are associated with increased risk (Michaud et  al., 2010; Nöthlings et  al., 2005; Stolzenberg-​Solomon et  al., 2002b). Limiting intake of these foods may therefore reduce risk. Long-​standing diabetes has been associated with increased PC risk. For diabetics, better glycemic control involving dietary modification and regular physical activity may reduce insulin resistance and its attendant hyperinsulinemia, which are associated with increased PC risk (Bao et  al., 2011; Li, 2012). While inconclusive, there are suggestions that smokeless tobacco use and secondhand tobacco smoke may increase risk (Alguacil and Silverman, 2004; Boffetta et al., 2008; Vrieling et al., 2010); therefore, avoiding these exposures may reduce individual risk.

Future Directions The development of robust risk prediction models that combine genomic and risk-​factor information is central to the identification of individuals at high risk of developing PC for timely preventive interventions, including chemoprevention and early detection. Epidemiologic and risk factors data as reviewed in this chapter provide important clues for risk stratification. With the increasing obesity epidemic, especially among youth, and the strong association between obesity and PC, it can be expected that obesity-​related PC rates will increase over the coming decades. Dietary results regarding PC risk have largely been inconsistent, with the potential exception of certain fatty acids and well-​done red meat. Dietary data have been fraught with measurement error, and often a large percentage of the data is missing for participants. Technology provides a potential solution for a more accurate assessment of dietary habits and a better understanding of how diet influences PC risk, but not for reconstructing past dietary exposures. As smoking rates continue to decrease, cigarette smoking–​related PC will also decrease. The role that e-​cigarettes may play in PC has yet to be determined, while the effect of environmental exposure, especially in early childhood, needs further exploration. Alcohol appears to be a risk for PC only among those in the heaviest consumption category; however, remaining potential confounding of smoking behaviors, even after adjustment, is a concern. Genetic data can assist in identifying individuals at high risk of developing PC; new statistical and epidemiological methods or processes are needed to pinpoint the responsible genetic variants or regions and their interaction with modifiable risk factors. Large and well-​designed prospective studies investigating the potential chemopreventive effects of metformin, statins, and aspirin, both independently and in combination, may inform strategies for addressing the rising incidence of PC. Epidemiologic research is also needed to increase precision in methods for identifying and targeting individuals for earlier detection. Evaluation of biomarkers for early detection of PC will require particular focus on designs to deliver highly specific and sensitive assays. New research directions involving advanced technology and better understanding of the biology of pancreatic carcinogenesis in molecular

627

and cell biology, assays using proteomics, microarray analysis of gene expression profiles, and detection of circulating tumor-​secreted DNA and exosomes hold promise. References Affatato AA, Wolfe KJ, Lopez MS, et al. 2004. Effect of XPD/​ERCC2 polymorphisms on chromosome aberration frequencies in smokers and on sensitivity to the mutagenic tobacco-​specific nitrosamine NNK. Environ Mol Mutagen, 44(1), 65–​73. PMID: 15199548. Ahlgren JD. 1996. Epidemiology and risk factors in pancreatic cancer. Semin Oncol, 23(2), 241–​250. PMID: 8623060. Alguacil J, and Silverman DT. 2004. Smokeless and other noncigarette tobacco use and pancreatic cancer:  a case-​control study based on direct interviews. Cancer Epidemiol Biomarkers Prev, 13(1), 55–​58. American Cancer Society. 2013. Cancer Facts & Figures. Atlanta, GA: American Cancer Society. American Cancer Society. 2016. Cancer Facts & Figures. Atlanta, GA: American Cancer Society. Amin S, and Kim MK. 2016. Islet cell tumors of the pancreas. Gastroenterol Clin North Am, 45(1), 83–​100. PMID: 26895682. Amundadottir LT. 2016. Pancreatic cancer genetics. Int J Biol Sci, 12(3), 314–​ 325. PMCID: PMC4753160. Amundadottir L, Kraft P, Stolzenberg-​Solomon RZ, et  al. 2009. Genome-​ wide association study identifies variants in the ABO locus associated with susceptibility to pancreatic cancer. Nat Genet, 41(9), 986–​ 990. PMCID: PMC2839871. Anderson KE, Johnson TW, Lazovich D, and Folsom AR. 2002. Association between nonsteroidal anti-​inflammatory drug use and the incidence of pancreatic cancer. JNCI, 94(15), 1168–​1171. PMID: 12165642. Anderson KE, Mack TM, and Silverman DT. 2006. Cancer of the pancreas. In:  Schottenfeld D, and Fraumeni FJ (Eds.), Cancer Epidemiology and Prevention (3rd ed., pp. 721–​763). New York: Oxford University Press. Anderson KE, Mongin SJ, Sinha R, et al. 2012. Pancreatic cancer risk: associations with meat-​ derived carcinogen intake in the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO) cohort. Mol Carcinog, 51(1), 128–​137. PMCID: PMC3516181. Andreotti G, and Silverman DT. 2012. Occupational risk factors and pancreatic cancer:  a review of recent findings. Mol Carcinog, 51(1), 98–​108. PMID: 22162234. Antonello D, Gobbo S, Corbo V, et al. 2009. Update on the molecular pathogenesis of pancreatic tumors other than common ductal adenocarcinoma. Pancreatology, 9(1-​2), 25–​33. PMID: 19077452. Antwi SO, Eckert EC, Sabaque CV, et  al. 2015. Exposure to environmental chemicals and heavy metals, and risk of pancreatic cancer. Cancer Causes & Control, 26(11), 1583–​1591. PMCID: PMC4624268. Antwi SO, Oberg AL, Shivappa N, et al. 2016. Pancreatic cancer: associations of inflammatory potential of diet, cigarette smoking and long-​standing diabetes. Carcinogenesis. PMID: 26905587. PMCID: PMC4843052. Appleman LJ, Berezovskaya A, Grass I, and Boussiotis VA. 2000. CD28 costimulation mediates T cell expansion via IL-​2-​independent and IL-​ 2-​dependent regulation of cell cycle progression. J Immunol, 164(1), 144–​151. Arnold LD, Patel AV, Yan Y, et  al. 2009. Are racial disparities in pancreatic cancer explained by smoking and overweight/​obesity? Cancer Epidemiol Biomarkers Prev, 18(9), 2397–​2405. PMCID: PMC3630792. Askari MDF, Tsao M-​S, and Schuller HM. 2005. The tobacco-​specific carcinogen, 4-​(methylnitrosamino)-​1-​(3-​pyridyl)-​1-​butanone stimulates proliferation of immortalized human pancreatic duct epithelia through beta-​adrenergic transactivation of EGF receptors. J Cancer Res Clin Oncol, 131, 639–​648. PMID: 16091975. Aune D, Greenwood DC, Chan DS, et al. 2012. Body mass index, abdominal fatness and pancreatic cancer risk:  a systematic review and non-​linear dose-​response meta-​analysis of prospective studies. Ann Oncol, 23(4), 843–​852. PMID: 21890910. Austin MA, Kuo E, Van Den Eeden SK, et  al. 2013. Family history of diabetes and pancreatic cancer as risk factors for pancreatic cancer:  the PACIFIC study. Cancer Epidemiol Biomarkers Prev, 22(10), 1913–​1917. PMCID: PMC3808252. Ayaz L, Ercan B, Dirlik M, Atik U, and Tamer L. 2008. The association between N-​acetyltransferase 2 gene polymorphisms and pancreatic cancer. Cell Biochem Funct, 26(3), 329–​333. PMID: 18027363. Baan R, Grosse Y, Straif K, et al. 2009. A review of human carcinogens: Part F:  chemical agents and related occupations. Lancet Oncol, 10(12), 1143–​1144. Banke MG, Mulvihill JJ, and Aston CE. 2000. Inheritance of pancreatic cancer in pancreatic cancer-​prone families. Med Clin North Am, 84(3), 677–​690, x-​xi. PMID: 10872424.

628

628

PART IV:  Cancers by Tissue of Origin

Bao B, Wang Z, Li Y, et al. 2011. The complexities of obesity and diabetes with the development and progression of pancreatic cancer. Biochim Biophys Acta, 1815(2), 135–​146. PMCID: PMC3056906. Bao Y, and Michaud DS. 2008. Physical activity and pancreatic cancer risk: a systematic review. Cancer Epidemiol Biomarkers Prev, 17(10), 2671–​ 2682. PMCID: PMC4896495. Bao Y, Prescott J, Yuan C, et al. 2016. Leucocyte telomere length, genetic variants at the TERT gene region and risk of pancreatic cancer. Gut, 1–​7. doi:10.1136/​gutjnl-​2016-​312510. PMID: 27797938. Bardeesy N, and DePinho RA. 2002. Pancreatic cancer biology and genetics. Nat Rev Cancer, 2(12), 897–​909. PMID: 12459728. Bartsch H, Malaveille C, Lowenfels AB, et  al. 1998. Genetic polymorphism of N-​ acetyltransferases, glutathione S-​ transferase M1 and NAD(P) H:quinone oxidoreductase in relation to malignant and benign pancreatic disease risk. The International Pancreatic Disease Study Group. Eur J Cancer Prev, 7(3), 215–​223. PMID: 9696930. Basturk O, Hong SM, Wood LD, et al. 2015. A revised classification system and recommendations from the Baltimore Consensus Meeting for neoplastic precursor lesions in the Pancreas. Am J Surg Pathol, 39(12), 1730–​ 1741. PMCID: PMC4646710. Batabyal P, Vander Hoorn S, Christophi C, and Nikfarjam M. 2014. Association of diabetes mellitus and pancreatic adenocarcinoma: a meta-​analysis of 88 studies. Ann Surg Oncol, 21(7), 2453–​2462. PMID: 24609291. Ben Q, Xu M, Ning X, et al. 2011. Diabetes mellitus and risk of pancreatic cancer: a meta-​analysis of cohort studies. Eur J Cancer, 47(13), 1928–​1937. PMID: 21458985. Berrington de Gonzalez A, Spencer EA, Bueno-​de-​Mesquita HB, et al. 2006. Anthropometry, physical activity, and the risk of pancreatic cancer in the European Prospective Investigation into Cancer and Nutrition. Cancer Epidemiol Biomarkers Prev, 15(5), 879–​885. PMID: 16702364. Berrington de Gonzalez A, Sweetland S, and Spencer E. 2003. A meta-​analysis of obesity and the risk of pancreatic cancer. Br J Cancer, 89(3), 519–​523. PMCID: PMC2394383. Bertuccio P, La Vecchia C, Silverman DT, et al. 2011. Cigar and pipe smoking, smokeless tobacco use and pancreatic cancer: an analysis from the International Pancreatic Cancer Case-​Control Consortium (PanC4). Ann Oncol, 22(6), 1420–​1426. Biankin AV, Waddell N, Kassahn KS, et al. 2012. Pancreatic cancer genomes reveal aberrations in axon guidance pathway genes. Nature, 491(7424), 399–​405. PMCID: PMC3530898. Bilimoria KY, Bentrem DJ, Merkow RP, et al. 2007. Application of the pancreatic adenocarcinoma staging system to pancreatic neuroendocrine tumors. J Am Coll Surg, 205(4), 558–​563. PMID: 17903729. Blackburn EH, Epel ES, and Lin J. 2015. Human telomere biology: a contributory and interactive factor in aging, disease risks, and protection. Science, 350(6265), 1193–​1198. PMID: 26785477. Blackford A, Parmigiani G, Kensler TW, et al. 2009. Genetic mutations associated with Cigarette smoking in pancreatic cancer. Cancer Res, 69, 3681–​ 3688. PMID: 19351817. Bocci G, Fioravanti A, Orlandi P, et al. 2005. Fluvastatin synergistically enhances the antiproliferative effect of gemcitabine in human pancreatic cancer MIAPaCa-​2 cells. Br J Cancer, 93(3), 319–​330. PMCID: PMC2361561. Bodmer M, Becker C, Meier C, Jick SS, and Meier CR. 2012. Use of antidiabetic agents and the risk of pancreatic cancer: a case-​control analysis. Am J Gastroenterol, 107(4), 620–​626. PMID: 22290402. Boffetta P, Hecht S, Gray N, Gupta P, and Straif K. 2008. Smokeless tobacco and cancer. Lancet Oncol, 9(7), 667–​675. PMID: 18598931. Bonovas S, Filioussi K, and Sitaras NM. 2008. Statins are not associated with a reduced risk of pancreatic cancer at the population level, when taken at low doses for managing hypercholesterolemia: evidence from a meta-​analysis of 12 studies. Am J Gastroenterol, 103(10), 2646–​2651. PMID: 18684187. Bosetti C, Lucenteforte E, Bracci PM, et al. 2013. Ulcer, gastric surgery and pancreatic cancer risk:  an analysis from the International Pancreatic Cancer Case-​Control Consortium (PanC4). Ann Oncol, 24(11), 2903–​ 2910. PMCID: PMC3811904. Bosetti C, Lucenteforte E, Silverman DT, et al. 2012a. Cigarette smoking and pancreatic cancer: an analysis from the International Pancreatic Cancer Case-​Control Consortium (Panc4). Ann Oncol, 23(7), 1880–​1888. Bosetti C, Rosato V, Gallus S, Cuzick J, and La Vecchia C. 2012b. Aspirin and cancer risk: a quantitative review to 2011. Ann Oncol, 23(6), 1403–​1415. PMID: 22517822. Bosetti C, Rosato V, Li D, et  al. 2014. Diabetes, antidiabetic medications, and pancreatic cancer risk: an analysis from the International Pancreatic Cancer Case-​ Control Consortium. Ann Oncol, 25(10), 2065–​ 2072. PMCID: PMC4176453.

Brenner DR, Wozniak MB, Feyt C, et al. 2014. Physical activity and risk of pancreatic cancer in a central European multicenter case–​control study. Cancer Causes & Control, 25(6), 669–​681. Bryant KL, Mancias JD, Kimmelman AC, and Der CJ. 2014. KRAS: feeding pancreatic cancer proliferation. Trends Biochem Sci, 39(2), 91–​100. Calle EE, and Kaaks R. 2004. Overweight, obesity and cancer: epidemiological evidence and proposed mechanisms. Nat Rev Cancer, 4(8), 579–​591. PMID: 15286738. Campa D, Mergarten B, De Vivo I, et al. 2014. Leukocyte telomere length in relation to pancreatic cancer risk: a prospective study. Cancer Epidemiol Biomarkers Prev, 23(11), 2447–​2454. PMID: 25103821. Capurso G, Schunemann HJ, Terrenato I, et al. 2007. Meta-​analysis: the use of non-​steroidal anti-​inflammatory drugs and pancreatic cancer risk for different exposure categories. Aliment Pharmacol Ther, 26(8), 1089–​1099. PMID: 17894651. Carter BD, Freedman ND, and Jacobs EJ. 2015. Smoking and mortality: beyond established causes. N Engl J Med, 372(22), 2170. PMID: 26017836. Center M, Siegel R, and Jemal A. 2011. Global Cancer Facts & Figures. Atlanta, GA: American Cancer Society. Chaffee, K.G., et al. Genetic heterogeneity and survival among pancreatic adenocarcinoma (PDAC) patients with positive family history, in American Society of Clinical Oncology Annual Meeting. 2016: Chicago, IL. Chang AH, and Parsonnet J. 2010. Role of bacteria in oncogenesis. Clin Microbiol Rev, 23(4), 837–​857. PMCID: PMC2952975. Chari ST. 2007. Detecting early pancreatic cancer:  problems and prospects. Semin Oncol, 34(4), 284–​294. PMCID: PMC2680914. Chari ST, Leibson CL, Rabe KG, et  al. 2008. Pancreatic cancer-​associated diabetes mellitus: prevalence and temporal association with diagnosis of cancer. Gastroenterology, 134(1), 95–​101. PMCID: PMC2271041. Chari ST, Mohan V, Pitchumoni CS, et al. 1994. Risk of pancreatic carcinoma in tropical calcifying pancreatitis: an epidemiologic study. Pancreas, 9(1), 62–​66. PMID: 8108373. Childs EJ, Mocci E, Campa D, et al. 2015. Common variation at 2p13.3, 3q29, 7p13 and 17q25.1 associated with susceptibility to pancreatic cancer. Nat Genet, 47(8), 911–​916. PMCID: PMC4520746. Chuang S-​C, Gallo V, Michaud D, et  al. 2011. Exposure to environmental tobacco smoke in childhood and incidence of cancer in adulthood in never smokers in the EuropeanProspective Investigation into Cancer and Nutrition. Cancer Causes & Control, 22(3), 487–​494. Cogliano VJ, Baan R, Straif K, et al. 2011. Preventable exposures associated with human cancers. JNCI, 103(24), 1827–​1839. PMCID: PMC3243677. Collins K, and Mitchell JR. 2002. Telomerase in the human organism. Oncogene, 21(4), 564–​579. PMID: 11850781. Coogan PF, Rosenberg L, Palmer JR, et al. 2000. Nonsteroidal anti-​inflammatory drugs and risk of digestive cancers at sites other than the large bowel. Cancer Epidemiol Biomarkers Prev, 9(1), 119–​123. PMID: 10667472. Costa FF. 2010. Epigenomics in cancer management. Cancer Manag Res, 2, 255–​265. PMCID: PMC3004578. Cullen JJ, Weydert C, Hinkhouse MM, et  al. 2003. The role of manganese superoxide dismutase in the growth of pancreatic adenocarcinoma. Cancer Res, 63(6), 1297–​1303. Curado M-​P, Edwards B, Shin HR, et  al. 2007. Cancer Incidence in Five Continents, Volume IX. Lyon, France: IARC Press, International Agency for Research on Cancer. Currie CJ, Poole CD, and Gale EA. 2009. The influence of glucose-​lowering therapies on cancer risk in type 2 diabetes. Diabetologia, 52(9), 1766–​ 1777. PMID: 19572116. Dat NM, and Sontag SJ. 1982. Pancreatic carcinoma in brothers. Ann Intern Med, 97(2), 282. PMID: 7103290. de Martel C, Llosa AE, Friedman GD, et al. 2008a. Helicobacter pylori infection and development of pancreatic cancer. Cancer Epidemiol Biomarkers Prev, 17(5), 1188–​1194. PMID: 18483341. de Martel C, Llosa AE, Friedman GD, et al. 2008b. Helicobacter pylori infection and development of pancreatic cancer. Cancer Epidemiol Biomarkers Prev, 17(5), 1188–​1194. PMID: 18483341. Ding XZ, Fehsenfeld DM, Murphy LO, Permert J, and Adrian TE. 2000. Physiological concentrations of insulin augment pancreatic cancer cell proliferation and glucose utilization by activating MAP kinase, PI3 kinase and enhancing GLUT-​1 expression. Pancreas, 21(3), 310–​320. PMID: 11039477. Dong X, Li Y, Tang H, et al. 2012. Insulin-​like growth factor axis gene polymorphisms modify risk of pancreatic cancer. Cancer Epidemiol, 36(2), 206–​211. PMID: 21852217. Drury SS, Esteves K, Hatch V, et al. 2015. Setting the trajectory: racial disparities in newborn telomere length. J Pediatrics, 166(5), 1181–​1186.

 629



Cancer of the Pancreas

Duell EJ, Holly EA, Bracci PM, et al. 2002. A population-​based, case-​control study of polymorphisms in carcinogen-​metabolizing genes, smoking, and pancreatic adenocarcinoma risk. JNCI, 94(4), 297–​306. PMID: 11854392. Duell EJ, Holly EA, Bracci PM, Wiencke JK, and Kelsey KT. 2002. A population-​based study of the Arg399Gln polymorphism in x-​ray repair cross-​complementing group 1 (XRCC1) and risk of pancreatic adenocarcinoma. Cancer Res, 62(16), 4630–​4636. Duell EJ, Lucenteforte E, Olson SH, et  al. 2012. Pancreatitis and pancreatic cancer risk:  a pooled analysis in the International Pancreatic Cancer Case-​ Control Consortium (PanC4). Ann Oncol, 23(11), 2964–​ 2970. PMCID: PMC3477881. Dutruel C, Bergmann F, Rooman I, et al. 2014. Early epigenetic downregulation of WNK2 kinase during pancreatic ductal adenocarcinoma development. Oncogene, 33(26), 3401–​3410. Edderkaoui M, and Thrower E. 2013. Smoking and pancreatic disease. J Cancer Ther, 4, 34–​40. PMID: 24660091. Ehrenthal D, Haeger L, Griffin T, and Compton C. 1987. Familial pancreatic adenocarcinoma in three generations: a case report and a review of the literature. Cancer, 59(9), 1661–​1664. PMID: 3828965. Elena JW, Steplowski E, Yu K, et  al. 2013. Diabetes and risk of pancreatic cancer: a pooled analysis from the pancreatic cancer cohort consortium. Cancer Causes & Control, 24(1), 13–​25. PMCID: PMC3529822. Everhart J, and Wright D. 1995. Diabetes mellitus as a risk factor for pancreatic cancer: a meta-​analysis. JAMA, 273(20), 1605–​1609. PMID: 7745774. Falk RT, Pickle LW, Fontham ET, Correa P, and Fraumeni JF, Jr. 1988. Life-​ style risk factors for pancreatic cancer in Louisiana: a case-​control study. Am J Epidemiol, 128(2), 324–​336. PMID: 3394699. Farrell JJ, Zhang L, Zhou H, et al. 2012. Variations of oral microbiota are associated with pancreatic diseases including pancreatic cancer. Gut, 61(4), 582–​588. PMCID: PMC3705763. Feldman JM, Plonk JW, Bivens CH, and Lebovitz HE. 1975. Glucose intolerance in the carcinoid syndrome. Diabetes, 24(7), 664–​671. PMID: 125668. Ferlay J SI, Ervik M, Dikshit R, Eser S, Mathers C, Rebelo M, Parkin DM, Forman D, Bray F. 2013. GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality Worldwide:  IARC CancerBase No. 11. Lyon, France:  International Agency for Research on Cancer; 2013. Available from: http://​globocan.iarc.fr. Accessed October 5, 2016. Fernandez E, La Vecchia C, D’Avanzo B, Negri E, and Franceschi S. 1994. Family history and the risk of liver, gallbladder, and pancreatic cancer. Cancer Epidemiol Biomarkers Prev, 3(3), 209–​212. PMID: 8019368. Fiorino S, Chili E, Bacchi-​Reggiani L, et al. 2013. Association between hepatitis B or hepatitis C virus infection and risk of pancreatic adenocarcinoma development:  a systematic review and meta-​ analysis. Pancreatology, 13(2), 147–​160. PMID: 23561973. Fong PY, Fesinmeyer MD, White E, et al. 2010. Association of diabetes susceptibility gene calpain-​10 with pancreatic cancer among smokers. J Gastrointest Cancer, 41(3), 203–​208. PMCID: PMC2952047. Forman D, Burley V, Cade J, et al. 2007. The associations between food, nutrition and physical activity and the risk of pancreatic cancer and underlying mechanisms. World Cancer Res Fund. Food, nutrition, physical activity, and the prevention of cancer: a global perspective. Washington, DC: AICR. Forsmark CE. 2007. Pancreatitis and Its Complications. New York: Springer Science & Business Media. Friedman JM, and Fialkow PJ. 1976. Familial carcinoma of the pancreas. Clin Genet, 9(5), 463–​469. PMID: 1269168. Fryzek JP, Garabrant DH, Harlow SD, et  al. 1997. A case-​control study of self-​reported exposures to pesticides and pancreas cancer in southeastern Michigan. Int J Cancer, 72(1), 62–​67. PMID: 9212224. Fuchs CS, Colditz GA, Stampfer MJ, et al. 1996. A prospective study of cigarette smoking and the risk of pancreatic cancer. Arch Intern Med, 156(19), 2255–​2260. PMID: 8885826. Gallicchio L, Kouzis A, Genkinger JM, et al. 2006. Active cigarette smoking, household passive smoke exposure, and the risk of developing pancreatic cancer. Prev Med, 42(3), 200–​205. Gandini S, Lowenfels AB, Jaffee EM, Armstrong TD, and Maisonneuve P. 2005. Allergies and the risk of pancreatic cancer:  a meta-​analysis with review of epidemiology and biological mechanisms. Cancer Epidemiol Biomarkers Prev, 14(8), 1908–​1916. PMID: 16103436. Garcea G, Dennison AR, Steward WP, and Berry DP. 2005. Role of inflammation in pancreatic carcinogenesis and the implications for future therapy. Pancreatology, 5(6), 514–​529. PMID: 16110250. Garcia C, Feve B, Ferre P, et al. 2010. Diabetes and inflammation: fundamental aspects and clinical implications. Diabetes Metab, 36(5), 327–​338. PMID: 20851652. Gbelcova H, Lenicek M, Zelenka J, et al. 2008. Differences in antitumor effects of various statins on human pancreatic cancer. Int J Cancer, 122(6), 1214–​1221. PMID: 18027870.

629

Genkinger JM, Spiegelman D, Anderson KE, et al. 2009. Alcohol intake and pancreatic cancer risk: a pooled analysis of fourteen cohort studies. Cancer Epidemiol Biomarkers Prev, 18(3), 765–​776. PMCID: PMC2715951. Genkinger JM, Spiegelman D, Anderson KE, et al. 2011. A pooled analysis of 14 cohort studies of anthropometric factors and pancreatic cancer risk. Int J Cancer, 129(7), 1708–​1717. PMCID: PMC3073156. Gerken T, Girard CA, Tung Y-​CL, et al. 2007. The obesity-​associated FTO gene encodes a 2-​oxoglutarate-​dependent nucleic acid demethylase. Science, 318(5855), 1469–​1472. Ghadirian P, Simard A, and Baillargeon J. 1987. Cancer of the pancreas in two brothers and one sister. Int J Pancreatol, 2(5-​6), 383–​391. PMID: 3693982. Ghadirian P, Liu G, Gallinger S, et al. 2002. Risk of pancreatic cancer among individuals with a family history of cancer of the pancreas. Int J Cancer, 97(6), 807–​810. PMID: 11857359. Giardiello FM, Brensinger JD, Tersmette AC, et  al. 2000. Very high risk of cancer in familial Peutz-​Jeghers syndrome. Gastroenterology, 119(6), 1447–​1453. PMID: 11113065. Giovannucci E. 2007. Metabolic syndrome, hyperinsulinemia, and colon cancer: a review. Am J Clin Nutr, 86(3), s836–​s842. PMID: 18265477. Gold EB, and Goldin SB. 1998. Epidemiology of and risk factors for pancreatic cancer. Surg Oncol Clin N Am, 7(1), 67–​91. PMID: 9443987. Goldacre MJ, Abisgold JD, Seagroatt V, and Yeates D. 2005. Cancer after cholecystectomy: record-​linkage cohort study. Br J Cancer, 92(7), 1307–​1309. Goldstein AM, Struewing JP, Fraser MC, Smith MW, and Tucker MA. 2004. Prospective risk of cancer in CDKN2A germline mutation carriers. J Med Genet, 41(6), 421–​424. PMCID: PMC1735800. Grant RC, Selander I, Connor AA, et al. 2015. Prevalence of germline mutations in cancer predisposition genes in patients with pancreatic cancer. Gastroenterology, 148(3), 556–​564. PMCID: PMC4339623. Greer JB, and Whitcomb DC. 2009. Inflammation and pancreatic cancer:  an evidence-​based review. Curr Opin Pharmacol, 9(4), 411–​418. PMID: 19589727. Grutzmann R, Niedergethmann M, Pilarsky C, Kloppel G, and Saeger HD. 2010. Intraductal papillary mucinous tumors of the pancreas: biology, diagnosis, and treatment. Oncologist, 15(12), 1294–​1309. PMCID: PMC3227924. Gukovskaya AS, Mouria M, Gukovsky I, et  al. 2002. Ethanol metabolism and transcription factor activation in pancreatic acinar cells in rats. Gastroenterology, 122(1), 106–​118. PMID: 11781286. Gullo L, Migliori M, Falconi M, et al. 2003. Nonfunctioning pancreatic endocrine tumors:  a multicenter clinical study. Am J Gastroenterol, 98(11), 2435–​2439. PMID: 14638345. Gullo L, Pezzilli R, and Morselli-​Labate AM. 1994. Diabetes and the risk of pancreatic cancer. N Engl J Med, 331(2), 81–​84. Guo M, Jia Y, Yu Z, et al. 2014. Epigenetic changes associated with neoplasms of the exocrine and endocrine pancreas. Discovery Med, 17(92), 67. Haber PS, Apte MV, Moran C, et al. 2004. Non-​oxidative metabolism of ethanol by rat pancreatic acini. Pancreatology, 4(2), 82–​89. PMID: 15056978. Halfdanarson TR, Bamlet WR, McWilliams RR, et al. 2014. Risk factors for pancreatic neuroendocrine tumors:  a clinic-​ based case-​ control study. Pancreas, 43(8), 1219–​1222. PMCID: PMC4267883. Halfdanarson TR, Rabe KG, Rubin J, and Petersen GM. 2008a. Pancreatic neuroendocrine tumors (PNETs):  incidence, prognosis and recent trend toward improved survival. Ann Oncol, 19(10), 1727–​ 1733. PMCID: PMC2735065. Halfdanarson TR, Rubin J, Farnell MB, Grant CS, and Petersen GM. 2008b. Pancreatic endocrine neoplasms:  epidemiology and prognosis of pancreatic endocrine tumors. Endocr Relat Cancer, 15(2), 409–​ 427. PMCID: PMC2693313. Hallet J, Law CH, Cukier M, et  al. 2015. Exploring the rising incidence of neuroendocrine tumors:  a population-​ based analysis of epidemiology, metastatic presentation, and outcomes. Cancer, 121(4), 589–​597. PMID: 25312765. Hang B. 2010. Formation and repair of tobacco carcinogen-​derived bulky DNA adducts. J Nucleic Acids, 2010, 709521. PMCID: PMC3017938. Harley CB. 1997. Human ageing and telomeres. Ciba Found Symp, 211, 129–​ 139; discussion 139–​144. PMID: 9524755. Hashimoto Y, Murakami Y, Uemura K, et al. 2008. Telomere shortening and telomerase expression during multistage carcinogenesis of intraductal papillary mucinous neoplasms of the pancreas. J Gastrointest Surg, 12(1), 17–​28; discussion 28–​29. PMID: 17960465. Hassan MM, Abbruzzese JL, Bondy ML, et al. 2007. Passive smoking and the use of noncigarette tobacco products in association with risk for pancreatic cancer: a case-​control study. Cancer, 109(12), 2547–​2556. Hassan MM, Bondy ML, Wolff RA, et  al. 2007. Risk factors for pancreatic cancer:  case-​control study. Am J Gastroenterol, 102(12), 2696–​2707. PMCID: PMC2423805.

630

630

PART IV:  Cancers by Tissue of Origin

Hassan MM, Phan A, Li D, et al. 2008a. Family history of cancer and associated risk of developing neuroendocrine tumors:  a case-​control study. Cancer Epidemiol Biomarkers Prev, 17(4), 959–​965. PMID: 18398037. Hassan MM, Phan A, Li D, et al. 2008b. Risk factors associated with neuroendocrine tumors: A U.S.-​based case-​control study. Int J Cancer, 123(4), 867–​873. PMID: 18491401. Hauptmann N, and Cadenas E. 1997. The oxygen paradox:  biochemistry of active oxygen. Cold Spring Harbor Monogrh Arch, 34, 1–​20. Heard E, and Martienssen RA. 2014. Transgenerational epigenetic inheritance: myths and mechanisms. Cell, 157 95–​109. Hecht SS. 2003. Tobacco carcinogens, their biomarkers and tobacco-​induced cancer. Nat Rev Cancer, 3(10), 733–​744. Hein DW, Doll MA, Fretland AJ, et al. 2000. Molecular genetics and epidemiology of the NAT1 and NAT2 acetylation polymorphisms. Cancer Epidemiol Biomarkers Prev, 9(1), 29–​42. PMID: 10667461. Heinen MM, Verhage BA, Goldbohm RA, and van den Brandt PA. 2012. Intake of vegetables, fruits, carotenoids and vitamins C and E and pancreatic cancer risk in The Netherlands Cohort Study. Int J Cancer, 130(1), 147–​ 158. PMID: 21328344. Hering J, Garrean S, Dekoj TR, et  al. 2007. Inhibition of proliferation by omega-​3 fatty acids in chemoresistant pancreatic cancer cells. Ann Surg Oncol, 14(12), 3620–​3628. PMID: 17896154. Herman WH. 2007. Diabetes epidemiology: guiding clinical and public health practice:  the Kelly West Award Lecture, 2006. Diabetes Care, 30(7), 1912–​1919. PMID: 17496237. Hernandez AV, Pasupuleti V, Benites-​Zapata VA, et al. 2015. Insulin resistance and endometrial cancer risk: a systematic review and meta-​analysis. Eur J Cancer, 51(18), 2747–​2758. PMID: 26597445. Hinney A, Nguyen TT, Scherag A, et  al. 2007. Genome wide association (GWA) study for early onset extreme obesity supports the role of fat mass and obesity associated gene (FTO) variants. PloS One, 2(12), e1361. Hiripi E, Bermejo JL, Sundquist J, and Hemminki K. 2009. Familial gastrointestinal carcinoid tumours and associated cancers. Ann Oncol, 20(5), 950–​954. PMID: 19150948. Hirvonen A. 1999. Polymorphic NATs and cancer predisposition. IARC Sci Publ 148, 251–​270. PMID: 10493262. Hoffmann D, Hoffmann I, and El-​Bayoumy K. 2001. The less harmful cigarette:  a controversial issue. a tribute to Ernst L.  Wynder. Chem Res Toxicol, 14(7), 767–​790. PMID: 11453723. Hou L, Zhang X, Wang D, and Baccarelli A. 2012. Environmental chemical exposures and human epigenetics. Int J Epidemiol, 41(1), 79–​105. PMID: 22253299. PMCID: PMC3304523. Howes N, and Neoptolemos JP. 2002. Risk of pancreatic ductal adenocarcinoma in chronic pancreatitis. Gut, 51(6), 765–​766. PMCID: PMC1773485. Hruban R, Petersen GM, Ha P, and Kern S. 1998. Genetics of pancreatic cancer: from genes to families. Surg Oncol Clinics North Am, 7(1), 1–​23. Hruban RH, Adsay NV, Albores-​Saavedra J, et al. 2001. Pancreatic intraepithelial neoplasia: a new nomenclature and classification system for pancreatic duct lesions. Am J Surg Pathol, 25(5), 579–​586. PMID: 11342768. Hruban RH, Maitra A, and Goggins M. 2008. Update on pancreatic intraepithelial neoplasia. Int J Clin Exp Pathol, 1(4), 306–​ 316. PMCID: PMC2480542. Hruban RH, van Mansfeld AD, Offerhaus GJ, et al. 1993. K-​ras oncogene activation in adenocarcinoma of the human pancreas: a study of 82 carcinomas using a combination of mutant-​enriched polymerase chain reaction analysis and allele-​specific oligonucleotide hybridization. Am J Pathol, 143(2), 545–​554. PMICD: PMC1887038. Hu C, Hart SN, Bamlet WR, et al. 2016. Prevalence of pathogenic mutations in cancer predisposition genes among pancreatic cancer patients. Cancer Epidemiol Biomarkers Prev, 25(1), 207–​211. PMCID: PMC4754121. Hu Z-​H, Connett JE, Yuan J-​M, and Anderson KE. 2016. Role of survivor bias in pancreatic cancer case-​control studies. Ann Epidemiol, 26, 50–​56. PMID: 26688282. Huang JY, Butler LM, Wang R, et  al. 2016. Dietary intake of one-​carbon metabolism-​related nutrients and pancreatic cancer risk: The Singapore Chinese Health Study. Cancer Epidemiol Biomarkers Prev, 25, 417–​424. Humphris JL, Johns AL, Simpson SH, et  al. 2014. Clinical and pathologic features of familial pancreatic cancer. Cancer, 120(23), 3669–​ 3675. PMID: 25313458. Huxley R, Ansary-​ Moghaddam A, Berrington de Gonzalez A, Barzi F, and Woodward M. 2005. Type-​ II diabetes and pancreatic cancer:  a meta-​analysis of 36 studies. Br J Cancer, 92(11), 2076–​2083. PMCID: PMC2361795. Inoue M, Tajima K, Takezaki T, et  al. 2003. Epidemiology of pancreatic cancer in Japan:  a nested case-​control study from the Hospital-​based Epidemiologic Research Program at Aichi Cancer Center (HERPACC). Int J Epidemiol, 32(2), 257–​262. PMID: 12714546.

Iodice S, Gandini S, Maisonneuve P, and Lowenfels AB. 2008. Tobacco and the risk of pancreatic cancer: a review and meta-​analysis. Langenbeck’s Arch Surgery, 393(4), 535–​545. Iqbal J, Ragone A, Lubinski J, et al. 2012. The incidence of pancreatic cancer in BRCA1 and BRCA2 mutation carriers. Br J Cancer, 107(12), 2005–​2009. PMCID: PMC3516682. Jacobs EJ, Chanock SJ, Fuchs CS, et al. 2010. Family history of cancer and risk of pancreatic cancer: a pooled analysis from the Pancreatic Cancer Cohort Consortium (PanScan). Int J Cancer, 127(6), 1421–​ 1428. PMCID: PMC2926939. Jang JH, Cotterchio M, Borgida A, Gallinger S, and Cleary SP. 2012. Genetic variants in carcinogen-​ metabolizing enzymes, cigarette smoking and pancreatic cancer risk. Carcinogenesis, 33(4), 818–​ 827. PMCID: PMC3324443. Jansen RJ, Robinson DP, Frank RD, et  al. 2013a. Meat-​ related mutagens and pancreatic cancer:  null results from a clinic-​ based case-​ control study. Cancer Epidemiol Biomarkers Prev, 22(7), 1336–​1339. PMCID: PMC3702664. Jansen RJ, Robinson DP, Frank RD, et  al. 2014. Fatty acids found in dairy, protein and unsaturated fatty acids are associated with risk of pancreatic cancer in a case-​control study. Int J Cancer, 134(8), 1935–​1946. PMCID: PMC3942799. Jansen RJ, Robinson DP, Stolzenberg-​Solomon RZ, et al. 2011. Fruit and vegetable consumption is inversely associated with having pancreatic cancer. Cancer Causes Control, 22(12), 1613–​1625. PMCID: PMC3522747. Jansen RJ, Robinson DP, Stolzenberg-​ Solomon RZ, et  al. 2013b. Polymorphisms in metabolism/​antioxidant genes may mediate the effect of dietary intake on pancreatic cancer risk. Pancreas, 42(7), 1043–​1053. PMCID: PMC3779344. Jemal A, Simard EP, Dorell C, et  al. 2013. Annual Report to the Nation on the Status of Cancer, 1975–​2009, Featuring the Burden and Trends in Human Papillomavirus (HPV)–​Associated Cancers and HPV Vaccination Coverage Levels. JNCI, 105(3), 175–​201. Jennings BJ, Ozanne SE, and Hales CN. 2000. Nutrition, oxidative damage, telomere shortening, and cellular senescence:  individual or connected agents of aging? Mol Genet Metab, 71(1–​2), 32–​42. Jiao L, Bondy ML, Hassan MM, et  al. 2007. Glutathione S-​transferase gene polymorphisms and risk and survival of pancreatic cancer. Cancer, 109(5), 840–​848. PMCID: PMC1892189. Jiao L, Hassan MM, Bondy ML, et al. 2008. XRCC2 and XRCC3 gene polymorphism and risk of pancreatic cancer. Am J Gastroenterol, 103(2), 360–​367. PMCID: PMC2268638. Jones S, Hruban RH, Kamiyama M, et al. 2009. Exomic sequencing identifies PALB2 as a pancreatic cancer susceptibility gene. Science, 324(5924), 217. PMCID: PMC2684332. Jones S, Zhang X, Parsons DW, et  al. 2008. Core signaling pathways in human pancreatic cancers revealed by global genomic analyses. Science, 321(5897), 1801–​1806. PMCID: PMC2848990. Kaaks R, Lukanova A, and Kurzer MS. 2002. Obesity, endogenous hormones, and endometrial cancer risk:  a synthetic review. Cancer Epidemiol Biomarkers Prev, 11(12), 1531–​1543. PMID: 12496040. Kaerlev L, Teglbjaerg PS, Sabroe S, et al. 2002. The importance of smoking and medical history for development of small bowel carcinoid tumor: a European population-​based case-​control study. Cancer Causes Control, 13(1), 27–​34. PMID: 11899115. Kahn BB, and Flier JS. 2000. Obesity and insulin resistance. J Clin Invest, 106(4), 473–​481. PMCID: PMC380258. Karlson BM, Ekbom A, Josefsson S, et  al. 1997. The risk of pancreatic cancer following pancreatitis:  an association due to confounding? Gastroenterology, 113(2), 587–​592. PMID: 9247480. Kastrinos F, Mukherjee B, Tayob N, et al. 2009. Risk of pancreatic cancer in families with Lynch syndrome. Jama, 302(16), 1790–​ 1795. PMCID: PMC4091624. Khawja SN, Mohammed S, Silberfein EJ, et al. 2015. Pancreatic cancer disparities in African Americans. Pancreas, 44(4), 522–​527. PMID: 25872128. Klein AP. 2012. Genetic susceptibility to pancreatic cancer. Mol Carcinog, 51(1), 14–​24. PMCID: PMC3570154. Klein AP, Beaty TH, Bailey-​Wilson JE, et al. 2002. Evidence for a major gene influencing risk of pancreatic cancer. Genet Epidemiol, 23(2), 133–​149. PMID: 12214307. Klöting N, Schleinitz D, Ruschke K, et al. 2008. Inverse relationship between obesity and FTO gene expression in visceral adipose tissue in humans. Diabetologia, 51(4), 641–​647. Kokawa A, Kondo H, Gotoda T, et  al. 2001. Increased expression of cyclooxygenase-​2 in human pancreatic neoplasms and potential for chemoprevention by cyclooxygenase inhibitors. Cancer, 91(2), 333–​338. PMID: 11180079.

 631



Cancer of the Pancreas

Kollarova H, Azeem K, Tomaskova H, et al. 2014. Is physical activity a protective factor against pancreatic cancer? Bratisl Lek Listy, 115(8), 474–​478. PMID: 25246281. Korsse SE, Harinck F, van Lier MG, et al. 2013. Pancreatic cancer risk in Peutz-​ Jeghers syndrome patients: a large cohort study and implications for surveillance. J Med Genet, 50(1), 59–​64. PMID: 23240097. Kovi J, and Heshmat MY. 1972. Incidence of cancer in negroes in Washington, D.C.  and selected African cities. Am J Epidemiol, 96(6), 401–​ 413. PMID: 4643672. Kowalska I, Straczkowski M, Nikolajuk A, et al. 2008. Insulin resistance, serum adiponectin, and proinflammatory markers in young subjects with the metabolic syndrome. Metabolism, 57(11), 1539–​1544. PMID: 18940391. Krain LS. 1970. The rising incidence of carcinoma of the pancreas:  real or apparent? J Surg Oncol, 2(2), 115–​124. PMID: 5520829. Kusama T, Mukai M, Iwasaki T, et  al. 2002. 3-​hydroxy-​3-​methylglutaryl-​ coenzyme a reductase inhibitors reduce human pancreatic cancer cell invasion and metastasis. Gastroenterology, 122(2), 308–​ 317. PMID: 11832446. Lakhani VT, You YN, and Wells SA. 2007. The multiple endocrine neoplasia syndromes. Annu Rev Med, 58, 253–​265. PMID: 17037976. Larsson SC, Giovannucci E, Bergkvist L, and Wolk A. 2006. Aspirin and nonsteroidal anti-​inflammatory drug use and risk of pancreatic cancer:  a meta-​analysis. Cancer Epidemiol Biomarkers Prev, 15(12), 2561–​2564. PMID: 17164387. Larsson SC, Orsini N, and Wolk A. 2007. Body mass index and pancreatic cancer risk: a meta-​analysis of prospective studies. Int J Cancer, 120(9), 1993–​1998. PMID: 17266034. Lawrence B, Gustafsson BI, Chan A, et al. 2011. The epidemiology of gastroenteropancreatic neuroendocrine tumors. Endocrinol Metab Clin North Am, 40(1), 1–​18, vii. PMID: 21349409. Levitzky YS, Guo CY, Rong J, et al. 2008. Relation of smoking status to a panel of inflammatory markers:  the framingham offspring. Atherosclerosis, 201(1), 217–​224. PMCID: PMC2783981. Li D. 2012. Diabetes and pancreatic cancer. Mol Carcinog, 51(1), 64–​74. PMCID: PMC3238796. Li D, Day RS, Bondy ML, et al. 2007. Dietary mutagen exposure and risk of pancreatic cancer. Cancer Epidemiol Biomarkers Prev, 16(4), 655–​661. PMCID: PMC1892159. Li D, Jiao L, Li Y, et al. 2005. Polymorphisms of cytochrome P4501A2 and N-​ acetyltransferase genes, smoking, and risk of pancreatic cancer. Carcinogenesis, 27(1), 103–​111. Li D, Tang H, Hassan MM, et al. 2011. Diabetes and risk of pancreatic cancer: a pooled analysis of three large case-​control studies. Cancer Causes Control, 22(2), 189–​197. PMID: 21104117. Li D, Yeung SC, Hassan MM, Konopleva M, and Abbruzzese JL. 2009. Antidiabetic therapies affect risk of pancreatic cancer. Gastroenterology, 137(2), 482–​488. PMCID: PMC2735093. Li J, Zhu J, Hassan MM, et al. 2007. K-​ras mutation and p16 and preproenkephalin promoter hypermethylation in plasma DNA of pancreatic cancer patients:  in relation to cigarette smoking. Pancreas, 34(1), 55–​62. PMCID: PMC1905887. Lichtenstein P, Holm NV, Verkasalo PK, et al. 2000. Environmental and heritable factors in the causation of cancer:  analyses of cohorts of twins from Sweden, Denmark, and Finland. N Engl J Med, 343(2), 78–​85. PMID: 10891514. Lin Y, Yagyu K, Egawa N, et  al. 2011. An overview of genetic polymorphisms and pancreatic cancer risk in molecular epidemiologic studies. J Epidemiol, 21(1), 2–​12. PMCID: PMC3899511. Lindkvist B, Johansen D, Borgstrom A, and Manjer J. 2008. A prospective study of Helicobacter pylori in relation to the risk for pancreatic cancer. BMC Cancer, 8, 321. PMCID: PMC2613155. Liu G, Ghadirian P, Vesprini D, et al. 2000. Polymorphisms in GSTM1, GSTT1 and CYP1A1 and risk of pancreatic adenocarcinoma. Br J Cancer, 82(10), 1646–​1649. PMCID: PMC2374522. Lo AC, Soliman AS, El-​Ghawalby N, et al. 2007. Lifestyle, occupational, and reproductive factors in relation to pancreatic cancer risk. Pancreas, 35(2), 120–​129. PMID: 17632317. Lonser RR, Glenn GM, Walther M, et  al. 2003. von Hippel-​Lindau disease. Lancet, 361(9374), 2059–​2067. PMID: 12814730. Luo J, Margolis KL, Adami HO, et  al. 2008. Obesity and risk of pancreatic cancer among postmenopausal women:  the Women’s Health Initiative (United States). Br J Cancer, 99(3), 527–​531. PMCID:  PMC2527801. PMID: 18628761 Lowenfels AB, and Maisonneuve P. 2004. Epidemiology and prevention of pancreatic cancer. Jpn J Clin Oncol, 34(5), 238–​244. Lowenfels AB, and Maisonneuve P. 2006. Epidemiology and risk factors for pancreatic cancer. Best Pract Res Clin Gastroenterol, 20(2), 197–​209. PMID: 16549324.

631

Lowenfels AB, Maisonneuve P, and Whitcomb DC. 2000. Risk factors for cancer in hereditary pancreatitis. International Hereditary Pancreatitis Study Group. Med Clin North Am, 84(3), 565–​573. PMID: 10872414. Lowenfels AB, Maisonneuve P, Cavallini G, et al. 1993. Pancreatitis and the risk of pancreatic cancer. International Pancreatitis Study Group. N Engl J Med, 328(20), 1433–​1437. PMID: 8479461. Lowenfels AB, Maisonneuve P, DiMagno EP, et al. 1997. Hereditary pancreatitis and the risk of pancreatic cancer. International Hereditary Pancreatitis Study Group. JNCI, 89(6), 442–​446. PMID: 9091646. Lynch SM, Major JM, Cawthon R, et al. 2013. A prospective analysis of telomere length and pancreatic cancer in the alpha-​tocopherol beta-​carotene cancer (ATBC) prevention study. Int J Cancer, 133(11), 2672–​2680. PMID: 23674344. Lynch SM, Vrieling A, Lubin JH, et al. 2009. Cigarette smoking and pancreatic cancer: a pooled analysis from the pancreatic cancer cohort consortium. Am J Epidemiol, 170(4), 403–​413. Ma J, Siegel R, and Jemal A. 2013. Pancreatic cancer death rates by race among US men and women, 1970–​ 2009. JNCI, 105(22):1694–​1700. PMID: 24203988. MacDermott RP, and Kramer P. 1973. Adenocarcinoma of the pancreas in four siblings. Gastroenterology, 65(1), 137–​139. PMID: 4720820. Magruder JT, Elahi D, and Andersen DK. 2011. Diabetes and pancreatic cancer: chicken or egg? Pancreas, 40(3), 339–​351. PMID: 21412116. Maisonneuve P, Lowenfels AB, Bueno-​de-​Mesquita HB, et al. 2010. Past medical history and pancreatic cancer risk:  results from a multicenter case-​ control study. Ann Epidemiol, 20(2), 92–​98. PMID: 20123159. Maitra A, and Hruban RH. 2008. Pancreatic cancer. Annu Rev Pathol, 3, 157–​ 188. PMCID: PMC2666336. Majumder S, Bockorny B, Baker WL, and Dasanu CA. 2014. Association between HBsAg positivity and pancreatic cancer:  a meta-​ analysis. J Gastrointest Cancer, 45(3), 347–​352. PMID: 24788082. McCarty MF. 2001. Insulin secretion as a determinant of pancreatic cancer risk. Med Hypotheses, 57(2), 146–​150. PMID: 11461162. McKeown-​Eyssen G. 1994. Epidemiology of colorectal cancer revisited:  are serum triglycerides and/​or plasma glucose associated with risk? Cancer Epidemiol Biomarkers Prev, 3(8), 687–​695. PMID: 7881343. McWilliams RR, Rabe KG, Olswold C, De Andrade M, and Petersen GM. 2005. Risk of malignancy in first-​degree relatives of patients with pancreatic carcinoma. Cancer, 104(2), 388–​394. PMID: 15912495. McWilliams RR, Bamlet WR, Cunningham JM, et al. 2008. Polymorphisms in DNA repair genes, smoking, and pancreatic adenocarcinoma risk. Cancer Res, 68(12), 4928–​4935. McWilliams RR, Petersen GM, Rabe KG, et al. 2010. Cystic fibrosis transmembrane conductance regulator (CFTR) gene mutations and risk for pancreatic adenocarcinoma. Cancer, 116(1), 203–​209. PMCID: PMC2807917. McWilliams RR, Wieben ED, Rabe KG, et al. 2011. Prevalence of CDKN2A mutations in pancreatic cancer patients: implications for genetic counseling. Eur J Hum Genet, 19(4), 472–​478. PMCID: PMC3060321. Mettu NB, and Abbruzzese JL. 2016. Clinical insights into the biology and treatment of pancreatic cancer. J Oncol Pract, 12(1), 17–​23. PMID: 26759461. Michaud DS, Giovannucci E, Willett WC, et al. 2001. Physical activity, obesity, height, and the risk of pancreatic cancer. JAMA, 286(8), 921–​929. PMID: 11509056. Michaud DS, Joshipura K, Giovannucci E, and Fuchs CS. 2007. A prospective study of periodontal disease and pancreatic cancer in US male health professionals. JNCI, 99(2), 171–​175. PMID: 17228001. Michaud DS, Vrieling A, Jiao L, et  al. 2010. Alcohol intake and pancreatic cancer:  a pooled analysis from the pancreatic cancer cohort consortium (PanScan). Cancer Causes Control, 21(8), 1213–​ 1225. PMCID: PMC3098295. Min Y, Adachi Y, Yamamoto H, et al. 2003. Genetic blockade of the insulin-​like growth factor-​I receptor: a promising strategy for human pancreatic cancer. Cancer Res, 63(19), 6432–​6441. PMID: 14559833. Mohelnikova-​Duchonova B, Vrana D, Holcatova I, et  al. 2010. CYP2A13, ADH1B, and ADH1C gene polymorphisms and pancreatic cancer risk. Pancreas, 39(2), 144–​148. Molina MA, Sitja-​Arnau M, Lemoine MG, Frazier ML, and Sinicrope FA. 1999. Increased cyclooxygenase-​2 expression in human pancreatic carcinomas and cell lines: growth inhibition by nonsteroidal anti-​inflammatory drugs. Cancer Res, 59(17), 4356–​4362. PMID: 10485483. Muscat JE, Stellman SD, Hoffmann D, and Wynder EL. 1997. Smoking and pancreatic cancer in men and women. Cancer Epidemiol Biomarkers Prev, 6(1), 15–​19. Nagy C, and Turecki G. 2015. Transgenerational epigenetic inheritance:  an open discussion. Epigenomics, 7, 781–​790. PMID: 26344807. Nakao M, Hosono S, Ito H, et al. 2011. Interaction between IGF-​1 polymorphisms and overweight for the risk of pancreatic cancer in Japanese. Int J Mol Epidemiol Genet, 2(4), 354–​366. PMCID: PMC3243451.

632

632

PART IV:  Cancers by Tissue of Origin

Nilsen TIL, and Vatten LJ. 2000. A prospective study of lifestyle factors and the risk of pancreatic cancer in Nord-​Trøndelag, Norway. Cancer Causes Control, 11(7), 645–​652. Nones K, Waddell N, Song S, et al. 2014. Genome-​wide DNA methylation patterns in pancreatic ductal adenocarcinoma reveal epigenetic deregulation of SLIT-​ROBO, ITGA2 and MET signaling. Int J Cancer, 135(5), 1110–​1118. Nöthlings U, Wilkens LR, Murphy SP, et  al. 2005. Meat and fat intake as risk factors for pancreatic cancer: The Multiethnic Cohort Study. JNCI, 97(19), 1458–​1465. PMID: 16204695. Ojajarvi IA, Partanen TJ, Ahlbom A, et al. 2000. Occupational exposures and pancreatic cancer: a meta-​analysis. Occup Environ Med, 57(5), 316–​324. PMCID: PMC1739949. Okuda K, Bardeguez A, Gardner JP, et al. 2002. Telomere length in the newborn. Pediatr Res, 52(3), 377–​381. PMID: 12193671. Olson SH. 2012. Selected medical conditions and risk of pancreatic cancer. Mol Carcinog, 51(1), 75–​97. PMID: 22162233. Olson SH, Hsu M, Satagopan JM, et  al. 2013. Allergies and risk of pancreatic cancer: a pooled analysis from the Pancreatic Cancer Case-​Control Consortium. Am J Epidemiol, 178(5), 691–​700. PMCID: PMC3755648. O’Rorke MA, Cantwell MM, Cardwell CR, Mulholland HG, and Murray LJ. 2010. Can physical activity modulate pancreatic cancer risk? a systematic review and meta-​analysis. Int J Cancer, 126(12), 2957–​2968. PMID: 19856317. Ough M, Lewis A, Zhang Y, et al. 2004. Inhibition of cell growth by overexpression of manganese superoxide dismutase (MnSOD) in human pancreatic carcinoma. Free Radical Res, 38(11), 1223–​1233. Pannala R, Basu A, Petersen GM, and Chari ST. 2009. New-​onset diabetes: a potential clue to the early diagnosis of pancreatic cancer. Lancet Oncol, 10(1), 88–​95. PMCID: PMC2795483. Park C-​H, Lee I-​S, Grippo P, et al. 2013. Akt kinase mediates the pro-​survival effect of smoking compounds in pancreatic ductal cells. Pancreas, 42, 655–​662. PMID: 23271397. Parkin D, Whelan S, Ferlay J, Teppo L, and Thomas D. 2002. Cancer Incidence in Five Continents, Vol. VIII. Lyon, France:  IARC Scientific Publications, 155. Parkin DM, Boyd L, and Walker LC. 2011. 16. The fraction of cancer attributable to lifestyle and environmental factors in the UK in 2010. Br J Cancer, 105 Suppl 2, S77–​S81. PMCID: PMC3252065. Patel AV, Rodriguez C, Bernstein L, et al. 2005. Obesity, recreational physical activity, and risk of pancreatic cancer in a large U.S. Cohort. Cancer Epidemiol Biomarkers Prev, 14(2), 459–​466. PMID: 15734973. Permuth-​Wey J, and Egan KM. 2009. Family history is a significant risk factor for pancreatic cancer: results from a systematic review and meta-​analysis. Fam Cancer, 8(2), 109–​117. PMID: 18763055. Perugini RA, McDade TP, Vittimberga FJ, Jr, Duffy AJ, and Callery MP. 2000. Sodium salicylate inhibits proliferation and induces G1 cell cycle arrest in human pancreatic cancer cell lines. J Gastrointest Surg, 4(1), 24–​32, discussion 32–​23. PMID: 10631359. Petersen GM, Amundadottir L, Fuchs CS, et al. 2010. A genome-​wide association study identifies pancreatic cancer susceptibility loci on chromosomes 13q22.1, 1q32.1 and 5p15.33. Nat Genet, 42(3), 224–​228. PMCID: PMC2853179. Petersen GM, de Andrade M, Goggins M, et al. 2006. Pancreatic cancer genetic epidemiology consortium. Cancer Epidemiol Biomarkers Prev, 15(4), 704–​710. PMID: 16614112. Petersen GM. 2015. Familial pancreatic adenocarcinoma. Hematol Oncol Clin North Am, 29(4), 641–​653. PMCID: PMC4522044. Potjer TP, Kranenburg HE, Bergman W, et al. 2015. Prospective risk of cancer and the influence of tobacco use in carriers of the p16-​Leiden germline variant. Eur J Hum Genet, 23(5), 711–​714. PMCID: PMC4402641. Pradhan AD, Manson JE, Rifai N, Buring JE, and Ridker PM. 2001. C-​reactive protein, interleukin 6, and risk of developing type 2 diabetes mellitus. JAMA, 286(3), 327–​334. PMID: 11466099. Prokopczyk B, Hoffmann D, Bologna M, et al. 2002. Identification of tobacco-​ derived compounds in human pancreatic juice. Chem Res Toxicol, 15(5), 677–​685. Purohit V, Bode JC, Bode C, et al. 2008. Alcohol, intestinal bacterial growth, intestinal permeability to endotoxin, and medical consequences:  summary of a symposium. Alcohol, 42(5), 349–​361. PMCID: PMC2614138. Raderer M, Wrba F, Kornek G, et al. 1998. Association between Helicobacter pylori infection and pancreatic cancer. Oncology, 55(1), 16–​ 19. PMID: 9428370. Rahman F, Cotterchio M, Cleary SP, and Gallinger S. 2015. Association between alcohol consumption and pancreatic cancer risk: a case-​control study. PLoS One, 10(4), e0124489. PMCID: PMC4391718. Raimondi S, Lowenfels AB, Morselli-​Labate AM, Maisonneuve P, and Pezzilli R. 2010. Pancreatic cancer in chronic pancreatitis; aetiology, incidence,

and early detection. Best Pract Res Clin Gastroenterol, 24(3), 349–​358. PMID: 20510834. Raimondi S, Maisonneuve P, Lohr JM, and Lowenfels AB. 2007. Early onset pancreatic cancer:  evidence of a major role for smoking and genetic factors. Cancer Epidemiol Biomarkers Prev, 16(9), 1894–​1897. PMID: 17855711. Rebours V, Boutron-​Ruault MC, Schnee M, et al. 2008. Risk of pancreatic adenocarcinoma in patients with hereditary pancreatitis: a national exhaustive series. Am J Gastroenterol, 103(1), 111–​119. PMID: 18184119. Renehan AG, Tyson M, Egger M, Heller RF, and Zwahlen M. 2008. Body-​mass index and incidence of cancer:  a systematic review and meta-​analysis of prospective observational studies. Lancet, 371(9612), 569–​ 578. PMID: 18280327. Reports of the US Surgeon General. 2014. The Health Consequences of Smoking—​50 Years of Progress: A Report of the Surgeon General. Atlanta, GA: US Department of Health and Human Services, Centers for Disease Control and Prevention, Coordinating Center for Health Promotion, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. Available at http://​www.surgeongeneral. gov/​library/​reports/​50-​years-​of-​progress/​full-​report.pdf. Resta N, Pierannunzio D, Lenato GM, et  al. 2013. Cancer risk associated with STK11/​ LKB1 germline mutations in Peutz-​ Jeghers syndrome patients:  results of an Italian multicenter study. Dig Liver Dis, 45(7), 606–​611. PMID: 23415580. Risch HA. 2003. Etiology of pancreatic cancer, with a hypothesis concerning the role of N-​nitroso compounds and excess gastric acidity. JNCI, 95(13), 948–​960. Risch HA, Yu H, Lu L, and Kidd MS. 2010. ABO blood group, Helicobacter pylori seropositivity, and risk of pancreatic cancer: a case-​control study. JNCI, 102(7), 502–​505. PMCID: PMC2902822. Rivenson A, Hoffmann D, Prokopczyk B, Amin S, and Hecht SS. 1988. Induction of lung and exocrine pancreas tumors in F344 rats by tobacco-​specific and Areca-​derived N-​nitrosamines. Cancer Res, 48(23), 6912–​6917. Roberts NJ, Jiao Y, Yu J, et  al. 2012. ATM mutations in patients with hereditary pancreatic cancer. Cancer Discov, 2(1), 41–​46. PMCID: PMC3676748. Roberts NJ, Norris AL, Petersen GM, et al. 2016. Whole genome sequencing defines the genetic heterogeneity of familial pancreatic cancer. Cancer Discov, 6(2), 166–​175. PMCID: PMC4744563. Rosato V, Polesel J, Bosetti C, et  al. 2015. Population attributable risk for pancreatic cancer in Northern Italy. Pancreas, 44(2), 216–​ 220. PMID: 25479588. Ryan DP, Hong TS, and Bardeesy N. 2014. Pancreatic adenocarcinoma. N Engl J Med, 371(11), 1039–​1049. PMID: 25207767. Scheipers P, and Reiser H. 1998. Fas-​independent death of activated CD4+ T lymphocytes induced by CTLA-​4 crosslinking. Proc Natl Acad of Sci, 95(17), 10083–​10088. Schenk M, Schwartz AG, O’Neal E, et  al. 2001. Familial risk of pancreatic cancer. J Natl Cancer Inst, 93(8), 640–​644. PMID: 11309441. Schernhammer ES, Kang JH, Chan AT, et al. 2004. A prospective study of aspirin use and the risk of pancreatic cancer in women. JNCI, 96(1), 22–​28. PMID: 14709735. Schernhammer ES, Michaud DS, Leitzmann MF, et al. 2002. Gallstones, cholecystectomy, and the risk for developing pancreatic cancer. Br J Cancer, 86(7), 1081–​1084. PMCID: PMC2364180. Schulte A, Pandeya N, Fawcett J, et  al. 2015. Association between helicobacter pylori and pancreatic cancer risk: a meta-​analysis. Cancer Causes Control, 26(7), 1027–​1035. PMID: 25951801. Scott LJ, Mohlke KL, Bonnycastle LL, et al. 2007. A genome-​wide association study of type 2 diabetes in Finns detects multiple susceptibility variants. Science, 316(5829), 1341–​1345. Shen Q, Tian Y, Li K, et al. 2015. Association of single nucleotide polymorphisms of DNA repair gene and susceptibility to pancreatic cancer. Int J Clin Exp Pathol, 8(3), 3180–​3185. PMCID: PMC4440146. Shirts BH, Burt RW, Mulvihill SJ, and Cannon-​ Albright LA. 2010. A population-​based description of familial clustering of pancreatic cancer. Clin Gastroenterol Hepatol, 8(9), 812–​816. PMID: 20570637. Siegel RL, Miller KD, and Jemal A. 2015. Cancer statistics, 2015. CA Cancer J Clin, 65(1), 5–​29. Silverman DT, Hoover RN, Brown LM, et al. 2003. Why do Black Americans have a higher risk of pancreatic cancer than White Americans? Epidemiology, 14(1), 45–​54. PMID: 12500045. Silverman DT, Schiffman M, Everhart J, et  al. 1999. Diabetes mellitus, other medical conditions and familial history of cancer as risk factors for pancreatic cancer. Br J Cancer, 80(11), 1830–​1837. PMCID: PMC2363127.

 63



Cancer of the Pancreas

Skinner HG, Gangnon RE, Litzelman K, et al. 2012. Telomere length and pancreatic cancer: a case-​control study. Cancer Epidemiol Biomarkers Prev, 21(11), 2095–​2100. PMCID: PMC3493789. Song S, Wang B, Zhang X, et al. 2015. Long-​term diabetes mellitus is associated with an increased risk of pancreatic cancer: a meta-​analysis. PLoS One, 10(7), e0134321. PMCID: PMC4519136. Sponsiello-​Wang Z, Weitkunat R, and Lee PN. 2008. Systematic review of the relation between smokeless tobacco and cancer of the pancreas in Europe and North America. BMC Cancer, 8, 356–​356. PMCID: PMC2612688. Stevens RJ, Roddam AW, and Beral V. 2007. Pancreatic cancer in type 1 and young-​onset diabetes: systematic review and meta-​analysis. Br J Cancer, 96(3), 507–​509. PMCID: PMC2360024. Stolzenberg-​Solomon RZ, Adams K, Leitzmann M, et  al. 2008. Adiposity, physical activity, and pancreatic cancer in the National Institutes of Health-​AARP Diet and Health Cohort. Am J Epidemiol, 167(5), 586–​597. PMID: 18270373. Stolzenberg-​Solomon RZ, Blaser MJ, Limburg PJ, et  al. 2001. Helicobacter pylori seropositivity as a risk factor for pancreatic cancer. JNCI, 93(12), 937–​941. PMID: 11416115. Stolzenberg-​Solomon RZ, Cross AJ, Silverman DT, et al. 2007. Meat and meat-​ mutagen intake and pancreatic cancer risk in the NIH-​AARP cohort. Cancer Epidemiol Biomarkers Prev, 16(12), 2664–​2675. PMID: 18086772. Stolzenberg-​Solomon RZ, Dodd KW, Blaser MJ, et al. 2003. Tooth loss, pancreatic cancer, and Helicobacter pylori. Am J Clin Nutr, 78(1), 176–​181. PMID: 12816788. Stolzenberg-​Solomon RZ, Graubard BI, Chari S, et  al. 2005. Insulin, glucose, insulin resistance, and pancreatic cancer in male smokers. JAMA, 294(22), 2872–​2878. PMID: 16352795. Stolzenberg-​Solomon RZ, Pietinen P, Taylor PR, et  al. 2002a. A prospective study of medical conditions, anthropometry, physical activity, and pancreatic cancer in male smokers (Finland). Cancer Causes Control, 13(5), 417–​426. PMID: 12146846. Stolzenberg-​Solomon RZ, Pietinen P, Taylor PR, et  al. 2002b. Prospective study of diet and pancreatic cancer in male smokers. Am J Epidemiol, 155(9):783–​792. PMID: 11978580. Streicher SA, Yu H, Lu L, Kidd MS, and Risch HA. 2014. Case-​control study of aspirin use and risk of pancreatic cancer. Cancer Epidemiol Biomarkers Prev, 23(7), 1254–​1263. PMCID: PMC4091763. Strosberg JR, Cheema A, Weber J, et al. 2011. Prognostic validity of a novel American Joint Committee on Cancer Staging Classification for pancreatic neuroendocrine tumors. J Clin Oncol, 29(22), 3044–​ 3049. PMID: 21709192. Sumi S, Beauchamp RD, Townsend CM, Jr, et al. 1992. Inhibition of pancreatic adenocarcinoma cell growth by lovastatin. Gastroenterology, 103(3), 982–​989. PMID: 1499946. Surveillance Epidemiology and End Results (SEER) Program Fact Sheets: Pancreas Cancer; 2016. http://​seer.cancer.gov/​canques/​incidence.html. Accessed September 15, 2016. Suzuki H, Li Y, Dong X, et al. 2008a. Effect of insulin-​like growth factor gene polymorphisms alone or in interaction with diabetes on the risk of pancreatic cancer. Cancer Epidemiol Biomarkers Prev, 17(12), 3467–​3473. PMCID: PMC2600618. Suzuki H, Morris JS, Li Y, et al. 2008. Interaction of the cytochrome P4501A2, SULT1A1 and NAT gene polymorphisms with smoking and dietary mutagen intake in modification of the risk of pancreatic cancer. Carcinogenesis, 29(6), 1184–​1191. Suzuki H, Morris JS, Li Y, et  al. 2008b. Interaction of the cytochrome P4501A2, SULT1A1 and NAT gene polymorphisms with smoking and dietary mutagen intake in modification of the risk of pancreatic cancer. Carcinogenesis, 29(6), 1184–​1191. PMCID: PMC2443278. Talamini G, Falconi M, Bassi C, et  al. 1999. Incidence of cancer in the course of chronic pancreatitis. Am J Gastroenterol, 94(5), 1253–​1260. PMID: 10235203. Tan XL, Reid Lombardo KM, Bamlet WR, et al. 2011. Aspirin, nonsteroidal anti-​inflammatory drugs, acetaminophen, and pancreatic cancer risk:  a clinic-​based case-​control study. Cancer Prev Res (Phila), 4(11), 1835–​ 1841. PMCID: PMC3208748. Tang H, Dong X, Day RS, Hassan MM, and Li D. 2010. Antioxidant genes, diabetes and dietary antioxidants in association with risk of pancreatic cancer. Carcinogenesis, 31(4):607–​613.3 PMCID: PMC2847085. Tang H, Dong X, Hassan MM, Abbruzzese JL, and Li D. 2011. Body mass index and obesity-​ and diabetes-​ associated genes and risk for pancreatic cancer. Cancer Epidemiol Biomarkers Prev, 20(5):779–​792. PMCID: PMC3089680. Tang H, Wei P, Duell EJ, et al. 2014a. Axonal guidance signaling pathway interacting with smoking in modifying the risk of pancreatic cancer: a gene-​and

633

pathway-​based interaction analysis of GWAS data. Carcinogenesis, 35, 1039–​1045. PMID: 24419231. PMCID: PMC4004205. Tang H, Wei P, Duell EJ, et  al. 2014b. Genes-​environment interactions in obesity-​and diabetes-​associated pancreatic cancer:  a GWAS data analysis. Cancer Epidemiol Biomarkers Prev, 23, 98–​106. PMID: 24136929. Thiebaut AC, Jiao L, Silverman DT, et al. 2009. Dietary fatty acids and pancreatic cancer in the NIH-​AARP diet and health study. JNCI, 101(14), 1001–​1011. PMCID: PMC2724851. Thompson PA, Seyedi F, Lang NP, et al. 1999. Comparison of DNA adduct levels associated with exogenous and endogenous exposures in human pancreas in relation to metabolic genotype. Mutat Res, 424(1–​2), 263–​274. Torre LA, Bray F, Siegel RL, et al. 2015. Global cancer statistics, 2012. CA Cancer J Clin, 65(2), 87–​108. Tramacere I, Scotti L, Jenab M, et al. 2010. Alcohol drinking and pancreatic cancer risk: a meta-​analysis of the dose-​risk relation. Int J Cancer, 126(6), 1474–​1486. PMID: 19816941. Tranah GJ, Holly EA, Wang F, and Bracci PM. 2011. Cigarette, cigar and pipe smoking, passive smoke exposure, and risk of pancreatic cancer: a population-​based study in the San Francisco Bay Area. BMC Cancer, 11(1), 138. Trikudanathan G, Philip A, Dasanu CA, and Baker WL. 2011. Association between Helicobacter pylori infection and pancreatic cancer: a cumulative meta-​analysis. JOP, 12(1), 26–​31. PMID: 21206097. Ura H, Makino T, Ito S, et  al. 1986. Combined effects of cholecystectomy and lithocholic acid on pancreatic carcinogenesis of N-​ nitrosobis(2-​ hydroxypropyl)amine in Syrian golden hamsters. Cancer Res, 46(9), 4782–​4786. PMID: 3731125. Verma M. 2005. Pancreatic cancer epidemiology. Technol Cancer Res Treat, 4(3), 295–​301. PMID: 15896085. Verma M. 2013. Cancer control and prevention: nutrition and epigenetics. Curr Opin Clin Nutr, 16(4), 376–​384. Verma M, Rogers S, Divi RL, et al. 2014. Epigenetic research in cancer epidemiology:  trends, opportunities, and challenges. Cancer Epidemiol Biomarkers Prev, 223–​233. PMID: 24326628. Villeneuve PJ, Johnson KC, Mao Y, and Hanley AJ. 2004. Environmental tobacco smoke and the risk of pancreatic cancer: findings from a Canadian population-​based case-​control study. Can J Public Health, 95(1), 32–​37. PMID: 14768739. Vincent A, Hong S-​M, Hu C, et  al. 2014. Epigenetic silencing of EYA2 in pancreatic adenocarcinomas promotes tumor growth. Oncotarget, 5(9), 2575–​2587. Vrana D, Pikhart H, Mohelnikova-​Duchonova B, et al. 2009. The association between glutathione S-​transferase gene polymorphisms and pancreatic cancer in a central European Slavonic population. Mutat Res, 680(1–​2), 78–​81. PMID: 19786118. Vrieling A, Bueno-​de-​Mesquita HB, Boshuizen HC, et  al. 2010. Cigarette smoking, environmental tobacco smoke exposure and pancreatic cancer risk in the European Prospective Investigation into Cancer and Nutrition. Int J Cancer, 126(10), 2394–​2403. PMID: 19790196. Vrieling A, Verhage BA, van Duijnhoven FJ, et al. 2009. Fruit and vegetable consumption and pancreatic cancer risk in the European Prospective Investigation into Cancer and Nutrition. Int J Cancer, 124(8), 1926–​1934. PMID: 19107929. Walker ARP, Walker BF, and Segal I. 1993. Cancer patterns in three African populations compared with the United States Black population. Eur J Cancer Prev, 2(4), 313–​320. PMID: 00008469-​199307000-​00005. Wang F, Gupta S, and Holly EA. 2006. Diabetes mellitus and pancreatic cancer in a population-​based case-​control study in the San Francisco Bay Area, California. Cancer Epidemiol Biomarkers Prev, 15(8), 1458–​1463. PMID: 16896032. Wang M, Abbruzzese JL, Friess H, et al. 1998. DNA adducts in human pancreatic tissues and their potential role in carcinogenesis. Cancer Res, 58(1), 38–​41. Wang Z, Lai ST, Xie L, et al. 2014. Metformin is associated with reduced risk of pancreatic cancer in patients with type 2 diabetes mellitus: a systematic review and meta-​analysis. Diabetes Res Clin Pract, 106(1), 19–​26. PMID: 24837144. Waterhouse M, Risch HA, Bosetti C, et  al. 2016. Vitamin D and pancreatic cancer:  a pooled analysis from the Pancreatic Cancer Case-​ Control Consortium. Ann Oncol, 27(1), 208. PMCID: PMC4701119. Weiderpass E, Partanen T, Kaaks R, et al. 1998. Occurrence, trends and environment etiology of pancreatic cancer. Scand J Work Environ Health, 24(3), 165–​174. PMID: 9710368. Weydert C, Roling B, Liu J, et al. 2003. Suppression of the malignant phenotype in human pancreatic cancer cells by the overexpression of manganese superoxide dismutase1. Mol Cancer Ther, 2(4), 361–​369.

634

634

PART IV:  Cancers by Tissue of Origin

Wideroff L, Gridley G, Mellemkjaer L, et  al. 1997. Cancer incidence in a population-​based cohort of patients hospitalized with diabetes mellitus in Denmark. JNCI, 89(18), 1360–​1365. PMID: 9308706. Wiseman M. 2008. The second World Cancer Res Fund/​American Institute for Cancer Res expert report: food, nutrition, physical activity, and the prevention of cancer: a global perspective. Proc Nutr Soc, 67(3), 253–​256. PMID: 18452640. Wogan GN, Hecht SS, Felton JS, Conney AH, and Loeb LA. 2004. Environmental and chemical carcinogenesis. Semin Cancer Biol, 14(6), 473–​486. Wolfgang CL, Herman JM, Laheru DA, et al. 2013. Recent progress in pancreatic cancer. CA Cancer J Clin, 63(5), 318–​348. Wolpin BM, Bao Y, Qian ZR, et al. 2013. Hyperglycemia, insulin resistance, impaired pancreatic beta-​cell function, and risk of pancreatic cancer. JNCI, 105(14), 1027–​1035. PMCID: PMC3714020. Wolpin BM, Rizzato C, Kraft P, et al. 2014. Genome-​wide association study identifies multiple susceptibility loci for pancreatic cancer. Nat Genet, 46, 994–​1000. PMID: 25086665. Wong WW, Tan MM, Xia Z, et al. 2001. Cerivastatin triggers tumor-​specific apoptosis with higher efficacy than lovastatin. Clin Cancer Res, 7(7), 2067–​2075. PMID: 11448925. Xiao M, Wang Y, and Gao Y. 2013. Association between Helicobacter pylori infection and pancreatic cancer development: a meta-​analysis. PLoS One, 8(9), e75559. PMCID: PMC3784458. Yadav D, and Lowenfels AB. 2013. The epidemiology of pancreatitis and pancreatic cancer. Gastroenterology, 144(6), 1252–​1261. PMCID: PMC3662544. Yamada T, Alpers DH, and Owyang C. 2009. Textbook of Gastroenterology, Vol. 5. Oxford: Wiley-​Blackwell. Yamauchi R. 2007. Addition products of α-​tocopherol with lipid-​derived free radicals. Vitam Horm, 76, 309–​327. Yan AW, and Schnabl B. 2012. Bacterial translocation and changes in the intestinal microbiome associated with alcoholic liver disease. World J Hepatol, 4(4), 110–​118. PMCID: PMC3345535. Yang J-​P, Li X, Wang F, et al. 2015. Association analysis of genetic variants of adiponectin gene and risk of pancreatic cancer. Int J Clin Exp Med, 8, 8094–​8100. PMID: 26221375. Yang M, Sun T, Zhou Y, et al. 2012. The functional cytotoxic T lymphocyte-​ associated Protein 4 49G-​to-​A genetic variant and risk of pancreatic cancer. Cancer, 118(19), 4681–​4686. PMID: 22359319. Yao CJ, Lai GM, Chan CF, et al. 2006. Dramatic synergistic anticancer effect of clinically achievable doses of lovastatin and troglitazone. Int J Cancer, 118(3), 773–​779. PMID: 16094629.

Yao JC, Hassan M, Phan A, et  al. 2008. One hundred years after “carcinoid”: epidemiology of and prognostic factors for neuroendocrine tumors in 35,825 cases in the United States. J Clin Oncol, 26(18), 3063–​3072. PMID: 18565894. Ye W, Lagergren J, Nyren O, and Ekbom A. 2001. Risk of pancreatic cancer after cholecystectomy: a cohort study in Sweden. Gut, 49(5), 678–​681. PMCID: PMC1728505. Ying H, Dey P, Yao W, et al. 2016. Genetics and biology of pancreatic ductal adenocarcinoma. Genes Dev, 30(4), 355–​385. PMCID: PMC4762423. Yun JE, Jo I, Park J, et al. 2006. Cigarette smoking, elevated fasting serum glucose, and risk of pancreatic cancer in Korean men. Int J Cancer, 119(1), 208–​212. Zambirinis CP, Pushalkar S, Saxena D, and Miller G. 2014. Pancreatic cancer, inflammation, and microbiome. Cancer J, 20(3), 195–​202. PMCID: PMC4112373. Zatonski WA, Boyle P, Przewozniak K, et al. 1993. Cigarette smoking, alcohol, tea and coffee consumption and pancreas cancer risk: a case-​control study from Opole, Poland. Int J Cancer, 53(4), 601–​607. PMID: 8436433. Zhang J, Dhakal I, Yan H, Phillips M, and Kesteloot H. 2007. Trends in pancreatic cancer incidence in nine SEER Cancer Registries, 1973–​2002. Ann Oncol, 18(7), 1268–​1279. PMID: 17488731. Zhang J, Prizment AE, Dhakal IB, and Anderson KE. 2014. Cholecystectomy, gallstones, tonsillectomy, and pancreatic cancer risk: a population-​based case-​ control study in Minnesota. Br J Cancer, 110(9), 2348–​ 2353. PMCID: PMC4007236. Zhang J, Zhang X, Dhakal IB, et  al. 2011. Sequence variants in antioxidant defense and DNA repair genes, dietary antioxidants, and pancreatic cancer risk. Int J Mol Epidemiol Genet, 2(3), 236. Zhang R, Zhao J, Xu J, and Liu F. 2016. Association of peripheral leukocyte telomere length and its variation with pancreatic cancer and colorectal cancer risk in Chinese population. Oncotarget, 7(25), 38579–​ 38585. PMCID: PMC5122412. Zhen DB, Rabe KG, Gallinger S, et al. 2015. BRCA1, BRCA2, PALB2, and CDKN2A mutations in familial pancreatic cancer:  a PACGENE study. Genet Med, 17(7), 569–​577. PMCID: PMC4439391. Zheng W, McLaughlin JK, Gridley G, et al. 1993. A cohort study of smoking, alcohol consumption, and dietary factors for pancreatic cancer (United States). Cancer Causes Control, 4(5), 477–​482. PMID: 8218880. Zheng Z, Zheng R, He Y, et  al. 2016. Risk factors for pancreatic cancer in China: a multicenter case-​control study. Epidemiology, 26, 1–​7.

 635

33 Liver Cancer W. THOMAS LONDON, JESSICA L. PETRICK, AND KATHERINE A. MCGLYNN

OVERVIEW Primary liver cancer is the sixth most frequently occurring cancer in the world and the second most common cause of cancer mortality. The global burden of liver cancer is borne principally by countries in East Asia and Africa, where 80% of liver cancer arises. Incidence rates of liver cancer, however, have begun to decline in Asia, while rates are increasing in low-​rate areas such as Europe and North America. The dominant histology of liver cancer in almost all countries is hepatocellular carcinoma (HCC). The major risk factors for HCC—​ chronic infection with either hepatitis B virus (HBV) or hepatitis C virus (HCV), aflatoxin B1 (AFB1) contamination of foodstuffs, excessive alcohol consumption, and diabetes/​obesity/​fatty liver disease—​all result in chronic inflammation in the liver. HBV infection accounts for 75%–​80% of virus-​associated HCC, while HCV is responsible for 10%–​20%. HBV infection is preventable by immunization, and HCV infection is largely preventable by public health measures and now is curable with new antiviral therapies. Control of AFB1 contamination is difficult because it must be implemented at a governmental level. However, as the greatest risk of HCC occurs when AFB1 and HBV co-​occur, eradicating HBV via vaccination will reduce the risk associated with AFB1. One of the looming problems with HCC will be how to reduce the prevalence of obesity/​diabetes/​fatty liver disease in the world, as this constellation of disorders will likely be the major risk factor in the future, and may already be responsible for the increasing incidence of HCC in Europe and North America.

TUMOR CLASSIFICATION Errors in classification of liver tumors are common because the liver is a frequent site of metastasis for tumors originating in other organs, in particular, the colorectum, stomach, lung, pancreas, and breast. Metastases to the liver outnumber primary tumors of the liver by 30 to 1. Among persons without cirrhosis, only 2% of tumors in the liver are primary liver tumors. In contrast, among persons with cirrhosis, 77% of tumors in the liver are primary liver tumors (Goodman, 2007). Death certificates and hospital charts cannot be relied on to distinguish primary from secondary tumors; thus liver cancer mortality data are less reliable than liver cancer incidence data. Even incidence statistics based on criteria other than histopathology can be unreliable, however. Much of the literature, particularly from developing countries, is not derived from histologically verified diagnoses. There are both benign and malignant liver tumors, but most benign tumors are quite rare. Table 33–​1 provides a simplified classification of liver tumors, with a brief description of their histopathology, etiology, and epidemiology.

Benign Liver Tumors Hepatocellular adenomas are usually subcapsular, well circumscribed, 2–​20  cm in diameter, and located predominantly in the right lobe of the liver. Because they are composed entirely of hepatocytes, the blood supply to the center of the tumor is limited, and thus adenomas are prone to central necrosis and hemorrhage. They may present with acute abdominal pain resulting from bleeding into the peritoneal

cavity (Vijay et al., 2015). Bile duct adenomas are also mainly subcapsular, but they are generally small lesions (1  cm in diameter or less) found at autopsy. Hemangiomas are usually solitary lesions of less than 4 cm in diameter that can occur anywhere in the liver. They are often detected as incidental findings on ultrasonograms or computerized tomograms of the liver. Giant hemangiomas, defined as larger than 10 cm in diameter, are mainly subcapsular. They are usually solitary, but can be multiple. They may spontaneously involute or they may, more rarely, give rise to angiosarcomas. Infantile hemangioendotheliomas may also occur anywhere in the liver, can be either solitary or multicentric, and vary in size from 1 to 15 cm in diameter. They are usually detected during the first 6 months of life. Despite their size, they almost always involute spontaneously. Mesenchymal hamartomas are uncommon tumors that are usually diagnosed before the age of 2 years. They range in size from 15 to 30 cm in diameter and may occur anywhere in the liver.

Malignant Liver Tumors Hepatoblastomas, the most common childhood liver tumors, vary in size from 5 to 17 cm and usually present as large, well-​circumscribed solitary masses. The majority of hepatoblastomas (57%) arise in the right lobe, with only 15% arising in the left lobe and 27% in both lobes (Ishak et  al., 2001). Angiosarcomas, the most common type of sarcoma in the liver, are high-​grade neoplasms of endothelial cells (Ishak et al., 2001). They frequently involve the entire liver and rarely arise on a background of cirrhosis. Hepatocellular carcinomas (HCCs), tumors of hepatocytes, the parenchymal cells of the liver, are the most common malignant neoplasm of the liver. HCCs begin as solitary nodules that can arise anywhere in the liver. The right lobe is more frequently involved than the left, but this may be related to the larger size of the right lobe. A diffuse form of HCC has been described among patients in Africa, but, as few HCCs present at early stages in Africa, it is possible that these tumors also begin as single nodules. HCCs commonly invade the portal and hepatic venous systems, producing tumor thrombi in the portal vein and its tributaries. Tumor thrombi undoubtedly occur in hepatic veins, but are more difficult to identify (Kojiro and Nakashima, 1987). There are several histologic variants of HCC, including scirrhous, spindle cell, clear cell, pleomorphic, and fibrolamellar types. All appear to have common etiologic and clinical characteristics, with the exception of fibrolamellar HCC, which is characterized by fibrous lamellae and polygonal tumor cells. Unlike other types of HCC, most fibrolamellar HCCs do not arise on a background of liver disease, can occur at younger ages, are more likely to be resectable and have better survival after resection or transplant (Eggert et al., 2013). They most often present in the left lobe of the liver, typically as a single mass (Ishak et al., 2001). Cholangiocarcinomas, the second most common malignant liver tumor, arise from cholangiocytes that line the bile ducts and may occur anywhere in either the intrahepatic or the extrahepatic bile duct. Intrahepatic cholangiocarcinomas (ICCs) are classified as primary liver cancers, while extrahepatic cholangiocarcinomas are classified as biliary tract cancers. These tumors are more often diffuse and multicentric, but they can be solitary or multinodular. The diffuse types may be densely fibrotic (De Groen et  al., 1999; Edmondson, 1958). The

635

63

636

Part IV:  Cancers by Tissue of Origin

Table 33–​1.  Classification of Primary Liver Tumors Liver Tumors BENIGN Simple hepatic cyst Hepatocellular adenoma

Hemangioma Infantile hemangioendothelioma Epithelioid hemangioendothelioma Bile duct hamartoma Mesenchymal hamartoma MALIGNANT Hepatoblastoma

Histology

Comments

Unilocular lesion lined by simple cuboidal epithelium Composed entirely of well-​differentiated hepatocytes with abundant eosinophilic cytoplasm, arranged in sheets, cords. No lobules, portal areas, bile ducts. No capsule, but clearly demarcated from surrounding liver. Differentiated endothelial cells Type I: Multiple small vascular channels. Type II: Pleomorphic endothelial cells Dendritic and epithelioid cells. Not necessarily benign. May invade hepatic veins, metastasize Usually multiple lesions; dense stroma containing dilated bile ducts Multicystic masses of loose mesenchymal tissue containing small bile ducts

80% in women Almost all cases in women. Exposure to oral contraceptives, androgens, clomiphene. Associated with glycogen storage disease Type I, tyrosinemia, galactosemia. Most common benign tumor. Prevalence 1%–​5% of adults. Giant hemangiomas (> 10 cm) more common in women. Diagnosed at birth to 6 months. Spontaneously involute

Epithelial type (56%): fetal and/​or embryonal hepatocytes, rarely macrotrabecular or anaplastic. Mixed type (44%): epithelial and mesenchymal components (osteoid, cartilage, other spindle cells).

US incidence = 1/​100,000; M:F = 1.5–​2.0:1; most common childhood hepatic tumor. Associated with Beckwith-​Wiedemann syndrome, hemihypertrophy, familial adenomatous polyposis, precocious puberty. Cholangiocarcinoma Can be classified into mass-​forming, periductal infiltrating, and US incidence = 0.8/​100,000; M:F = 1.4:1. International intraductal types incidence = 2–​6/​100,000 Angiosarcoma Multicentric ill-​defined hemorrhagic nodules composed of anaplastic US incidence = 0.02/​100,000; M:F = 2:1. In adults, endothelial derived spindle cells. associated with exposure to thorotrast, arsenicals, vinyl chloride monomer. Hepatocellular carcinoma Well, moderately, or poorly differentiated parenchymal cells. US incidence = 6.5/​100,000; M:F = 3.8:1 Trabeculae 2–​8 cells thick, separated by sinusoids. Clear cell variant contains excess glycogen. Fibrolamellar variant: Large polygonal, eosinophilic cells circumscribed by bundles of acellular, creating large tumor islands or trabeculae. TUMOR-​LIKE NON-​NEOPLASTIC GROWTHS Focal nodular hyperplasia Solitary, circumscribed, unencapsulated, nodule with fibrotic central M:F = 1:2. 8% of primary liver “tumors” in US. Not (stellate) scar with radiating septae. Contains disorganized associated with oral contraceptives; estrogens stimulate hepatocytes, bile ducts, Kuppfer cells among fibrous bands. growth. Nodular regenerative Multiple nodules without fibrosis containing normal hepatocytes, bile Rare disease, M < F. Associated with renal transplantation, hyperplasia ducts, Kuppfer cells auto-​immune diseases.

tumor cells are usually arranged in tubules and glands, but they can also form nests, solid cords, or papillary structures (Goodman, 2007).

Precursor Neoplastic Lesions of Hepatocellular Carcinomas Animal studies of experimental hepatocarcinogenesis have revealed a sequence of events beginning with foci of phenotypically altered hepatocytes, proceeding to dysplastic foci and nodules (Farber and Cameron, 1980). This sequence has been confirmed in woodchucks chronically infected with the woodchuck hepatitis virus (Toshkov et al., 1990) and in humans chronically infected with HBV or HCV (Mason et al., 2010; Sun et al., 2006). The time from initial infection with HBV or HCV to the development of HCC is between two and eight decades. Dysplastic nodules almost always occur in cirrhotic livers. They may evolve into a carcinoma by developing an arterial blood supply and stromal invasion (Kojiro and Roskams, 2005). HCC that arises as a consequence of hemochromatosis begins with iron deposition early in life, but HCC generally does not develop until the fifth to seventh decade. During this long interval of exposure, many changes may occur in the liver, including chronic inflammation, fibrosis, cirrhosis, increased hepatocyte death rates, and regeneration. The continuous cycle of cell death and regeneration may eventually result in the proliferation of hepatic stem cells morphologically recognizable as oval cells (Fausto and Campbell, 2003; Vessey and de la Hall, 2001), although mature hepatocytes are the major source of cell replacement in the damaged liver (Summers et al., 2003). Oval cells may give rise to the phenotypically altered foci, but whether the oval cell is the

precursor cell of HCC is unproven. Many of the phenotypically altered foci and the dysplastic nodules are composed of monoclonal populations of hepatocytes (Takaishi et al., 2000).

Molecular Genetic Characteristics of Hepatocellular Carcinomas With current technology it is now possible to examine, in a comprehensive manner, the genomic landscape of hepatocarcinogenesis. Like other solid tumors, HCC appears to arise via alterations in numerous genes that modify multiple biologic processes. Whole-​genome sequencing has identified an average of 9718 nucleotide alterations, 271 insertions/​deletions, and 41 structural variations per tumor, with noted variability from tumor to tumor (Fujimoto et al., 2016). Within coding sequences, it has been reported that there are an average of 21 synonymous and 64 non-​synonymous mutations per tumor (Schulze et al., 2015). Tumors of larger size are observed to have greater numbers of point mutations, which are speculated to contribute to heterogeneity within the tumors. Across studies of HCC, somatic mutations most commonly (47%) occur in telomerase reverse transcriptase (TERT) (Lee, 2015). A gene frequently activated in many tumor types, TERT is essential for telomere maintenance and continued proliferation. Another commonly mutated gene, p53 (TP53), is mutated in 29% of HCCs. Similarly, β-​ catenin (CTNNB1) is mutated almost as frequently as p53, in 27% of tumors. AT-​rich interactive domain-​containing protein 1A (ARID1A) and AT-​ rich interactive domain-​ containing protein 2 (ARID2), genes associated with chromatin remodeling, which regulates DNA

 637

Liver Cancer accessibility, are mutated in 10% of HCCs. In addition to the more common somatic mutations, HCCs demonstrate a large number of lower frequency mutations in numerous genes (Totoki et al., 2014). Beyond individual somatic mutations, HCCs commonly display genomic copy number variation (Shibata and Aburatani, 2014). These affect well-​known oncogenes and tumor suppressor genes such as myelocytomatosis viral oncogene (MYC), cyclin-​D1 (CCND1), retinoblastoma 1 (RB1) and TP53. Copy number alterations are also evident in other regions that appear to target additional biologic processes including xenobiotic metabolism (e.g., MTDH, 8q22.1), proteolysis (e.g., COPS5, 8q13.1), mitochondria (COX6C, 8q22.1), immune response (TNFRSF14, 1p36.33) and DNA repair (BRCA2, 13q13.1), among others (Shibata and Aburatani, 2014). HCC genomes also show a number of structural variations including tandem duplications, deletions, and translocations. The genetic abnormalities associated with these structural variants differ from those observed from single nucleotide mutations (Fujimoto et al., 2016) but are similar to those from copy number variations. They appear to occur preferentially in early replicating regions. In addition to TERT and CCND1, structural variants affect loci such as low-​density lipoprotein receptor-​related protein 1B (LRP1B), and cyclin-​ dependent kinase inhibitor 2A (CDKN2A). In HCCs related to HBV, the genome can be altered by integration of the virus. An average number of five independent integrations per individual has been observed. While there is no specific genomic sequence associated with viral integration, frequent sites of integration are reported to include TERT and myeloid/​lymphoid or mixed-​lineage leukemia 4 (MLL4), G1/​S-​specific cyclin-​E1 (CCNE1) and fibronectin (FN1) (Sung et al., 2012). The molecular alterations observed in HCC are not random with respect to each other or to the likely etiology. A number of efforts have been made to identify mutational signatures. These mutational signatures appear to vary by geographic population and etiology (Fujimoto et al., 2016). For example, within European populations, TERT mutational frequency is observed to vary from 37% to 80%, depending on mutational signatures associated with different etiologies (Schulze et  al., 2015). Analysis of the molecular alterations elucidates key biologic processes underlying HCC. Telomere reactivation appears to be a critical component of hepatocarcinogenesis, as evidenced by somatic mutations, copy number changes, and HBV insertion events. The β-catenin pathway, critical in metabolic control of the liver, is also a frequent target of somatic alteration through mutations in CTNNB1 and AXIN1. The p53 cell-​cycle pathway is altered in at least 50% of HCCs (Zucman-​Rossi et al., 2015). However, with the exception of the signature R249S mutation associated with AFB1, no other TP53 recurrent mutation hot spot has been identified. The control of progression from G1 to S phase of the cell cycle is also commonly altered in HCC through mutations within the retinoblastoma pathway. Epigenetic control through histone methylation is commonly altered through mutations in ARID1A, ARID2, and MLL4. In summary, a large proportion of hepatocytes are affected by one or more genetic or epigenetic alterations during the long period of HCC development. Because HCCs are clonal and usually single tumors, only rare altered hepatocytes give rise to a malignant neoplasm. The recent identification of a very few genetic lesions in dysplastic nodules and early HCCs has revealed for the first time the possibility of a common initial lesion in the progression of such nodules to HCC.

Biomarkers of Hepatocellular Carcinoma Noninvasive biomarkers have been investigated for the early detection, surveillance, diagnosis, and prognosis of HCC (Schutte et al., 2015). From an epidemiologic standpoint, biomarkers for early detection and surveillance are of greatest interest. The first biomarker identified for the detection or diagnosis of HCC was α-​fetoprotein (AFP) (Abelev et al., 1967; Alpert et al., 1968). Although AFP is currently the most widely used biomarker for early detection, it is no longer recommended for such use because of both low sensitivity (~60%) (Marrero

637

et al., 2009) and specificity (80%–​90%) (Lok et al., 2010); 70%–​80% of small HCCs do not have elevated AFP levels (Saffroy et al., 2007). Newer biomarkers, such as des-​ gamma-​ carboxy-​ prothrombin (DCP), osteopontin, interleukin-​6 (Il-​6), and Golgi protein 73 (GP73), have been tested singly and in combination for early detection. Of these, GP-​73 is, currently, the most promising but is still not sufficiently sensitive or specific. Whether combinations of biomarkers can overcome these deficiencies is currently under investigation. Surveillance at 6-​month intervals using ultrasonography is currently recommended for persons with cirrhosis, chronic HBV infection, or chronic HCV infection (Bruix and Sherman, 2011). The ability of ultrasound is, however, dependent on the quality of the machine, the experience and skill of the examiner, and several patient-​related factors. Obesity impairs the sensitivity of tumor detection and, among persons with cirrhosis, regenerative and dysplastic nodules are difficult to distinguish from small HCCs (Schutte et  al., 2015). In summary, although surveillance of high-​risk persons with ultrasound and AFP monitoring is the current standard of care, their use for detection of small tumors remains of limited value.

DEMOGRAPHIC PATTERNS International Patterns of Incidence, Mortality, and Survival Globally, HCC is the dominant histology of liver cancer in almost all countries, accounting for approximately 80% of primary liver cancers. ICC is the second most common histologic type, accounting for approximately 15% of cases (Okuda et  al., 2002). In Northeast Thailand, unlike other geographic regions, ICC is the dominant histologic type of liver cancer, due to chronic infection with liver flukes (Chapter 24). There ICC accounts for 74% of microscopically verified liver cancers among men and 85% among women (Parkin et al., 2014). Large geographic variability exists in incidence and mortality rates for all types of liver cancer. The highest incidence rates in the world are in Asia and Africa (Figure 33–​1) (Ferlay et al., 2013). Approximately 75% of all liver cancers arise in Asia, with China accounting for over 50% of the world’s burden. The country with the single highest incidence rate, however, is Mongolia, with an age-​standardized rate (ASR) per 100,000 persons of 78.1 (97.8 in males and 61.1 in females) (Ferlay et al., 2013). In contrast, the lowest incidence rates in the world occur in countries of Northern Europe, the Middle East, Oceania, and North and South America, while countries in Central Europe have intermediate rates. Within specific geographic regions there is great variability, however, particularly in Asia and Africa. In Asian countries with cancer registries, the ASR of males range from 2.0 in Bhopal and Dindigul, India, to 77.5 in Qidong City, China. In African countries, the ASR of males range from 1.7 in Tunisia to 24.8 in Gharbiah, Egypt (Parkin et al., 2014). Incidence rates of liver cancer have been increasing in many low-​ rate areas and decreasing in high-​and intermediate-​rate areas. In the interval between 1983–​1987 and 2003–​2007, liver cancer incidence increased in North and South America, Oceania, and India, as well as in most European countries (McGlynn et al., 2015b; Petrick et al., 2016a). While the reasons for increasing incidence rates in many areas of the world are not entirely certain, HCV and the associated metabolic disorders of obesity, diabetes, and metabolic syndrome, as well as improved survival after diagnosis of cirrhosis, are likely to be associated. In contrast, liver cancer incidence rates have declined in some Asian countries, in Spain, and in Italy among men (McGlynn et  al., 2015b; Petrick et  al., 2016a). The decreasing incidence rates in China are likely due primarily to programs to reduce aflatoxin B1 (AFB1) exposure and HBV transmission, as well as other public health efforts, such as reducing smoking rates, eating a healthier diet, and improving water quality (Gao et al., 2012). Among younger persons, HBV vaccination has already resulted in decreased HCC incidence; as the cohort of vaccinated persons ages, rates of liver cancer should decline even further. In Japan, the decreasing incidence of HCC is

638

638

Part IV:  Cancers by Tissue of Origin

Liver cancer 9.2+ 5.4–9.2 4.2–5.4 3.0–4.2 /​=1.0 cm): clinicopathologic and immunohistochemical analysis of 123 cases. Am J Surg Pathol, 36(9), 1279–​1301. PMID: 22895264. Albores-​Saavedra J, Adsay NV, Crawford JM, et al. 2010. Carcinoma of the gallbladder and extrahepatic bile ducts. In Bosman FT, Carneiro F, Hruban R, et al. (Eds.), World Health Organization classification of tumors: tumors of digestive system (4th ed., pp. 266–​274). Lyon: IARC Press. Andersen JB, Spee B, Blechacz BR, et al. 2012. Genomic and genetic characterization of cholangiocarcinoma identifies therapeutic targets for tyrosine kinase inhibitors. Gastroenterology, 142(4), 1021–​1031 e1015. PMCID: 3413201. Andia ME, Hsing AW, Andreotti G, and Ferreccio C. 2008. Geographic variation of gallbladder cancer mortality and risk factors in Chile:  a population-​ based ecologic study. Int J Cancer, 123(6), 1411–​ 1416. PMCID: 2864002. Andreotti G, Hou L, Gao YT, et al. 2010. Reproductive factors and risks of biliary tract cancers and stones: a population-​based study in Shanghai, China. Br J Cancer, 102(7), 1185–​1189. PMCID: 2853091. Andreotti G, Liu E, Gao YT, et al. 2011. Medical history and the risk of biliary tract cancers in Shanghai, China: implications for a role of inflammation. Cancer Causes Control, 22(9), 1289–​1296. PMID: 21744094. Basu S, Singh MK, Singh TB, et al. 2013. Heavy and trace metals in carcinoma of the gallbladder. World J Surg, 37(11), 2641–​2646. PMID: 23942528. Benavides M, Anton A, Gallego J, et al. 2015. Biliary tract cancers: SEOM clinical guidelines. Clin Transl Oncol, 17(12), 982–​987. PMID: 26607930. Bhaskaran K, Douglas I, Forbes H, et al. 2014. Body-​mass index and risk of 22 specific cancers: a population-​based cohort study of 5.24 million UK adults. Lancet, 384(9945), 755–​765. PMCID: 4151483. Bizama C, Garcia P, Espinoza JA, et  al. 2015. Targeting specific molecular pathways holds promise for advanced gallbladder cancer therapy. Cancer Treat Rev, 41(3), 222–​234. PMID: 25639632. Bouvard V, Baan R, Straif K, et al. 2009. A review of human carcinogens—​Part B: biological agents. Lancet Oncol, 10(4), 321–​322. PMID: 19350698. Buch S, Schafmayer C, Volzke H et  al. 2007. A genome-​wide association scan identifies the hepatic cholesterol transporter ABCG8 as a susceptibility factor for human gallstone disease. Nat Genet, 39(8), 995–​999. PMID: 17632509. Castro FA, Koshiol J, Hsing AW, and Devesa SS. 2013. Biliary tract cancer incidence in the United States: demographic and temporal variations by anatomic site. Int J Cancer, 133(7), 1664–​1671. PMCID: 3713194. Castro FA, Liu X, Forsti A, et al. 2014. Increased risk of hepatobiliary cancers after hospitalization for autoimmune disease. Clin Gastroenterol Hepatol, 12(6), 1038–​1045 e1037. PMID: 24246767. CEPAL and OCDE. 2005. Evaluaciones del desempeno ambiental. Available from:  http://​sinia.mma.gob.cl/​. Accessed February 3, 2016. (Vol. 2016). Chile: Sistema Nacional de Informacion Ambiental, Gobierno de Chile. Cha PC, Zembutsu H, Takahashi A, et  al. 2012. A genome-​wide association study identifies SNP in DCC is associated with gallbladder cancer in the Japanese population. J Hum Genet, 57(4), 235–​237. PMID: 22318345. Dorjgochoo T, Shu XO, Li HL, et al. 2009. Use of oral contraceptives, intrauterine devices and tubal sterilization and cancer risk in a large prospective study, from 1996 to 2006. Int J Cancer, 124(10), 2442–​2449. PMCID: 2666967. Ehling J, and Tacke F. 2015. Role of chemokine pathways in hepatobiliary cancer. Cancer Lett, 379(2), 173–​183. PMID: 26123664. Ellinghaus D, Jostins L, Spain SL, et  al. 2016. Analysis of five chronic inflammatory diseases identifies 27 new associations and highlights disease-​ specific patterns at shared loci. Nat Genet, 48(5), 510–​ 518. PMCID: PMC4848113. Espinoza JA, García P, Bizama C, and Roa JC. 2015. Molecular Genetics of Gallbladder Cancer. eLS, 1–7. Feakins RM. 2015. Obesity and metabolic syndrome: pathological effects on the gastrointestinal tract. Histopathology, 68(5), 630–​640. PMID: 26599607. Ferlay J, Soerjomataram I, Ervik M, et  al. 2013. GLOBOCAN 2012 v1.0, Cancer incidence and mortality worldwide:  IARC CancerBase No. 11 [Internet]. International Agency for Research on Cancer. Available from: http://​globocan.iarc.fr. Accessed December 23, 2015. Forman D, Bray F, Brewster DH, et  al. 2013. Cancer incidence in five continents, Vol. X (electronic version). Available from:  http://​ci5.iarc.fr. Accessed May 28, 2015.

 69

Biliary Tract Cancer Funabiki T, Matsubara T, Miyakawa S, and Ishihara S. 2009. Pancreaticobiliary maljunction and carcinogenesis to biliary and pancreatic malignancy. Langenbecks Arch Surg, 394(1), 159–​169. PMID: 18500533. Garcia P, Manterola C, Araya JC, et  al. 2009. Promoter methylation profile in preneoplastic and neoplastic gallbladder lesions. Mol Carcinog, 48(1), 79–​89. PMID: 18543280. Gaya-​Vidal M, Athanasiadis G, Carreras-​Torres R, et al. 2012. Apolipoprotein E/​ C1/​ C4/​ C2 gene cluster diversity in two native Andean populations:  Aymaras and Quechuas. Ann Hum Genet, 76(4), 283–​ 295. PMID: 22681517. Grunhage F, Acalovschi M, Tirziu S, et  al. 2007. Increased gallstone risk in humans conferred by common variant of hepatic ATP-​binding cassette transporter for cholesterol. Hepatology, 46(3), 793–​801. PMID: 17626266. Hariharan D, Saied A, and Kocher HM. 2008. Analysis of mortality rates for gallbladder cancer across the world. HPB (Oxford), 10(5), 327–​331. PMCID: PMC2575684. Hemminki K, Mousavi SM, Brandt A, Ji J, and Sundquist J. 2010. Liver and gallbladder cancer in immigrants to Sweden. Eur J Cancer, 46(5), 926–​ 931. PMID: 20064704. Henley SJ, Weir HK, Jim MA, Watson M, and Richardson LC. 2015. Gallbladder cancer incidence and mortality, United States 1999–​ 2011. Cancer Epidemiol Biomarkers Prev, 24(9), 1319–​1326. PMID: 26070529. Hoblinger A, and Lammert F. 2008. Genetics of biliary tract diseases:  new insights into gallstone disease and biliary tract cancers. Curr Opin Gastroenterol, 24(3), 363–​371. PMID: 18408466. Howlader N, Noone AM, Krapcho M, et  al. 2015. SEER cancer statistics review, 1975–​2012. Bethesda, MD: National Cancer Institute. Hsing AW, Gao YT, Devesa SS, Jin F, and Fraumeni JF, Jr. 1998. Rising incidence of biliary tract cancers in Shanghai, China. Int J Cancer, 75(3), 368–​370. PMID: 9455795. Hsing AW, Gao YT, Han TQ, et al. 2007. Gallstones and the risk of biliary tract cancer: a population-​based study in China. Br J Cancer, 97(11), 1577–​ 1582. PMCID: 2360257. Hundal R, and Shaffer EA. 2014. Gallbladder cancer: epidemiology and outcome. Clin Epidemiol, 6, 99–​109. PMCID: 3952897. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. A review of human carcinogens Part F: Chemical agents and related occupations. Lyon, France: IARC; 2012. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. Alcohol Consumption and Ethyl Carbamate. Lyon, France: IARC; 2010. Karlsen TH, and Chung BK. 2015. Genetic risk and the development of autoimmune liver disease. Dig Dis, 33 Suppl 2, 13–​24. PMID: 26641277. Katsika D, Magnusson P, Krawczyk M, et  al. 2010. Gallstone disease in Swedish twins: risk is associated with ABCG8 D19H genotype. J Intern Med, 268(3), 279–​285. PMID: 20497293. Kilander C, Mattsson F, Lu Y, et  al. 2015. Reproductive factors and risk of biliary tract cancer in a population-​ based study. Acta Oncol, 1–​7. PMID: 25752973. Koshiol J, Castro F, Kemp TJ, et al. 2016a. Association of inflammatory and other immune markers with gallbladder cancer: Results from two independent case-​control studies. Cytokine, 83, 217–​225. PMCID: PMC4876019. Koshiol J, Pawlish K, Goodman MT, McGlynn KA, and Engels EA. 2014. Risk of hepatobiliary cancer after solid organ transplant in the United States. Clin Gastroenterol Hepatol, 12(9), 1541–​1549 e1543. PMCID: 4064001. Koshiol J, Wozniak A, Cook P, et al. 2016b. Salmonella Typhi carriage and gallbladder cancer: a case-​control and meta-​analysis. Cancer Med, 5(11), 3310–3235. PMCID: PMC5119987. Krawczyk M, Miquel JF, Stokes CS, et al. 2013. Genetics of biliary lithiasis from an ethnic perspective. Clin Res Hepatol Gastroenterol, 37(2), 119–​ 125. PMID: 23340007. Krawczyk M, Mullenbach R, Weber SN, Zimmer V, and Lammert F. 2010. Genome-​wide association studies and genetic risk assessment of liver diseases. Nat Rev Gastroenterol Hepatol, 7(12), 669–​681. PMID: 21045792. Krawczyk M, Wang DQ, Portincasa P, and Lammert F. 2011. Dissecting the genetic heterogeneity of gallbladder stone formation. Semin Liver Dis, 31(2), 157–​172. PMID: 21538282. Kubo S. 2015. Re: Occupational cholangiocarcinoma. J Hepatobiliary Pancreat Sci, 22(1), E2–​3. PMID: 25521324. Lai CH, and Lau WY. 2008. Gallbladder cancer: a comprehensive review. Surgeon, 6(2), 101–​110. PMID: 18488776. Lammert F, Gurusamy K, Ko CW, et al. 2016. Gallstones. Nat Rev Dis Prim, 2, 16024. PMID: 27121416. Lee BS, Park EC, Park SW, Nam CM, and Roh J. 2015. Hepatitis B virus infection, diabetes mellitus, and their synergism for cholangiocarcinoma development: a case-​control study in Korea. World J Gastroenterol, 21(2), 502–​510. PMCID: 4292281.

669

Mangtani P, Maringe C, Rachet B, Coleman MP, and dos Santos Silva I. 2010. Cancer mortality in ethnic South Asian migrants in England and Wales (1993–​2003):  patterns in the overall population and in first and subsequent generations. Br J Cancer, 102(9), 1438–​1443. PMCID: 2865755. Maringhini A, Moreau JA, Melton LJ, 3rd, et al. 1987. Gallstones, gallbladder cancer, and other gastrointestinal malignancies: an epidemiologic study in Rochester, Minnesota. Ann Intern Med, 107(1), 30–​35. PMID: 3592446. Nakamura H, Arai Y, Totoki Y, et  al. 2015. Genomic spectra of biliary tract cancer. Nat Genet, 47(9), 1003–​1010. PMID: 26258846. Nath G, Gulati AK, and Shukla VK. 2010. Role of bacteria in carcinogenesis, with special reference to carcinoma of the gallbladder. World J Gastroenterol, 16(43), 5395–​5404. PMCID: 2988230. Nogueira L, Foerster C, Groopman J, et  al. 2015. Association of aflatoxin with gallbladder cancer in Chile. JAMA, 313(20), 2075–​ 2077. PMID: 26010638. Nogueira L, Freedman ND, Engels EA, et al. 2014. Gallstones, cholecystectomy, and risk of digestive system cancers. Am J Epidemiol, 179(6), 731–​ 739. PMID: 24470530. Norat T, Rosenblatt DN, Vingeliene S, and Aune D. 2014. The associations between food, nutrition and physical activity and the risk of gallbladder cancer. London: World Cancer Research Fund International Systematic Literature Review, Imperial College. Nyberg U, Nilsson B, Travis LB, Holm LE, and Hall P. 2002. Cancer incidence among Swedish patients exposed to radioactive thorotrast:  a forty-​year follow-​up survey. Radiat Res, 157(4), 419–​425. PMID: 11893244. Panda D, Sharma A, Shukla NK, et al. 2013. Gall bladder cancer and the role of dietary and lifestyle factors: a case-​control study in a North Indian population. Eur J Cancer Prev, 22(5), 431–​437. PMID: 23462456. Pesatori AC, Grillo P, Consonni D, et al. 2013. Update of the mortality study of workers exposed to polychlorinated biphenyls (Pcbs) in two Italian capacitor manufacturing plants. Med Lav, 104(2), 107–​114. PMID: 23789517. Randi G, Franceschi S, and La Vecchia C. 2006. Gallbladder cancer worldwide:  geographical distribution and risk factors. Int J Cancer, 118(7), 1591–​1602. PMID: 16397865. Ren HB, Yu T, Liu C, and Li YQ. 2011. Diabetes mellitus and increased risk of biliary tract cancer: systematic review and meta-​analysis. Cancer Causes Control, 22(6), 837–​847. PMID: 21424210. Roa I, de Aretxabala X, Araya JC, and Roa J. 2006. Preneoplastic lesions in gallbladder cancer. J Surg Oncol, 93(8), 615–​623. PMID: 16724345. Roa JC, Roa I, Correa P, et al. 2005. Microsatellite instability in preneoplastic and neoplastic lesions of the gallbladder. J Gastroenterol, 40(1), 79–​86. PMID: 15692793. Roa JC, Tapia O, Cakir A, et al. 2011. Squamous cell and adenosquamous carcinomas of the gallbladder: clinicopathological analysis of 34 cases identified in 606 carcinomas. Mod Pathol, 24(8), 1069–​1078. PMID: 21532545. Scanu T, Spaapen RM, Bakker JM, et  al. 2015. Salmonella manipulation of host signaling pathways provokes cellular transformation associated with gallbladder carcinoma. Cell Host Microbe, 17(6), 763–​774. PMID: 26028364. Schnelldorfer T. 2013. Porcelain gallbladder: a benign process or concern for malignancy? J Gastrointest Surg, 17(6), 1161–​1168. PMID: 23423431. SEER. 2015a. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Incidence—​SEER 18 Regs Research Data + Hurricane Katrina Impacted Louisiana Cases, Nov 2014 Sub (1973–​2012 varying)—​Linked to County Attributes—​Total U.S., 1969–​2013 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2015, based on the November 2014 submission. SEER. 2015b. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database:  Incidence—​ SEER 18 Regs Research Data + Hurricane Katrina Impacted Louisiana Cases, Nov 2014 Sub (2000–​ 2012) Katrina/​ Rita Population Adjustment—​ Linked to County Attributes—​Total U.S., 1969–​2013 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2015, based on the November 2014 submission. Segura-​Lopez FK, Guitron-​Cantu A, and Torres J. 2015. Association between Helicobacter spp. infections and hepatobiliary malignancies:  a review. World J Gastroenterol, 21(5), 1414–​1423. PMCID: 4316084. Serra I, Yamamoto M, Calvo A, et al. 2002. Association of chili pepper consumption, low socioeconomic status and longstanding gallstones with gallbladder cancer in a Chilean population. Int J Cancer, 102(4), 407–​ 411. PMID: 12402311. Shukla VK, Rastogi AN, Adukia TK et al. 2001. Organochlorine pesticides in carcinoma of the gallbladder:  a case-​control study. Eur J Cancer Prev, 10(2), 153–​156. PMID: 11330456.

670

670

PART IV:  Cancers by Tissue of Origin

Soares KC, Arnaoutakis DJ, Kamel I, et al. 2014. Choledochal cysts: presentation, clinical differentiation, and management. J Am Coll Surg, 219(6), 1167–​1180. PMCID: 4332770. Sobue T, Utada M, Makiuchi T, et al. 2015. Risk of bile duct cancer among printing workers exposed to 1,2-​dichloropropane and/​or dichloromethane. J Occup Health, 57(3), 230–​236. PMID: 25739336. Songserm N, Promthet S, Sithithaworn P, et  al. 2012. Risk factors for cholangiocarcinoma in high-​risk area of Thailand: role of lifestyle, diet and methylenetetrahydrofolate reductase polymorphisms. Cancer Epidemiol, 36(2), e89–​94. PMID: 22189445. Srivastava K, Srivastava A, Sharma KL, Mittal B. 2011. Candidate gene studies in gallbladder cancer:  a systematic review and meta-​analysis. Mutation Research, 728, 67–​79. Stender S, Frikke-​ Schmidt R, Nordestgaard BG, and Tybjaerg-​ Hansen A. 2011. Sterol transporter adenosine triphosphate-​ binding cassette transporter G8, gallstones, and biliary cancer in 62,000 individuals from the general population. Hepatology, 53(2), 640–​648. PMID: 21274884. Stinton LM, and Shaffer EA. 2012. Epidemiology of gallbladder disease: cholelithiasis and cancer. Gut Liver, 6(2), 172–​187. PMCID: 3343155. Stokes CS, Krawczyk M, and Lammert F. 2011. Gallstones: environment, lifestyle and genes. Dig Dis, 29(2), 191–​201. PMID: 21734384. Travis LB, Hauptmann M, Gaul LK et al. 2003. Site-​specific cancer incidence and mortality after cerebral angiography with radioactive thorotrast. Radiat Res, 160(6), 691–​706. PMID: 14640794. Travis LB, Land CE, Andersson M et  al. 2001. Mortality after cerebral angiography with or without radioactive Thorotrast:  an international cohort of 3,143 two-​ year survivors. Radiat Res, 156(2), 136–​ 150. PMID: 11448234. Torre LA, Bray F, Siegel RL, et al. 2015. Global cancer statistics, 2012. CA Cancer J Clin, 65(2), 87–​108. PMID: 25651787. Tsuchiya Y, Terao M, Okano K, et al. 2011. Mutagenicity and mutagens of the red chili pepper as gallbladder cancer risk factor in Chilean women. Asian Pac J Cancer Prev, 12(2), 471–​476. PMID: 21545215.

U.S. Department of Health and Human Services. The Health Consequences of Smoking—​50 Years of Progress:  A  Report of the Surgeon General. Atlanta, GA 2014. van Kaick G, Dalheimer A, Hornik S et  al. 1999. The german thorotrast study: recent results and assessment of risks. Radiat Res, 152(6 Suppl), S64–​71. PMID: 10564940. von Kampen O, Buch S, Nothnagel M, et al. 2013. Genetic and functional identification of the likely causative variant for cholesterol gallstone disease at the ABCG5/​8 lithogenic locus. Hepatology (Baltimore, Md.), 57(6), 2407–​ 2417. PMID: 22898925. Weber H, Leal P, Stein S, et  al. 2015. Rapamycin and WYE-​354 suppress human gallbladder cancer xenografts in mice. Oncotarget, 6(31), 31877–​ 31888. PMID: 26397134. Wenbin D, Zhuo C, Zhibing M, et al. 2012. The effect of smoking on the risk of gallbladder cancer:  a meta-​analysis of observational studies. Eur J Gastroenterol Hepatol, 25(3), 373–​379. PMID: 23085578. Wistuba, II, Miquel JF, Gazdar AF, and Albores-​Saavedra J. 1999. Gallbladder adenomas have molecular abnormalities different from those present in gallbladder carcinomas. Hum Pathol, 30(1), 21–​25. PMID: 9923922. Xiao M, Gao Y, and Wang Y. 2014. Helicobacter species infection may be associated with cholangiocarcinoma: a meta-​analysis. Int J Clin Pract, 68(2), 262–​270. PMID: 24373128. Ye XH, Huai JP, Ding J, Chen YP, and Sun XC. 2013. Smoking, alcohol consumption, and the risk of extrahepatic cholangiocarcinoma: a meta-​ analysis. World J Gastroenterol, 19(46), 8780–​8788. PMCID: 3870528. Zatonski WA, La Vecchia C, Przewozniak K, et al. 1992. Risk factors for gallbladder cancer: a Polish case-​control study. Int J Cancer, 51(5), 707–​711. PMID: 1612778. Zen Y, Adsay NV, Bardadin K, et al. 2007. Biliary intraepithelial neoplasia: an international interobserver agreement study and proposal for diagnostic criteria. Mod Pathol, 20(6), 701–​709. PMID: 17431410. Zhou M, Wang H, Zhu J, et al. 2015. Cause-​specific mortality for 240 causes in China during 1990–​2013: a systematic subnational analysis for the Global Burden of Disease Study 2013. Lancet, 387(10015), 251–​272. PMID: 26510778.

 671

35

Small Intestine Cancer JENNIFER L. BEEBE-​DIMMER, FAWN D. VIGNEAU, AND DAVID SCHOTTENFELD

OVERVIEW The small intestine extends on average 6–​7 meters from the gastric pylorus to its insertion into the large intestine. Its mucosal surface contains 90% of the absorptive surface area of the digestive tract. Remarkably, in 2015, only about 3% of digestive system cancers and less than 1% of cancer deaths in the United States were observed in the small intestine. In contrast, approximately 50% of cancers in the digestive system were diagnosed in the large intestine, which measures just 1.5 meters in length. Cancers of the small intestine are among the most histologically heterogeneous of gastrointestinal neoplasms, encompassing pathologic subtypes of neuroendocrine carcinoid, adenocarcinoma, lymphoma, and gastrointestinal stromal (GIST) tumors. Adenocarcinoma, accounts for ~25%–​35% of cancers in the small intestine, in contrast to over 90% of cancers in the large intestine. Genetic syndromes, such as familial adenomatous polyposis (FAP) and hereditary non-​polyposis colorectal cancer (HNPCC), predispose to adenocarcinoma in the small intestine. Patients with a history of celiac disease or gluten-​sensitive enteropathy are at increased risk of T-​cell, and to a lesser extent, B-​cell non-​Hodgkin lymphoma, and adenocarcinoma in the small intestine. Crohn`s disease (CD) is a chronic inflammatory disease presenting prominently in the distal small intestine and is also associated with an increased risk of adenocarcinoma. Research on the pathogenesis of CD is focusing on the complex interplay of environmental triggers, innate and adaptive immunity, and genetic susceptibility. The biologic implications of the infrequency of cancers in the small intestine, when compared with that in the large intestine, require consideration of the contrasting concentration and diversity of the intestinal microflora, differences in transit time, tumorigenic effects of bile salt metabolism, in conjunction with innate differences in mutational susceptibility expressed in terms of rate and cumulative number of stem cell divisions and the stochastic risks of mutational events.

DISEASE BURDEN An estimated 10,090 cases and 1330 deaths from cancers of the small intestine are projected in 2016 in the United States, comprising just 0.6% of total cancer cases and 0.2% of total cancer deaths. It is estimated that 53% of new cases will occur in men and 47% in women (Siegel et  al., 2015). The average annual age-​ adjusted incidence per 100,000 for cancers of the small intestine was 2.6, when age-​ standardized to the US 2000 population and based on 2008–​2012 data. The rates were 3.1 in men, and 2.2 in women. Age-​ adjusted rates per 100,000 were highest in African American men (4.6) and women (3.3) (Surveillance, Epidemiology, and End Results [SEER], 2015a). Generally, men have a higher incidence of both neuroendocrine carcinoid tumors and adenocarcinomas than women (Figure 35–1), and blacks have higher rates for both pathologic subtypes than whites (Figure 35–2). Whites have the highest rates of non-​Hodgkin lymphoma compared with all other races, but the highest incidence of stromal tumors is found in American Indian/​Alaskan Natives and Asian/​Pacific Islanders men (0.3 per 100,000) (Table 35–1). Using population-​ based cancer registry data, Goodman et  al. demonstrated that in both males and females, age-​specific incidence rates of small intestine cancers were greater in blacks than whites or

Asian/​Pacific Islanders and greater in non-​Hispanics than Hispanics (Goodman et al., 2013; SEER, 2015c). Using data from the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) program, Qubaiah et al. (2010) showed that age-​specific rates were generally higher in males than females for carcinomas, neuroendocrine tumors, sarcomas, and lymphomas. Age-​specific incidence rates for adenocarcinomas tend to increase after age 30, after age 35 for lymphomas and carcinoids, and after age 45 for stromal tumors. Furthermore, the age-​specific incidence of adenocarcinomas and carcinoids increases more sharply with increasing age in blacks than in whites or other races. The reverse is true for lymphomas, where the incidence in whites and other races increase more sharply than in blacks. For stromal tumors, overall sharper increases are seen in other races, compared to either whites or blacks (SEER, 2015c). Between 2008 and 2012, black men (0.7) and black women (0.5) incurred the highest overall mortality of small intestine cancers (mortality data excludes lymphomas) (Table 35–1) (SEER, 2015c). Black males have worse 5-​year relative survival (56.6%; 95% CI = 51.1%–​ 61.7%) than white males (66.9%; 95% CI = 64.8%–​68.8%). Histology also plays a role in survival, with carcinoid/​neuroendocrine cases having better 5-​year relative survival (84.4%; 95% CI = 82.6%–​86.0%) than those with stromal tumors (82.0%; 95% CI = 77.8%–​85.5%), non-​Hodgkin lymphoma (70.1%; 95% CI = 66.8%–​73.0%) and adenocarcinoma (29.2%; 95% CI = 26.9%–​31.5%) (SEER, 2015b). The observed survival advantage among those diagnosed with carcinoid/​ neuroendocrine tumors is consistent with an earlier study of SEER data from 1992–​2005 (Qubaiah et al., 2010).

GEOGRAPHIC VARIATION The burden of small intestine cancers internationally was generally higher in males than females, as evidenced by the age-​standardized (world) incidence rates per 100,000 in 2003–​2007 (Figure 35–3). Male rates tended to be highest in North America, Brazil, Iceland, Norway, Sweden, and parts of Argentina, Switzerland, France, Germany, and Italy, and in Oceania Indigenous and New Zealand Pacific Islanders. Female rates were highest in North America, Tierra del Fuego, the Graubunden and Glarus area of Switzerland, and parts of Oceania. The lowest rates for males were found in the Indian subcontinent, areas of Thailand, Uganda, and Yangcheng County in China. Female rates were likewise lowest in parts of India and Thailand and also parts of Korea and among Malaysians, African Zimbabwe Harares, and in Setif, Algeria. In North America, Canadian rates were lowest in the Northwest Territories for males (0.7) and New Brunswick for females (0.7) and highest in Manitoba Province for males (1.7) and Yukon Territory for females (2.2). Referencing the consolidated rates for ethnic groups across 18 population-​based cancer registries in the SEER program and 42 registries in the National Program of Cancer Registries (NPCR), the overall rates for Asian/​Pacific Islanders were lower (0.5–​0.9) than rates for whites (1.0–​1.5), black females (1.8), and black males (2.4). In Central and South America, of 25 areas reporting, 19 for males and 20 for females had rates less than 1.0, while six areas for males and five for females reported rates from 1.0–​2.6. The higher rates were found most frequently in Brazil, but also in Cordoba, Argentina;

671

672

672

PART IV:  Cancers by Tissue of Origin Islanders (males:  2.4); and in Hawaii among native Hawaiian males (2.9); although the rate among non-​ indigenous females in the Australian Northern Territory was also high (2.3).

16.0 Rates per 1,000,000

14.0 12.0 10.0

TEMPORAL TRENDS

8.0 6.0 4.0 2.0 0.0 2008

2009

2010

2011

2012

Year of Diagnosis Male-Carcinoid/NET Male-Adenocarcinoma

Female-Carcinoid/NET Female-Adenocarcinoma

Figure 35–1.  Age-​adjusted incidence rates of adenocarcinoma and carcinoid/​neuroendocrine of the small intestine by gender in all races in the United States. SEER-​18 (2008–​2012).

Martinique; Tierra del Fuego, Argentina; Valdivia, Chile; and Manizales, Colombia. In European males, rates ranged from the lowest in Syracuse, Italy (0.2), Belarus (0.3), Cyprus (0.3), Kielce, Poland (0.3), Ukraine (0.3), Bulgaria (0.4), Latvia (0.4), Podkarpackie, Poland (0.4), and St. Petersburg, Russian Federation (0.4) to the largest in Iceland (2.0) and Ticino, Switzerland (2.0). European females had the lowest rates in Italy (Nuoro, 0.1), Kielce Poland (0.1), Bulgaria (2.0), Cyprus (2.0), and Ukraine (0.2), and highest rates in Graubunden and Glarus, Switzerland (1.6), Iceland (1.3), and Norway (1.1). In Asia males, rates ranged from the lowest in Yangcheng County, China (0.0), and (0.1) in India (Dindigul, Ambilikkai; Poona and Trivandrum), and Kuwait, among non-​Kuwaitis (0.1) and Thailand (Khon Kaen and Lampang), to the largest in Iran (1.4). In Asian females, rates ranged from the lowest (0.0) in Trivandrum, India; Chonburi, Thailand and (0.1) in India (Dindigul, Ambilikkai; Karunagappally, Mizoram and New Delhi), Korea (Jejudo), Malaysia (Penang: Malay), Singapore: Malay and Thailand (Khon Kaen) to the largest in Singapore: Indian (1.2). In the African continent, incidence rates were lowest among males in Kyadonda County, Uganda (0.1), and among females in Setif, Algeria (0.1), and Africans in Harare, Zimbabwe (0.1). The highest rates were among male Africans of Harare, Zimbabwe (0.8). In the Oceania continent, highest rates were found among male indigenous populations:  Australia, Northern Territory–​ Indigenous (males:  3.3); New Zealand–​Maori (males:  2.7, females:  2.2) and New Zealand–​Pacific

Rates per 1,000,000

25.0 20.0 15.0 10.0 5.0 0.0 2008

2009

2010

2011

2012

Year of Diagnosis Black -Carcinoid/NET

White-Carcinoid/NET

Black-Adenocarcinoma

White-Adenocarcinoma

Figure  35–​2. Age-​adjusted incidence rates of adenocarcinoma and carcinoid/​neuroendocrine of the small intestine by race in white and black populations in the United States. SEER-​18 (2008–​2012).

During the 40-​year period from 1973 to 2012, age-​adjusted incidence rates of small intestine cancer reported by the original nine registries in the US SEER program increased by 141%. The annual percent change (APC) in incidence over this period was 2.2%, while death rates decreased (APC = –​0.1%) (SEER, 2015a). The annual increases in the incidence rate (per 100,000) were observed in both males (2.2%) and females (2.2%), and in whites (2.3%), blacks (2.5%), and other races combined (American Indian/​Alaskan Native and Asian/​Pacific Islander, 1.4%). This increasing trend was evident when race groups were stratified by gender, for males in each race—​white males (2.3%), black males (2.1%), other race males (1.7%)—​and for white (2.1%) and black females (2.9%). During the most recent time period, 2008–​ 2012, overall incidence and mortality rates have remained relatively stable (SEER, 2015b). The trend patterns reflect in part, stable collection criteria since implementation of the revised International Classification of Diseases for Oncology (ICD-​O-​2) in 1990, when carcinoid tumors became classified as malignant, and since 1992, when they were reported to the US SEER Program (Qubaiah et al., 2010).

ANATOMY The small intestine is a coiled tubular organ that extends on average 6–​ 7 meters from the gastric pylorus to its insertion into the large intestine at the junction of the cecum and ascending colon. Three subdivisions of the small intestine—​duodenum, jejunum and ileum—​are defined by various anatomical and histological features. The duodenum, the most proximal and shortest portion (about 0.25 meters), extends from the gastric pylorus to the duodenal-​jejunal junction. The biliary and main pancreatic ducts empty their contents into the descending segment of the duodenum, which forms a C-​shape around the head of the pancreas. The small intestine distal to the duodenal-​jejunal junction is divided somewhat arbitrarily into the jejunum (about 2.5 meters) and the ileum (about 3.6 meters). The small intestine comprises about 90% of the mucosal absorptive surface area of the gastrointestinal system. As noted previously, in 2015, only 3.2% of digestive system cancer cases and 0.8% of deaths in the United States were observed in the small intestine.

HISTOPATHOLOGY There are at least 40 pathologic subtypes of neoplasms diagnosed in the small intestine; however, more than 90% encompass neuroendocrine carcinoid, adenocarcinoma, lymphoma, and stromal tumors (Weiss and Yang, 1987). The classification of epithelial cancers includes adenocarcinoma, mucinous carcinoma, small cell carcinoma, adeno-​squamous carcinoma, medullary carcinoma, and undifferentiated carcinoma. Adenocarcinomas are relatively more prominent in the duodenum (0.4) compared with the jejunum (0.1) and the ileum (0.1) per 100,000 population (Table 35–2; see also Figure 35–4 [per million]). Adenocarcinomas in the small intestine accounted for 26% of the total of cancers in the small intestine registered in SEER (2008–​ 2012). In contrast, the proportion of total cancers in the colon classified as adenocarcinomas exceeded 92%. Carcinoid tumors, or argentaffinomas, are neuroendocrine tumors derived from enterochromaffin or Kulchitsky cells that are capable of producing serotonin (5-​hydroxytryptamine). Neuroendocrine tumors represent a heterogeneous group of neoplasms that exhibit distinctive morphologic and biologic features (Sorbye et al., 2014). Such tumors originate from pancreatic islet cells, gastroenteric tissues, gallbladder and biliary duct system, respiratory epithelium, and thyroid parafollicular cells (Kloppel et al., 2004). The neuroendocrine cells are capable

 673



673

Small Intestine Cancer

Table 35–​1.  Age-​Adjusted Incidence and Mortality Rates and Proportions of Cancers of the Small Intestine and Colon by Histology, Race, and Sex, SEER-​18 and US Mortality Data (2008–​2012) Rate per 100,000 (%) Site and Histology Small Intestine Incidence Carcinoid/​Neuroendocrine Adenocarcinoma Non-​Hodgkin Lymphoma Stromal Tumors Other Histology Colon Incidence Carcinoid/​Neuroendocrine Adenocarcinoma Non-​Hodgkin Lymphoma Stromal Tumors Other Histology Incidence Rate Ratios^ Carcinoid/​Neuroendocrine Adenocarcinoma Non-​Hodgkin Lymphoma Stromal Tumors Other Histology Small Intestine Mortality Overall U.S.** Colon Mortality Overall U.S.**

White Men

Black Men

Other* Men

White Women

Black Women

Other* Women

Total

3.1 (80.4%) 1.4 (44.5%) 0.8 (25.4%) 0.6 (17.8%) 0.2 (8.2%) 0.1 (4.0%) 33.2 (79.2%) 0.6 (2.1%) 30.7 (92.6%) 0.4 (1.1%) 0.0 (0.1%) 1.5 (4.2%)

4.6 (13.5%) 2.3 (52.2%) 1.7 (33.1%) 0.2 (5.4%) 0.2 (5.0%) 0.2 (4.3%) 44.8 (12.1%) 0.8 (2.1%) 41.5 (92.8%) 0.3 (0.6%) 0.1 (0.1%) 2.3 (4.3%)

1.9 (6.1%) 0.5 (27.5%) 0.5 (26.2%) 0.5 (25.9%) 0.3 (14.9%) 0.1 (5.5%) 29.3 (8.8%) 0.3 (1.4%) 27.2 (93.0%) 0.5 (1.6%) 0.0 (0.1%) 1.3 (4.0%)

2.1 (77.8%) 1.0 (48.5%) 0.5 (25.0%) 0.3 (14.2%) 0.2 (8.3%) 0.1 (4.0%) 27.0 (78.4%) 0.6 (2.2%) 24.8 (91.6%) 0.2 (0.6%) 0.0 (0.0%) 1.3 (5.5%)

3.3 (15.9%) 1.8 (55.3%) 1.0 (31.1%) 0.2 (5.4%) 0.1 (4.5%) 0.1 (3.7%) 35.4 (12.8%) 0.7 (2.3%) 32.6 (92.2%) 0.1 (0.3%) 0.1 (0.2%) 1.9 (5.0%)

1.3 (6.3%) 0.3 (24.5%) 0.4 (31.8%) 0.2 (19.1%) 0.2 (17.9%) 0.1 (6.7%) 23.5 (8.7%) 0.4 (2.0%) 21.6 (92.2%) 0.2 (0.9%) 0.0 (0.1%) 1.2 (4.8%)

2.6 (100%) 1.2 (41.6%) 0.7 (26.4%) 0.4 (15.0%) 0.2 (8.2%) 0.1 (4.1%) 30.4 (100%) 0.6 (2.1%) 28.0 (92.2%) 0.3 (0.9%) 0.0 (0.1%) 1.4 (4.8%)

0.4 38.4 0.7 0.0 15.0

0.3 24.4 1.5 0.5 11.5

0.6 54.4 1.0 0.0 13.0

0.6 49.6 0.7 0.0 13.0

0.4 32.6 0.5 1.0 19.0

1.3 54.0 1.0 0.0 12.0

0.5 40.0 0.8 0.0 14.0

0.4

0.7

0.3

0.3

0.5

0.2

0.4

14.5

22.4

10.5

10.5

15.3

7.9

12.7

Note: Rates are per 100,000 and age-​adjusted to the 2000 US Std Population (19 age groups—​Census P25-​1130) standard. * Other race ^ Rate ratios = sex, race, histology-​specific colon rate divided by small intestine rate. ** Mortality of small intestine and colon site groups; lymphomas excluded. Source, Incidence: Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Incidence—​SEER 9 Regs Research Data, Nov 2014 Sub (1973–​2012) Katrina/​Rita Population Adjustment—​Linked to County Attributes—​Total U.S., 1969–​2013 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2015, based on the November 2014 submission. Source, Mortality: Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Mortality—​All COD, Aggregated With State, Total U.S. (1969–​2012) Katrina/​Rita Population Adjustment, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2015. Underlying mortality data provided by NCHS (www.cdc.gov/​nchs).

of producing various polypeptide hormones, including serotonin, gastrin, histamine, insulin, glucagon, vasoactive intestinal peptide, calcitonin, and somatostatin. The neuroendocrine tumors may be classified according to their embryologic origin into tumors of the “foregut” (bronchopulmonary, stomach, pancreas, duodenum), “mid gut” (jejunum, ileum, appendix, right colon) and “hind gut” (left colon, rectum) (Arnold et al., 2004). Within the small intestine, neuroendocrine carcinoid tumors occur predominantly in the ileum (0.5 per 100,000 population). In the duodenum, these tumors are classified as gastrinomas (Modlin et al., 2008). The non-​Hodgkin lymphomas comprise a broad spectrum of lymphoproliferative neoplasms arising from B cells or T cells. Primary lymphomas of the small intestine include mucosa-​associated lymphoid tissue lymphoma (MALT), diffuse large B-​cell lymphoma, immunoproliferative lymphoma, and enteropathy-​associated T-​cell lymphoma. Diffuse large B-​cell, immunoblastic, and Burkitt lymphomas have been associated with human immunodeficiency viral infection. The majority of enteric lymphomas are B-​cell lymphomas (Banks, 2007). The origin and progression of B-​cell lymphomas involve complex interactions of host immune cells to foreign antigens, accompanied by transforming molecular events. The duodenum is the most common site for primary follicular lymphoma. Diffuse large B-​cell lymphomas account for approximately 45% of primary lymphomas of the small intestine. Although the stomach is the site for the majority of Helicobacter pylori-​associated MALT lymphomas, the terminal ileum is the most common site in the small intestine. Immunoproliferative lymphoma (immunoproliferative small intestinal disease; IPSID) tends to involve the duodenum or jejunum; IPSID is considered a subtype of MALT lymphoma that has been reported in the Middle East and in the Cape region of South Africa. Enteropathy-​type T-​cell lymphoma may arise

in patients with celiac disease, although the majority of such tumors, while associated with villous atrophy, are not accompanied by clinical symptoms of malabsorption. The most commonly occurring sarcoma in the jejunum and ileum is the gastrointestinal stromal tumor (GIST) (Joensuu et  al., 2013). These mesenchymal tumors are most often diagnosed in the stomach, but have been diagnosed in the colon, esophagus, mesentery, omentum, and peritoneum. Before the use of immunohistochemistry and the demonstration of expression of cell markers such as platelet-​ derived growth factor receptor alpha (PDGFRA) and CD117/​KIT, many of these tumors were classified as leiomyosarcomas or neurogenic tumors (Ma et al., 2015). The cell of origin is the interstitial cell of Cajal, which is interspersed between the circular and longitudinal muscle layers, and normally functions within the autonomic nervous system. The development of GISTs is associated with oncogenic activating mutations of KIT and PDGFRA genes that function as tyrosine kinase receptors, and provide an important therapeutic target in the treatment of GISTs (Rubin et al., 2001).

ADENOMA: PRECURSOR OF ADENOCARCINOMA As in the colon, the adenoma in the small intestine is a precursor of adenocarcinoma (Sellner, 1990). Residual adenomatous tissue is observed commonly in the margins of sporadic carcinomas, and of carcinomas in patients with familial adenomatous polyposis. The spatial distribution of adenomas in the small intestine parallels the distribution of carcinomas. Adenomas are located more frequently in the duodenum than in the distal small intestine, particularly in the periampullary region in proximity to the common bile and pancreatic ducts.

674

674

PART IV:  Cancers by Tissue of Origin Table 35–​2. Incidence and Proportion of Cancers of the Small Intestine by Anatomic Subsite and Histology, SEER-​18 (2008–​2012) Anatomic Sub-site and Histology

Rate

Count

%

Duodenum (35%)

0.9

4,069

100.0%

Adenocarcinoma

0.4

1782

43.8%

NHL

0.1

463

11.4%

Carcinoid

0.3

1421

34.9%

Stromal Tumors

0.0

186

4.6%

Other Histologies

0.0

217

5.3%

0.3 0.1 0.1 0.1 0.1 0.0 0.7 0.1 0.1 0.5 0.0 0.0 0.7 0.1 0.1 0.3 0.1 0.0

1,192 413 228 256 231 64 3,009 384 390 2,072 106 57 3,214 458 647 1,550 423 136

100.0% 34.6% 19.1% 21.5% 19.4% 5.4% 100.0% 12.8% 13.0% 68.9% 3.5% 1.9% 100.0% 14.3% 20.1% 48.2% 13.2% 4.2%

Jejunum (10%) Adenocarcinoma NHL Carcinoid Stromal Tumors Other Histologies Ileum (26%) Adenocarcinoma NHL Carcinoid Stromal Tumors Other Histologies Other Small Intestine (28%) Adenocarcinoma NHL Carcinoid Stromal Tumors Other Histologies

Note: Rates are per 100,000 and age-​adjusted to the 2000 US Std Population (19 age groups—​Census P25-​1130) standard. Source: Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer. gov) SEER*Stat Database: Incidence—​SEER 18 Regs Research Data + Hurricane Katrina Impacted Louisiana Cases, Nov 2014 Sub (2000–​2012) Katrina/​Rita Population Adjustment—​Linked to County Attributes—​Total U.S., 1969–​2013 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2015, based on the November 2014 submission.

Small Intestine Rates per 1,000,000

The risk potential for neoplastic transformation of adenomas increases with components of villous histopathology, increasing size in cross-​ sectional diameter (≥ 1 cm), and with a higher grade of dysplasia. The pathogenesis of neoplastic transformation of adenomas and of invasive

4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 2008

2010

2009

2011

2012

Year of Diagnosis Duodenum Jejunum

Figure  35–​3. Age-​standardized (world population) incidence rates per 100,000 of small intestine cancer in males and females (2003–​2007). Cancer in Five Continents, Vol. IX, IARC (2014).

Ileum Other Small Intestine

Figure 35–​4.  Age-​adjusted incidence rates of adenocarcinoma of the small intestine by cancer site. SEER-​18 (2008–​2012).

 675



Small Intestine Cancer

adenocarcinomas in the small intestine involves mutational and epigenetic pathways that are similar to, but not identical with, multistep events in colorectal neoplasia (Wheeler et al., 2002). Although adenomatous polyposis coli mutations appear to be less common when compared to their significance in early transformational events in colon cancer, hypermethylation of mismatch repair genes and tumor suppressor genes and abnormal expression of E-​cadherin and β-​catenin appear to contribute to small intestinal and colorectal carcinogenesis (Arber et al., 1999; Spigelman et al., 1994).

ETIOLOGIC RISK FACTORS Host Factors Genetics

There are several genetic syndromes that predispose to multicentric cancers in the small and large intestines, and other organ sites. Familial cancers are typically diagnosed at a younger age than when diagnosed in individuals without a family history. Hereditary non-​ polyposis colorectal cancer (HNPCC), or Lynch syndrome, is characterized by high microsatellite instability indicative of DNA mismatch repair dysfunction (Lynch et  al., 2009). Most germline mutations reside at hMLH1 and hMSH2 loci (Vasen et al., 2013). HNPCC is associated with an increased risk of developing extra-​colonic carcinomas in the small intestine, endometrium, pancreas, renal pelvis, stomach, liver and biliary tract, ovary, prostate, breast, and central nervous system (Aarnio et al., 1999; Engel et al., 2012; Vasen et al., 2013; Win et al., 2012). The lifetime risk of small intestinal neoplasms in patients with HNPCC has been estimated to range from 1% to 7% (Lynch et  al., 1989; Vasen et al., 2001, 2006). The tumors are primarily located in the duodenum, and manifest a high level of microsatellite instability, associated with a distinctive phenotype of poor differentiation, prominent lymphocyte infiltration, and mucin production (Rodriguez-​Bigas et al., 1998; Schulmann et al., 2005). Microsatellites contain a limited number of DNA sequences repeated in tandem and exhibit polymorphisms in length throughout the genome. Because they are repetitive, microsatellites are prone to strand slippage and errors in replication. Familial adenomatous polyposis (FAP) is an autosomal dominant disorder caused by germline mutation in the adenomatous polyposis coli gene on chromosome 5q21, resulting in hundreds to thousands of adenomatous polyps throughout the colon and rectum, with an almost 100% lifetime risk of developing colorectal cancer (Arber and Moshkowitz, 2011). The duodenum is another common site for adenomas in FAP, occurring in 30%–​70% of patients (Arber and Moshkowitz, 2011). Risks of periampullary and duodenal adenocarcinomas in FAP patients are increased more than 100-​fold (Offerhaus et al., 1992). The Peutz-​ Jeghers syndrome is associated with hamartomatous polyps in the small intestine, colon, and stomach. The disorder is an autosomal dominant syndrome caused by a germline mutation in the serine/​threonine kinase gene, STK11/​LKB1, a tumor suppressor gene located on chromosome 19p13.3 (Gruber et  al., 1998; Jenne et  al., 1998; Sato et  al., 2001). Most mutations are small deletions, insertions, or base substitutions, resulting in an abnormal truncated protein. Another prominent feature is the distribution of melanin-​pigmented lesions on the lips, perioral region, hands, and buccal mucosa. Patients with the syndrome are at increased risk of small bowel, other gastrointestinal, breast, ovarian, uterine cervical, testicular, and lung cancers (Boardman et al., 1998; Giardiello et al., 2000; Konishi et al., 1987; Lim et al., 2004). Hereditary and sporadic gastrointestinal carcinoid tumors that are associated with point mutations or deletions of the MEN1 gene on chromosome 11q13 tend to be gastrin-​cell tumors of duodenal or proximal jejunal origin (D’Adda et al., 2002). The MEN1 gene encodes a protein product “menin” that is ubiquitously expressed in endocrine and non-​endocrine tissues. Menin exhibits tumor suppressor activity, presumably by modulating histone methylation in promoters of HOX genes and various cyclin-​dependent kinase inhibitors. Distal jejunal and ileal carcinoid tumors tend to be more specifically associated with

675

inactivation or losses in chromosomes 18p, 18q, 11q, 16q, and 9p, and to be associated with overexpression of transforming growth factors α and ß, epidermal growth factor, and fibroblast growth factor, and neural cell adhesion molecules and their receptors (Kidd et al., 2007; Kytola et al., 2001).

Celiac Disease: Perturbations in the Immune System Patients with a history of celiac disease or gluten-​sensitive enteropathy are at increased risk of T-​cell and, with lesser frequency, β-​cell non-​Hodgkin lymphoma, and of adenocarcinoma in the small intestine (Anderson et al., 2007). The relative risk of gluten-​induced enteropathy associated with T-​cell lymphoma is increased approximately 10-​fold in patients with untreated or protracted clinical celiac disease refractory to treatment (Cellier et al., 2000; Green and Jabri, 2006). The disease is precipitated in genetically susceptible individuals by the ingestion of enzyme-​modified alcohol-​soluble fractions of gluten, namely gliadin, in wheat, barley, and rye. Untreated patients usually have increased antibodies against gluten, other food antigens, and autoantigens (e.g., transglutaminase) (Mowat, 2003). In affected patients, there is a striking coexistence of autoimmune diseases, such as type 1 diabetes, Sjögren’s syndrome, autoimmune thyroiditis and hepatitis, and dermatitis herpetiformis (Anderson and Mackay, 2014). Predisposition to gluten sensitivity is influenced by a variant of HLA-​ DQ2 (HLA-​DQ2.5). Usually 90% or more of celiac disease patients carry this HLA molecule (Fallang et al., 2009; Lionetti et al., 2014). The strong genetic influence in celiac disease is suggested by the high concordance (about 75%) in monozygotic twins, and by the high prevalence (10%) among first-​degree relatives. Serologic screening using anti-​gliadin and anti-​endomysial antibody testing, followed by endoscopic biopsy, has estimated that the prevalence of celiac disease in Europe and the United States is about 0.5%–​1% (Dube et al., 2005; Rewers, 2005). The dynamic phases of the intestinal lesions in celiac disease are classified pathologically into three stages: infiltration of intraepithelial lymphocytes in the villous epithelium; mucosal hyperplasia associated with infiltration of CD4+ T cells, macrophages, dendritic cells, and neutrophils; and atrophy of villous epithelium. The extent of mucosal damage is correlated with serum titers of immunoglobin A (IgA) endomysial antibodies and tissue transglutaminase antibodies. The immune reaction to gliadin is accompanied by inflammatory cytokine-​mediated pathogenic responses that are ultimately destructive and oncogenic in the intestinal mucosa (Kagnoff, 2007).

Crohn’s Disease

Chronic inflammatory bowel diseases include two clinically and pathologically distinctive autoimmune diseases: Crohn’s disease (CD) and ulcerative colitis (UC). CD is a chronic inflammatory condition presenting principally in the distal small intestine, but may involve the proximal colon, perianal tissues, and other digestive organs. With increasing chronicity, the ulcerating inflammatory lesions coalesce and tend to be transmural (extending from mucosa to serosa) and granulomatous. The granulomatous lesions are composed of macrophages, lymphocytes, plasma cells, and multinucleated giant cells that are formed in response to the release of cytokines such as tumor necrosis factor. Genome-​wide association studies have linked CD with extra-​intestinal inflammatory diseases such as polyarticular arthritis, ankylosing spondylitis, primary sclerosing cholangitis, and associated autoimmune diseases (Fasano and Catassi, 2012). The estimated annual incidence of CD, greater than 15 per 100,000, is highest in North America, Northern and Western European countries, and among Ashkenazi Jews. The risk of cancer of the small intestine or of the colon in CD patients is correlated with early age at onset, duration and extent of disease, and the prevalence of dysplastic mucosal lesions (Beaugerie and Itzkowitz, 2015). In patients with CD, the risk of adenocarcinoma in the small intestine was increased 20 to 30 times compared to risk in patients without CD. The cumulative incidence at 10 years was estimated as 0.2% (95% CI = 0.0%–​0.8%), and at 25 years, 2.2% (95% CI = 0.7%–​6.4%). In a meta-​analysis of six population-​based cohort studies, the age-​standardized rate ratio for

67

676

PART IV:  Cancers by Tissue of Origin

colorectal cancer in patients with CD of the colon was estimated as 2.5 (95% CI = 1.7%–​3.5%) (Jess et al., 2005). The integrity of the mucosal barrier lining the intestinal tract, which is maintained by tight junctions between epithelial cells, and the relative impermeability of the apical villous brush border, are important components of the innate immune system. The intestinal tract is exposed continuously to commensal indigenous microbial flora and potentially to pathogenic organisms. Enteric epithelial cells, in particular Paneth cells, located in the basal portion of the intestinal crypts, release lysozyme, phospholipase A, and peptides called defensins, all of which bind and break down cell walls of microorganisms (Wehkamp et al., 2005). The pathophysiology of CD underscores the pathogenic significance of the permeability of the mucosal barrier, and of the aberrant immune response to indigenous microorganisms and foreign antigens. The pathogenesis of CD as shown in animal and human studies would suggest that protracted inflammation and its sequelae arise from the interaction of luminal microflora, immune-​ mediated tissue injury, and genetic susceptibility (Baumgart and Sandborn, 2012). The metabolic products of activated inflammatory cells are accompanied by excessive formation of reactive oxygen and nitrogen species that are potentially damaging to DNA and the integrity of cell surface membranes. Research on the pathogenesis of inflammatory bowel disease is focusing on the complex interplay of environmental triggers, innate and adaptive immunity, and genetic susceptibility factors (Ohman and Simren, 2010). Genome-​wide association studies have identified more than 90 susceptibility loci for CD. Approximately 50% of loci are also evident in autoimmune diseases. These putative loci are involved in gut homeostasis, maintenance of the mucosal epithelial barrier, integrity of innate and adaptive immune mechanisms, and autophagy. Autophagy is an adaptive process that involves the sequestration and degradation of cytoplasmic organelles, including microbial pathogens, by lysosomes. In this discussion, we will focus on NOD2 (nucleotide-​binding oligomerization domain) /​CARD15 susceptibility gene, located on chromosome 16q12.1. NOD2 is prevalent in 8%–​25% of CD patients. The gene functions as an intracellular sensor of muramyl dipeptide subunits of peptidoglycans in the bacterial cell wall and regulates Paneth cell expression of antimicrobial defensins (Hugot et al., 2001). It appears to modulate signaling through Toll-​like receptor pathways and to trigger autophagy after detecting intracellular bacteria. NOD2 polymorphic variants associated with CD patients may result in defective clearance of intracellular bacteria and dysfunction in inflammatory response mechanisms.

LIFESTYLE RISK FACTORS Tobacco and Alcohol Risk factors such as regular use of tobacco and alcohol, physical inactivity, obesity, and nutrition have been studied extensively in patients with colon cancer. In contrast, epidemiologic studies of patients with cancer of the small intestine have been limited due to the rarity of these tumors, coupled with biologic heterogeneity among the major histologic subtypes of tumor. Early studies of tobacco and alcohol that were based solely on the review of medical records (Chen et al., 1994; Neugut et al., 1998), or next-​of-​kin surrogate interviews (Chow et al., 1993), did not explore the effect of cumulative lifetime exposure to these carcinogens, nor did they examine induction-​latency intervals or dose–​response trends. The findings have been inconsistent with some suggestion of variation by histologic subtype in the risk associated with smoking, with increases in risk for malignant carcinoid tumors (Kaerlev et al., 2002), but no association with adenocarcinoma of the small intestine (Kaerlev et al., 2000). Wu et al. (1997) reported a gender difference in risk of adenocarcinoma of the small intestine associated with smoking cigarettes, after adjusting for ethanol use, with odds ratios of 2.6 in men and 1.1 in women. However, the estimates of risk were similar among former and current smokers, with no apparent difference in risk related to the age at the initiation of smoking, or dose (measured in number of cigarettes smoked per day, and total years of cigarette smoking) (Wu et al., 1997). More recent studies reported

no association with either subtype (Boffetta et al., 2012; Cross et al., 2013). In fact, in a recently published meta-​analysis of 148 cases and 3350 non-​cases, no association was noted between smoking history and risk of adenocarcinoma (RRpooled  =  1.24; 95% CI  =  0.71–​2.17) (Bennett et al., 2015). Consumption of alcohol has been associated with small bowel adenocarcinomas independent of smoking history (Chen et al., 1994; Kaerlev et  al., 2000; Wu et  al., 1997). Wu et  al. (1997) reported a nearly 3-​fold increase in risk (OR  =  2.9; 95% CI  =  1.2, 7.1) for those who consumed more than 80 g (more than 6 drinks) per day, regardless of type of alcoholic beverage. Kaerlev et al. (2000) found intake of beer and spirits (≥ 24 g/​day), but not wine, was associated with an approximate 3.5-​fold increase in risk for adenocarcinomas. A  meta-​analysis of existing studies indicated a modest increase in risk of adenocarcinoma comparing the highest to lowest quintile of alcohol intake (RR = 1.82; 95% CI = 1.05, 3.15) controlling for the heterogeneity of findings. The results of studies of alcohol intake and carcinoid tumors have been inconsistent (Chen et  al., 1994; Cross et al., 2013; Hassan et al., 2008; Kaerlev et al., 2002). Various tumorigenic mechanisms have been attributed to alcohol and its genotoxic metabolite, acetaldehyde, including impaired DNA methylation, inflammation-​mediated oxidative stress, and perturbation of the intestinal mucosal immune response (Boffetta and Hashibe, 2006; Seitz and Stickel, 2007).

Diet and Body Mass A number of epidemiologic studies indicate that diet plays a role in small intestine cancer, supported by early ecologic evidence in that global incidence rates correlate with per capita consumption of dietary fat and red meat (Chow et  al., 1993; Haselkorn et  al., 2005; Lowenfels and Anderson, 1977; Negri et  al., 1999). Among 494,000 participants included in the NIH-​AARP cohort, diets high in saturated fat were strongly associated with carcinoid tumors of the small intestine (HR = 3.18; 95% CI = 1.62–​6.25) with an elevated, but more modest association with adenocarcinomas (Cross et  al., 2008). No associations were observed for either histologic subtype with consumption of red or processed meats, or mono-​and polyunsaturated fats. Other investigations of dietary sugar, fiber, and intake of whole grains in the same cohort indicated that high fructose sugar consumption was associated with an increased risk for cancer of the small bowel, irrespective of subtype (RR = 2.10; 95% CI = 1.16–​4.16, comparing highest to lowest quintile of intake) (Tasevska et al., 2012), and fiber from grains and whole grain intake was associated with a reduced risk (RR  =  0.51; 95% CI  =  0.29–​ 0.89 and RR = 0.59; 95% CI = 0.33–​1.05, respectively) (Schatzkin et al., 2008). The observed associations between increased body mass and cancer of the small intestine have shown estimations of relative risk to vary by subtype, gender, race, and anatomic location (Boffetta et al., 2012; Bjorge, Tretli, and Engeland, 2005; Chow et al., 1993; Cross et al., 2013; Hassan et al., 2008; Samanic et al., 2004; Wolk et al., 2001). The observed stratum-​specific differences are worthy of further study to confirm findings. Wolk et  al. (2001) observed significant elevation in risk of small bowel cancer among obese men and women (SIR = 2.8; 95% CI = 1.6, 4.5). In a large cohort of US veterans, obesity (body mass index [BMI] ≥ 30  kg/​m2) was associated with an approximate 60% increase in risk of cancer of the small intestine among white men (RR = 1.58; 95% CI = 1.18, 2.12), but not among black men (RR = 1.07; 95% CI = 0.54, 2.08). This risk was further elevated among whites when restricted to duodenal cancers (RR = 2.10; 95% CI = 1.38, 3.22) (Samanic et al., 2004). Similarly, in a Norwegian cohort study of 1162 cases in 2 million men and women, obese men were 60% more likely to be diagnosed with small bowel cancer (RR  =  1.59; 95% CI  =  1.13, 2.23) compared to men with normal BMIs (18.5–​24.9 kg/​m2). No relationship between BMI and cancer risk was observed among women, except for cancers of the duodenum (RR  =  1.67; 95% CI  =  1.00, 2.80) (Bjorge et al., 2005).

 67



Small Intestine Cancer

BIOLOGIC IMPLICATIONS OF THE INFREQUENCY OF ADENOCARCINOMA IN THE SMALL INTESTINE The ratio of age-​standardized incidence for adenocarcinoma in the colon to that in the small intestine in 2012 (SEER, 2015b) indicated an approximate 40-​fold difference. The international incidence of adenocarcinoma of the small intestine is correlated positively with the geographic distribution of colorectal cancer incidence. The SEER cohort of patients who had survived 5 or more years after diagnosis of cancer of the small intestine experienced more than a two-​fold increase in risk of colorectal cancer (RR  =  2.62; 95% CI = 1.77, 3.75) (Neugut and Santos, 1993; Scelo et al., 2006). There was greater diversity of histopathologic types in the small intestine, with 26% of the total classified as adenocarcinoma, compared with more than 92% in the colon. The contrasting incidence may be viewed in the context of the differences in microbial density and the taxonomy of microorganisms in the fecal stream. The contents of the fecal stream remain liquid throughout the small intestine. Chyme, the resulting semifluid mixture of food and gastric secretions, is transported through the small intestine by peristaltic and segmenting waves, requiring about 3–​5 hours for passage from the pylorus to the ileocecal valve. The right colon carries out complex functions that include mixing and fermenting of the ileal effluent, and excretion and desiccation of the intraluminal contents, which are concluded in the left colon. Although the rate of transport in the small intestine is correlated with that in the colon, transit time through the small intestine is more rapid per unit length than in the colon. Slowing of the rate of intestinal transit augments the enterohepatic recycling of bile acids, the excreted fecal mass of microorganisms, anaerobic bacterial fermentation, and the rising pH of the fecal stream (Guarner and Malagelada, 2003; Lewis and Heaton, 1999). A diverse and dynamic microbial ecosystem is normally resident within the lumen or adherent to the mucosal surface of the small and large intestine. The prevalence of microbiota in different anatomic areas of the gastrointestinal tract is influenced by pH, bowel motility, interactions among the various species of microorganisms, mucosal immune mechanisms, and the oxidation–​reduction potential of the luminal contents. Decreased oxidation–​reduction potential predisposes to colonization with obligate anaerobic bacteria. Commensal bacteria of the normal intestinal microbiome have a regulatory effect on the emergence of potential microbial pathogens, a phenomenon described as “intestinal homeostasis.” The upper two-​thirds of the small intestine contain low concentrations of Gram-​positive bacteria such as lactobacilli and enterococci. The density of microbiota increases markedly in the distal ileum, presumably in part due to the reflux of fecal contents in the cecum through the ileocecal valve, and may contain up to 100 million organisms per gram of luminal contents. By contrast, the large intestine may harbor over 500 species of bacteria, mainly obligate anaerobic Gram-​negative and Gram-​positive organisms, at a concentration in excess of one trillion bacteria per gram of luminal contents, that may exhibit complex enzymatic and metabolic functions (Ramakrishna, 2007). The enormous number and diversity of microorganisms in the intestinal tract impact the development of gut-​ associated lymphoid tissue and immune capacity. Mucosal inflammation may arise as a result of persistent, inappropriate perturbation of the intestinal microbial flora, a phenomenon described as “microbial dysbiosis” (Mai and Morris, Jr., 2013; Sobhani et al., 2013). The anaerobic microorganisms generate metabolizing enzymes, such as β-​glucuronidase, β-​glucosidase, nitro reductases, and decarboxylases that act on various potential tumor-​promoting substrates, such as bile salts that are ionized forms of bile acid molecules. Greater than 95% of the bile salts that are synthesized in the liver are reabsorbed, either by passive diffusion in the proximal jejunum, or by active transport in the distal ileum. The enterohepatic recirculation of bile salts recycles about six to eight times daily. The carboxyl group in the side chain of the bile salt molecule, when activated, can form glycine or taurine amides known as conjugated bile salts. Intestinal anaerobic bacteria deconjugate and dehydroxylate the bile salts by removing glycine and taurine residues and the hydroxyl group. The

677

primary bile salts are then biochemically transformed into tumor-​ promoting secondary bile acids, deoxycholic and lithocholic acids, which are less soluble in intestinal chyme, and less readily absorbed from the intestinal lumen, than the primary bile salts (Imray et  al., 1992; Ochsenkuhn et al., 1999; Ross et al., 1991). The risk of developing cancer of the small intestine, especially of epithelial origin, is remarkably low given its large mucosal surface area and the abundance of crypts and stem cells. Tomasetti and Vogelstein (2015) have hypothesized that variations in cancer risk in various tissues may be explained by the variable lifetime number of stem cell divisions. When compared with the large intestine, they estimated that the numbers of stem cells per crypt in the small intestine were 15% (ileum) to 30% (duodenum and proximal jejunum) lower than in the colon. In addition, they estimated that stem cells in the duodenum and distal small intestine divide, respectively, only one-​third and two-​thirds as frequently as those in the large intestine (Tomasetti and Vogelstein, 2015). Each cell division would incur the stochastic risk of a mutational event. Presumably, the contrasting biochemical, microbial, and immunologic features of the small and large intestinal lumina would foster genetic and epigenetic events that are advantageous to neoplasia in the large intestine (Luzzatto and Pandolfi, 2015).

OPPORTUNITIES FOR PREVENTION Due to the rarity of carcinoma of the small intestine, preventive efforts should be focused on high-​risk groups in the population, namely those diagnosed with inflammatory bowel diseases, as well as patients with known germline mutations consistent with clinical genetic syndromes (i.e., HNPCC, PJS, and FAP). Most patients with symptomatic celiac disease improve by avoiding gluten-​containing grains. It has been shown that patients who adhere to a gluten-​free diet experience reduced risk of cancer, including enteropathy-​associated T-​cell lymphoma. However, of special concern are those individuals with refractory disease who do not respond to dietary intervention and continue to manifest incomplete histologic recovery. In patients with Crohn’s disease, the aim of medical therapy is to ameliorate the inflammatory response and to maintain remission with the use of immunomodulators, such as the thiopurine analogues, and “biological” agents, such as monoclonal antibody therapy directed against tumor necrosis factor-​alpha. Over the course of 5 years, only 25% of patients will remain in remission after diagnosis.

FUTURE RESEARCH The apparent similarities in the molecular genetic events of physiologic cell renewal and apoptosis, and the events of malignant cell transformation in the small and large intestines, contrast dramatically with global cancer incidence patterns. As stated previously, current research is focusing on the biochemistry and diversity of the intestinal microbial flora, enterohepatic bile acid metabolism, mucosal stem cell proliferation kinetics, and intestinal motility. Future research should focus on putative lifestyle risk factor and host factor components of the causal pathway in the adenoma–​carcinoma transformation and its pattern of anatomic distribution in the small intestine. The identification of specific familial germline mutations associated with polyposis and adenocarcinoma in the small intestine, large intestine, and other organ sites will serve to guide future research on the complexity of molecular mechanisms in intestinal carcinogenesis. Pathologic mechanisms associated with chronic inflammation and perturbations in immune mechanisms of response should continue to be investigated in studies of primary non-​Hodgkin lymphoma of the small intestine and associated autoimmune diseases. References Aarnio M, Sankila R, Pukkala E, et  al. 1999. Cancer risk in mutation carriers of DNA-​ mismatch-​ repair genes. Int J Cancer, 81(12), 214–​ 218. PMID:10188721. Anderson LA, McMillan SA, Watson RG, et al. 2007. Malignancy and mortality in a population-​based cohort of patients with coeliac disease or “gluten sensitivity.” World J Gastroenterol, 13(1), 146–​151. PMID:17206762.

678

678

PART IV:  Cancers by Tissue of Origin

Anderson WH, Mackay IR. 2014. Gut reactions: from celiac affection to autoimmune model. N Engl J Med, 371(1), 6–​7. PMID: 24988553. Arber N, Hibshoosh H, Yasui W, et  al. 1999. Abnormalities in the expression of cell cycle-​related proteins in tumors of the small bowel. Cancer Epidemiol Biomarkers Prev, 8(12), 1101–​1105. PMID: 10613343. Arber N, Moshkowitz M. 2011. Small bowel polyposis syndromes. Curr Gastroenterol Rep, 13(5), 435–​441. PMID: 21800071. Arnold CN, Sosnowski A, Blum HE. 2004. Analysis of molecular pathways in neuroendocrine cancers of the gastroenteropancreatic system. Ann NY Acad Sci, 1014, 218–​219. PMID:15153437. Banks PM. 2007. Gastrointestinal lymphoproliferative disorders. Histopathology, 50(1), 42–​54. PMID:17204020. Baumgart DC, Sandborn WJ. 2012. Crohn’s disease. Lancet, 380(9853), 1590–​ 1605. PMID: 22914295. Beaugerie L, Itzkowitz SH. 2015. Cancers complicating inflammatory bowel disease. N Engl J Med, 372(15), 1441–​1452. PMID: 26154801. Bennett CM, Coleman HG, Veal PG, Cantwell MM, Lau CC, Murray LJ. 2015. Lifestyle factors and small intestine adenocarcinoma risk:  a systematic review and meta-​analysis. Cancer Epidemiol, 39(3), 265–​273. PMID: 25736860. Bjorge T, Tretli S, Engeland A. 2005. Height and body mass index in relation to cancer of the small intestine in two million Norwegian men and women. Br J Cancer, 93(7), 807–​810. PMID: 16175183. Boardman LA, Thibodeau SN, Schaid DJ, et al. 1998. Increased risk for cancer in patients with the Peutz-​Jeghers syndrome. Ann Intern Med, 128(11), 896–​899. PMID: 9634427. Boffetta P, Hashibe M. 2006. Alcohol and cancer. Lancet Oncol, 7(2), 149–​156. PMID: 16455479. Boffetta P, Hazelton WD, Chen Y, et al. 2012. Body mass, tobacco smoking, alcohol drinking and risk of cancer of the small intestine: a pooled analysis of over 500,000 subjects in the Asia Cohort Consortium. Ann Oncol, 23(7), 1894–​1898. PMID: 22147734. Cellier C, Delabesse E, Helmer C, et al. 2000. Refractory sprue, coeliac disease, and enteropathy-​ associated T-​ cell lymphoma. French Coeliac Disease Study Group. Lancet, 356(9225), 203–​208. PMID: 10963198. Chen CC, Neugut AI, Rotterdam H. 1994. Risk factors for adenocarcinomas and malignant carcinoids of the small intestine:  preliminary findings. Cancer Epidemiol Biomarkers Prev, 3(3), 205–​207. PMID: 8019367. Chow WH, Linet MS, McLaughlin JK, Hsing AW, Chien HT, Blot WJ. 1993. Risk factors for small intestine cancer. Cancer Causes Control, 4(2), 163–​ 169. PMID: 8481495. Cross AJ, Hollenbeck AR, Park Y. 2013. A large prospective study of risk factors for adenocarcinomas and malignant carcinoid tumors of the small intestine. Cancer Causes Control, 24(9), 1737–​1746. PMID: 23812550. Cross AJ, Leitzmann MF, Subar AF, Thompson FE, Hollenbeck AR, Schatzkin A. 2008. A prospective study of meat and fat intake in relation to small intestinal cancer. Cancer Res, 68(22), 9274–​9279. PMID:19010900. D’Adda T, Pizzi S, Azzoni C, et  al. 2002. Different patterns of 11q allelic losses in digestive endocrine tumors. Hum Pathol, 33(3), 322–​ 329. PMID: 11979373. Dube C, Rostom A, Sy R, et  al. 2005. The prevalence of celiac disease in average-​ risk and at-​ risk Western European populations:  a systematic review. Gastroenterology, 128(1), S57–​S67. PMID:15825128. Engel C, Loeffler M, Steinke V, et  al. 2012. Risks of less common cancers in proven mutation carriers with lynch syndrome. J Clin Oncol, 30(35), 4409–​4415. PMID: 23091106. Fallang LE, Bergseng E, Hotta K, Berg-​Larsen A, Kim CY, Sollid LM. 2009. Differences in the risk of celiac disease associated with HLA-​DQ2.5 or HLA-​DQ2.2 are related to sustained gluten antigen presentation. Nat Immunol, 10(10), 1096–​1101. PMID: 19718029. Fasano A, Catassi C. 2012. Clinical practice:  celiac disease. N Engl J Med, 367(25), 2419–​2426. PMID: 23252527. Giardiello FM, Brensinger JD, Tersmette AC, et  al. 2000. Very high risk of cancer in familial Peutz-​Jeghers syndrome. Gastroenterology, 119(6), 1447–​1453. PMID:11113065. Goodman MT, Matsuno RK, Shvetsov YB. 2013. Racial and ethnic variation in the incidence of small-​bowel cancer subtypes in the United States, 1995–​ 2008. Dis Colon Rectum, 56(4), 441–​448. PMID: 23478611. Green PH, Jabri B. 2006. Celiac disease. Annu Rev Med, 57, 207–​ 221. PMID: 16409146. Gruber SB, Entius MM, Petersen GM, et al. 1998. Pathogenesis of adenocarcinoma in Peutz-​Jeghers syndrome. Cancer Res, 58(23), 5267–​5270. PMID: 9850045. Guarner F, Malagelada JR. 2003. Gut flora in health and disease. Lancet, 361(9356), 512–​519. PMID: 12583961.

Haselkorn T, Whittemore AS, Lilienfeld DE. 2005. Incidence of small bowel cancer in the United States and worldwide:  geographic, temporal, and racial differences. Cancer Causes Control, 16(7), 781–​ 787. PMID: 16132788. Hassan MM, Phan A, Li D, Dagohoy CG, Leary C, Yao JC. 2008. Risk factors associated with neuroendocrine tumors: a U.S.-​based case-​control study. Int J Cancer, 123(4), 867–​873. PMID:18491401. Hugot JP, Chamaillard M, Zouali H, Lesage S, Cezard JP, Belaiche J, et  al. 2001. Association of NOD2 leucine-​rich repeat variants with susceptibility to Crohn’s disease. Nature, 411 (6837), 599–​603. PMID:11385576. Imray CH, Radley S, Davis A, et  al. 1992. Faecal unconjugated bile acids in patients with colorectal cancer or polyps. Gut, 33(9), 1239–​1245. PMID: 1427378. Jenne DE, Reimann H, Nezu J, et al. 1998. Peutz-​Jeghers syndrome is caused by mutations in a novel serine threonine kinase. Nat Genet, 18(1), 38–​43. PMID: 9425897. Jess T, Gamborg M, Matzen P, Munkholm P, Sorensen TI. 2005. Increased risk of intestinal cancer in Crohn’s disease: a meta-​analysis of population-​based cohort studies. Am J Gastroenterol, 100(12), 2724–​2729. PMID: 16393226. Joensuu H, Hohenberger P, Corless CL. 2013. Gastrointestinal stromal tumour. Lancet, 382(9896), 973–​983. PMID: 23623056. Kaerlev L, Teglbjaerg PS, Sabroe S, et al. 2002. The importance of smoking and medical history for development of small bowel carcinoid tumor: a European population-​based case-​control study. Cancer Causes Control, 13(1), 27–​34. PMID: 11899115. Kaerlev L, Teglbjaerg PS, Sabroe S, et al. 2000. Is there an association between alcohol intake or smoking and small bowel adenocarcinoma? Results from a European multi-​center case-​control study. Cancer Causes Control, 11(9), 791–​797. PMID:11075867. Kagnoff MF. 2007. Celiac disease:  pathogenesis of a model immunogenetic disease. J Clin Invest, 117(1), 41–​49. PMID: 17200705. Kidd M, Modlin IM, Pfragner R, et al. 2007. Small bowel carcinoid (enterochromaffin cell) neoplasia exhibits transforming growth factor-​ beta1-​ mediated regulatory abnormalities including up-​regulation of C-​Myc and MTA1. Cancer, 109(12), 2420–​2431. PMID: 17469181. Kloppel G, Perren A, Heitz PU. 2004. The gastroenteropancreatic neuroendocrine cell system and its tumors: the WHO classification. Ann NY Acad Sci, 1014, 13–​27. PMID: 15153416. Konishi F, Wyse NE, Muto T, et al. (1987). Peutz-​Jeghers polyposis associated with carcinoma of the digestive organs: report of three cases and review of the literature. Dis Colon Rectum, 30(10), 790–​799. PMID: 3308369. Kytola S, Hoog A, Nord B, et  al. 2001. Comparative genomic hybridization identifies loss of 18q22-​qter as an early and specific event in tumorigenesis of midgut carcinoids. Am J Pathol, 158(5), 1803–​1808. PMID: 11337378. Lewis SJ, Heaton KW. 1999. The metabolic consequences of slow colonic transit. Am J Gastroenterol, 94(8), 2010–​2016. PMID: 10445521. Lim W, Olschwang S, Keller JJ, et al. 2004. Relative frequency and morphology of cancers in STK11 mutation carriers. Gastroenterology, 126(7), 1788–​1794. PMID: 15188174. Lionetti E, Castellaneta S, Francavilla R, et al. 2014. Introduction of gluten, HLA status, and the risk of celiac disease in children. N Engl J Med, 371(14), 1295–​1303. PMID: 25271602. Lowenfels AB, Anderson ME. 1977. Diet and cancer. Cancer, 39(4 suppl), 1809–​1814. PMID: 192435. Luzzatto L, Pandolfi PP. 2015. Causality and chance in the development of cancer. N Engl J Med, 373(1), 84–​88. PMID: 26132946. Lynch HT, Lynch PM, Lanspa SJ, Snyder CL, Lynch JF, Boland CR. 2009. Review of the Lynch syndrome:  history, molecular genetics, screening, differential diagnosis, and medicolegal ramifications. Clin Genet, 76(1), 1–​18. PMID: 19659756. Lynch HT, Smyrk TC, Lynch PM, et al. 1989. Adenocarcinoma of the small bowel in lynch syndrome II. Cancer, 64(10), 2178–​2183. PMID: 2804907. Ma GL, Murphy JD, Martinez ME, Sicklick JK. 2015. Epidemiology of gastrointestinal stromal tumors in the era of histology codes:  results of a population-​based study. Cancer Epidemiol Biomarkers Prev, 24(1), 298–​ 302. PMID: 25277795. Mai V, Morris JG, Jr. 2013. Need for prospective cohort studies to establish human gut microbiome contributions to disease risk. J Natl Cancer Inst, 105(24), 1850–​1851. PMID: 24316594. Modlin IM, Oberg K, Chung DC, et al. 2008. Gastroenteropancreatic neuroendocrine tumours. Lancet Oncol, 9(1), 61–​72. PMID: 18177818. Mowat AM. 2003. Coeliac disease: a meeting point for genetics, immunology, and protein chemistry. Lancet, 361(9365), 1290–​1292. PMID: 12699968. Negri E, Bosetti C, La VC, Fioretti F, Conti E, Franceschi S. 1999. Risk factors for adenocarcinoma of the small intestine. Int J Cancer, 82(2), 171–​174. PMID: 10389747.

 679



Small Intestine Cancer

Neugut AI, Jacobson JS, Suh S, Mukherjee R, Arber N. 1998. The epidemiology of cancer of the small bowel. Cancer Epidemiol Biomarkers Prev, 7(3), 243–​251. PMID: 9521441. Neugut AI, Santos J. 1993. The association between cancers of the small and large bowel. Cancer Epidemiol Biomarkers Prev, 2(6), 551–​553. PMID: 8268772. Ochsenkuhn T, Bayerdorffer E, Meining A, et al. 1999. Colonic mucosal proliferation is related to serum deoxycholic acid levels. Cancer, 85(8), 1664–​ 1669. PMID: 10223558. Offerhaus GJ, Giardiello FM, Krush AJ, et  al. 1992. The risk of upper gastrointestinal cancer in familial adenomatous polyposis. Gastroenterology, 102(6), 1980–​1982. PMID: 1316858. Ohman L, Simren M. 2010. Pathogenesis of IBS: role of inflammation, immunity and neuroimmune interactions. Nat Rev Gastroenterol Hepatol, 7(3), 163–​173. PMID: 20101257. Qubaiah O, Devesa SS, Platz CE, Huycke MM, Dores GM. 2010. Small intestinal cancer: a population-​based study of incidence and survival patterns in the United States, 1992 to 2006. Cancer Epidemiol Biomarkers Prev, 19(8), 1908–​1918. PMID: 20647399. Ramakrishna BS. 2007. Wanted:  diagnostic criteria for adult celiac disease. Indian J Gastroenterol, 26(4), 157–​158. PMID: 17986739. Rewers M. 2005. Epidemiology of celiac disease: what are the prevalence, incidence, and progression of celiac disease? Gastroenterology, 128(4 suppl 1), S47–​S51. PMID: 15825124. Rodriguez-​Bigas MA, Vasen HF, Lynch HT, et  al. 1998. Characteristics of small bowel carcinoma in hereditary nonpolyposis colorectal carcinoma. International Collaborative Group on HNPCC. Cancer, 83(2), 240–​244. PMID: 9669805. Ross RK, Hartnett NM, Bernstein L, Henderson BE. 1991. Epidemiology of adenocarcinomas of the small intestine: is bile a small bowel carcinogen? Br J Cancer, 63(1), 143–​145. PMID:1989654. Rubin BP, Singer S, Tsao C, et  al. 2001. KIT activation is a ubiquitous feature of gastrointestinal stromal tumors. Cancer Res, 61(22), 8118–​8121. PMID: 11719439. Samanic C, Gridley G, Chow WH, Lubin J, Hoover RN, Fraumeni JF, Jr. 2004. Obesity and cancer risk among white and black United States veterans. Cancer Causes Control, 15(1), 35–​43. PMID:14970733. Sato N, Rosty C, Jansen M, et al. 2001. STK11/​LKB1 Peutz-​Jeghers gene inactivation in intraductal papillary-​mucinous neoplasms of the pancreas. Am J Pathol, 159(6), 2017–​2022. PMID: 11733352. Scelo G, Boffetta P, Hemminki K, et  al. 2006. Associations between small intestine cancer and other primary cancers: an international population-​ based study. Int J Cancer, 118(1), 189–​196. PMID: 16003748. Schatzkin A, Park Y, Leitzmann MF, Hollenbeck AR, Cross AJ. 2008. Prospective study of dietary fiber, whole grain foods, and small intestinal cancer. Gastroenterology, 135(4), 1163–​1167. PMID: 18727930. Schulmann K, Mori Y, Croog V, et al. 2005. Molecular phenotype of inflammatory bowel disease-​associated neoplasms with microsatellite instability. Gastroenterology, 129(1), 74–​85. PMID: 16012936. SEER. 2015a. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Incidence—​SEER 18 Regs Research Data + Hurricane Katrina Impacted Louisiana Cases, Nov 2014 Sub (1973–​2012 varying)—​Linked to County Attributes—​Total U.S., 1969–​2013 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2015, based on the November 2014 submission. SEER. 2015b. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database:  Incidence—​ SEER 18 Regs Research Data + Hurricane Katrina Impacted Louisiana Cases, Nov 2014 Sub (2000–​2012) Katrina/​Rita Population Adjustment—​Linked

679

to County Attributes—​ Total U.S., 1969–​ 2013 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2015, based on the November 2014 submission. SEER. 2015c. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database:  Incidence—​SEER 9 Regs Research Data, Nov 2014 Sub (1973–​ 2012) Katrina/​ Rita Population Adjustment—​ Linked to County Attributes—​ Total U.S., 1969–​ 2013 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2015, based on the November 2014 submission. Seitz HK, Stickel F. 2007. Molecular mechanisms of alcohol-​mediated carcinogenesis. Nat Rev Cancer, 7(8), 599–​612. PMID: 17646865. Sellner F. 1990. Investigations on the significance of the adenoma-​carcinoma sequence in the small bowel. Cancer, 66(4), 702–​715. PMID: 2167140. Siegel RL, Miller KD, Jemal A. 2015. Cancer statistics, 2015. CA Cancer J Clin, 65(1), 5–​29. PMID: 25559415. Sobhani I, Amiot A, Le BY, et al. 2013. Microbial dysbiosis and colon carcinogenesis: could colon cancer be considered a bacteria-​related disease? Therap Adv Gastroenterol, 6(3), 215–​229. PMID: 23634186. Sorbye H, Strosberg J, Baudin E, Klimstra DS, Yao JC. 2014. Gastroenteropancreatic high-​grade neuroendocrine carcinoma. Cancer, 120(18), 2814–​2823. PMID: 24771552. Spigelman AD, Talbot IC, Penna C et  al. 1994. Evidence for adenoma-​ carcinoma sequence in the duodenum of patients with familial adenomatous polyposis. The Leeds Castle Polyposis Group (Upper Gastrointestinal Committee). J Clin Pathol, 47(8), 709–​710. PMID: 7962621. Tasevska N, Jiao L, Cross AJ, et al. 2012. Sugars in diet and risk of cancer in the NIH-​AARP Diet and Health Study. Int J Cancer, 130(1), 159–​169. PMCID: PMC3494407. Tomasetti C, Vogelstein B. 2015. Cancer risk:  role of environment-​response. Science, 347(6223), 729–​731. PMID: 25554788. Vasen HF, Blanco I, Aktan-​Collan K, et al. 2013. Revised guidelines for the clinical management of Lynch syndrome (HNPCC): recommendations by a group of European experts. Gut, 62(6), 812–​823. PMID: 23408351. Vasen HF, de Vos Tot Nederveen Cappel WH. 2006. An evidence-​based review on surveillance for Lynch syndrome. Dis Colon Rectum, 49(11), 1797–​ 1798. PMID: 17053686. Vasen HF, Stormorken A, Menko FH, et al. 2001. MSH2 mutation carriers are at higher risk of cancer than MLH1 mutation carriers: a study of hereditary nonpolyposis colorectal cancer families. J Clin Oncol, 19(20), 4074–​ 4080. PMID: 11600610. Wehkamp J, Salzman NH, Porter E, et  al. 2005. Reduced Paneth cell alpha-​ defensins in ileal Crohn’s disease. Proc Natl Acad Sci USA, 102(50), 18129–​18134. PMID: 16330776. Weiss NS, Yang CP. 1987. Incidence of histologic types of cancer of the small intestine. J Natl Cancer Inst, 78(4), 653–​656. PMID: 3470541. Wheeler JM, Warren BF, Mortensen NJ, et al. 2002. An insight into the genetic pathway of adenocarcinoma of the small intestine. Gut, 50(2), 218–​223. PMID: 11788563. Win AK, Lindor NM, Young JP, et al. 2012. Risks of primary extracolonic cancers following colorectal cancer in lynch syndrome. J Natl Cancer Inst, 104(18), 1363–​1372. PMID: 22933731. Wolk A, Gridley G, Svensson M, et  al. 2001. A prospective study of obesity and cancer risk (Sweden). Cancer Causes Control, 12(1), 13–​21. PMID: 11227921. Wu AH, Yu MC, Mack TM. 1997. Smoking, alcohol use, dietary factors and risk of small intestinal adenocarcinoma. Int J Cancer, 70(5), 512–​517. PMID: 9052748.

680

 681

36

Cancers of the Colon and Rectum KANA WU, NANA KEUM, REIKO NISHIHARA, AND EDWARD L. GIOVANNUCCI

OVERVIEW Worldwide, colorectal cancer (CRC) is the third most common cancer in men and second in women, with annual estimates of 1.4 million newly diagnosed cases and over 690,000 deaths. Incidence rates relate closely to economic development. Although incidence rates have stabilized at a high level in most economically developed countries, they continue to increase in many traditionally low-​risk countries, following the uptake of Western patterns of diet and physical inactivity. In principle, CRC is among the most preventable of all common cancers. Potentially modifiable risk factors include obesity, physical inactivity, high intake of red or processed meat, tobacco smoking, and heavy alcohol use. Several screening tests effectively reduce both the incidence and death rates of CRC through the detection of pre-​cancerous lesions and the treatment of early stage cancers. Despite the preventability of CRC, incidence rates over the last 20 years have decreased in only a few countries. In the United States this decrease occurred at ages 50 years and above, the age above which screening is recommended for average risk people. Further progress has been made in delineating the molecular pathogenesis of CRC. Molecular pathways that accelerate the development of CRC and various precursor lesions are increasingly used to classify molecular subtypes of CRC. Even though studies relating risk factors for CRC to molecular characteristics are still limited, studying CRC subtypes has greatly enhanced our understanding of the mechanisms underlying CRC development and progression. About 5%–​7% of incident CRC cases are attributed to high-​penetrance inherited genetic abnormalities. The majority of cases are sporadic, mostly resulting from interactions among potentially modifiable risk factors, low-​penetrance inherited susceptibility genes, predisposing medical conditions, and chance. Research on the microbiome has introduced new opportunities for etiologic research, as have efforts to understand the interactions of external, metabolic, and hormonal risk factors, and the effects of these on specific molecular pathways. Future efforts to clarify these relationships will require large-​ scale population studies that collect and archive tumor tissue and utilize multidisciplinary approaches.

INTRODUCTION This chapter will review the molecular features and epidemiology of CRC, focusing on adenocarcinomas, which comprise the vast majority of the cases. It will also consider the well-​established precursor lesions of CRC and how these have informed studies of etiology and molecular pathogenesis. Other types of cancer that occur much less commonly in the large bowel, such as carcinoid tumors, gastrointestinal stromal tumors, lymphomas, and sarcomas, are not discussed here, but are considered in Chapters 31, 35, 40, and 43, respectively. These tumors affect multiple organs but collectively account for less than 5% of cancers in the large intestine (Levin and Raijman, 1995).

CLASSIFICATION Anatomic Distribution The anatomic segments of the large bowel include the cecum, ascending colon, hepatic flexure, transverse colon, splenic flexure, descending

colon, sigmoid colon, and the rectum (Figure 36–​1, adapted from National Institutes of Health, 2014). Sections proximal to the small intestine and including the transverse colon are by convention considered right-​ sided or proximal, whereas the descending and sigmoid colon are designated as left-​sided or distal (National Institutes of Health, 2014). Information about the anatomic location of CRCs has been recorded by the US Surveillance, Epidemiology, and End Results (SEER) registries since 1973. The distribution of tumors by cancer location has been changing over time, but between 2006 and 2010, 42% of incident cases were located in the proximal colon, 28% in the rectum, 23% in the distal colon, and 7% elsewhere in the large intestine (Siegel et al., 2014; see Figure 36–​1, adapted from National Institutes of Health, 2014). A shift in the anatomic distribution of colon cancers has occurred in the United States, toward proximal, right-​sided tumors, as discussed later in the chapter. While the anatomic subdivisions are useful clinically and for research purposes, the frequency of molecular alterations changes as a continuum along the large bowel, rather than abruptly by anatomic subsite (Yamauchi et al., 2012).

Histopathology of Colorectal Polyps and Cancer Colorectal Adenoma and Serrated Polyps

Colorectal adenomas and serrated polyps are direct precursors to the majority of colorectal adenocarcinomas.

Adenomatous Polyps

Adenomatous polyps (or adenomas) are abnormal growths of tissue rising from the inner lining of the large intestine that usually protrude into the lumen. Three different histologic types of adenomatous polyps have been defined: tubular, tubulo-​villous, and villous. An estimated 75%–​90% of these are tubular adenomas (Lee et al., 1995). Adenomas are common in Western countries. About one third of asymptomatic average risk individuals aged 50–​75  years will harbor one or more adenomas (Heitman et  al., 2009). Individuals with adenomas, especially high-​risk adenomas, are at increased risk for CRC if these are not removed. Although the majority of sporadic CRC develop from adenomas (Neugut et al., 1993), fewer than 10% of them will transform into cancer. The rate at which adenomas progress to cancer varies, but typically takes at least 10  years (Carethers and Jung, 2015; Lev, 1990). Factors that predict a higher likelihood that an adenoma will progress into cancer include villous histology, larger size (≥ 1 cm in surface diameter), multiplicity, and degree of dysplasia (Carethers and Jung, 2015; Cotton et al., 1996; Morson, 1984). These morphologic characteristics are often correlated, because they may indicate an adenoma’s propensity for increased (and uncontrolled) growth. For example, adenomas with a villous histology often also tend to be large (Lee et al., 1995). An individual diagnosed with a small, tubular adenoma may not be considered at appreciably higher risk of subsequent cancer. In fact, in a linkage study of the Cancer Registry and the Cause of Death Registry of Norway, CRC mortality was lower among patients with low-​risk adenomas (expected deaths, 189; observed deaths, 141; standardized mortality ratio, 0.75; 95% CI: 0.63, 0.88), even though no surveillance was recommended for patients with low-​risk adenomas (Loberg et al., 2014). In contrast, a large (e.g., 2 cm) villous adenoma should be considered a proximate precursor to cancer, with a high likelihood to progress over time. In observational studies, the risk factors

681

682

682

PART IV:  Cancers by Tissue of Origin

Proximal Colon – 47% (Cecum, Ascending Colon, and Transverse Colon)

Transverse colon

Ascending colon

Descending colon

Distal Colon – 23% (Descending Colon and Sigmoid Colon)

Small intestine Cecum

Rectum 28%

Sigmoid colon

Figure 36–​1.  Distribution of CRC by location (%) adapted from (National Institutes of Health, 2014). The anatomic segments of the large bowel include the cecum, ascending colon, hepatic flexure, transverse colon, splenic flexure, descending colon, sigmoid colon, and rectum. Sections proximal to and including the transverse colon are by convention considered right-​sided or proximal, whereas the descending and sigmoid colon are designated as left-​sided or distal.

for large/​advanced adenoma often overlap with those of CRC (Chan and Giovannucci, 2010).

Serrated Lesions

“Hyperplastic polyps” are more common than adenomatous polyps and were historically considered to be benign (Lee et al., 1995). Emerging evidence indicates that hyperplastic polyps are not a single disease, but are a group of serrated lesions comprising at least three main subtypes, based on morphologic, genetic, and clinical characteristics. According to recent recommendations from the World Health Organization, the most common subtype retains the generic name “hyperplastic polyps” (about 70%), followed by sessile serrated adenoma and polyp (SSA/​ SSP, 25%) and traditional serrated adenoma (TSA, < 2%) (Rex et al., 2012). Flat SSAs/​SSPs usually occur in the proximal colon and are more difficult to detect through screening than adenomatous polyps. SSAs and TSAs have the potential to develop into CRCs via an alternate pathway, the so-​called serrated pathway (see later discussion) (Lochhead et al., 2014; Rosty et al., 2013; Snover, 2011; Tadros and Anderson, 2013). In a recent study, individuals diagnosed with large serrated polyps were at 2-​to 3-​fold increased risk of CRC, a risk comparable to that in individuals with advanced/​high-​risk adenomas (Holme et al., 2015).

Molecular Pathogenesis CRC was the first malignancy for which a multistage model was proposed to explain the progression from normal epithelium, first to precursor lesions and later to invasive and metastatic cancer (Fearon and Vogelstein, 1990; Vogelstein et al., 1988). Subsequent discoveries have identified alternative, often overlapping molecular pathways that accelerate the accumulation of genetic and epigenetic abnormalities in clonal cell lines. Because of the diversity of these pathways, CRC does not represent a single disease, but rather a heterogeneous group of molecular subtypes that vary in their genetic, epigenetic, and clinical characteristics (Ogino and Goel, 2008). Three global processes define the major molecular subtypes of CRC. These include chromosomal instability (CIN), which compromises the structural integrity of chromosomes; microsatellite instability (MSI), resulting from faulty DNA repair; and the CpG island methylator pathway (CIMP), which dysregulates gene expression by methylation of

promotor regulatory regions (Carethers and Jung, 2015; Kang, 2011; Markowitz and Bertagnolli, 2009; Samowitz, 2008; Simons et  al., 2013). These aberrant processes can be inherited, but are more often acquired through somatic events. They are global in the sense that they affect many different genes, not only those involved in colorectal carcinogenesis. Their net effect is to increase the likelihood that a clonal cell line will acquire the necessary genetic mutations, chromosomal abnormalities, and/​or epigenetic events to undergo malignant transformation (Ogino and Goel, 2008). The existence of multiple pathways was not recognized when Fearon and Vogelstein first proposed their model for the adenoma-​ carcinoma sequence in 1990 (Fearon and Vogelstein, 1990; Vogelstein et al., 1988). Their model identified the genetic mutations that cause progression from normal epithelium to early, intermediate, and late adenoma, and subsequently to carcinoma and metastasis (Kinzler and Vogelstein, 1996). Other types of CRC develop through the alternative and overlapping pathways, as discussed later in the chapter. The most common subtype is CIN, which typically begins with mutations in the APC (adenomatous polyposis coli) gene initiating the adenoma-​carcinoma sequence. Structural rearrangements of chromosomes caused by CIN result in frequent karyotype abnormalities and chromosomal gains and losses. Chromosomal gains can increase oncogene expression by increasing copy number; chromosomal losses can inactivate tumor suppressor genes. The causes of CIN are likely varied and may involve damage to genes that maintain the integrity of chromosomal structure during cell division. CIN cancers are more common in the distal colon and in men (Pino and Chung, 2010). The second major subtype is the CIMP pathway. The CIMP subtype involves a defect in the regulation of DNA methylation, resulting in excessive methylation (or hypermethylation) of cytosine at promotor regulatory regions. This inactivates transcription of tumor suppressor genes and is an important mechanism for human carcinogenesis in multiple organs, including CRC. The definition of CIMP is based on the frequency of methylation of target loci (e.g., RUNX3, CACNA1G, IGF2, MLH1, NEUROG1, CRABP1, SOCS1, CDKN2A) (Ogino et al., 2007; Samowitz et al., 2005; Toyota et al., 2000; Weisenberger et al., 2006). Tumors are designated as CIMP+ if ≥ 6/​8 promotors are methylated and the cancer is characterized by distinct clinical, pathological, and molecular features such as older age at onset, proximal tumor

 683



683

Cancers of the Colon and Rectum FAP-associated carcinoma CIMP-MSS 1%

Suppressor (chromosomal instability) pathway

CIMP+MSS carcinoma 20%

Familial adenomatous polyposis

CIMP+MSS pathway

TSA pathway Conventional adenomacarcinoma sequence Sporadic CIMP-MSS carcinoma 60%

Serrated (CIMP+ pathway Mutator (microsatellite instability) pathway

Lynch syndrome carcinoma CIMP-MSI-H 5%

MYH pathway

TSA-associated carcinoma (?CIMP +MSI-L) 1%

Sporadic CIMP+MSI-H carcinoma 13%

Figure  36–​2. The molecular pathways leading to CRC [reproduced from (Snover, 2011)]. Abbreviations:  FAP  =  familial adenomatous polyposis; CIMP+ = CpG island methylator phenotype positive; CIMP–​ = CpG island methylator phenotype negative; MSI-​H = high degree of microsatellite instability; MSI-​L = low degree of microsatellite instability; MSS = microsatellite stable; TSA = traditional serrated adenoma.

location, female gender, poor differentiation or high BRAF and low TP53 mutation rates (Barault et  al., 2008; Ogino and Goel, 2008; Samowitz et al., 2005; Toyota et al., 1999; Weisenberger et al., 2006). The third (MSI) pathway contributes to approximately 10%–​20% of sporadic CRC and almost all inherited cases of hereditary non-​ polyposis colon cancer (HNPCC) (discussed later in the chapter) (Herman et al., 1998; Muller and Fishel, 2002; Poynter et al., 2008). Mutations in mismatch repair genes such as MLH1 and MLH2 can interfere with the repair of deletions or insertions during DNA replication (Wheeler et al., 2000a). The CIMP+ and MSI high (MSI-​H) subtypes tend to co-​occur and are more common in proximally located CRCs (Ogino et al., 2007; Samowitz et al., 2005; Toyota et al., 2000; Weisenberger et al., 2006, 2015). Figure 36–​2 (reproduced from Snover, 2011) provides a schematic diagram of the molecular pathways leading to CRC. The complexity of these overlapping pathways is striking. The molecular events involved in these pathways and their downstream effects are considerably more complicated than those indicated by Figure 36–​2, however. For example, initiation of the CIN pathway requires inactivating mutations in both alleles of the APC gene. Inactivation of the APC gene results in increased concentrations of β-​catenin, a protein that adheres to the T-​cell factor (TCF) and mediates transcription of certain genes, including the oncogene C-​MYC (He et al., 1998). If the mutation in APC is inherited (autosomal dominant), it results in the familial adenomatous polyposis (FAP) syndrome, characterized by the development of multiple (up to thousands) colorectal adenomas. Subsequent progression from adenomas to carcinomas depends on the accumulation of additional genetic and epigenetic alterations. Commonly observed mutations include those in the KRAS and BRAF oncogenes, and loss in TP53, a critical cell-​cycle regulator (Pino and Chung, 2010; Thiel and Ristimaki, 2013).

Classification of Molecular Subtypes Clinical and epidemiologic studies increasingly seek to determine the etiology and prognosis of particular molecular subtypes of CRC. For

epidemiologic studies, this requires that tumor tissue be collected and archived on cases to allow a standardized approach to classification. Several classification systems have been proposed (Guinney et  al., 2015; Jass, 2007; Kang, 2011; Marisa et  al., 2013; Sinicrope et  al., 2015). However, the classification systems differ with respect to the number of subtypes and the criteria used to define them. A separate carcinogenesis pathway has been suggested for CRC in individuals with inflammatory bowel disease (IBD) such as ulcerative colitis and Crohn’s disease, who are at higher risk of developing CRC (colitis-​ associated colorectal carcinoma, CAC) (Rogler, 2014). Chronic inflammation from IBD causes genetic alterations that progress to dysplasia and subsequently to cancer (Murthy et al., 2002; Wong and Harrison, 2001). The molecular alterations observed in CACs support a separate, albeit rare, pathway for colorectal carcinogenesis (“dysplasia-​carcinoma sequence”). For example, TP53 mutations associated with ulcerative colitis tend to occur earlier than in sporadic CRCs (Brentnall et al., 1994), and development to CACs does not have to follow the adenoma-​carcinoma sequence (Murthy et al., 2002; Potter et al., 1999; Romano et al., 2016) (also see later section on host factors). Another alternate pathway has also been proposed to explain CRCs that develop through sessile SSAs/​SSPs, which account for about 15%–​35% of large bowel cancers. Commonly seen genetic and epigenetic alterations in SSA/​SSPs include BRAF and KRAS mutations and hypermethylation of CpG islands in the promoter region of tumor suppressor genes (Lochhead et al., 2014; Rosty et al., 2013; Tadros and Anderson, 2013). Another important molecular event that affects the progression from adenoma to cancer is induction of the prostaglandin-​endoperoxide synthase 2 (PTGS2) or cyclooxygenase-​2 (COX-​2) pathway, a pro-​ inflammatory cascade whose downstream signaling promotes cell proliferation, inhibits apoptosis, and stimulates angiogenesis (Rajakariar et  al., 2006; Sostres et  al., 2014). PTGS2 converts arachidonic acid to prostaglandins, which exert numerous pro-​inflammatory effects. Supportive of a role in cancer, PTGS2 is overexpressed in the majority of CRC (Soumaoro et  al., 2004)  and may predict adverse prognosis

684

684

PART IV:  Cancers by Tissue of Origin

(Ogino and Goel, 2008). Interestingly, this pathway may be inhibited by non-​steroidal anti-​inflammatory drugs (NSAIDs) including aspirin, although important questions remain about mechanism and dose (see later discussion). High PTGS2 expression is not limited to one molecular subtype of CRC, but the prevalence is higher in CRCs not characterized by CIMP+ or MSI high (Gala and Chan, 2015).

DISEASE BURDEN Overall Mortality and Incidence in the United States and Globally Overall, an estimated 1.4 million cases (9.7% of all cancers, excluding non-​melanoma skin cancers) are diagnosed worldwide, making CRC the third most common malignancy. For deaths, it ranks fourth in men (373,500) and third in women (320,300), representing about 8.5% of all cancer deaths worldwide (International Agency for Research on Cancer/​World Health Organization, 2015). The range of incidence and mortality rates in different geographic regions is shown in Figure 36–​3 (International Agency for Research on Cancer/​World Health Organization, 2015). In the United States, CRC is the third most common incident cancer for both men and women; approximately 70% of CRCs occur in the colon. According to the American Cancer Society, an estimated 95,270 cases of colon cancer and 39,220 cases of rectal cancer are projected to occur in 2016 (Siegel et al., 2016). The corresponding number of deaths from CRC was projected to be 49,190. Mortality data do not

reliably distinguish colon from rectal cancers. According to cancer registry data from 2010–​2012, the lifetime probability of developing CRC in the United States is about 4.7% for men and 4.4% for women (Siegel et al., 2016).

Detection CRC is a malignancy for which early detection and removal of adenomas has been shown to provide both primary and secondary prevention. The increased use of screening endoscopy, especially colonoscopy, in the United States has reduced both the incidence and mortality rate from CRC at ages 50 years and above by detecting and allowing treatment of precancerous lesions as well as early stage cancers (Cummings and Cooper, 2011; Lieberman et  al., 2012). Multiple screening tests are effective in detecting precursor lesions and cancer earlier than would otherwise be the case (see section on screening). Over-​detection from screening is of less concern for CRC than for other screened cancers such as prostate and breast cancer. Colonoscopy may also have contributed to the shift in the predominant location of tumors from left-​to right-​sided cancers, due to differential detection and removal of precursor lesions between the distal and proximal colon (Cress et al., 2006; Rim et al., 2009). Neoplasms that present as flat adenomas in the proximal colon (SSA/​Ps) are often undetected during routine screening, which may at least in part explain why CRCs that develop during intervals between screening are more likely to exhibit features consistent with the serrated pathway than non-​interval CRCs (Arain et al., 2010; Institute of Medicine National Cancer Policy, 2008; Nishihara et al., 2013b).

Colorectum ASR (W) per 100,000, all ages Male

Female

Incidence Mortality

Australia/New Zealand Western Europe Southern Europe Northern Europe More developed regions Central and Eastern Europe North America Micronesia Eastern Asia World Caribbean South America Western Asia South-Eastern Asia Less developed regions Southern Africa Polynesia Melanesia Central America Northern Africa Eastern Africa South-Central Asia Middle Africa Western Africa 60

40

20

0

20

40

60

Figure 36–​3.  Age-​standardized incidence and mortality rates for CRCs according to GLOBALCAN 2012 by country (reproduced from [International Agency for Research on Cancer/​World Health Organization, 2015]).

 685



Cancers of the Colon and Rectum

Precancerous lesions and localized CRC are usually asymptomatic, but as tumors progress they may cause a change in bowel habits (e.g., diarrhea, constipation, and narrow stools), persistent abdominal discomfort (e.g., cramps, gas, and pain), bloody stools, fatigue, and unintended weight loss, and occult or overt fecal blood. The prevalence of signs and symptoms also varies depending on the anatomic location of the tumor (Tsai and Gearhart, 2011).

Treatment, Prognosis, and Survival The three major treatment options for CRC are surgical removal, chemotherapy, and radiation therapy, used alone or in combination (Tsai and Gearhart, 2011). Prognosis and survival largely depend on the stage and grade of the tumor at the time of diagnosis and the availability of appropriate treatment. Survival rates decrease with greater extent of disease. The estimated 5-​year survival in the United States during 2005–​2011 was 90% for the least advanced CRC (i.e., stage I), but only 13% for the most advanced CRC (i.e., stage IV) (Surveillance, Epidemiology, and End Results, 2015). Overall survival rates are similar for both colon and rectal cancer. The liver is the most frequent site for metastases (Rodriguez-​Bigas et al., 2003).

Demographic Characteristics Age

CRC is rare under age 40 years. Cases diagnosed at younger ages are usually associated with hereditary syndromes. Incidence rates increase exponentially with age beyond age 50 years. About 10% of CRC cases occur before age 50 years, and approximately 20% occur before age 55 years (Siegel et al., 2014). Figure 36–​4 (adapted from Surveillance Epidemiology and End Results, 2015)  shows the age distribution of CRC in the United States. The median age at diagnosis is younger for rectal cancer than colon cancer in both men (63 years vs. 69 years) and women (65 vs. 73 years) (Siegel et al., 2014). The age at onset of a tumor affects the likelihood that it will be detected by screening under the current guidelines. Screening is not recommended for average-​risk adults until age 50  years; only those with a family history of CRC are advised to begin screening earlier (Cummings and Cooper, 2011). Most early-​onset CRCs are diagnosed through symptoms and hence have poorer prognosis than screening-​ detected cancers (Bailey et al., 2015; Dozois et al., 2008; Myers et al., 2013; Taggarshe et al., 2013). The current age cut point for routine screening in the United States may also explain why the incidence rates of CRC are increasing below age 50 years but decreasing at age 50 and above (Figure 36–​5) (Siegel et al., 2014). The reasons for the increase at younger ages are unclear,

but the increased prevalence of obesity at younger ages is a likely contributor (Siegel et al., 2009).

Sex

Globally, the age-​standardized worldwide incidence rate is about 1.44 times higher in men (746,000 cases in 2012) than in women (614,000 cases in 2012) (International Agency for Research on Cancer/​World Health Organization, 2015). There are more CRC deaths in men (373,631 per year) than in women (320,250 per year) (International Agency for Research on Cancer/​World Health Organization, 2015). The male to female ratio for CRC incidence increases with age; in addition, men are more likely to develop cancer in the left colon, whereas women are more likely to develop cancer in the right colon (Siegel et al., 2014). The reasons that the location of CRCs differs by sex and age are not clear, but appear to correspond with sex differences in molecular pathogenesis. Tumors in the CIMP pathway are associated with older age at onset, proximal location, and female sex. In contrast, CIN tumors are more common in the distal colon and in men (Carethers and Jung, 2015; Iacopetta, 2002). Differences in sex hormones may also contribute through their effects on bile acid metabolism, fecal transit time, and stool composition (Lampe et al., 1993; McMichael and Potter, 1985). Differences in exogenous sex hormones may, at least in part, help explain why the overall male to female ratio for CRC is lower before ages 50–​54 years (i.e., for premenopausal women) than after ages 50–​54 years (i.e., for postmenopausal women) (Bufill, 1990; McMichael and Potter, 1985). Estrogen replacement therapy in postmenopausal women has been suggested to reduce risk of CRC (Lin et al., 2012). The higher overall incidence rates of CRC in men than in women are consistent with the greater prevalence of risk factors such as smoking and higher alcohol intake in men, and with the stronger association of risk with obesity in men (Ning et al., 2010).

Race and Ethnicity

CRC incidence and mortality rates differ by race and ethnicity. Figure 36–​6 (Siegel et al., 2014) shows age-​standardized CRC incidence and mortality rates by race/​ethnicity and sex in the United States. The highest age-​standardized annual incidence rate is observed in African Americans (men: 63.8/​100,000 people; women: 47.6/​100,000 people), followed by American Indian/​ Alaska Native, non-​ Hispanic white, Hispanic and incidence rates are lowest in Asian/​Pacific Islander people (men:  40.8/​100,000, women:  31/​100,000) (Centers for Disease Control, 2015; Siegel et al., 2014). Among all racial and ethnic groups, African Americans also have the highest mortality rates (men: 29.4/​ 100,000, women: 19.4/​100,000). It is unclear why incidence and mortality rates differ by race/​ethnicity, but the high proportion of African Americans with low socioeconomic status (SES), lower utilization of screening, and less access to high-​quality treatment likely contribute to the disparities (Laryea et al., 2014; Murphy et al., 2015; Rao et al., 2015; Robbins et al., 2012).

Socioeconomic Status 30

Not preventable by colonoscopy and polypectomy ≥50 y (20% of CRC) cases

Percent of New Cases

25

23.9%

21.5%

22.6%

20 14.5%

15

12.1%

10 4.1%

5 0

0.1% < 20

1.3% 20–34

35–44

45–54

55–64

65–74

75–84

> 84

Age

Figure 36–​4.  Percentage of new CRC cases by age at diagnosis in the U.S. (reproduced from [Surveillance Epidemiology and End Results, 2015]).

685

Socioeconomic status (SES) may affect CRC incidence through differences in the prevalence of behavioral risk factors and screening. In the United States, individuals with lower SES generally have higher incidence rates of CRC (Manser and Bauerfeind, 2014). Contributing to this pattern is that individuals with lower SES are more likely to have higher prevalence of obesity, poor diet, tobacco smoking, and leisure-​time physical inactivity. In addition, they are less likely to be screened (Rao et al., 2015). In European countries, however, lower SES is more often associated with lower incidence rates of CRC or is unrelated to risk (Aarts et al., 2010). For example, in southern Europe the incidence of CRC is reduced among those with lower SES, possibly because of closer adherence to a Mediterranean diet, which is inversely associated with risk (Leufkens et al., 2012). Similar associations between low SES and lower risk of CRC have also been observed in Nordic countries such as Finland, Norway, and Sweden (Aarts et al., 2010). Contributing to the difference in the relationship of CRC to SES between Europe and the United States is that CRC screening is less common in Europe, even among higher SES groups.

68

686

PART IV:  Cancers by Tissue of Origin 0–49 years

8

50–64 years

120

65+ years

400

110 350

7 100

Rate per 100,000 people

6

300

90 80

250

5 70 4

200

60 Male Female

50 150

3 40 2

100

30 20

50

1 10 0

0

0 2002 2004 2006 2008 2010

2002 2004 2006 2008 2010

2002 2004 2006 2008 2010

Year of diagnosis

Figure 36–​5.  Trends in incidence rates of CRC by age and sex between 2002–​2010 in the US (reproduced from [Siegel et al., 2014]).

In contrast to the SES pattern for incidence, CRC-​specific mortality rates are higher among individuals with low SES in most studies conducted in Europe and the United States (Manser and Bauerfeind, 2014). This pattern is likely to reflect differences in participation in cancer screening programs as well as disparities in treatment (Laryea et al., 2014; Murphy et al., 2015; Robbins et al., 2012).

Incidence

70

Geographic Variation

Estimated incidence rates of CRC vary considerably by region, suggesting that cultural, behavioral, and/​or economic factors strongly influence colorectal carcinogenesis (Armstrong and Doll, 1975). According to GLOBOCAN 2012 (Figure 36–​3, International Agency for Research on Cancer/​ World Health Organization, 2015), the

Mortality

70

63.8 60

Rate per 100,000 people

50

40

60 51.7

50.9

51.7

40.8

50

47.6

47.3

42.7 39.1

38.8 31

30

40

32.6 29.4

30

20

20

19.2

18.7

19.5 16.1

13.1 10

0

0

Women

NHW

NHB

API

AI/AN†

15.4

13.6 9.7

10

Men

19.4

Men

Hispanic

13.9 10.2

Women

All persons

Figure  36–​6. Age-​adjusted incidence and mortality of CRC by race/​ethnicity and sex in the US, 2006–​2010 [reproduced from (Siegel et  al., 2014)]. Abbreviations: NHW = non-​Hispanic white; NHB = non-​Hispanic black; API = Asian/​Pacific Islander; AI/​AN = American Indian/​Alaska Native.

 687



687

Cancers of the Colon and Rectum

highest incidence rates are in Australia/​New Zealand, Europe, and North America, while the lowest are in Africa and South-​Central Asia. In general, incidence rates correspond closely to increases in economic development. Incidence rates vary by approximately 10-​fold from the highest to lowest risk countries. The ratio of colon to rectal cancer also varies substantially across countries. Some countries with low incidence of colon cancer often have relatively high rates of rectal cancer (Parkin et al., 1992). Mortality rates from CRC vary by about 6-​fold in men and 4-​fold in women from high-​to low-​risk countries. Although low-​and middle-​income countries have lower incidence rates of CRC than high-​income countries, about 50% of all CRC deaths worldwide occur in low-​and middle-​income countries due to their larger populations or differential access to treatment (International Agency for Research on Cancer/​World Health Organization, 2015).

During 2002–​ 2010, CRC incidence rates in men and women 15 years since diagnosis (Hu et al., 1999). Treatments of diabetes with insulin or metformin may also potentially influence risk of CRC by increasing or decreasing insulin levels, but published studies have not yet examined this issue (Giovannucci et al., 2010; Sehdev and O’Neil, 2015).

Specific Risk Factors Alcohol

Several meta-​analyses have shown that higher alcohol consumption increases risk of CRC, particularly in men (International Agency for Research on Cancer, 2010). In the Pooling Project of Prospective Studies of Diet and Cancer that included 4687 CRC cases among nearly 500,000 men and women, individuals who consumed ≥ 45 g/​ day of alcohol (i.e. ≥ 3 standard drinks/​day) had a 41% higher risk of CRC than non-​drinkers (Cho et al., 2004a). The association between heavy alcohol consumption and CRC risk has been observed in many studies for both colon and rectal cancer and for all types of alcoholic beverages. This relationship is not confounded by other known risk factors of CRC (International Agency for Research on Cancer, 2010). Whether risk is increased by intake levels below 30 g (about 2 drinks per day) is unclear. The association is noticeably stronger when folate status is poor, especially among men (Giovannucci, 2004). Alcohol antagonizes the effect of folate by impairing the intestinal absorption of folate or by suppressing intracellular folate metabolism (Hillman and Steinberg, 1982). Experimental studies suggest that the carcinogenic effect of alcohol in the large intestine may be mediated by acetaldehyde (Jokelainen et  al., 1997; Seitz and Becker, 2007), a direct

689

carcinogen (see Chapter  12). Acetaldehyde is the first metabolite of ingested alcohol. The predominant exposure of colonocytes results from intracolonic oxidation of alcohol to acetaldehyde by intestinal microflora that express alcohol dehydrogenase (ADH) enzyme, although some exposure occurs through the circulation (Jokelainen et al., 1997; Salaspuro, 2009; Seitz and Becker, 2007). Because intestinal epithelial cells lack the ability to detoxify acetaldehyde, its accumulation may cause recurrent damage, repair, and aberrant growth of the colonic mucosa (Jokelainen et al., 1997; Seitz and Becker, 2007). In addition, acetaldehyde may directly degrade intracellular folate in the large bowel mucosa (Giovannucci, 2004). More detail on potential mechanisms underlying the association between alcohol intake and cancer can be found in Chapter 12.

Tobacco Our understanding of the role of tobacco in colorectal carcinogenesis has evolved over the past several decades. The early classic studies of smoking and cancer did not find cigarette smoking associated with an increased risk of CRC (Doll et al., 1980; Doll and Peto, 1976; Hammond and Horn, 1958, 1966; Kahn, 1966; Rogot and Murray, 1980; Weir and Dunn, 1970). These studies focused on mortality from CRC, primarily in men during the 1950s and 1960s. However, when risk factors for colorectal adenomas, which had become established as cancer precursors, were first studied in the 1980s and 1990s, smokers were found to have a 2-​to 3-​fold elevated risk (Giovannucci, 2001b). Given the null findings for CRC mortality, the results for adenomas were strikingly positive. Giovannucci et al. hypothesized that smoking may act as an initiator of colorectal neoplasia in the large bowel and consequently, a long induction period may be required to observe an increase in cancer incidence (Giovannucci et  al., 1994). Many studies in the past two decades now support this hypothesis (Botteri et al., 2008a, 2008b; Hsing et  al., 1998; Liang et  al., 2009; Nyren et  al., 1996; Tsoi et al., 2009). For example, in a recent dose–​response meta-​ analysis, CRC risk decreased by 4% for each 10-​year delay in smoking initiation, and increased by 20% with each 40-​year increase in smoking duration (Liang et al., 2009). In another meta-​analysis, current smoking was associated with higher risk of adenomas, particularly high-​risk adenomas (Botteri et  al., 2008b). Thus, smoking appears to have a strong effect at early stages of carcinogenesis. Furthermore, a positive association appears to be stronger for rectal than colon cancer, and to be stronger in men than in women (Botteri et al., 2008a; Tsoi et al., 2009). Cigarette smoke contains numerous carcinogens, such as heterocyclic amines, N-​nitroso compounds (NOC), polycyclic aromatic hydrocarbons (PAH), among many others (Hoffmann and Hoffman, 1997), which can reach the colorectal mucosa and induce genetic mutations or epigenetic alterations (Giovannucci et  al., 1996). In experimental studies, tobacco triggers multistep epigenetic alterations at several sites within an exposed tissue field; over time, these may evolve to multifocal lesions and even cancer (DeMarini, 2004). The effect of tobacco on CRC was previously considered irreversible because most studies, though not all, suggested that past smokers carry a persistently elevated risk. However, recent studies have indicated reversibility of the effect; the mortality rate from CRC among former smokers approaches that of lifelong non-​smokers after 3 decades of cessation (Hannan et al., 2009). After a detailed review of the existing mechanistic and epidemiological evidence, the 2014 Surgeon General’s report on “the health consequences of tobacco smoking-​50  years of progress” concluded that “the evidence is sufficient to infer a causal relationship between smoking and colorectal adenomatous polyps and colorectal cancer,” and suggested that “clinicians and public health personnel should include both current and former smoking as potential risk factors for this disease” (Centers for Disease Control, 2014).

Molecular Subtypes. Recently, molecular insights have contrib-

uted to our knowledge of smoking and CRC. Studies consistently find that smokers have an increased risk of CRC with epigenetic alterations in genes associated with CIMP+ and MSI-​H. Interestingly, while the overall relative risk (RR) for CRC associated with current cigarette smoking

690

690

PART IV:  Cancers by Tissue of Origin

is approximately 1.2–​1.3, the RR for tumors with features of CIMP+ or MSI-​H is approximately 2.0 (Curtin et al., 2009; Limsui et al., 2010; Poynter et al., 2009). Furthermore, in one study, smoking cessation for at least 10 years was associated with an approximately 50% risk reduction for CIMP+, but not CIMP-​tumors in former compared to current smokers (Nishihara et al., 2013a). Yet, as smoking has also been associated with rectal cancers, which are rarely CIMP+/​MSI-​H, tobacco may additionally affect risk through other independent pathways (Giovannucci, 2001b).

Western Dietary Pattern

Most studies of diet and cancer traditionally focused on individual food items or nutrients. Because individuals eat nutrients and foods in combination, rather than in isolation, more recent studies have examined risk in relation to dietary patterns (Hu, 2002). Both healthful and deleterious dietary patterns have been identified and characterized through various approaches. In relation to CRC, one of the most commonly identified dietary patterns associated with increased risk is the Western dietary pattern, characterized by high consumption of red/​processed meats and refined grains, often accompanied by high-​fat dairy products, sugar-​ sweetened beverages, and French fries (Fung and Brown, 2013). In a recent meta-​analysis of 11 observational studies, the Western dietary pattern was associated with an increased risk of colon cancer, though not rectal cancer (Magalhaes et  al., 2012). Dietary patterns may influence carcinogenesis through their combined effect on pathways such as hyperinsulinemia and inflammation, rather than through any single nutrient.

Physical Activity One of the most consistent relationships observed for colon cancer has been an inverse association with physical activity (Boyle et al., 2012; Wolin et  al., 2011), which is now considered an established protective factor for CRC (Figure 36–​8, World Cancer Research Fund and American Institute for Cancer Research, 1997, 2011). Both prospective cohort and case-​control studies have consistently found that individuals who are physically more active have lower risk of colon cancer than those who are inactive or sedentary (Chapter 21). In addition, higher levels of physical activity have been associated with a reduced risk of colon adenoma, particularly large adenoma (Giovannucci et al., 1995, 1996; Kono et  al., 1991; Little et  al., 1993; Neugut et  al., 1996). In general, a relationship with rectal cancer has not been observed. The inverse association between higher physical activity level and lower colon cancer risk is likely to be causal because of (a) its consistency in many studies of various designs, (b) its relationship to large adenomas and cancer in men and women drawn from diverse populations, (c) its presence in all domains of physical activity (e.g. occupational, recreational, household) and (d) the apparent lack of confounding by dietary factors and obesity (Giovannucci et  al., 1995; Martinez et  al., 1996; Slattery et al., 1997; Thun et al., 1992). While greater risk reductions are possible with even higher levels of activity, even moderate levels of physical activity (e.g., brisk walking for 3–​4 hours/​week) are associated with substantial benefits (Giovannucci et al., 1996; Wolin et al., 2007). Recent data also suggest a beneficial role of post-​diagnosis physical activity on CRC prognosis (Van Blarigan and Meyerhardt, 2015). The mechanism for the action of physical activity is not known, but may involve in part its ability to reduce insulin levels (for details, see Chapter 21) (Turcotte and Fisher, 2008). Controlled trials showed that moderate-​intensity physical activity (Ross et  al., 2000, 2004), particularly vigorous activity, reduced visceral adiposity (Irving et  al., 2008) and improved insulin sensitivity, even in the absence of measureable weight loss (Ross et al., 2000, 2004). Other proposed mechanisms by which physical activity may affect the risk of colon cancer include enhanced immune function and reduced chronic inflammation (Harriss et al., 2007; Slattery, 2004). Physical activity may also increase colonic motility, though colonic motility has not been definitely linked to colon cancer risk (Friedenreich and Orenstein, 2002).

Molecular Subtypes.  Only a few studies have examined asso-

ciations between physical activity and CRC by molecular subtypes. There is some evidence that CRC tumors in physically inactive individuals are more likely to have weak or no expression of fatty acid

synthase (Kuchiba et al., 2012) or nuclear ß -​catenin (Morikawa et al., 2013) compared to those in physically active study participants.

Sedentary Behavior

Economic development typically leads to a decrease in occupational physical activity and an increase in sedentary behavior. The latter is characterized by prolonged sitting, which differs from physical inactivity in that it initiates qualitatively different cellular and molecular responses in skeletal muscle. In recent years, sedentary behavior has emerged as an independent risk factor for CRC (Hamilton et al., 2007; Schmid and Leitzmann, 2014). The largest study of this topic was conducted in the NIH-​AARP Diet and Health study, in which sedentary behavior was captured by time spent watching TV. Even controlling for physical activity, people who reported ≥ 63 hours/​week of TV watching time had an approximately 1.5-​fold higher risk of colon cancer compared to those who reported < 21 hours/​week of TV watching (Howard et  al., 2008). A recent meta-​analysis examined sedentary behavior time separately for TV watching and sitting at work. Comparing the highest versus lowest levels in each category, the RR for colon cancer was 1.54 (95% CI: 1.19, 1.98) for TV viewing time, 1.24 (95% CI: 1.09, 1.41) for occupational sitting time, and 1.24 (95% CI: 1.03, 1.50) for both combined (Shen et al., 2014).

Folate (Vitamin B9)

Folate is a B vitamin critical in maintaining the integrity of DNA synthesis and repair, and it is in the pathway for methylation reactions, including DNA methylation. Given its functions, folate has long been proposed as beneficial for CRC prevention. Yet, precisely because it may help provide DNA precursors, a carcinogenesis-​enhancing role of high folate in pre-​malignant neoplastic lesions has been proposed (Kim, 2006; Mason et al., 2007; Osterhues et al., 2009; Ulrich, 2007). Timing and dose are important when considering the potential dual role of folate. Specifically, folate adequacy during the early stages of the development of pre-​malignant neoplastic cells may be beneficial, while excess folate at later stages when advanced pre-​malignant lesions such as large adenomas are present may be deleterious. During the pre-​neoplastic phase, folate deficiency increases the misincorporation of uracil into DNA and hypomethylation of DNA (McGlynn et al., 2013). This field defect, which predisposes to adenoma formation, can be reversed with folic acid supplementation, as documented in an intervention study, strengthening the evidence for causality (O’Reilly et al., 2016). Because the progression from precursor legions to cancer diagnosis generally takes at least 10 years (Winawer, 1999), macroscopically undetectable abnormalities related to folate deficiency would precede the diagnosis of CRC by at least a decade. The epidemiologic evidence tends to support a beneficial role of folate. For example, a pooled analysis of 13 prospective studies with follow-​up periods ranging from 7 to 20 years found that total folate intake of ≥ 560 mcg/​day relative to < 240 mcg/​day was significantly associated with a 13% lower risk of colon cancer (Kim et al., 2010). Moreover, analyses of serial measurements of diet over three decades in the Nurses’ Health Study and Health Professionals Follow-​Up Study found a statistically significant inverse association when a lag of at least 12 years was required between folate intake and CRC ascertainment (Lee et al., 2011). The requirement of a minimum latency period in analyses of cancer is consistent with mechanistic data and with the natural history of colorectal carcinogenesis, and may explain why short-​term (≤ 5 years) randomized controlled trials (RCTs) have not seen a benefit of folic acid supplementation on CRC risk. Interestingly, a recent RCT of folic acid supplementation in a population with low folate levels in China showed a reduced risk of initial adenomas (not cancer) after 3 years of intervention (Gao et al., 2013). Although the concept of excessive folic acid increasing CRC risk has experimental support, the human evidence to date does not support this hypothesis. First, CRC incidence did not appear to increase in the United States following periods of widespread higher intakes of folic acid following first supplementation and then fortification (Keum and Giovannucci, 2014). Second, observational studies of long-​term intakes of folate and folic acid do not show an increased risk of adenoma or cancer. Third, in a recent meta-​analysis of 13 RCTs, folic

 691



Cancers of the Colon and Rectum

691

FOOD, NUTRITION, PHYSICAL ACTIVITY AND CANCERS OF THE COLON AND THE RECTUM 2011 DECREASES RISK

INCREASES RISK

Convincing

Physical activity1,2 Foods containing dietary fiber3

Red meat4,5 Processed meat4,5 Alcoholic drinks (men)7 Body fatness Abdominal fatness Adult attained height8

Probable

Garlic Milk9 Calcium10

Alcoholic drinks (women)7

Substantial effect on risk unlikely

None identified

1 Physical activity of all types: occupational, household, transport and recreational. 2 The Panel judges that the evidence for colon cancer is convincing. No conclusion was drawn for rectal cancer. 3 Includes both foods naturally containing the constituent and foods which have the constituent added. Dietary fiber is contained in plant foods. 4 Although red and processed meats contain iron, the general category of ‘foods containing iron’ comprises many other foods, including those of plant origin. 5 The term ‘red meat’ refers to beef, pork, lamb, and goat from domesticated animals. 6 The term ‘processed meat’ refers to meats preserved by smoking, curing, or salting, or addition of chemical preservatives. 7 The judgments for men and women are different because there are fewer data for women. For colorectal and colon cancers the effect appears stronger in men than in women. 8 Adult attained height is unlikely directly to modify the risk of cancer. It is a marker for genetic, environmental, hormonal, and also nutritional factors affecting growth during the period from preconception to completion of linear growth. 9 Milk from cows. Most data are from high-income populations, where calcium can be taken to be a marker for milk/dairy consumption. The Panel judges that a higher intake of dietary calcium is one way in which milk could have a protective effect. 10 The evidence is derived from studies using supplements at a dose of 1200mg/day.

Figure 36–​8.  Summary of conclusions of the “Continuous Update Project Report. Food, Nutrition, Physical Activity, and the Prevention of Colorectal Cancer” (reproduced from [World Cancer Research Fund, and American Institute or Cancer Research, 2011]). Note: Continuous Update Project report on colorectal cancer is currently being updated (publication expected in 2017).

acid supplementation at doses of 500–​40,000 mcg/​day across studies did not substantially increase the risk of CRC during the first 5 years of intervention compared to placebo (Vollset et  al., 2013). Of note, since subjects in these RCTs generally did not have a screening colonoscopy at baseline, most of the cancers diagnosed in the first 5 years originated from covert advanced adenomas or latent cancers. Thus, if folic acid supplementation enhanced the progression of existing neoplasms, an increased risk of CRC would be expected in those allocated to treatment. Moreover, in the subgroup analysis restricted to three trials conducted among participants with a prior history of adenomas, no evidence of an increased risk was observed (Vollset et al., 2013). While it is reasonable to caution against high intakes of folic acid without clear benefits, the human data to date are reassuring that folate intake does not increase risk of CRC, at least in the United States.

Calcium

Experimental studies in both animals and humans indicate that calcium may bind secondary bile acids or ionized fatty acids in the lumen

of the large bowel, diminishing their carcinogenic effects (Newmark et  al., 1984; Wargovich et  al., 1984). Experiments also indicate that calcium may reduce cell proliferation and promote cell differentiation by influencing cell signaling (Buset et al., 1986; Lamprecht and Lipkin, 2001). In adenoma patients, supplementation with 2000 mg/​ day of calcium induced favorable changes on gene expression in the APC/​β-​catenin pathway in normal-​appearing rectal mucosa (Ahearn et al., 2012). Additionally, calcium may inhibit heme-​iron-​induced formation of NOC, a known potent carcinogen. In a recent feeding study, increased fecal NOC levels in humans and rats seen after ingestion of processed meats returned to normal after the addition of calcium carbonate (Pierre et al., 2013; Santarelli et al., 2013). In a meta-​analysis of RCTs on patients with a diagnosis of colorectal adenoma, subjects assigned to 1200–​2000 mg of calcium supplements over 3–​4 years had an approximately 20% reduced risk of adenoma recurrence compared to the placebo group (Shaukat et al., 2005). In the Calcium Polyp Prevention Study, supplementation of 1200 mg/​day calcium over 4 years was particularly protective against histologically

692

692

PART IV:  Cancers by Tissue of Origin

advanced neoplasms, the proximal precursors to cancer (Wallace et al., 2004). However, a recent RCT of calcium and vitamin D did not confirm these findings (Baron et al., 2015). The reasons for inconsistencies in the RCTs are unclear. The average BMI was high in the most recent RCT and a subgroup analysis found a protective association among individuals with BMI in the normal range (< 25 kg/​m2). In contrast to the adenoma studies, a meta-​analysis of eight RCTs showed no benefit of calcium on cancer risk over 4 years (Bristow et  al., 2013). It is unclear why the results are inconsistent between colorectal adenoma and CRC in some studies, although this may relate to the duration of follow-​up. The prevention of an adenoma, especially of an advanced adenoma, would presumably lead to the prevention of CRC. Yet, the progression from adenoma to CRC spans at least 10 years. In a pooled analysis of 10 cohort studies, whose follow-​up period ranged between 6 and 16 years, a significant inverse association was observed between total calcium intake and CRC, suggesting a long induction period (Cho et al., 2004b). While the optimal dose for prevention remains unclear, a dose–​response meta-​analysis of prospective studies suggested that CRC risk may continue to decrease at intakes beyond 1000 mg/​day, though the slope became attenuated at higher intakes (Keum et al., 2014).

Vitamin D Status

In 1980, Garland and Garland observed that colon cancer mortality rates in the United States were generally higher in the northern regions, which have low exposure to solar ultraviolet B (UV-​B) radiation. As solar UV-​B radiation is the main source of vitamin D synthesis for most individuals, they hypothesized that vitamin D protects against this cancer (Garland and Garland, 1980). Subsequent molecular studies demonstrated that vitamin D influences cellular signaling pathways, generally leading to an inhibition of proliferation and angiogenesis and an activation of apoptosis (Feldman et  al., 2014). The relationship between vitamin D status and CRC has now been studied through various measures of vitamin D status, including solar UV-​B radiation exposure, dietary and supplementary vitamin D intake, and predicted or measured circulating 25(OH)vitamin D (25(OH)D), with generally supportive results (Giovannucci, 2009; 2013). In the largest study on circulating 25(OH)D and CRC risk, the EPIC study, pre-​diagnostic 25(OH)D concentration in the highest quintile compared to the level in the lowest quintile was associated with an approximately 40% reduced risk of incident CRC (Jenab et al., 2010) and with 31% improved survival of patients with CRC (Fedirko et al., 2012). From the observational studies, the optimal levels of 25(OH)D for CRC protection and survival may be in the range of 30–​40 ng/​mL (Giovannucci, 2013). To date, RCTs of vitamin D supplements are inadequate in terms of dose and duration to evaluate the potential benefit on CRC. A recent RCT of colorectal adenoma found that supplementing 1000 IU of vitamin D (with and without calcium) over a 3-​to 5-​year period did not reduce the risk of subsequent adenomas among those with a previously resected adenoma (Baron et al., 2015). The mean net increase in serum 25(OH)D levels was 7.83 ng/​mL, relative to subjects given placebo. Future RCTs are needed to provide more definitive evidence.

Fiber

The hypothesis that higher fiber intakes protect against CRC has remained popular for half a century. This hypothesis was first proposed by Burkitt in the late 1960s, following his observations of low incidence of CRC in southern Africa, where dietary fiber intake was high (Burkitt, 1969). Various mechanisms involving both insoluble and soluble fiber have been put forth to account for a lower risk of CRC among those with high-​fiber intakes. Insoluble fibers increase stool bulk, shorten stool transit time, and possibly decrease exposure of the colorectal mucosa to potential carcinogens (Kritchevsky, 1995). Soluble fibers are fermented by the anaerobic intestinal microbiota into short-​chain fatty acids, such as acetate, propionate, and butyrate. Butyrate inhibits proliferation and induces apoptosis of CRC cells (Kritchevsky, 1995; Sebastian and Mostoslavsky, 2014). Despite the intuitive appeal of this hypothesis, data from epidemiologic studies have been inconsistent. The cumulative evidence from observational studies suggests a protective association. Yet, the

association is relatively weak and less consistent than for other established etiologic factors. In some studies, adjustment for physical activity and dietary intakes of folate and calcium attenuated or eliminated the association (Michels et  al., 2005; Otani et  al., 2006; Schatzkin et al., 2007; Wakai et al., 2007). Furthermore, no protective association was observed in most RCTs of fiber supplements in relation to adenoma (Baron, 2005), although RCTs have limitations regarding the type and level of fiber intake, duration, and appropriate end point. However, a recent dose–​response meta-​analysis of prospective studies estimated an approximately 10% reduced risk of CRC for a 10g/​ day increment of total fiber intake, with cereal fiber intake appearing particularly protective (Aune et  al., 2011). The Continuous Update Project/​World Cancer Research Fund (WCRF) 2011 report on CRC concluded that there is “convincing” evidence that “foods containing dietary fiber” lower risk of CRC (World Cancer Research Fund and American Institute for Cancer Research, 2011). It may be unreasonable to expect fiber supplements to provide the same benefits as a diet naturally high in fiber-​rich foods. In modern diets, the amount and quality of fiber is much lower than in traditional diets. Given that fiber is diverse and that different types or sources of fiber may have different effects, a role of fiber cannot be dismissed; however, fiber is unlikely to have a dominant role in the prevention of CRC. Many risk factors other than fiber have been established for CRC that likely contribute to the higher rates in Western countries, including obesity, sedentary behavior, red or processed meat intake, tobacco and alcohol, and possibly micronutrients such as calcium, vitamin D, and folate. Future research addressing differences in the relationship by tumor molecular subtypes and interactions with the gut microbiome will shed more light into the role of fiber in colorectal carcinogenesis.

Animal Products Within populations, total protein intake has not been consistently related to higher risk of CRC. However, different sources of protein may have different effects (Carr et al., 2016).

Red Meat and Processed Meat.  Intakes of red meat and pro-

cessed meat have been examined in numerous epidemiological studies, including several meta-​analyses and systematic reviews (Alexander et  al., 2010, 2015; Carr et  al., 2016; Chan et  al., 2011; Hjartaker et  al., 2013; Huxley et  al., 2009; Johnson et  al., 2013; Larsson and Wolk, 2006; Norat et al., 2002; Pham et al., 2014; Sandhu et al., 2001; Santarelli et  al., 2008; World Cancer Research Fund and American Institute for Cancer Research, 2011; World Cancer Research Fund/​ American Institute for Cancer Research, 2007). Studies of red meat and CRC have generally examined total red meat (i.e., unprocessed and processed meat combined) or unprocessed and processed meat separately. The 2007 WCRF/​ American Institute for Cancer Research (AICR) report concluded that the evidence for a causal relationship between higher red and processed meat and CRC is “convincing” (American Institute for Cancer Research, 2007; World Cancer Research Fund/​ American Institute for Cancer Research, 2007). An updated meta-​analysis by Chan et al. (2011), conducted as part of the Continuous Update Project, confirmed their previous findings (Figure 36–​8) (World Cancer Research Fund and American Institute for Cancer Research, 1997, 2011). Consistent with the WCRF/​AICR reports, other meta-​analyses have also reported modest positive associations between red meat intake and CRC (Alexander et al., 2010; Hjartaker et al., 2013; Huxley et al., 2009; Johnson et al., 2013; Larsson and Wolk, 2006; Norat et al., 2002; Pham et al., 2014; Sandhu et al., 2001), while for processed meat intake some variations in effect size were noted across meta-​analyses. Overall, associations between processed meat and CRC are generally stronger than those for unprocessed red meat (Chan et al., 2011; Santarelli et al., 2008). In addition, evidence from most prospective studies (Bernstein et al., 2015; Hjartaker et al., 2013; Larsson and Wolk, 2006), albeit not all, (Parr et  al., 2013), suggests that associations between processed meat and CRC may be stronger for distally located cancers. In 2015, the International Agency for Research on Cancer (IARC) Monograph Working Group classified “processed meat as carcinogenic to humans (Group 1) based on the sufficient evidence in humans

 693



Cancers of the Colon and Rectum

that the consumption of processed meat causes colorectal cancer” (Bouvard et al., 2015). The Working Group also classified “the consumption of red meat as probably carcinogenic based on limited evidence that the consumption of red meat causes cancers and the strong mechanistic evidence (Group 2A).” The reason why red meat, particularly processed meat, as opposed to other sources of protein, tends to be associated with increased risk of colon cancer remains unclear. Some of the relevant hypotheses have stated that red meat is a major source of total fat, saturated fat or heme iron (Cross et al., 2003; Cross and Sinha, 2004). Based on the current evidence, the most promising pathways include (a) those related to meat mutagens, such as heterocyclic amines and other mutagens formed when meat is cooked at very high temperature and for long duration (Sinha and Norat, 2002; Sinha et al., 2005); and (b) factors related to the production of carcinogenic NOCs, such as nitrate/​nitrite added to processed meat or heme iron in red meat (Cross et al., 2003; Cross and Sinha, 2004). The role of heterocyclic amines in CRC remains unclear and cohort studies that have examined the effect of heterocyclic amines per se (rather than using cooking methods or doneness of meat as proxy) are limited and inconsistent (Cross et al., 2010; Le et al., 2016; Ollberding et al., 2012). Emerging evidence also suggests a role of heme iron, but not non-​ heme iron, in colorectal carcinogenesis. Heme iron in red meat can facilitate endogenous nitrosation as well as lipid peroxidation (Bastide et al., 2011; Cross et al., 2003), which may, at least in part, explain positive associations observed between red meat and CRC. In a meta-​ analysis of five prospective studies, higher intake of heme iron was modestly associated with elevated risk of colon cancer (Bastide et al., 2011). NOCs are among the most potent known carcinogens and agents that can alkylate DNA. Prospective data relating dietary nitrate, nitrite, and NOC intake per se to CRC risk are sparse and inconsistent (Cross et al., 2010; Cross and Sinha, 2004; Dellavalle et al., 2014; Knekt et al., 1999). One possible reason for these inconsistencies is the difficulty to assess exposure to NOCs using questionnaire data alone and the lack of an established long-​term biomarker for NOC exposure (Cross and Sinha, 2004; Hord et al., 2009).

Other Sources of Animal Protein.  In contrast, sources of ani-

mal protein other than red or processed meat, including low-​fat dairy products, fresh fish, and poultry, are generally associated with lower risk of CRC, even though some studies have also observed no association between poultry intake and CRC (Carr et al., 2016; Shi et al., 2015). In two meta-​analyses of poultry intake and CRC risk published in the same year, one found a significant, albeit very modest, inverse association (Shi et  al., 2015), while the other observed no association (Carr et al., 2016). In the latter study, however, higher intake of poultry was associated with a moderately lower risk of rectal (but not colon) cancers. These results do not support an adverse effect of protein, and possibly suggest a benefit. The underlying mechanism for this potential benefit is unknown, but these foods are good sources of methionine, which may be beneficial in regard to DNA methylation (see earlier section on folate). Because red and white meat are major protein sources in the human diet, it is also important to assess risk of CRC in the context of the replacement of red meat with white meat or other sources of protein (e.g., by employing substitution analysis; Bernstein et al., 2010), but data are limited. Findings from the Danish Diet, Cancer and Health cohort study suggest that, while substituting red meat with poultry was not associated with risk of CRC, substituting red meat with fish was associated with decreased risk of colon but not rectal cancer (Egeberg et al., 2013).

Genetic Susceptibility.  A pooled analysis of 11 studies (Anantha­

krishnan et  al., 2015)  found no evidence of an interaction between red meat intake and N-​acetyltransferase (NAT)2 genotype, an enzyme involved in the metabolism of heterocyclic amines (Turesky, 2004). In another meta-​analysis, the NAT1 phenotype also did not appear to modify the association between meat and CRC. With regard to NAT2 phenotype, however, a positive association between meat consumption

693

and CRC risk was observed among those with a fast acetylator phenotype, while there was no association in participants with the slow acetylator phenotype. However, the p-​value for interaction did not reach statistical significance (Andersen et al., 2013). Overall, studies that have examined potential interactions between meat intake and other single nucleotide polymorphisms (SNPs) related to xenobiotic enzymes or other hypothesized (meat) related pathways on CRC risk are inconclusive (Andersen et al., 2013; Andersen and Vogel, 2015).

Molecular Subtypes.  Only a few prospective studies have exam-

ined the association between processed meat and risk of CRC by tumor sub-​types. However, processed meat was positively and strongly associated with tumors found to have alkylating DNA damage in key CRC genes such as APC or TP53. This indicates a potential role of NOCs in CRC development (Bingham, 1999; Gay et al., 2012; Park et al., 2010; Slattery, 2000; Slattery et al., 2002).

ENERGY AND OBESITY Body Mass and Fat Distribution As a manifestation of a positive energy imbalance between energy intake and energy expenditure, excess adiposity is an established risk factor for CRC (Figure 36–​8) (World Cancer Research Fund and American Institute for Cancer Research, 1997, 2011). The effect is stronger for colon than for rectal cancer and is more pronounced in men than in women. In epidemiologic studies, adiposity has been assessed with diverse anthropometric metrics including body mass index (BMI), waist circumference (WC) or waist-​ to-​ hip ratio (WHR), and adult weight change. While these metrics are correlated with each other, they capture different aspects of adiposity. Specifically, BMI represents overall body fatness, WC and WHR abdominal fatness, and adult weight change time-​ integrated fat accumulation. BMI has been the most frequently used metric; a meta-​analysis of cohort studies found an approximately 20% increased risk of CRC for obese individuals (BMI ≥ 30 kg/​m2) compared with normal or underweight individuals (BMI < 25 kg/​m2) (Moghaddam et al., 2007). Substantial evidence indicates that abdominal adiposity may be more relevant to colorectal carcinogenesis than other indices. First, an elevated colon cancer risk associated with WC (or WHR) remained significant after adjustment for BMI, whereas that associated with BMI became non-​significant after adjustment for WC (or WHR) (MacInnis et  al., 2004; Pischon et  al., 2006). Second, annual weight gain (kg/​ year) during adulthood (approximately from age 20 to 50) was most strongly associated with colon cancer risk if it involved an increase in WC (Aleksandrova et al., 2013). Third, men have tendency toward abdominal obesity (Despres, 2006; Geer and Shen, 2009), and the relationship between excess adiposity and colon cancer risk is stronger in men than in women (Keum et al., 2015). Of note, abdominal fat can be further classified into visceral adipose tissue (VAT), which surrounds the internal organs, and subcutaneous adipose tissue (SAT), which is located beneath the skin. Relative to SAT, VAT is more strongly associated with insulin resistance, which promotes colorectal carcinogenesis, as discussed elsewhere in this chapter. Thus, it is biologically plausible that VAT may underlie the association between excess adiposity and colorectal neoplasia. Indeed, studies that used computed tomography to measure VAT and SAT in the abdominal area found that VAT, but not SAT, was associated with adenomas, particularly with advanced colorectal adenomas (Keum et al., 2015).

Molecular Subtypes

Colorectal tumors in obese patients are more likely to harbor mutations in KRAS (Slattery et al., 2001) and have little or no expression of fatty acid synthase (Kuchiba et al., 2012) or nuclear beta-​catenin (Morikawa et  al., 2013)  than CRC that develop in non-​ obese participants.

694

694

PART IV:  Cancers by Tissue of Origin

NSAIDS INCLUDING ASPIRIN Beginning in the 1980s, case-​control and cohort studies demonstrated with remarkable consistency that long-​term use of aspirin and other NSAIDs was associated with reduced risk of CRC. This association is now established as causal through RCTs of adenomas and CRCs (Chan et  al., 2005; Cook et  al., 2013; Rothwell et  al., 2010). Two important features, which affect the utility of aspirin as a chemopreventive agent, involve the dose and duration of treatment needed to see an effect. While some data suggest that relatively large doses may be required, the Women’s Health Study RCT demonstrated that even low doses (100 mg) on alternate days could confer protection against CRC (Cook et al., 2013). Yet, at least 10 years of follow-​up appear to be necessary to observe the benefit. This conclusion is based on observational studies of CRC, the observation that aspirin reduces the risk of colorectal adenomas (Cole et al., 2009), which generally require at least 10 years to progress to cancer, and at least one RCT of cancer (Cook et  al., 2005, 2013). In a recent large pooled analysis of five RCTs with a median follow-​up of 18.3 years, regular aspirin use of at least 75 mg (and duration of 5 years or longer) was associated with lower CRC incidence and mortality. Interestingly, when associations were examined by subsites within the colon (when available), inverse associations appeared to be strongest for proximal cancers (Rothwell et al., 2010; Thun et al., 2012).

not with risk of small adenoma (Chan and Giovannucci, 2010). It is likely that these factors act on adenoma progression but probably not on their initiation. To date, there are limited data relating specific etiologic factors to molecular events in tumors, but this is a very active area of study by many investigators. While a growing number of studies support a possible role of early life exposures on risk of breast cancer (Fuemmeler et al., 2009; Potischman and Linet, 2013), data relating early life exposures to CRC risk later in life are sparse and are often limited by small size or lack of data on covariates. Despite the dearth of prospective data, emerging evidence suggests that early life exposures may be etiologically important for CRC. In a historical cohort study (“Boyd Orr Cohort”), subjects in the highest quartile of body fatness during childhood had a non-​significantly 36% increased risk of CRC; however, this finding was based on only 76 cases (Jeffreys et al., 2004). In a study of 1.1 million Israeli men, adolescent BMI was positively associated with colon but not rectal cancers (Levi et al., 2011). However, neither of these studies collected data on important covariates of CRC. Among female nurses in the Nurses’ Health Study II, higher body fatness at age 5 was associated with higher risk of colorectal adenoma, particularly in the distal colon (Nimptsch et al., 2011). In some northern European countries, extreme calorie restriction in children and adolescents during time periods of famine was associated with lower CRC risk later in life, suggesting a role of dietary factors in early life on CRC (Dirx et al., 2003; Hughes et al., 2010; Svensson et al., 2005).

Molecular Subtypes The mechanism underlying the actions of these compounds may involve inhibition of the actions of PTGS2 or COX-​2 enzyme (discussed earlier). In one large study, regular aspirin use was associated with an approximately 40% lower risk for CRCs that overexpressed PTGS2, but was not associated with CRCs that did not overexpress PTGS2 (Chan et al., 2007). Moreover, the post-​diagnostic use of aspirin improved overall survival among CRC patients with PTGS2-​overexpression but not those without PTGS2 overexpression (Chan et al., 2009). Other molecular markers associated with CRC survival may also be relevant to the actions of NSAIDs. Phosphatidylinositol 3-​ kinase (PI3K) signaling activity stimulates cell growth and mutations in PI3K that increase kinase activity and may enhance carcinogenesis (Samuels et al., 2004). Among CRC patients with mutated-​PIK3CA, regular use of aspirin after diagnosis was associated with an over 5-​ fold improvement in CRC-​specific survival. In contrast, CRC patients with wild-​type PIK3CA experienced no benefit from regular use of aspirin (Domingo et al., 2013; Liao et al., 2012). If aspirin use can be targeted to patients most likely to benefit, the potential adverse effects may be minimized.

TIMING OF THE EFFECT OF  MODIFIABLE RISK FACTORS Colorectal carcinogenesis is a long process and can take up to several decades. Therefore, for certain exposures, the timing of their actions on important molecular events can be related to particular developmental stages of CRC. For example, tobacco is strongly related to risk of adenomas, but only affects cancer risk after a time lag of several decades (Botteri et al., 2008a, 2008b; Giovannucci et al., 1994; Hsing et al., 1998; Liang et al., 2009; Nyren et al., 1996; Tsoi et al., 2009). Exposure to growth factors such as IGF-​1 during childhood and adolescence (Juul et  al., 1994)  is thought to at least partly explain why tallness is consistently associated with higher risk of CRC (Engeland et al., 2005; Pischon et al., 2006; Wei et al., 2004). Certain chemical carcinogens used in animal experiments appear to affect both early and later events in carcinogenesis, possibly through APC and β-​catenin mutations in early stages, and other events that affect progression toward malignant transformation (Bissonnette et al., 2000; Takahashi and Wakabayashi, 2004; Yamada and Mori, 2007). In contrast, factors such as obesity, physical activity and circulating IGF-​1 during adulthood tend to be associated with risk of large adenoma and cancer, but

HOST FACTORS Acromegaly Acromegaly is a rare clinical condition characterized by excessive production of growth hormone and IGF-​1. It is also associated with insulin resistance and hyperinsulinemia, independent of obesity. Circulating levels of growth hormone and IGF-​1 correlate with the cell proliferation rate (Cats et al., 1996; Jenkins, 1999). Numerous small studies indicate that acromegalic patients have increased risk of developing both benign and malignant colorectal tumors. In a meta-​analysis, acromegalic patients had a higher risk of hyperplastic polyps (3.6-​fold), colon adenomas (2.5-​fold), and colon cancer (4.4-​fold) (Rokkas et al., 2008). These RRs were all highly statistically significant, with no statistical evidence of publication bias or heterogeneity. Acromegaly is also associated with severe insulin resistance. In one study among acromegalic patients, an increase in fasting insulin levels was associated with a 9-​to 15-​fold increased risk of colonic adenoma (Colao et al., 2007).

Genetic and Familial Susceptibility High-​Penetrance Hereditary Syndromes

The two inherited syndromes that strongly predispose to CRC are familial adenomatous polyposis (FAP) (Lynch et al., 1995), and hereditary non-​polyposis CRC (HNPCC) (Lynch and Smyrk, 1998).

Familial Adenomatous Polyposis (FAP)

FAP, also called familial polyposis coli, or adenomatous polyposis of the colorectum, is a rare inherited, autosomal dominant syndrome (Bisgaard et  al., 1994). FAP accounts for less than 1% of all CRC. Clinically, the syndrome is characterized by the occurrence of multiple colorectal adenomas (usually a minimum of 100 up to several thousand) in individuals in their twenties and thirties. If the condition is not treated by prophylactic colectomy or proctocolectomy, most affected people develop CRC (Fearnhead et al., 2001; Lynch et al., 1995). The risk of CRC appears to be positively related to the number of adenomas (Debinski et  al., 1996). The genetic mutation underlying this inherited disorder involves the loss of function of both alleles of the APC tumor suppressor gene (Fearnhead et al., 2001; Lynch et al., 1995). An

 695



Cancers of the Colon and Rectum

acquired somatic mutation of APC is an important early event in the development of sporadic CRC (see earlier section on molecular genetic characteristics) (Fearon and Vogelstein, 1990). By age 50, penetrance is almost 100%, but substantial differences in phenotypic expression have also been reported (Fearnhead et  al., 2001). Some individuals develop a modified form of FAP, characterized by the combination of colorectal adenomas and extracolonic tumors. In Turcot’s syndrome, additional tumors occur in the central nervous system (Hamilton et al., 1995; Lynch et al., 1995). In Gardner’s syndrome, affected individuals may also develop osteomas, epidermal cysts, or abnormalities of the teeth (Lynch et al., 1995). Some FAP patients may also develop desmoid (fibrous) tumors, commonly in the abdominal wall or retroperitoneum, while others may present with other gastrointestinal tumors, such as periampullary carcinomas (Fearnhead et al., 2001; Houlston and Tomlinson, 2001). A milder form of FAP, called attenuated FAP (AAPC), is characterized by fewer adenomas and mutations of APC that occur closer to the 5′ end than is typical (Houlston and Tomlinson, 2001; Spirio et  al., 1993). Individuals with a family history of FAP should be routinely screened for colorectal adenomas starting in their teens (Lynch et al., 1995).

Lynch Syndrome (HNPCC)

The HNPCC syndrome, also called Lynch syndrome, is another autosomal dominant inherited disorder that predisposes to CRC, and that may account for approximately 1%–​5% of colorectal cancer cases (Lynch and Smyrk, 1996). HNPCC-​related cancers usually develop at an early age (approximately mid-​forties) and about two-​ thirds of cancers occur in the proximal colon (Lynch and Smyrk, 1996; Vasen et al., 1999). HNPCC is more difficult to diagnose than FAP because the occurrence of adenomas is uncommon. However, if adenomas are present, they are more likely to be villous adenomas and have a higher grade of dysplasia than sporadic adenomas (Lynch and Smyrk, 1996; Vasen et  al., 1999). Lynch I  syndrome is associated only with CRC. Some patients with other forms of HNPCC can also develop cancers outside the colorectum, including tumors of the endometrium, ovary, stomach, pancreas, small bowel, hepatobiliary tract, ureter, and renal pelvis (Lynch II syndrome) (Wheeler et al., 2000b). Increased microsatellite instability of mismatch repair genes has been found in approximately 90% of HNPCC (Wheeler et al., 2000b). Genetic alterations have been identified in several mismatch repair genes, including hMSH2, hMLH1, hPMS1, hPMS2, and MSH6. Mutations in the hMLH1 and hMLH2 genes are most common, affecting approximately 70% of patients with HNPCC (Peltomaki, 2001; Wheeler et al., 2000b).

695

et al., 1994). Thus, the relationship between tallness and colon cancer risk is consistent with the hypothesis that high concentrations of circulating IGF-​1 during growth predisposes to higher risk of colon cancer later in life.

Predisposing Medical Conditions Inflammatory Bowel Disease (IBD)

IBD includes ulcerative colitis and Crohn’s disease (granulomatous colitis). These relatively rare conditions have distinct clinical and pathologic features. Chronic colonic inflammation triggers compensatory proliferation to replace damaged tissue, which increases opportunities for mutations (Coussens and Werb, 2002). Accumulation of genetic alterations can lead to dysplasia and subsequently to cancer. Tumor development in these patients usually does not involve an adenoma precursor. The mean age at diagnosis is between ages 40–​50 years. The magnitude of the association among individuals with ulcerative colitis has been estimated to be a 4-​to 20-​fold increase in the risk of CRC (Eaden, 2004; Murthy et al., 2002; Triantafillidis et al., 2009; Wong and Harrison, 2001; Rogler et al., 2014). The magnitude of the association appears to be related to early onset of the disease, the extent of involvement of mucosa, the duration of the symptoms, and the presence of multicentric foci of dysplasia (Rogler, 2014). For patients with ulcerative colitis, left untreated, the risk of CRC has been reported to approach 18% about 30 years after the onset of the disease (Eaden, 2004). Crohn’s disease affects the ileum and sometimes the large bowel. While it is widely accepted that ulcerative colitis can increase the risk of CRC, the association between Crohn’s disease and CRC has been unclear. However, in a recent meta-​analysis Crohn’s disease was associated with a 2.5-​fold higher risk of CRC and a 4.5-​fold higher risk of colon cancer (Canavan et al., 2006; Lennerz et al., 2016). Thus, patients with IBD require intensive surveillance.

Cholecystectomy

Family history of CRC in one or more first-​degree relatives (parents, siblings, or children) is associated with increased risk of CRC. A family history of CRC is found in approximately 15%–​20% of CRC patients. In a meta-​analysis of 27 case-​control and prospective studies (Johns and Houlston, 2001), the relative risk of developing CRC was 2.25 higher for individuals with one first-​degree relative compared to those with no affected family member. Associations between family history and CRC also tend to be slightly stronger for colon than for rectal cancer. The risk is even higher if the first-​degree relative was diagnosed at young ages or if more than one first-​degree relative are affected (Butterworth et  al., 2006; Johns and Houlston, 2001). The increased risk may be attributable to inherited genes, shared environmental factors, or combinations of the two.

Numerous studies have examined the relationship between cholecystectomy and risk of CRC. The interest in cholecystectomy primarily stems from the observation that individuals whose gall bladders have been removed have a continuous rather than periodic excretion of bile acids into the intestine. The constant exposure of bile acids to colonic bacteria may increase the production of potentially toxic secondary bile acids (Bernstein et al., 2005). A meta-​analysis of case control studies suggested some increase in the risk of proximal colon cancer (RR = 1.34), but this was not replicated in prospective studies (Giovannucci et al., 1993). Moreover, the evidence was strongest for case-​control studies that used hospital-​based rather than population-​ based controls, raising concern about selection bias. Additionally, the increased risk that is limited to proximal cancer could reflect detection bias from greater exposure to screening. Some studies have suggested that risk increases over time since cholecystectomy (Ekbom et  al., 1993; Goldbohm et  al., 1993). However, many of the potential risk factors for gallstones overlap with those for colon cancer (e.g., obesity, physical inactivity, insulin resistance, diabetes); thus, the association may represent confounding by these other factors, which few studies have taken into account. A recent meta-​analysis did not indicate a consistent association between cholecystectomy and colorectal adenoma (Zhao et al., 2012). Overall, there is no compelling evidence that individuals who have undergone cholecystectomy should be screened more frequently.

Height

Low-​Penetrance Susceptibility Genes

Family History of CRC in First-​Degree Relative

Tallness is an independent risk factor for colon cancer and rectal cancer in most studies. A recent meta-​analysis found that, for each 10 cm increase in height, the risk increased by 25% for colon cancer and by 14% for rectal cancer (Green et al., 2011). Tallness may be a surrogate for some other factor related to height, such as IGF-​1. Childhood and adolescent levels of IGF-​1 influence linear growth (particularly leg length) and correlate with attained height (Juul

Increasing numbers of SNPs have been identified in genome-​wide association studies (GWAS) as potential low-​penetrance susceptibility loci for CRC. Most of these SNPs are located in non-​coding regions, but several SNPs are in regulatory regions known to be involved in carcinogenic pathways of CRC, including the Wnt signaling pathway (Peters et al., 2013; Schumacher et al., 2015). Studies have identified more than 50 SNPs associated with CRC; several research consortia

69

696

PART IV:  Cancers by Tissue of Origin

have combined data from different studies to elucidate undiscovered susceptibility loci (Lemire et al., 2015; Orlando et al., 2016; Schmit et al., 2014; Schumacher et al., 2015; Whiffin et al., 2014; Zhang et al., 2014). Meta-​analyses have replicated the associations with more than 40 SNPs, for which the odds ratios are generally less than 1.25 (Peters et al., 2012, 2013; Schumacher et al., 2015). Taken together, these low penetrance variants account for only a limited fraction of the estimated heritability. Gene-​environment (G x E) interactions could explain some of the missing heritability of CRC (Kantor et al., 2014; Thomas, 2010). For example, a genome-​wide investigation of G x E interaction found that the inverse relation between aspirin and/​or NSAIDs and CRC differed by genotype (Nan et al., 2015). The identified SNPs were downstream from the MGST1 gene, possibly related to the pro-​ inflammatory effects of prostaglandin E2 (PGE2) synthase. Thus, NSAID inhibition of CRC may be modified by genetic variation in the PGE2-​induced Wnt signaling pathway. Similarly, polymorphisms in the genes that encode xenobiotic metabolizing enzymes involved in bioactivation or detoxification of carcinogens (Turesky, 2004) may modify the effects of meat consumption, as discussed earlier.

PREVENTIVE MEASURES Primary Prevention The potential for primary prevention of CRC is substantial. Population-​ level strategies to reduce tobacco use, prevent excessive weight gain, and increase physical activity are discussed in Chapters 62.1 and 62.2 of this volume. The dietary recommendations to prevent CRC mirror the dietary guidelines for overall health (Chapter 19). These emphasize not only healthful patterns of diet, but also the avoidance of tobacco use and excessive consumption of alcohol, maintenance of a healthy body weight, and regular physical activity (Kushi et al., 2012; World Cancer Research Fund/​American Institute for Cancer Research, 2007). They encourage the consumption of mostly plant-​based foods, non-​starchy vegetables, whole fruit, and whole (vs. refined) grains, and discourage consumption of refined sugars, highly processed starches, red and processed meat, and foods low in fiber and micronutrients. The recommended primary sources of protein are poultry, fish, and plants, rather than red or processed meats or high-​fat dairy products. Carbohydrates should come principally from unrefined grains, legumes, and fruits, rather than highly refined grains, starches, and sugars. Mono-​ unsaturated and poly-​unsaturated fats are the preferred sources of fat. The potential benefits of primary prevention of CRC are substantial. A study of colon cancer in male health professionals estimated that at least 70% could potentially be avoided by moderate changes in diet and behavior (Platz et al., 2000). A study in US female nurses estimated that about 40% of CRC could be prevented by modifying six dietary and behavioral factors, i.e., maintenance of a BMI < 25 kg/​m2, physical activity ≥ 21 metabolic equivalents per week, alcohol consumption of ≤ 30 g/​day, cigarette smoking < 10 pack-​years before the age of 30, current use of multivitamins for ≥ 15 years and total calcium intake ≥ 700 mg/​day (Erdrich et al., 2015). In a European study, 16% of the new CRC cases (22% in men and 11% in women) were attributed to not adhering to a combination of all five healthy behaviors (healthy weight, physical activity, non-​smoking, limited alcohol intake and healthy diet) (Aleksandrova et al., 2014). The observed discrepancy in the proportion of preventable cases between these cohorts may be related to difference in definition of exposure. For example, in the European study, both never and past smokers were assigned the same score.

Chemoprevention A major challenge in identifying drugs or other agents to prevent CRC is the precarious balance of benefits to risks, especially when treatment is administered to large numbers of apparently healthy people who are at average risk of the disease being prevented (see Chapter 63). For

example, aspirin is a widely used, non-​prescription drug that has been definitively shown to reduce the risk of CRC in randomized clinical trials (Cook et  al., 2013; Flossmann and Rothwell, 2007; Rothwell et  al., 2010). For individuals at average risk of CRC, however, the potential benefits of treatment in preventing CRC are offset by aspirin-​ induced bleeding. Even low-​dose aspirin (generally ≤ 100mg/​day), which has been shown in RCTs to be as effective as standard-​dose aspirin (approximately 300 mg) in reducing the risk of CRC, causes significant risk of bleeding. Thus, the formal approval of aspirin as an anti-​cancer agent will depend on the net balance of benefits (in reducing both cardiovascular events and cancer) and risks (mostly from bleeding) in populations at varying background risk of these end points. Ultimately, the recommendation for prophylactic treatment with low-​dose aspirin to protect against cardiovascular events may be broadened to consider protection against CRC (Rothwell et al., 2010; Thun et al., 2012). The balance of benefit to risk is far more favorable in populations at high risk of CRC, such as those with FAP or HNPCC. Here the use of conventional NSAIDs or selective PTGS2 inhibitors may be justified. Some evidence also exists for a protective effect from other agents, including non-​aspirin NSAIDs (sulindac sulfone, celecoxib), estrogen replacement therapy in women, ursodeoxycholic acid (a bile acid modifier), and difluoromethylornithine (DFMO) (Alberts et al., 2005; Asano and McLeod, 2004; Lin et  al., 2012; Meyskens et  al., 2008). While some results have been intriguing, the balance of potential efficacy versus side effects has not been settled to date, and future studies are necessary. Chemoprevention trials for CRC frequently assess whether drugs or other agents prevent the occurrence of new adenomas in individuals with previous adenomatous polyps. In these trials, patients who have undergone resection of one or more adenomas are randomized to either intervention or placebo treatment. Colonoscopy is then repeated after a period of 2–​5 years. The model has several limitations, however. It only detects benefits in preventing early events of carcinogenesis (the appearance of adenomas). The prevention of these may not truly represent the prevention of cancer, since most adenomas do not progress to cancer. False negative findings may also result from insufficient follow-​up. A substantial number of adenomas may be missed at the initial colonoscopy, as suggested by the very high rate of adenoma detection on follow-​up colonoscopies within a few years after initiation of these studies.

Screening As mentioned earlier, screening for adenomatous polyps and early-​ stage CRC is highly effective for both the primary and secondary prevention of CRC (Chapter  63). Various methods of screening are discussed in the following. The early detection and removal of colorectal adenomas reduces both the incidence and mortality rates of CRC (Winawer et al., 1993; Zauber et al., 2012). Similarly, the diagnosis and treatment of early-​stage cancers reduce mortality. Several screening modalities have been shown to be effective in detecting and allowing the removal of pre-​neoplastic lesions (Bibbins-​Domingo et  al., 2016). Screening implementations vary widely across the world, depending on the prevalence of CRC, the healthcare system, and economic resources (Schreuders et al., 2015). Currently available stool-​based tests are more effective in detecting cancer than pre-​malignant lesions. These include the guaiac-​ based fecal occult blood test (gFOBT), fecal immunochemical test (FIT), and multitargeted stool DNA test (FIT-​DNA). Stool tests are non-​invasive and do not require bowel preparation or anesthesia. In gFOBT, the test detects the presence of a small amount of blood by identifying human and non-​human hemoglobin in the stool. The specificity of this test is relatively low, as many factors can produce false positive results. In addition, adenomas often do not bleed, reducing the sensitivity of this test for pre-​cancerous lesions. A meta-​analysis of RCTs shows that annual or biennial screening of gFOBT was associated with decreased CRC mortality but not incidence (Hewitson et al., 2008). FIT is more specific in that it detects human hemoglobin

 697



Cancers of the Colon and Rectum

but not hemoglobin from dietary sources. Like gFOBT, FIT is more sensitive for detecting cancer than adenomas, but FIT screening has a higher detection rate for CRC (Hol et al., 2010). Observational studies report lower mortality from CRC in populations screened with FIT than in unscreened populations (Zorzi et  al., 2015). FIT-​DNA screening is a multi-​targeted stool test that detects the combination of methylation markers, genetic mutations, and human hemoglobin. The performance of FIT-​DNA is reportedly more sensitive than FIT for both advanced neoplasia and cancer; the test also has more false positives (Imperiale et al., 2014). Direct visualization tests for detection of adenomas and cancer include flexible sigmoidoscopy, colonoscopy, double-​ contrast barium enema (DCBE), and computed tomographic (CT) colonography (also known as virtual colonoscopy). These tests require bowel preparation of distal colorectum using an enema or clearing of entire colon using a strong laxative. Flexible sigmoidoscopy directly visualizes the lower half or third of the colorectum using a flexible sigmoidoscope. Colonoscopy visualizes the entire colon and rectum using a colonoscope. The detection of an adenoma by flexible sigmoidoscopy should lead to follow-​up testing by colonoscopy to evaluate the proximal regions of the large bowel. Screening with sigmoidoscopy reduces both the incidence and death rates from CRC in RCTs (Elmunzer et al., 2012; Littlejohn et al., 2012; Schoen et al., 2012). Screening with colonoscopy has not been tested in RCT, but observational studies support its efficacy in reducing incidence and mortality rates (Brenner et  al., 2011, 2012; Nishihara et  al., 2013b). DCBE involves barium and air to visualize the colorectal mucosal profile using X-​rays. In patients with prior polypectomy, DCBE is reported to be less sensitive for screening than colonoscopy (Winawer et al., 2000). CT colonography visualizes precursor lesions and cancer using abdominal CT. The colon must be evacuated and air inserted prior to the examination. Follow-​up colonoscopy is necessary if adenomas are detected by CT colonography. The sensitivity of this test is lower for small polyps that are less than 10 mm (Pickhardt et al., 2011). Non-​invasive stool tests are generally low in cost and have fewer side effects than more invasive tests that may require sedation. In low-​ income countries, population-​ based colonoscopy screening is generally unaffordable given budgetary and clinical constraints. An advantage of FIT screening in population-​based programs is that the threshold of positivity can be modified based on available resources and the burden of CRC. Harms from screening tests include perforation and major bleeding during endoscopic examinations, although clinically significant complications are rare (Warren et al., 2009). Radiologic tests involve exposure to radiation. FOBT shows a low positive predictive value, suggesting an unfavorably high false-​positive ratio (Hewitson et  al., 2007). A positive test result leads to further diagnostic colonoscopy or other invasive tests. Beginning at age 50, CRC screening for asymptomatic adults at average risk is widely recommended by major professional medical societies and organizations (Bibbins-​Domingo et al., 2016). The US Preventive Services Task Force recommends that this population receive one of the following:  gFOBT or FIT every year; FIT-​DNA every one or three years; flexible sigmoidoscopy or CT colonography every five years; colonoscopy every 10  years; or flexible sigmoidoscopy every 10  years plus FIT every year (Bibbins-​Domingo et  al., 2016). For adults over age 75 or those with life expectancy of less than 10 years, decisions about screening should be modified based on individual circumstances. Screening recommendations for high-​risk individuals should match the expected risk of CRC with the type of screening, age at initiation, and frequency (Roncucci and Mariani, 2015). High-​risk conditions include FAP, HNPPC, chronic ulcerative colitis, and Crohn’s disease. Other indications of increased risk include a history of small, large, or serrated adenomas, a family history of colorectal adenoma or cancer in first-​degree relatives, or a diagnosis of CRC. The possibility of considering multiple low-​ penetrance susceptibility loci, identified by GWAS, to individualize screening recommendations for adults under age 50  years warrants further investigation.

697

FUTURE DIRECTIONS Although important questions remain, cases and deaths from CRC are largely preventable through existing knowledge. Application research is essential for improving interventions to reduce tobacco use, limit alcohol consumption, facilitate maintenance of a healthy body weight, increase physical activity and improve diet. Etiologic research is needed to further our understanding of the relationship of insulin, IGF, and inflammation to specific events in carcinogenesis. The effects on the microbiologic milieu of the large intestine is another important area to be elucidated. Studies involving the microbiome are just now becoming feasible. Finally, evidence suggests a role of some micronutrients, especially vitamin D, calcium, and folate. Some of these may act early in carcinogenesis, so standard RCTs for cancer end point may not be feasible because of the long follow-​up period required to observe an effect. Alternative ways need to be developed and utilized to reach more definitive conclusions. Despite the remaining questions, a substantial proportion of CRCs can be averted through prudent behavioral measures. The long natural history of CRC suggests that some of the initiating events occur well before the diagnosis, possibly during childhood, adolescence, and early adulthood. The effects of early life exposures are another area open to discovery. Yet, this topic is challenging because of the long time lag between the exposure and its overt consequences. The main hurdles are cost, underutilization, and lack of an accepted optimal surveillance schedule when an adenoma is identified. Further research should focus on factors better characterizing the risk (e.g., genetic, smoking, obesity) to optimize the age when colonoscopy should be initiated. Currently, only family history is taken into account, but demographic, behavioral, and genetic factors can be used to characterize large differences in risk. Current recommendations will miss a relatively large number of early-​onset cases. Targeted screening approaches should be considered. Large opportunities already exist for the primary and secondary prevention of CRC. These involve the increased use of approaches to modify behavioral risk factors and to implement affordable screening. In the future, chemopreventive agents such as aspirin may prove to be effective as supplemental measures. However, the establishment of efficacy and the risk–​benefit ratio requires additional research before chemopreventive agents can be widely utilized. References Aarts MJ, Lemmens VE, Louwman MW, Kunst AE, and Coebergh JW. 2010. Socioeconomic status and changing inequalities in colorectal cancer? A  review of the associations with risk, treatment and outcome. Eur J Cancer, 46(15), 2681–​2695. PMID: 20570136. Ahearn TU, Shaukat A, Flanders WD, Rutherford RE, and Bostick RM. 2012. A  randomized clinical trial of the effects of supplemental calcium and vitamin D3 on the APC/​B-​catenin pathway in the normal mucosa of colorectal adenoma patients. Cancer Prev Res (Phila), 5(10), 1247–​1256. PMCID: PMC3466388. Alberts DS, Martínez ME, Hess LM, et al. 2005. Phase III trial of ursodeoxycholic acid to prevent colorectal adenoma recurrence. J Natl Cancer Inst, 97(11), 846–​853. PMID: 15928305. Aleksandrova K, Pischon T, Buijsse B, et al. 2013. Adult weight change and risk of colorectal cancer in the European Prospective Investigation into Cancer and Nutrition. Eur J Cancer, 49(16), 3526–​3536. PMID: 23867126. Aleksandrova K, Pischon T, Jenab M, et al. 2014. Combined impact of healthy lifestyle factors on colorectal cancer: a large European cohort study. BMC Med, 12, 168. PMCID: PMC4192278. Alexander DD, Miller AJ, Cushing CA, and Lowe KA. 2010. Processed meat and colorectal cancer: a quantitative review of prospective epidemiologic studies. Eur J Cancer Prev, 19(5), 328–​341. PMID: 20495462. Alexander DD, Weed DL, Miller P, and Mohamed M. 2015. Red meat and colorectal cancer: aquantitative update on the state of the epidemiologic science. J Am Coll Nutrition, 34(6), 521–​543. PMCID: PMC4673592. American Institute for Cancer Research. 2007. Chapter 7.9. Colon and Rectum. Food, Nutrition, Physical Activity, and the Prevention of Cancer: A Global Perspective. Washington, DC: AICR, pages 280–​288. Ananthakrishnan AN, Du M, Berndt SI, et al. 2015. Red meat intake, NAT2, and risk of colorectal cancer:  a pooled analysis of 11 studies. Cancer Epidemiol Biomarkers Prev. PMCID: PMC4294960.

698

698

PART IV:  Cancers by Tissue of Origin

Andersen V, Holst R, and Vogel U. 2013. Systematic review: diet-​gene interactions and the risk of colorectal cancer. Aliment Pharmacol Ther, 37(4), 383–​391. PMCID: PMC3565452. Andersen V, and Vogel U. 2015. Interactions between meat intake and genetic variation in relation to colorectal cancer. Genes Nutr, 10(1), 448. PMCID: PMC4261072. Arain MA, Sawhney M, Sheikh S, et al. 2010. CIMP status of interval colon cancers: another piece to the puzzle. Am J Gastroenterol, 105(5), 1189–​ 1195. PMID: 20010923. Armstrong B, and Doll R. 1975. Environmental factors and cancer incidence and mortality in different countries, with special reference to dietary practices. Int J Cancer, 15(4), 617–​631. PMID: 1140864. Asano TK, and McLeod RS. 2004. Nonsteroidal anti-​inflammatory drugs and aspirin for the prevention of colorectal adenomas and cancer: a systematic review. Dis Colon Rectum, 47(5), 665–​673. PMID: 15054679. Aune D, Chan DS, Lau R, et al. 2011. Dietary fibre, whole grains, and risk of colorectal cancer: systematic review and dose-​response meta-​analysis of prospective studies. BMJ, 343, d6617. PMCID: PMC3213242. Azcárate-​Peril MA, Sikes M, and Bruno-​Bárcena JM. 2011. The intestinal microbiota, gastrointestinal environment and colorectal cancer:  a putative role for probiotics in prevention of colorectal cancer? Am J Physiol Gastrointest Liver Physiol, 301(3), G401-​G424. PMCID: PMC3774253. Bailey CE, Hu CY, You YN, et al. 2015. Increasing disparities in the age-​related incidences of colon and rectal cancers in the United States, 1975–​2010. JAMA Surg, 150(1), 17–​22. PMCID: PMC4666003. Barault L, Charon-​Barra C, Jooste V, et al. 2008. Hypermethylator phenotype in sporadic colon cancer: study on a population-​based series of 582 cases. Cancer Res, 68(20), 8541–​8546. PMID: 18922929. Baron JA. 2005. Dietary fiber and colorectal cancer: an ongoing saga. JAMA, 294(22), 2904–​2906. PMID: 16352800. Baron JA, Barry EL, Mott LA, et al. 2015. A trial of calcium and vitamin d for the prevention of colorectal adenomas. N Engl J Med, 373(16), 1519–​ 1530. PMCID: PMC4643064. Bastide NM, Pierre FH, and Corpet DE. 2011. Heme iron from meat and risk of colorectal cancer:  a meta-​analysis and a review of the mechanisms involved. Cancer Prev Res (Phila), 4(2), 177–​184. PMID: 21209396. Baxter RC. 2014. IGF binding proteins in cancer:  mechanistic and clinical insights. Nat Rev Cancer, 14(5), 329–​341. PMID: 24722429. Bernstein AM, Song M, Zhang X, et al. 2015. Processed and unprocessed red meat and risk of colorectal cancer: analysis by tumor location and modification by time. PLoS One, 10(8), e0135959. PMCID: PMC4549221. Bernstein AM, Sun Q, Hu FB, et al. 2010. Major dietary protein sources and risk of coronary heart disease in women. Circulation, 122(9), 876–​883. PMCID: PMC2946797. Bernstein H, Bernstein C, Payne CM, Dvorakova K, and Garewal H. 2005. Bile acids as carcinogens in human gastrointestinal cancers. Mutat Res, 589(1), 47–​65. PMID: 15652226. Bibbins-​Domingo K, Grossman D, Curry S, et al. 2016. Screening for colorectal cancer:  US Preventitive Services Task Force recommendation statement. JAMA, 315(23), 2564–​2575. PMID: 27304597. Bingham SA. 1999. High-​meat diets and cancer risk. Proc Nutr Soc, 58(2), 243–​248. PMID: 10466162. Bisgaard ML, Fenger K, Bulow S, Niebuhr E, and Mohr J. 1994. Familial adenomatous polyposis (FAP): frequency, penetrance, and mutation rate. Hum Mutat, 3(2), 121–​125. PMID: 8199592. Bissonnette M, Khare S, von Lintig FC, et  al. 2000. Mutational and nonmutational activation of p21ras in rat colonic azoxymethane-​induced tumors:  effects on mitogen-​activated protein kinase, cyclooxygenase-​2, and cyclin D1. Cancer Res, 60(16), 4602–​4609. PMID: 10969813. Bonser AM, and Garcia-​Webb P. 1984. C-​peptide measurement: methods and clinical utility. Crit Rev Clin Lab Sci, 19(4), 297–​352. PMID: 6373142. Botteri E, Iodice S, Bagnardi V, et al. 2008a. Smoking and colorectal cancer: a meta-​analysis. JAMA, 300(23), 2765–​2778. PMID: 19088354. Botteri E, Iodice S, Raimondi S, Maisonneuve P, and Lowenfels AB. 2008b. Cigarette smoking and adenomatous polyps:  a meta-​ analysis. Gastroenterology, 134(2), 388–​395. PMID: 18242207. Bouvard V, Loomis D, Guyton KZ, et al. 2015. Carcinogenicity of consumption of red and processed meat. Lancet Oncol, 16(16), 1599–​ 1600. PMID: 26514947. Boyle T, Keegel T, Bull F, Heyworth J, and Fritschi L. 2012. Physical activity and risks of proximal and distal colon cancers:  a systematic review and meta-​analysis. J Natl Cancer Inst, 104(20), 1548–​1561. PMID: 22914790. Brenner H, Chang-​Claude J, Rickert A, Seiler CM, and Hoffmeister M. 2012. Risk of colorectal cancer after detection and removal of adenomas at colonoscopy: population-​based case-​control study. J Clin Oncol, 30(24), 2969–​2976. PMID: 22826281.

Brenner H, Chang-​Claude J, Seiler CM, and Hoffmeister M. 2011. Long-​term risk of colorectal cancer after negative colonoscopy. J Clin Oncol, 29(28), 3761–​3767. PMID: 21876077. Brentnall TA, Crispin DA, Rabinovitch PS, et al. 1994. Mutations in the p53 gene:  an early marker of neoplastic progression in ulcerative colitis. Gastroenterology, 107(2), 369–​378. PMID: 8039614. Bristow SM, Bolland MJ, MacLennan GS, et al. 2013. Calcium supplements and cancer risk:  a meta-​analysis of randomised controlled trials. Br J Nutr, 110(8), 1384–​1393. PMID: 23601861. Bufill JA. 1990. Colorectal cancer:  evidence for distinct genetic categories based on proximal or distal tumor location. Ann Intern Med, 113(10), 779–​788. PMID: 2240880. Burkitt DP. 1969. Related disease: related cause? Lancet, 2(7632), 1229–​1231. PMID: 4187817. Buset M, Kipkin M, Winawer S, Swaroop S, and Friedman E. 1986. Inhibition of human colonic epithelial cell proliferation in vivo and in vitro by calcium. Cancer Res, 46(10), 5426–​5430. PMID: 3756891. Butterworth AS, Higgins JP, and Pharoah P. 2006. Relative and absolute risk of colorectal cancer for individuals with a family history: a meta-​analysis. Eur J Cancer, 42(2), 216–​227. PMID: 16338133. Canavan C, Abrams KR, and Mayberry J. 2006. Meta-​analysis:  colorectal and small bowel cancer risk in patients with Crohn’s disease. Aliment Pharmacol Ther, 23(8), 1097–​1104. PMID: 16611269. Carethers JM, and Jung BH. 2015. Genetics and genetic biomarkers in sporadic colorectal cancer. Gastroenterology, 149(5), 1177–​ 1190 PMCID: PMC4589489. Carr PR, Walter V, Brenner H, and Hoffmeister M. 2016. Meat subtypes and their association with colorectal cancer:  systematic review and meta-​ analysis. Int J Cancer, 138(2), 293–​302. PMID: 25583132. Cats A, Dullaart RP, Kleibeuker JH, et al. 1996. Increased epithelial cell proliferation in the colon of patients with acromegaly. Cancer Res, 56(3), 523–​526. PMID: 8564965. Center MM, Jemal A, and Ward E. 2009. International trends in colorectal cancer incidence rates. Cancer Epidemiol Biomarkers Prev, 18(6), 1688–​ 1694. PMID: 19505900. Centers for Disease Control. 2014. The Health Consequences of Smoking—​ 50  years of Progress:  A  Report of the Surgeon General. Atlanta, GA: Centers for Disease Control and Prevention. Centers for Disease Control. 2015. Colorectal cancer rates by race and ethnicity. Available from:  http://​www.cdc.gov/​cancer/​colorectal/​statistics/​race. htm Accessed January 2016. Chan AT, and Giovannucci EL. 2010. Primary prevention of colorectal cancer. Gastroenterology, 138(6), 2029–​2043 PMCID: PMC2947820. Chan AT, Giovannucci EL, Meyerhardt JA, et al. 2005. Long-​term use of aspirin and nonsteroidal anti-​inflammatory drugs and risk of colorectal cancer. JAMA, 294(8), 914–​923. PMCID: PMC1550973. Chan AT, Ogino S, Fuchs CS. 2007. Aspirin and the risk of colorectal cancer in relation to the expression of Cox-2. N Engl J Med, 356(21), 2131–2142. PMID: 17522398. Chan AT, Ogino S, and Fuchs CS. 2009. Aspirin use and survival after diagnosis of colorectal cancer. JAMA, 302(6), 649–​659. PMCID: PMC2848289. Chan DS, Lau R, Aune D et al. 2011. Red and processed meat and colorectal cancer incidence: meta-​analysis of prospective studies. PLoS One, 6(6), e20456. PMCID: PMC3108955. Chen L, Li L, Wang Y et al. 2013. Circulating C-​peptide level is a predictive factor for colorectal neoplasia: evidence from the meta-​analysis of prospective studies. Cancer Causes Control, 24(10), 1837–​1847. PMID: 23846284. Cho E, Smith-​Warner SA, Ritz J, et al. 2004a. Alcohol intake and colorectal cancer:  a pooled analysis of 8 cohort studies. Ann Intern Med, 140(8), 603–​613. PMID: 15096331. Cho E, Smith-​Warner SA, Spiegelman D, et  al. 2004b. Dairy foods and calcium and colorectal cancer: a pooled analysis of 10 cohort studies. J Natl Cancer Inst, 96(13), 1015–​1022. PMID: 15240785. Citronberg J, Kantor ED, Potter JD, and White E. 2014. A prospective study of the effect of bowel movement frequency, constipation, and laxative use on colorectal cancer risk. Am J Gastroenterol, 109(10), 1640–​1649. PMCID: PMC4734897. Clemmons DR. 1997. Insulin-​like growth factor binding proteins and their role in controlling IGF actions. Cytokine Growth Factor Rev, 8(1), 45–​62. PMID: 9174662. Colao A, Pivonello R, Auriemma RS, et  al. 2007. The association of fasting insulin concentrations and colonic neoplasms in acromegaly:  a colonoscopy-​ based study in 210 patients. J Clin Endocrinol Metab, 92(10), 3854–​3860. PMID: 17652220. Cole BF, Logan RF, Halabi S, et  al. 2009. Aspirin for the chemoprevention of colorectal adenomas:  meta-​analysis of the randomized trials. J Natl Cancer Inst, 101(4), 256–​266. PMID: 19211452.

 69



Cancers of the Colon and Rectum

Cook NR, Lee IM, Gaziano JM, et al. 2005. Low-​dose aspirin in the primary prevention of cancer the women’s health study: a randomized controlled trial. JAMA, 294(1), 47–​55. PMID: 15998890. Cook NR, Lee IM, Zhang SM, Moorthy MV, and Buring JE. 2013. Alternate-​day, low-​dose aspirin and cancer risk: long-​term observational follow-​up of a randomized trial. Ann Intern Med, 159(2), 77–​85. PMCID: PMC3713531. Cotton S, Sharp L, and Little J. 1996. The adenoma-​carcinoma sequence and prospects for the prevention of colorectal neoplasia. Crit Rev Oncog, 7(5-​ 6), 293–​342. PMID: 9467664. Coussens LM, and Werb Z. 2002. Inflammation and cancer. Nature, 420(6917), 860–​867. PMCID: PMC2803035. Cress RD, Morris C, Ellison GL, and Goodman MT. 2006. Secular changes in colorectal cancer incidence by subsite, stage at diagnosis, and race/​ethnicity, 1992–​2001. Cancer, 107(5 Suppl), 1142–​1152. PMID: 16835912. Cross AJ, Ferrucci LM, Risch A, et al. 2010. A large prospective study of meat consumption and colorectal cancer risk:  an investigation of potential mechanisms underlying this association. Cancer Res, 70(6), 2406–​2414. PMCID: PMC2840051. Cross AJ, Pollock JR, and Bingham SA. 2003. Haem, not protein or inorganic iron, is responsible for endogenous intestinal N-​nitrosation arising from red meat. Cancer Res, 63(10), 2358–​2360. PMID: 12750250. Cross AJ, and Sinha R. 2004. Meat-​ related mutagens/​ carcinogens in the etiology of colorectal cancer. Environ Mol Mutagen, 44(1), 44–​ 55. PMID: 15199546. Cucino C, Buchner AM, and Sonnenberg A. 2002. Continued rightward shift of colorectal cancer. Dis Colon Rectum, 45(8), 1035–​ 1040. PMID: 12195187. Cummings LC, and Cooper GS. 2011. Colorectal cancer screening: update for 2011. Semin Oncol, 38(4), 483–​489. PMID: 21810507. Curtin K, Samowitz WS, Wolff RK, et al. 2009. Somatic alterations, metabolizing genes and smoking in rectal cancer. Int J Cancer, 125(1), 158–​164. PMCID: PMC2782655. Debinski HS, Love S, Spigelman AD, and Phillips RK. 1996. Colorectal polyp counts and cancer risk in familial adenomatous polyposis. Gastroenterology, 110(4), 1028–​1030. PMID: 8612989. DeFronzo RA. 1992. Pathogenesis of type 2 (non-​insulin dependent) diabetes mellitus:  a balanced overview. Diabetologia, 35(4), 389–​ 397. PMID: 1516769. Dellavalle CT, Xiao Q, Yang G, et al. 2014. Dietary nitrate and nitrite intake and risk of colorectal cancer in the Shanghai Women’s Health Study. Int J Cancer, 134(12), 2917–​2926. PMCID: PMC3980001. DeMarini DM. 2004. Genotoxicity of tobacco smoke and tobacco smoke condensate: a review. Mutat Res, 567(2–​3), 447–​474. PMID: 15572290. Despres JP. 2006. Abdominal obesity:  the most prevalent cause of the metabolic syndrome and related cardiometabolic risk. Eur Heart J Suppl, 8(B), B4–​B12. Dirx MJ, van den Brandt PA, Goldbohm RA, and Lumey LH. 2003. Energy restriction early in life and colon carcinoma risk:  results of The Netherlands Cohort Study after 7.3  years of follow-​up. Cancer, 97(1), 46–​55. PMID: 12491504. Doll R, Gray R, Hafner B, and Peto R. 1980. Mortality in relation to smoking:  22  years’ observations on female British doctors. Br Med J, 280(6219), 967–​971. PMCID: PMC1601142. Doll R, and Peto R. 1976. Mortality in relation to smoking: 20 years’ observations on male British doctors. Br Med J, 2(6051), 1525–​ 1536. PMCID: PMC1690096. Domingo E, Church DN, Sieber O, et al. 2013. Evalution of PIK3CA mutation as a predictor of benefit from nonsteroidal anti-​inflammatory drug therapy in colorectal cancer. J Clin Oncol, 31(34), 4297–​4305. PMID: 24062397. Dozois EJ, Boardman LA, Suwanthanma W, et al. 2008. Young-​onset colorectal cancer in patients with no known genetic predisposition: can we increase early recognition and improve outcome? Medicine (Baltimore), 87(5), 259–​263. PMCID: PMC4437192. Duffy LC, Raiten DJ, Hubbard VS, and Starke-​Reed P. 2015. Progress and challenges in developing metabolic footprints from diet in human gut microbial cometabolism. J Nutr, 145(5), 1123S–​1130S. PMCID: PMC4410496. Eaden J. 2004. Review article: colorectal carcinoma and inflammatory bowel disease. Aliment Pharmacol Ther, 20(4), 24–​30. PMID: 15352890. Egeberg R, Olsen A, Christensen J, et al. 2013. Associations between red meat and risks for colon and rectal cancer depend on the type of red meat consumed. J Nutr, 143(4), 464–​472. PMID: 23427329. Ekbom A, Yuen J, Adami HO, et al. 1993. Cholecystectomy and colorectal cancer. Gastroenterology, 105(1), 142–​147. PMID: 8514031. Elmunzer BJ, Hayward RA, Schoenfeld PS, et  al. 2012. Effect of flexible sigmoidoscopy-​based screening on incidence and mortality of colorectal cancer: a systematic review and meta-​analysis of randomized controlled trials. PLoS Med, 9(12), e1001352. PMCID: PMC3514315.

699

Engeland A, Tretli S, Austad G, and Bjorge T. 2005. Height and body mass index in relation to colorectal and gallbladder cancer in two million Norwegian men and women. Cancer Causes Control, 16(8), 987–​996. PMID: 16132807. Erdrich J, Zhang X, Giovannucci E, and Willett W. 2015. Proportion of colon cancer attributable to lifestyle in a cohort of US women. Cancer Causes Control, 26(9), 1271–​1279. PMCID: PMC4545459. Fearnhead NS, Britton MP, and Bodmer WF. 2001. The ABC of APC. Hum Mol Genet, 10(7), 721–​733. PMID: 11257105. Fearon ER, and Vogelstein B. 1990. A genetic model for colorectal tumorigenesis. Cell, 61(5), 759–​767. PMID: 2188735. Fedirko V, Riboli E, Tjonneland A, et al. 2012. Prediagnostic 25-​hydroxyvitamin D, VDR and CASR polymorphisms, and survival in patients with colorectal cancer in western European ppulations. Cancer Epidemiol Biomarkers Prev, 21(4), 582–​593. PMCID: PMC3378514. Feldman D, Krishnan AV, Swami S, Giovannucci E, and Feldman BJ. 2014. The role of vitamin D in reducing cancer risk and progression. Nat Rev Cancer, 14(5), 342–​357. PMID: 24705652. Flood DM, Weiss NS, Cook LS, et  al. 2000. Colorectal cancer incidence in Asian migrants to the United States and their descendants. Cancer Causes Control, 11(5), 403–​411. PMID: 10877333. Flossmann E, and Rothwell PM. 2007. Commentary:  aspirin and colorectal cancer an epidemiological success story. Int J Epidemiol, 36(5), 962–​965. PMID: 17954716. Friedenreich CM, and Orenstein MR. 2002. Physical activity and cancer prevention:  etiologic evidence and biological mechanisms. J Nutr, 132(11 Suppl), 3456S–​3464S. PMID: 12421870. Fuemmeler BF, Pendzich MK, and Tercyak KP. 2009. Weight, dietary behavior, and physical activity in childhood and adolescence: implications for adult cancer risk. Obes Facts, 2(3), 179–​186. PMCID: PMC2924235. Fung TT, and Brown LS. 2013. Dietary patterns and the risk of colorectal cancer. Curr Nutr Rep, 2(1), 48–​55. PMCID: PMC3649851. Gagniere J, Raisch J, Veziant J, et  al. 2016. Gut microbiota imbalance and colorectal cancer. World J Gastroenterol, 22(2), 501–​ 518. PMCID: PMC4716055. Gala MK, and Chan AT. 2015. Molecular pathways: aspirin and Wnt signaling-​ a molecularly targeted approach to cancer prevention and treatment. Clin Cancer Res, 21(7), 1543–​1548. PMCID: PMC4383688. Gao QY, Chen HM, Chen YX, et al. 2013. Folic acid prevents the initial occurrence of sporadic colorectal adenoma in Chinese older than 50 years of age: a randomized clinical trial. Cancer Prev Res (Phila), 6(7), 744–​752. PMID: 23682073. Garland CF, and Garland FC. 1980. Do sunlight and vitamin D reduce the likelihood of colon cancer? Int J Epidemiol, 9(3), 227–​231. PMID: 7440046. Gay LJ, Mitrou PN, Keen J, et al. 2012. Dietary, lifestyle and clinicopathological factors associated with APC mutations and promoter methylation in colorectal cancers from the EPIC-​Norfolk study. J Pathol, 228(3), 405–​ 415. PMID: 22864938. Geer EB, and Shen W. 2009. Gender differences in insulin resistance, body composition, and energy balance. Gend Med, 6(Suppl 1), 60–​ 75. PMCID: PMC2908522. Giovannucci E. 2001a. Insulin, insulin-​like growth factors and colon cancer:  a review of the evidence. J Nutr, 131(11 Suppl), 3109S–​3120S. PMID: 11694656. Giovannucci E. 2001b. An updated review of the epidemiological evidence that cigarette smoking increases risk of colorectal cancer. Cancer Epidemiol Biomarkers Prev, 10(7), 725–​731. PMID: 11440957. Giovannucci E. 2004. Alcohol, one-​carbon metabolism, and colorectal cancer: recent insights from molecular studies. J Nutr, 134(9), 2475S–​2481S. PMID: 15333745. Giovannucci E. 2007. Metabolic syndrome, hyperinsulinemia, and colon cancer: a review. Am J Clin Nutr, 86(3), s836–​s842. PMID: 18265477. Giovannucci E. 2009. Vitamin D and cancer incidence in the Harvard cohorts. Ann Epidemiol, 19(2), 84–​88. PMID: 18291673. Giovannucci E. 2013. Epidemiology of vitamin D and colorectal cancer. Anticancer Agents Med Chem, 13(1), 11–​19. PMID: 23094917. Giovannucci E, Ascherio A, Rimm EB, et al. 1995. Physical activity, obesity, and risk for colon cancer and adenoma in men. Ann Intern Med, 122(5), 327–​334. PMID: 7847643. Giovannucci E, Colditz GA, and Stampfer MJ. 1993. A meta-​analysis of cholecystectomy and risk of colorectal cancer. Gastroenterology, 105(1), 130–​ 141. PMID: 8514030. Giovannucci E, Colditz GA, Stampfer MJ, et al. 1994. A prospective study of cigarette smoking and risk of colorectal adenoma and colorectal cancer in U.S. women. J Natl Cancer Inst, 86(3), 192–​199. PMID: 8283491. Giovannucci E, Colditz GA, Stampfer MJ, and Willett WC. 1996. Physical activity, obesity, and risk of colorectal adenoma in women (United States). Cancer Causes Control, 7(2), 253–​263. PMID: 8740738.

70

700

PART IV:  Cancers by Tissue of Origin

Giovannucci E, Harlan DM, Archer MC, et al. 2010. Diabetes and cancer: a consensus report. CA Cancer J Clin, 60(4), 207–​221. PMID: 20554718. Giovannucci E, Pollak MN, Platz EA, et al. 2000. A prospective study of plasma insulin-​like growth factor-​I and binding protein-​3 and risk of colorectal neoplasia in women. Cancer Epidemiol Biomarkers Prev, 9(4), 345–​349. PMID: 10794477. Giovannucci E, and Wu K. 2006. Cancers of the colon and rectum. In:  Schottenfeld D, and Fraumeni J (Eds.), Cancer Epidemiology and Prevention (3rd ed.). New York: Oxford University Press. Goldbohm RA, Van den Brandt PA, van’t Veer P, et al. 1993. Cholecystectomy and colorectal cancer: evidence from a cohort study on diet and cancer. Int J Cancer, 53(5), 735–​739. PMID: 8449596. Green J, Cairns BJ, Casabonne D, et  al. 2011. Height and cancer incidence in the Million Women Study:  prospective cohort, and meta-​analysis of prospective studies of height and total cancer risk. Lancet Oncol, 12(8), 785–​794. PMCID: PMC3148429. Grundy SM. 1999. Hypertriglyceridemia, insulin resistance, and the metabolic syndrome. Am J Cardiol, 83(9B), 25F–​29F. PMID: 10357572. Guinney J, Dienstmann R, Wang X, et  al. 2015. The consensus molecular subtypes of colorectal cancer. Nat Med, 21(11), 1350–​ 1356. PMCID: PMC4636487. Haenszel W, and Kurihara M. 1968. Studies of Japanese migrants. I. Mortality from cancer and other diseases among Japanese in the United States. J Natl Cancer Inst, 40(1), 43–​68. PMID: 5635018. Hamilton MT, Hamilton DG, and Zderic TW. 2007. Role of low energy expenditure and sitting in obesity, metabolic syndrome, type 2 diabetes, and cardiovascular disease. Diabetes, 56(11), 2655–​2667. PMID: 17827399. Hamilton SR, Liu B, Parsons RE, et al. 1995. The molecular basis of Turcot’s syndrome. N Engl J Med, 332(13), 839–​847. PMID: 7661930. Hammond CE, and Horn D. 1966. Smoking in relation to the death rates of one million men and women. Natl Cancer Inst Monog, 19, 127–​204. PMID: 5905667. Hammond EC, and Horn D. 1958. Smoking and death rates: report on forty-​ four months of followup of 187,783 men. II. Death rates by causality. J Am Med Assoc, 166(11), 1294–​1308. PMID: 12308037. Hannan LM, Jacobs EJ, and Thun MJ. 2009. The association between cigarette smoking and risk of colorectal cancer in a large prospective cohort from the United States. Cancer Epidemiol Biomarkers Prev, 18(12), 3362–​ 3367. PMID: 19959683. Harriss DJ, Cable NT, George K, et  al. 2007. Physical activity before and after diagnosis of colorectal cancer:  disease risk, clinical outcomes, response pathways and biomarkers. Sports Med, 37(11), 947–​ 960. PMID: 17953466. Hasler WL, and Owyang C. 2012. Approach to the patient with gastrointestinal disease. In:  Longo D, Fauci A, Kasper D, Hauser S, Jameson J, and Loscalzo J (Eds.), Harrison’s Principles of Internal Medicine, 19e. New  York:  McGraw-​Hill. http://​accessmedicine.mhmedical. com.ezp-​ p rod1.hul.harvard.edu/ ​ c ontent.aspx?bookid=1130&Sectio nid=79747222. Accessed November 21, 2016. He TC, Sparks AB, Rago C, et al. 1998. Identification of c-​MYC as a target of the APC pathway. Science, 281(5382), 1509–​1512. PMID: 9727977. Heitman SJ, Ronksley PE, Hilsden RJ, et  al. 2009. Prevalence of adenomas and colorectal cancer in average risk individuals:  a systematic review and meta-​analysis. Clin Gastroenterol Hepatol, 7(12), 1272–​ 1278. PMID: 19523536. Herman JG, Umar A, Polyak LK, et al. 1998. Incidence and functional consequences of hMLH1 promoter hypermethylation in colorectal carcinoma. Proc Natl Acad Sci U S A, 95(12), 6870–​6875. PMCID: PMC22665. Hewitson P, Glasziou P, Irwig L, Towler B, and Watson E. 2007. Screening for colorectal cancer using the faecal occult blood test, Hemoccult. Cochrane Database Syst Rev 1, CD001216. PMID: 17253456. Hewitson P, Glasziou P, Watson E, Towler B, and Irwig L. 2008. Cochrane systematic review of colorectal cancer screening using the fecal occult blood test (hemoccult):  an update. Am J Gastroenterol, 103(6), 1541–​ 1549. PMID: 18479499. Hillman RS, and Steinberg SE. 1982. The effects of alcohol on folate metabolism. Ann Rev Med, 33, 345–​354. PMID: 6805415. Hjartaker A, Aagnes B, Robsahm TE, et al. 2013. Subsite-​specific dietary risk factors for colorectal cancer: a review of cohort studies. J Oncol, 2013, 703854. PMCID: PMC3610350. Hoffmann D, and Hoffman I. 1997. The changing cigarette, 1950–​1995. J Toxicol Environ Health, 50(4), 307–​364. PMID: 9120872. Hol L, van Leerdam ME, van Ballegooijen M et al. 2010. Screening for colorectal cancer: randomised trial comparing guaiac-​based and immunochemical faecal occult blood testing and flexible sigmoidoscopy. Gut, 59(1), 62–​68. PMID: 19671542.

Holme O, Bretthauer M, Eide TJ, et  al. 2015. Long-​term risk of colorectal cancer in individuals with serrated polyps. Gut, 64(6), 929–​ 936. PMID: 25399542. Hord NG, Tang Y, and Bryan NS. 2009. Food sources of nitrates and nitrites: the physiologic context for potential health benefits. Am J Clin Nutr, 90(1), 1–​10. PMID: 19439460. Houlston RS, and Tomlinson IP. 2001. Polymorphisms and colorectal tumor risk. Gastroenterology, 121(2), 282–​301. PMID: 11487538. Howard RA, Freedman DM, Park Y, et  al. 2008. Physical activity, sedentary behavior, and the risk of colon and rectal cancer in the NIH-​AARP Diet and Health Study. Cancer Causes Control, 19(9), 939–​ 953. PMCID: PMC2574900. Hsing AW, McLaughlin JK, Chow WH, et al. 1998. Risk factors for colorectal cancer in a prospective study among U.S. white men. Int J Cancer, 77(4), 549–​553. PMID: 9679757. Hu FB. 2002. Dietary pattern analysis: a new direction in nutritional epidemiology. Curr Opin Lipidol, 13(1), 3–​9. PMID: 11790957. Hu FB, Manson JE, Liu S, et al. 1999. Prospective study of adult onset diabetes mellitus (Type 2) and risk of colorectal cancer in women. J Natl Cancer Inst, 91(6), 542–​547. PMID: 10088625. Hughes LA, van den Brandt PA, Goldbohm RA, et al. 2010. Childhood and adolescent energy restriction and subsequent colorectal cancer risk:  results from the Netherlands Cohort Study. Int J Epidemiol, 39(5), 1333–​1344. PMID: 20427463. Huxley RR, Ansary-​Moghaddam A, Clifton P, et al. 2009. The impact of dietary and lifestyle risk factors on risk of colorectal cancer: a quantitative overview of the epidemiological evidence. Int J Cancer, 125(1), 171–​180. PMID: 19350627. Iacopetta B. 2002. Are there two sides to colorectal cancer? Int J Cancer, 101(5), 403–​408. PMID: 12216066. Imperiale TF, Ransohoff DF, and Itzkowitz SH. 2014. Multitarget stool DNA testing for colorectal-​cancer screening. N Engl J Med, 371(2), 187–​188. PMID: 25006736. Institute of Medicine National Cancer Policy. 2008. Implementing Colorectal Cancer Screening:  Workshop Summary. Washington, DC:  National Academies Press. International Agency for Research on Cancer. 2010. IARC Working Group on the Evaluation of Carcinogenic Risk to Humans Alcohol Consumption and Ethyl Carbamate. Lyon, France: International Agency for Research on Cancer. International Agency for Research on Cancer/​ World Health Organization. 2015. GLOBOCAN 2012:  Estimated cancer incidence, mortality and prevalence worldwide in 2012. Available from:  http://​globocan.iarc.fr/​ Pages/​fact_​sheets_​cancer.aspx. Accessed January 5, 2016. Irving BA, Davis CK, Brock DW et al. 2008. Effect of exercise training intensity on abdominal visceral fat and body composition. Med Sci Sports Exerc, 40(11), 1863–​1872. PMCID: PMC2730190. Jass JR. 2007. Molecular heterogeneity of colorectal cancer: implications for cancer control. Surg Oncol, 16(Suppl 1), S7–​S9. PMID: 18023574. Jeffreys M, Smith GD, Martin RM, Frankel S, and Gunnell D. 2004. Childhood body mass index and later cancer risk: a 50-​year follow-​up of the Boyd Orr study. Int J Cancer, 112(2), 348–​351. PMID: 15352051. Jenab M, Bueno-​de-​Mesquita HB, Ferrari P, et al. 2010. Association between pre-​diagnostic circulating vitamin D concentration and risk of colorectal cancer in European populations:a nested case-​control study. BMJ, 340, b5500. PMCID: PMC2809840. Jenkins PJ. (1999). Acromegaly and Colon Cancer. Paper presented at the First International Workshop on Growth Hormone, Insulin-​like Growth Factors and Neoplasia, Boston, MA. PMID: 10984286. Johns LE, and Houlston RS. 2001. A systematic review and meta-​analysis of familial colorectal cancer risk. Am J Gastroenterol, 96(10), 2992–​3003. PMID: 11693338. Johnson CM, Wei C, Ensor JE, et  al. 2013. Meta-​ analyses of colorectal cancer risk factors. Cancer Causes Control, 24(6), 1207–​ 1222. PMCID: PMC4161278. Jokelainen K, Nosova T, Koivisto T, et al. 1997. Inhibition of bacteriocolonic pathway for ethanol oxidation by ciprofloxacin in rats. Life Sci, 61(18), 1755–​1762. PMID: 9365222. Juul A, Main K, Blum WF, et  al. 1994. The ratio between serum levels of insulin-​like growth factor (IGF)-​I and the IGF binding proteins (IGFBP-​1, 2 and 3)  decreases with age in healthy adults and is increased in acromegalic patients. Clin Endocrinol (Oxf), 41(1), 85–​93. PMID: 7519529. Kaaks R, and Lukanova A. 2001. Energy balance and cancer:  the role of insulin and insulin-​like growth factor-​I. Proc Nutr Soc, 60(1), 91–​106. PMID: 11310428.

 701



Cancers of the Colon and Rectum

Kaaks R, Toniolo P, Akhmedkhanov A, et al. 2000. Serum C-​peptide, insulin-​ like growth factor (IGF)-​1, IGF-​binding proteins, and colorectal cancer risk in women. J Natl Cancer Inst, 92(19), 1592–​1600. PMID: 11018095. Kahn HA. 1966. The Dorn study of smoking and mortality among U.S. veterans: report on eight and one-​half years of observation. Natl Cancer Inst Monogr, 19, 1–​125. PMID: 5905668. Kang GH. 2011. Four molecular subtypes of colorectal cancer and their precursor lesions. Arch Pathol Lab Med, 135(6), 698–​703. PMID: 21631262. Kantor ED, Hutter CM, Minnier J, et al. 2014. Gene-​environment interaction involving recently identified colorectal cancer susceptibility loci. Cancer Epidemiol Biomarkers Prev, 23(9), 1824–​1833. PMCID: PMC4209726. Keum N, Aune D, Greenwood DC, Ju W, and Giovannucci EL. 2014. Calcium intake and colorectal cancer risk: dose-​response meta-​analysis of prospective observational studies. Int J Cancer, 135(8), 1940–​1948. PMID: 24623471. Keum N, and Giovannucci EL. 2014. Folic acid fortification and colorectal cancer risk. Am J Prev Med, 46(3 Suppl 1), S65–​S72. PMID: 24512932. Keum N, Greenwood DC, Lee DH et al. 2015. Adult weight gain and adiposity-​ related cancers:  a dose-​response meta-​analysis of prospective observational studies. J Natl Cancer Inst, 107(3). PMID: 25618901. Keum N, Lee DH, Kim R, Greenwood DC, and Giovannucci EL. 2015. Visceral adiposity and colorectal adenomas:  dose-​ response meta-​ analysis of observational studies. Ann Oncol, 26(6), 1101–​ 1109. PMID: 25480876. Kim DH, Smith-​Warner SA, Spiegelman D, et  al. 2010. Pooled analyses of 13 prospective cohort studies on folate intake and colon cancer. Cancer Causes Control, 21(11), 1919–​1930. PMCID: PMC3082430. Kim YI. 2006. Folate: a magic bullet or a double edged sword for colorectal cancer prevention? Gut, 55(10), 1387–​1389. PMCID: PMC1856406. Kinzler KW, and Vogelstein B. 1996. Lessons from hereditary colorectal cancer. Cell, 87(2), 159–​170. PMID: 8861899. Knekt P, Jarvinen R, Dich J, and Hakulinen T. 1999. Risk of colorectal and other gastro-​intestinal cancers after exposure to nitrate, nitrite and N-​ nitroso compounds:  a follow-​up study. Int J Cancer, 80(6), 852–​856. PMID: 10074917. Kojima M, Wakai K, Tokudome S, et  al. 2004. Bowel movement frequency and risk of colorectal cancer in a large cohort study of Japanese men and women. Br J Cancer, 90(7), 1397–​1401. PMCID: PMC2409677. Kono S, Shinchi K, Ikeda N, Yanai F, and Imanishi K. 1991. Physical activity, dietary habits and adenomatous polyps of the sigmoid colon: a study of self-​defense officials in Japan. 44(11), 1255–​1261. PMID: 1941019. Kritchevsky D. 1995. Epidemiology of fibre, resistant starch and colorectal cancer. Eur J Cancer Prev, 4(5), 345–​352. PMID: 7496322. Kuchiba A, Morikawa T, Yamauchi M, et  al. 2012. Body mass index and risk of colorectal cancer according to fatty acid synthase expression in the nurses’ health study. J Natl Cancer Inst, 104(5), 415–​420. PMCID: PMC3295745. Kune GA. 2010. The Melbourne Colorectal Cancer Study: reflections on a 30-​ year experience. Med J Aust, 193(11-​12), 648–​652. PMID: 21143050. Kushi LH, Doyle C, McCullough M, et  al. 2012. American Cancer Society Guidelines on nutrition and physical activity for cancer prevention: reducing the risk of cancer with healthy food choices and physical activity. CA Cancer J Clin, 62(1), 30–​67. PMID: 22237782. Lampe JW, Fredstrom SB, Slavin JL, and Potter JD. 1993. Sex differences in colonic function:  a randomised trial. Gut, 34(4), 531–​ 536. PMCID: PMC1374316. Lamprecht SA, and Lipkin M. 2001. Cellular mechanisms of calcium and vitamin D in the inhibition of colorectal carcinogenesis. Ann N Y Acad Sci, 952, 73–​87. PMID: 11795445. Larsson SC, and Wolk A. 2006. Meat consumption and risk of colorectal cancer: a meta-​analysis of prospective studies. Int J Cancer, 119(11), 2657–​ 2664. PMID: 16991129. Laryea JA, Siegel E, and Klimberg S. 2014. Racial disparity in colorectal cancer: the role of equal treatment. Dis Colon Rectum, 57(3), 295–​302. PMID: 24509450. Le NT, Michels FA, Song M, et  al. 2016. A prospective analysis of meat mutagens and colorectal cancer in the Nurses’ Health Study and Health Professional Follow-​up Study. Environ Health Perspect, (Epub ahead of print). PMID: 27105317. Lee C, Kozlowski JM, and Grayhack JT. 1995. Etiology of benign prostatic hyperplasia. Urol Clin North Am, 22(2), 237–​246. PMID: 7539172. Lee JE, Willett WC, Fuchs CS, et al. 2011. Folate intake and risk of colorectal cancer and adenoma: modification by time. Am J Clin Nutr, 93(4), 817–​ 825. PMCID: PMC3057549. Lemire M, Qu C, Loo LW, et al. 2015. A genome-​wide association study for colorectal cancer identifies a risk locus in 14q23.1. Hum Genet, 134(11–​ 12), 1249–​1262. PMCID: PMC4687971.

701

Lennerz JK, Sloot KW, Le LP, et al. 2016. Colorectal cancer in Crogn’s colitis is comparable to sporadic colorectal cancer. Int J Colorectal Dis, 31(5), 973–​982. PMID: 27026089. Leufkens AM, Duijnhoven FJ, Boshuizen HC, et  al. 2012. Educational level and risk of colorectal cancer in EPIC with specific reference to tumor location. Int J Cancer, 130(3), 622–​630. PMID: 21412763. Lev R. 1990. Adenomatous Polyps of the Colon (p. 61). New York: Springer-​Verlag. Levi Z, Kark JD, Barchana M, et al. 2011. Measured body mass index in adolescence and the incidence of colorectal cancer in a cohort of 1.1  million males. Cancer Epidemiol Biomarkers Prev, 20(12), 2524–​ 2531. PMID: 22056504. Levin B, and Raijman I. 1995. Malignant tumors of the colon and rectum. In: Haubrich W S, Schaffner F, and Berk J (Eds.), Bockus Gastroenterology (5th ed., pp. 1744–​1772). Philadelphia: Saunders. Liang PS, Chen TY, and Giovannucci E. 2009. Cigarette smoking and colorectal cancer incidence and mortality: systematic review and meta-​analysis. Int J Cancer, 124(10), 2406–​2415. PMID: 19142968. Liao XY, Lochhead P, Nishihara R, et  al. 2012. Aspirin use, tumor PIK3CA mutation, and colorectal-​cancer survival. N Engl J Med, 367(17), 1596–​ 1606. PMCID: PMC3532946. Lieberman DA, Rex DK, Winawer SJ, et al. 2012. Guidelines for colonoscopy surveillance after screening and polypectomy: a consensus update by the US Multi-​Society Task Force on Colorectal Cancer. Gastroenterology, 143(3), 844–​857. PMID: 22763141. Lilienfeld AM, Levin ML, and Kessler II. 1972a. Mortality among the foreign born and their countries of origin. In:  Cancer in the United States (pp. 233–​278). Cambridge, MA: Harvard University Press. Lilienfeld AM, Levin ML, and Kessler II. 1972b. Summary of mortality data for selected sites. In: Cancer in the United States (pp. 279–​297). Cambridge, MA: Harvard University Press. Limsui D, Vierkant RA, Tillmans LS, et  al. 2010. Cigarette smoking and colorectal cancer risk by molecularly defined subtypes. J Natl Cancer Inst, 102(14), 1012–​1022. PMCID: PMC2915616. Lin KJ, Cheung WY, Lai JY, and Giovannucci EL. 2012. The effect of estrogen vs. combined estrogen-​progestogen therapy on the risk of colorectal cancer. Int J Cancer, 130(2), 419–​430. PMID: 21365647. Little J, Logan RF, Hawtin PG, Hardcastle JD, and Turner ID. 1993. Colorectal adenomas and energy intake, body size and physical activity: a case-​control study of subjects participating in the Nottingham faecal occult blood screening programme. Br J Cancer, 67(1), 172–​176. PMCID: PMC1968212. Littlejohn C, Hilton S, Macfarlane GJ, and Phull P. 2012. Systematic review and meta-​analysis of the evidence for flexible sigmoidoscopy as a screening method for the prevention of colorectal cancer. Br J Surg, 99(11), 1488–​1500. PMID: 23001715. Loberg M, Kalager M, Holme O, et  al. 2014. Long-​term colorectal-​cancer mortality after adenoma removal. N Engl J Med, 371(9), 799–​ 807. PMID: 25162886. Lochhead P, Chan AT, Giovannucci E, et al. 2014. Progress and opportunities in molecular pathological epidemiology of colorectal premalignant lesions. Am J Gastroenterol, 109(8), 1205–​1214. PMCID: PMC4125459. Luo S, Li JY, Zhao LN, et  al. 2016. Diabetes mellitus increases the risk of colorectal neoplasia:  an updated meta-​ analysis. Clin Res Hepatol Gastroenterol, 40(1), 110–​123. PMID: 26162991. Lynch HT, and Smyrk T. 1996. Hereditary nonpolyposis colorectal cancer (Lynch syndrome):  an updated review. Cancer, 78(6), 1149–​1167. PMID: 8826936. Lynch HT, and Smyrk T. 1998. An update on Lynch syndrome. Curr Opin Oncol, 10(4), 349–​356. PMID: 9702403. Lynch HT, Smyrk T, and Jass JR. 1995. Hereditary nonpolyposis colorectal cancer and colonic adenomas: aggressive adenomas? Semin Surg Oncol, 11(6), 406–​410. PMID: 8607009. Ma J, Pollak MN, Giovannucci E, et al. 1999. Prospective study of colorectal cancer risk in men and plasma levels of insulin-​like growth factor (IGF)-​1 and IGF-​ binding protein-​3. J Natl Cancer Inst, 91(7), 620–​625. PMID: 10203281. MacInnis RJ, English DR, Hopper JL, et al. 2004. Body size and composition and colon cancer risk in men. Cancer Epidemiol Biomarkers Prev, 13(4), 553–​559. PMID: 15066919. Magalhaes B, Peleteiro B, and Lunet N. 2012. Dietary patterns and colorectal cancer: systematic review and meta-​analysis. Eur J Cancer Prev, 21(1), 15–​23. PMID: 21946864. Manser CN, and Bauerfeind P. 2014. Impact of socioeconomic status on incidence, mortality, and survival of colorectal cancer patients: a systematic review. Gastrointest Endosc, 80(1), 42–​60. PMID: 24950641. Marisa L, Reynies A, Duval A, et al. 2013. Gene expression classification of colon cancer into molecular subtypes:  characterization, validation, and prognostic value. PLoS Med, 10(5), e1001453. PMCID: PMC3660251.

702

702

PART IV:  Cancers by Tissue of Origin

Markowitz SD, and Bertagnolli MM. 2009. Molecular origins of cancer: molecular basis of colorectal cancer. N Engl J Med, 361(25), 2449–​ 2460. PMCID: PMC2843693. Martinez ME, Giovannucci E, Spiegelman D, et  al. 1996. Physical activity, body size, and colorectal cancer in women (Abstract). Am J Epidemiol, 143, S73. Maskarinec G, and Noh JJ. 2004. The effect of migration on cancer incidence among Japanese in Hawaii. Ethn Dis, 14(3), 431–​ 439. PMID: 15328946. Mason JB, Dickstein A, Jacques PF, et  al. 2007. A  temporal association between folic acid fortification and an increase in colorectal cancer rates may be illuminating important biological principles: a hypothesis. Cancer Epidemiol Biomarkers Prev, 16(7), 1325–​1329. PMID: 17626997. McCredie M, Williams S, and Coates M. 1999. Cancer mortality in migrants from the British Isles and continental Europe to New South Wales, Australia, 1975–​1995. Int J Cancer, 83(2), 179–​185. PMID: 10471524. McGlynn AP, Wasson GR, O’Reilly SL, et  al. 2013. Low colonocyte folate is associated with uracil misincorporation and global DNA hypomethylation in human colorectum. J Nutr, 143(1), 27–​33. PMID: 23190761. McMichael AJ, and Potter JD. 1985. Host factors in carcinogenesis:  certain bile-​acid metabolic profiles that selectively increase the risk of proximal colon cancer. J Natl Cancer Inst, 75(2), 185–​191. PMID: 3894750. Meyskens FL, McLaren CE, Pelot D, et al. 2008. Difluoromethylornithine plus sulindac for the prevention of sporadic colorectal adenomas: a randomized placebo-​controlled, double-​blind trial. Cancer Prev Res (Phila), 1(1), 32–​38. PMCID: PMC2562024. Michels KB, Fuchs CS, Giovannucci E, et al. 2005. Fiber intake and incidence of colorectal cancer among 76,947 women and 47,279 men. Cancer Epidemiol Biomarkers Prev, 14(4), 842–​849. PMID: 15824154. Moghaddam AA, Woodward M, and Huxley R. 2007. Obesity and risk of colorectal cancer: a meta-​analysis of 31 studies with 70,000 events. Cancer Epidemiol Biomarkers Prev, 16(12), 2533–​2547. PMID: 18086756. Morikawa T, Kuchiba A, Lochhead P, et  al. 2013. Prospective analysis of body mass index, physical activity, and colorectal cancer risk associated with β-​catenin (CTNNB1) status. Cancer Res, 73(5), 1600–​1610. PMCID: PMC3594537. Morson BC. 1984. The evolution of colorectal carcinoma. Clin Radiol, 35(6), 425–​431. PMID: 6499378. Muller A, and Fishel R. 2002. Mismatch repair and the hereditary non-​ polyposis colorectal cancer syndrome (HNPCC). Cancer Invest, 20(1), 102–​109. PMID: 11852992. Murphy CC, Harlan LC, Warren JL, and Geiger AM. 2015. Race and insurance differences in the receipt of adjuvant chemotherapy among patients with stage III colon cancer. J Clin Oncol, 33(23), 2530–​2536. PMCID: PMC4525047. Murthy S, Flanigan A, and Clearfield H. 2002. Colorectal cancer in inflammatory bowel disease:  molecular and clinical features. Gastroenterol Clin North Am, 31(2), 551–​564. PMID: 12134618. Myers EA, Feingold DL, Forde KA, et al. 2013. Colorectal cancer in patients under 50 years of age: a retrospective analysis of two institutions’ experience. World J Gastroenterol, 19(34), 5651–​5657. PMCID: PMC3769901. Nan H, Hutter CM, Lin Y, et  al. 2015. Association of aspirin and NSAID use with risk of colorectal cancer according to genetic variants. JAMA, 313(11), 1133–​1142. PMCID: PMC4382867. National Institutes of Health. 2014. Tests to detect colorectal cancer and polyps. Available from:  http://​www.cancer.gov/​types/​colorectal/​screening-​fact-​ sheet Accessed January 5, 2016. Neugut AI, Jacobson JS, and De Vivo I. 1993. Epidemiology of colorectal adenomatous polyps. Cancer Epidemiol Biomarkers Prev, 2(2), 159–​176. PMID: 8467251. Neugut AI, Terry MB, Hocking G, et al. 1996. Leisure and occupational physical activity and risk of colorectal adenomatous polyps. Int J Cancer, 68(6), 744–​748. PMID: 8980177. Newmark HL, Wargovich MJ, and Bruce WR. 1984. Colon cancer and dietary fat, phosphate, and calcium:  a hypothesis. J Natl Cancer Inst, 72(6), 1323–​1325. PMID: 6587152. Nimptsch K, Giovannucci E, Willett WC, et  al. 2011. Body fatness during childhood and adolescence, adult height, and risk of colorectal adenoma in women. Cancer Prev Res (Phila), 4(10), 1710–​1718. PMCID: PMC3188350. Ning Y, Wang L, and Giovannucci EL. 2010. A quantitative analysis of body mass index and colorectal cancer: findings from 56 observational studies. Obes Rev, 11(1), 19–​30. PMID: 19538439. Nishihara R, Morikawa T, Kuchiba A, et  al. 2013a. A  prospective study of duration of smoking cessation and colorectal cancer risk by epigenetics-​ related tumor classification. Am J Epidemiol, 178(1), 84–​ 100. PMCID: PMC3698990.

Nishihara R, Wu K, Lochhead P, et  al. 2013b. Long-​term colorectal-​cancer incidence and mortality after lower endoscopy. N Engl J Med, 369(12), 1095–​1105. PMCID: PMC3840160. Norat T, Lukanova A, Ferrari P, and Riboli E. 2002. Meat consumption and colorectal cancer risk:  dose-​response meta-​analysis of epidemiological studies. Int J Cancer, 98(2), 241–​256. PMID: 11857415. Nyren O, Bergstrom R, Nystrom L, et al. 1996. Smoking and colorectal cancer: a 20-​year follow-​up study of Swedish construction workers. J Natl Cancer Inst, 88(18), 1302–​1307. PMID: 8797770. O’Reilly SL, McGlynn AP, McNulty H, et al. 2016. Folic acid supplementation in postpolypectomy patients in a randomized controlled trial increases tissue folate concentrations and reduces aberrant DNA biomarkers in colonic tissues adjacent to the former polyp site. J Nutr, 146(5), 933–​939. PMID: 27075913. Oettle GJ. 1991. Effect of moderate exercise on bowel habit. Gut, 32(8), 941–​ 944. PMCID: PMC1378967. Ogino S, and Goel A. 2008. Molecular classification and correlates in colorectal cancer. J Mol Diagn, 10(1), 13–​27. PMCID: PMC2175539. Ogino S, Kawasaki T, Kirkner GJ, et  al. 2007. Evaluation of markers for CpG island methylator phenotype (CIMP) in colorectal cancer by a large population-​ based sample. J Mol Diagn, 9(3), 305–​ 314. PMCID: PMC1899428. Ollberding NJ, Cheng I, Wilkens LR, et al. 2012. Genetic variants, prediagnostic circulating levels of insulin-​like growth factors, insulin, and glucose and the risk of colorectal cancer: the Multiethnic Cohort study. Cancer Epidemiol Biomarkers Prev, 21(5), 810–​820. PMCID: PMC4494075. Ollberding NJ, Wilkens LR, Henderson BE, Kolonel LN, and Le Marchand L. 2012. Meat consumption, heterocyclic amines and colorectal cancer risk: the Multiethnic Cohort Study. Int J Cancer, 131(7), E1125–​E1133. PMCID: PMC3553660. Orlando G, Law PJ, Palin K, et  al. 2016. Variation at 2q35 (PNKD and TMBIM1) influences colorectal cancer risk and identifies a pleiotropic effect with inflammatory bowel disease. Hum Mol Genet, (Epub ahead of print). PMID: 27005424. Osterhues A, Holzgreve W, and Michels KB. 2009. Shall we put the world on folate? Lancet, 374(9694), 959–​961. PMID: 19766871. Otani T, Iwasaki M, Ishihara J et al. 2006. Dietary fiber intake and subsequent risk of colorectal cancer: the Japan Public Health Center-​based prospective study. Int J Cancer, 119(6), 1475–​1480. PMID: 16642466. Otani T, Iwasaki M, Sasazuki S, et  al. 2007. Plasma C-​peptide, insulin-​like growth factor-​ I, insulin-​ like growth factor binding proteins and risk of colorectal cancer in a nested case-​control study:  the Japan public health center-​based prospective study. Int J Cancer, 120(9), 2007–​2012. PMID: 17266031. Palmqvist R, Hallmans G, Rinaldi S, et al. 2002. Plasma insulin-​like growth factor 1, insulin-​like growth factor binding protein 3, and risk of colorectal cancer: a prospective study in northern Sweden. Gut, 50(5), 642–​646. PMCID: PMC1773192. Park JY, Mitrou PN, Keen J, et  al. 2010. Lifestyle factors and p53 mutation patterns in colorectal cancer patients in the EPIC-​Norfolk study. Mutagenesis, 25(4), 351–​358. PMID: 20228093. Parkin D, Muir C, Whelan S, et al. 1992. Cancer incidence in five continents, comparability and quality of data. IARC Sci Publ, 143(120), 45–​173. PMID: 1284606. Parr CL, Hjartaker A, Lund E, and Veierod MB. 2013. Meat intake, cooking methods and risk of proximal colon, distal colon and rectal cancer: the Norwegian Women and Cancer (NOWAC) cohort study. Int J Cancer, 133(5), 1153–​1163. PMID: 23401013. Peltomaki P. 2001. Deficient DNA mismatch repair: a common etiologic factor for colon cancer. Hum Mol Genet, 10(7), 735–​740. PMID: 11257106. Peters U, Hutter CM, Hsu L, et al. 2012. Meta-​analysis of new genome-​wide association studies of colorectal cancer risk. Hum Genet, 131(2), 217–​ 234. PMCID: PMC3257356. Peters U, Jiao S, Schumacher FR, et  al. 2013. Identification of genetic susceptibility loci for colorectal tumors in a genome-​wide meta-​analysis. Gastroenterology, 144(4), 799–​807 PMCID: PMC3636812. Pham NM, Mizoue T, Tanaka K, et al. 2014. Meat consumption and colorectal cancer risk: an evaluation based on a systematic review of epidemiologic evidence among the Japanese population. Jpn J Clin Oncol, 44(7), 641–​ 650. PMID: 24842864. Pickhardt PJ, Hassan C, Halligan S, and Marmo R. 2011. Colorectal cancer: CT colonography and colonoscopy for detection: systematic review and meta-​analysis. Radiology, 259(2), 393–​405. PMCID: PMC3079122. Pierre FH, Martin OC, Santarelli RL, et al. 2013. Calcium and alpha-​tocopherol suppress cured-​meat promotion of chemically induced colon carcinogenesis in rats and reduce associated biomarkers in human volunteers. Am J Clin Nutr, 98(5), 1255–​1262. PMCID: PMC3798078.

 703



Cancers of the Colon and Rectum

Pino MS, and Chung DC. 2010. The chromosomal instability pathway in colon cancer. Gastroenterology, 138(6), 2059–​2072. PMCID: PMC4243705. Pischon T, Lahmann PH, Boeing H, et al. 2006. Body size and risk of colon and rectal cancer in the European Prospective Investigation Into Cancer and Nutrition (EPIC). J Natl Cancer Inst, 98(13), 920–​931. PMID: 16818856. Platz EA, Willett WC, Colditz GA, et al. 2000. Proportion of colon cancer risk that might be preventable in a cohort of middle-​aged US men. Cancer Causes Control, 11(7), 579–​588. PMID: 10977102. Pollak M. 2008. Insulin and insulin-​like growth factor signalling in neoplasia. Nat Rev Cancer, 8(12), 915–​928. PMID: 19029956. Potischman N, and Linet MS. 2013. Invited commentary: are dietary intakes and other exposures in childhood and adolescence important for adult cancers? Am J Epidemiol, 178(2), 184–​189. PMCID: PMC3816339. Potter JD, Bigler J, Fosdick L, et al. 1999. Colorectal adenomatous and hyperplastic polyps:  smoking and N-​ acetyltransferase 2 polymorphisms. Cancer Epidemiol Biomarkers Prev, 8(1), 69–​75. PMID: 9950242. Poynter JN, Haile RW, Siegmund KD, et al. 2009. Associations between smoking, alcohol consumption, and colorectal cancer, overall and by tumor microsatellite instability status. Cancer Epidemiol Biomarkers Prev, 18(10), 2745–​2750. PMCID: PMC2759847. Poynter JN, Siegmund KD, Weisenberger DJ, et  al. 2008. Molecular characterization of MSI-​ H colorectal cancer by MLHI promoter mythylation, immunohistochemistry, and mismatch repair germline mutation screening. Cancer Epidemiol Biomarkers Prev, 17(11), 3208–​ 3215. PMCID: PMC2628332. Probst-​Hensch NM, Yuan JM, Stanczyk FZ, et  al. 2001. IGF-​1, IGF-​2 and IGFBP-​3 in prediagnostic serum:  association with colorectal cancer in a cohort of Chinese men in Shanghai. Br J Cancer, 85(11), 1695–​1699. PMCID: PMC2363974. Rajakariar R, Yaqoob MM, and Gilroy DW. 2006. COX-​2 in inflammation and resolution. Mol Interv, 6(4), 199–​207. PMID: 16960142. Rao SR, Breen N, and Graubard BI. 2015. Trends in black-​white disparities in breast and colorectal cancer screening rates in a changing screening environment: the Peters-​Belson approach using United States National Health Interview Surveys 2000-​2010. Med Care, 54(2), 133–​139. PMID: 26565527. Renehan AG, Painter JE, Atkin WS, et  al. 2001. High-​risk colorectal adenomas and serum insulin-​like growth factors. Br J Surg, 88(1), 107–​113. PMID: 11136321. Rex DK, Ahnen DJ, Baron JA, et  al. 2012. Serrated lesions of the colorectum:  review and recommendations from an expert panel. Am J Gastroenterol, 107(9), 1315–​1329. PMCID: PMC3629844. Ridlon JM, Harris SC, Bhowmik S, Kang DJ, and Hylemon PB. 2016. Consequences of bile salt biotransformations by intestinal bacteria. Gut Microbes, 7(1), 22–​39. PMCID: PMC4856454. Ries LA, Wingo PA, Miller DS, et  al. 2000. The annual report to the nation on the status of cancer, 1973–​1997, with a special section on colorectal cancer. Cancer, 88(10), 2398–​2424. PMID: 10820364. Rim SH, Seeff L, Ahmed F, King JB, and Coughlin SS. 2009. Colorectal cancer incidence in the United States, 1999–​2004: an updated analysis of data from the National Program of Cancer Registries and the Surveillance, Epidemiology, and End Results Program. Cancer, 115(9), 1967–​1976. PMID: 19235249. Rinaldi S, Cleveland R, Norat T, et al. 2010. Serum levels of IGF-​I, IGFBP-​3 and colorectal cancer risk:  results from the EPIC cohort, plus a meta-​ analysis of prospective studies. Int J Cancer, 126(7), 1702–​ 1715. PMID: 19810099. Robbins AS, Siegel RL, and Jemal A. 2012. Racial disparities in stage-​specific colorectal cancer mortality rates from 1985 to 2008. J Clin Oncol, 30(4), 401–​405. PMID: 22184373. Rodriguez-​Bigas MA, Lin EH, and Crane CH. 2003. Stage IV colorectal cancer. In:  Kufe D, Pollock R, Weichselbaum R, et  al. (Eds.), Cancer Medicine (6th ed.). Rogler G. 2014. Chronic ulcerative colitis and colorectal cancer. Cancer Lett, 345(2), 235–​241. PMID: 23941831. Rogot E, and Murray JL. 1980. Smoking and causes of death among US Veterans:  16  years of observation. Public Health Rep, 95(3), 213–​222. PMCID: PMC1422715. Rokkas T, Pistiolas D, Sechopoulos P, Margantinis G, and Koukoulis G. 2008. Risk of colorectal neoplasm in patients with acromegaly: a meta-​analysis. World J Gastroenterol, 14(22), 3484–​3489. PMCID: PMC2716609. Romano M, Francesco F, Zarantonello L, et  al. 2016. From inflammation to cancer in inflammatory bowel disease: molecular perspectives. Anticancer Res, 36(4), 1447–​1460. PMID: 27069120. Roncucci L, and Mariani F. 2015. Prevention of colorectal cancer: How many tools do we have in our basket? Eur J Intern Med, 26(10), 752–​756. PMID: 26499755. Ross R, Dagnone D, Jones PJ, et  al. 2000. Reduction in obesity and related comorbid conditions after diet-​induced weight loss or exercise-​induced

703

weight loss in men:  a randomized, controlled trial. Ann Intern Med, 133(2), 92–​103. PMID: 10896648. Ross R, Janssen I, Dawson J, et al. 2004. Exercise-​induced reduction in obesity and insulin resistance in women: a randomized controlled trial. Obes Res, 12(5), 789–​798. PMID: 15166299. Rosty C, Hewett DG, Brown IS, Leggett BA, and Whitehall VL. 2013. Serrated polyps of the large intestine: current understanding of diagnosis, pathogenesis, and clinical management. J Gastroenterol, 48(3), 287–​ 302. PMCID: PMC3698429. Rothwell PM, Wilson M, Elwin CE, et al. 2010. Long-​term effect of aspirin on colorectal cancer incidence and mortality: 20-​year follow-​up of five randomised trials. Lancet, 376(9754), 1741–​1750. PMID: 20970847. Salaspuro M. 2009. Acetaldehyde as a common denominator and cumulative carcinogen in digestive tract cancers. Scand J Gastroenterol, 44(8), 912–​ 925. PMID: 19396661. Samowitz WS. 2008. Genetic and epigenetic change in colon cancer. Exp Mol Pathol, 85(1), 64–​67. PMID: 18482722. Samowitz WS, Albertsen H, Herrick J et  al. 2005. Evaluation of a large, population-​based sample supports a CpG island methylator phenotype in colon cancer. Gastroenterology, 129(3), 837–​845. PMID: 16143123. Samuels Y, Wang Z, Bardelli A et  al. 2004. High frequency of mutations of the PIK3CA gene in human cancers. Science, 304(5670), 554. PMID: 15016963. Sandhu MS, Dunger DB, and Giovannucci EL. 2002. Insulin, insulin-​like growth factor-​I (IGF-​I), IGF binding proteins, their biologic interactions, and colorectal cancer. J Natl Cancer Inst, 94(13), 972–​ 980. PMID: 12096082. Sandhu MS, White IR, and McPherson K. 2001. Systematic review of the prospective cohort studies on meat consumption and colorectal cancer risk: a meta-​analytic approach. Cancer Epidemiol Biomarkers Prev, 10(5), 439–​ 446. PMID: 11352852. Santarelli RL, Naud N, Tache S, et al. 2013. Calcium inhibits promotion by hot dog of 1,2-​dimethylhydrazine-​induced mucin-​depleted foci in rat colon. Int J Cancer, 133(11), 2533–​2541. PMCID: PMC3788046. Santarelli RL, Pierre F, and Corpet DE. 2008. Processed meat and colorectal cancer:  a review of epidemiologic and experimental evidence. Nutr Cancer, 60(2), 131–​144. PMCID: PMC2661797. Schatzkin A, Mouw T, Park Y, et al. 2007. Dietary fiber and whole-​grain consumption in relation to colorectal cancer in the NIH-​AARP Diet and Health Study. Am J Clin Nutr, 85(5), 1353–​1360. PMID: 17490973. Schmid D, and Leitzmann MF. 2014. Television viewing and time spent sedentary in relation to cancer risk: a meta-​analysis. J Natl Cancer Inst, 106(7). PMID: 24935969. Schmit SL, Schumacher FR, Edlund CK et  al. 2014. A  novel colorectal cancer risk locus at 4q32.2 identified from an international genome-​wide association study. Carcinogenesis, 35(11), 2512–​2519. PMCID: PMC4271131. Schoen RE, Pinsky PF, Weissfeld JL, et al. 2012. Colorectal-​cancer incidence and mortality with screening flexible sigmoidoscopy. N Engl J Med, 366(25), 2345–​2357. PMCID: PMC3641846. Schoen RE, Tangen CM, Kuller LH, et al. 1999. Increased blood glucose and insulin, body size, and incident colorectal cancer. J Natl Cancer Inst, 91(13), 1147–​1154. PMID: 10393723. Schreuders EH, Ruco A, Rabeneck L, et  al. 2015. Colorectal cancer screening: a global overview of existing programmes. Gut, 64(10), 1637–​1649. PMID: 26041752. Schumacher FR, Schmit SL, Jiao S, et al. 2015. Corrigendum: Genome-​wide association study of colorectal cancer identifies six new susceptibility loci. Nat Commun, 6, 8739. PMID: 26498495. Sebastian C, and Mostoslavsky R. 2014. Untangling the fiber yarn:  butyrate feeds warburg to suppress colorectal cancer. Cancer Discov, 4(12), 1368–​ 1370. PMCID: PMC4258833. Sehdev A, and O’Neil BH. 2015. The role of aspirin, vitamin D, exercise, diet, statins, and metformin in the prevention and treatment of colorectal cancer. Curr Treat Options Oncol, 16(9), 43. PMID: 26187794. Seitz HK, and Becker P. 2007. Alcohol metabolism and cancer risk. Alcohol Res Health, 30(1), 38–​41, 44–​37. PMCID: PMC3860434. Shaukat A, Scouras N, and Schunemann HJ. 2005. Role of supplemental calcium in the recurrence of colorectal adenomas:  a metaanalysis of randomized controlled trials. Am J Gastroenterol, 100(2), 390–​ 394. PMID: 15667497. Shen D, Mao W, Liu T, et  al. 2014. Sedentary behavior and incident cancer:  a meta-​analysis of prospective studies. Plos One, 9(8), e105709. PMCID: PMC4143275. Shi Y, Yu PW, and Zeng DZ. 2015. Dose-​response meta-​analysis of poultry intake and colorectal cancer incidence and mortality. Eur J Nutr, 54(2), 243–​250. PMID: 24788671.

704

704

PART IV:  Cancers by Tissue of Origin

Siegel R, Desantis C, and Jemal A. 2014. Colorectal cancer statistics, 2014. CA Cancer J Clin, 64(2), 104–​117. PMID: 24639052. Siegel RL, Jemal A, and Ward EM. 2009. Increase in incidence of colorectal cancer among young men and women in the United States. Cancer Epidemiol Biomarkers Prev, 18(6), 1695–​1698. PMID: 19505901. Siegel RL, Miller KD, and Jemal A. 2015. Cancer statistics, 2015. CA Cancer J Clin, 65(1), 5–​29. PMID: 25559415. Siegel RL, Miller KD, and Jemal A. 2016. Cancer statistics, 2016. CA Cancer J Clin, 66(1), 7–​30. PMID: 26742998. Simons CC, Hughes LA, Smits KM, et  al. 2013. A  novel classification of colorectal tumors based on microsatellite instability, the CpG island methylator phenotype and chromosomal instability:  implications for prognosis. Ann Oncol, 24(8), 2048–​2056. PMID: 23532114. Sinha R, and Norat T. 2002. Meat cooking and cancer risk. IARC Sci Publ, 156, 181–​186. PMID: 12484160. Sinha R, Peters U, Cross AJ, et  al. 2005. Meat, meat cooking methods and preservation, and risk for colorectal adenoma. Cancer Res, 65(17), 8034–​ 8041. PMID: 16140978. Sinicrope FA, Shi Q, Smyrk TC, et  al. 2015. Molecular markers identify subtypes of stage III colon cancer associated with patient outcomes. Gastroenterology, 148(1), 88–​99. PMCID: PMC4274188. Slattery ML. 2000. Diet, lifestyle, and colon cancer. Semin Gastrointest Dis, 11(3), 142–​146. PMID: 10950460. Slattery ML. 2004. Physical activity and colorectal cancer. Sports Med, 34(4), 239–​252. PMID: 15049716. Slattery ML, Anderson K, Curtin K, et  al. 2001. Lifestyle factors and Ki-​ ras mutations in colon cancer tumors. Mutat Res, 483(1-​2), 73–​81. PMID: 11600135. Slattery ML, Benson J, Berry TD, et al. 1997. Dietary sugar and colon cancer. Cancer Epidemiol Biomarkers Prev, 6(9), 677–​685. PMID: 9298574. Slattery ML, Curtin K, Ma K, et al. 2002. Diet activity, and lifestyle associations with p53 mutations in colon tumors. Cancer Epidemiol Biomarkers Prev, 11(6), 541–​548. PMID: 12050095. Snover DC. 2011. Update on the serrated pathway to colorectal carcinoma. Hum Pathol, 42(1), 1–​10. PMID: 20869746. Sostres C, Gargallo CJ, and Lanas A. 2014. Aspirin, cyclooxygenase inhibition and colorectal cancer. World J Gastrointest Pharmacol Ther, 5(1), 40–​49. PMCID: PMC3944468. Soumaoro LT, Uetake H, Higuchi T, et  al. 2004. Cyclooxygenase-​2 expression: a significant prognostic indicator for patients with colorectal cancer. Clin Cancer Res, 10(24), 8465–​8471. PMID: 15623626. Spirio L, Olschwang S, Groden J, et al. 1993. Alleles of the APC gene: an attenuated form of familial polyposis. Cell, 75(5), 951–​957. PMID: 8252630. Surveillance Epidemiology and End Results. 2015. Surveillance, Epidemiology, and End Results (SEER) Stat Fact Sheets:  Colon and Rectum Cancer. Available from:  http://​seer.cancer.gov/​statfacts/​html/​colorect.html Accessed January 5, 2016. Svensson E, Moller B, Tretli S, et al. 2005. Early life events and later risk of colorectal cancer:  age-​period-​cohort modelling in the Nordic countries and Estonia. Cancer Causes Control, 16(3), 215–​223. PMID: 15947873. Tadros M, and Anderson JC. 2013. Serrated polyps: clinical implications and future directions. Curr Gastroenterol Rep, 15(9), 342. PMID: 23934652. Taggarshe D, Rehil N, Sharma S, Flynn JC, and Damadi A. 2013. Colorectal cancer: are the “young” being overlooked? Am J Surg, 205(3), 312–​316. PMID: 23414955. Takahashi M, and Wakabayashi K. 2004. Gene mutations and altered gene expression in azoxymethane-​induced colon carcinogenesis in rodents. Cancer Sci, 95(6), 475–​480. PMID: 15182426. Thiel A, and Ristimaki A. 2013. Toward a molecular classification of colorectal cancer: the role of BRAF. Front Oncol, 3, 281. PMCID: PMC3828559. Thissen JP, Ketelslegers JM, and Underwood LE. 1994. Nutritional regulation of the insulin-​like growth factors. Endocr Rev, 15(1), 80–​101. PMID: 8156941. Thomas D. 2010. Gene–​ environment-​ wide association studies:  emerging approaches. Nat Rev Genet, 11(4), 259–​272. PMCID: PMC2891422. Thun MJ, Calle EE, Namboodiri MM, et al. 1992. Risk factors for fatal colon cancer in a large prospective study. J Natl Cancer Inst, 84(19), 1491–​ 1500. PMID: 1433333. Thun MJ, Jacobs EJ, and Patrono C. 2012. The role of aspirin in cancer prevention. Nat Rev Clin Oncol, 9(5), 259–​267. PMID: 22473097. Tomasetti C, and Vogelstein B. 2015. Cancer etiology: variation in cancer risk among tissues can be explained by the number of stem cell divisions. Science, 347(6217), 78–​81. PMCID: PMC4446723. Torre LA, Siegel RL, Ward EM, and Jemal A. 2016. Global cancer incidence and mortality rates and trends: an update. Cancer Epidemiol Biomarkers Prev, 25(1), 16–​27. PMID: 26667886.

Toyota M, Ahuja N, Ohe-​Toyota M, et al. 1999. CpG island methylator phenotype in colorectal cancer. Proc Natl Acad Sci U S A, 96(15), 8681–​8686. PMCID: PMC17576. Toyota M, Ohe-​Toyota M, Ahuja N, and Issa JP. 2000. Distinct genetic profiles in colorectal tumors with or without the CpG island methylator phenotype. Proc Natl Acad Sci U S A, 97(2), 710–​715. PMCID: PMC15395. Triantafillidis JK, Nasioulas G, and Kosmidis PA. 2009. Colorectal cancer and inflammatory bowel disease: epidemiology, risk factors, mechanisms of carcinogenesis and prevention strategies. Anticancer Res, 29(7), 2727–​ 2737. PMID: 19596953. Tsai S, and Gearhart S. 2011. Presentation and initial evaluation of colorectal cancer. In:  Gearhart S, and Ahuja N (Eds.), Early Diagnosis and Treatment of Cancer (pp. 13–​19). Philadelphia: Saunders Elsevier. Tsoi KK, Pau CY, Wu WK, et  al. 2009. Cigarette smoking and the risk of colorectal cancer:  a meta-​analysis of prospective cohort studies. Clin Gastroenterol Hepatol, 7(6), 682–​688. PMID: 19245853. Tuomilehto J, Knowler WC, and Zimmet P. 1992. Primary prevention of non-​ insulin-​dependent diabetes mellitus. Diabetes Metab Rev, 8(4), 339–​353. PMID: 1307523. Turcotte LP, and Fisher JS. 2008. Skeletal muscle insulin resistance:  roles of fatty acid metabolism and exercise. Phys Ther, 88(11), 1279–​1296. PMCID: PMC2579902. Turesky RJ. 2004. The role of genetic polymorphisms in metabolism of carcinogenic heterocyclic aromatic amines. Curr Drug Metab, 5(2), 169–​180. PMID: 15078194. Turner ND, Ritchie LE, Bresalier RS, and Chapkin RS. 2013. The microbiome and colorectal neoplasia: environmental modifiers of dysbiosis. Curr Gastroenterol Rep, 15(9), 346. PMCID: PMC3807587. Ulrich CM. 2007. Folate and cancer prevention:  a closer look at a complex picture. Am J Clin Nutr, 86(2), 271–​273. PMID: 17684194. Van Blarigan EL, and Meyerhardt JA. 2015. Role of physical activity and diet after colorectal cancer diagnosis. J Clin Oncol, 33(16), 1825–​1834. PMCID: PMC4438267. Vasen HF, Watson P, Mecklin JP, and Lynch HT. 1999. New clinical criteria for hereditary nonpolyposis colorectal cancer (HNPCC, Lynch syndrome) proposed by the International Collaborative group on HNPCC. Gastroenterology, 116(6), 1453–​1456. PMID: 10348829. Vogelstein B, Fearon ER, Hamilton SR et  al. 1988. Genetic alterations during colorectal-​ tumor development. N Engl J Med, 319(9), 525–​ 532. PMID: 2841597. Vollset SE, Clark R, Lewington S, et al. 2013. Effects of folic acid supplementation on overall and site-​specific cancer incidence during the randomised trials:  meta-​analyses of data on 50,000 individuals. Lancet, 381(9871), 1029–​1036. PMCID: PMC3836669. Wakai K, Date C, Fukui M, et al. 2007. Dietary fiber and risk of colorectal cancer in the Japan collaborative cohort study. Cancer Epidemiol Biomarkers Prev, 16(4), 668–​675. PMID: 17416756. Wallace K, Baron JA, Cole BF, et al. 2004. Effect of calcium supplementation on the risk of large bowel polyps. J Natl Cancer Inst, 96(12), 921–​925. PMID: 15199111. Wargovich MJ, Eng VW, and Newmark H. 1984. Calcium inhibits the damaging and compensatory proliferative effects of fatty acids on mouse colon epithelium. Cancer Lett, 23(3), 253–​258. PMID: 6744249. Warren JL, Klabunde CN, Mariotto AB, et al. 2009. Adverse events after outpatient colonoscopy in the Medicare population. Ann Intern Med, 150(12), 849–​857, W152. PMID: 19528563. Wei EK, Giovannucci E, Wu K, et al. 2004. Comparison of risk factors for colon and rectal cancer. Int J Cancer, 108(3), 433–​442. PMCID: PMC2903217. Wei EK, Ma J, Pollak MN et al. 2005. A prospective study of C-​peptide, insulin-​ like growth factor-​I, insulin-​like growth factor binding protein-​1, and the risk of colorectal cancer in women. Cancer Epidemiol Biomarkers Prev, 14(4), 850–​855. PMID: 15824155. Weir JM, and Dunn JE, Jr. 1970. Smoking and mortality: a prospective study. Cancer, 25(1), 105–​112. PMID: 5410301. Weisenberger DJ, Levine AJ, Long TI, et al. 2015. Association of the colorectal CpG island methylator phenotype with molecular features, risk factors, and family history. Cancer Epidemiol Biomarkers Prev, 24(3), 512–​519. PMCID: PMC4355081. Weisenberger DJ, Siegmund KD, Campan M, et al. 2006. CpG island methylator phenotype underlies sporadic microsatellite instability and is tightly associated with BRAF mutation in colorectal cancer. Nat Genet, 38(7), 787–​793. PMID: 16804544. Wheeler JM, Bodmer WF, and Mortensen NJ. 2000a. DNA mismatch repair genes and colorectal cancer. Gut, 47(1), 148–​153. PMCID: PMC1727951. Wheeler JM, Loukola A, Aaltonen LA, Mortensen NJ, and Bodmer WF. 2000b. The role of hypermethylation of the hMLH1 promoter region in HNPCC

 705



Cancers of the Colon and Rectum

versus MSI+ sporadic colorectal cancers. J Med Genet, 37(8), 588–​592. PMCID: PMC1734667. Whiffin N, Hosking FJ, Farrington SM, et al. 2014. Identification of susceptibility loci for colorectal cancer in a genome-​wide meta-​analysis. Hum Mol Genet, 23(17), 4729–​4737. PMCID: PMC4133584. Winawer SJ. 1999. Natural history of colorectal cancer. Am J Med, 106(1A), 3S–​6S; discussion 50S–​51S. PMID: 10089106. Winawer SJ, Stewart ET, Zauber AG, et al. 2000. A comparison of colonoscopy and double-​contrast barium enema for surveillance after polypectomy. National Polyp Study Work Group. N Engl J Med, 342(24), 1766–​1772. PMID: 10852998. Winawer SJ, Zauber AG, O’Brien MJ, et  al. 1993. Randomized comparison of surveillance intervals after colonoscopic removal of newly diagnosed adenomatous polyps. The National Polyp Study Workgroup. N Engl J Med, 328(13), 901–​906. PMID: 8446136. Wolin KY, Lee IM, Colditz GA, et al. 2007. Leisure-​time physical activity patterns and risk of colon cancer in women. Int J Cancer, 121(12), 2776–​ 2781. PMCID: PMC2291204. Wolin KY, Yan Y, and Colditz GA. 2011. Physical activity and risk of colon adenoma:  a meta-​ analysis. Br J Cancer, 104(5), 882–​ 885. PMCID: PMC3048199. Wong NA, and Harrison DJ. 2001. Colorectal neoplasia in ulcerative colitis-​ recent advances. Histopathology, 39(3), 221–​234. PMID: 11532032. World Cancer Research Fund, and American Institute for Cancer Research. 1997. Food, Nutrition and the Prevention of Cancer: A Global Perspective. Washington, DC: AICR. World Cancer Research Fund, and American Institute for Cancer Research. 2007. Food, Nutrition, Physical Activity, and the Prevention of Cancer: A Global Perspective. Washington, DC: AICR World Cancer Research Fund, and American Institute for Cancer Research. 2011. Continuous Update Project Report: Food, Nutrition, Physical Activity, and the Prevention of Colorectal Cancer. Washington, DC: AICR. Wu K, Feskanich D, Fuchs CS, et  al. 2011. Interactions between plasma levels of 25-​hydroxyvitamin D, insulin-​like growth factor (IGF)-​1 and C-​ peptide with risk of colorectal cancer. PLoS One, 6(12), e28520. PMCID: PMC3247212.

705

Yamada Y, and Mori H. 2007. Multistep carcinogenesis of the colon in ApcMin/​ + mouse. Cancer Sci, 98(1), 6–​10. PMID: 17052257. Yamauchi M, Lochhead P, Morikawa T, et  al. 2012. Colorectal cancer:  a tale of two sides or a continuum? Gut, 61(6), 794–​ 797. PMCID: PMC3345045. Yao X, and Tian Z. 2015. Dyslipidemia and colorectal cancer risk:  a meta-​ analysis of prospective studies. Cancer Causes Control, 26(2), 257–​268. PMID: 25488827. Yoon YS, Keum N, Zhang X, Cho E, and Giovannucci EL. 2015a. Circulating levels of IGF-​1, IGFBP-​3, and IGF-​1/​IGFBP-​3 molar ratio and colorectal adenomas:  a meta-​analysis. Cancer Epidemiol, 39(6), 1026–​1035. PMID: 26388613. Yoon YS, Keum N, Zhang X, Cho E, and Giovannucci EL. 2015b. Hyperinsulinemia, insulin resistance and colorectal adenomas:  a meta-​ analysis. Metabolism, 64(10), 1324–​1333. PMID: 26169471. Zauber AG, Winawer SJ, O’Brien MJ, et al. 2012. Colonoscopic polypectomy and long-​term prevention of colorectal-​cancer deaths. N Engl J Med, 366(8), 687–​696. PMCID: PMC3322371. Zhang B, Jia WH, Matsuda K et  al. 2014. Large-​scale genetic study in East Asians identifies six new loci associated with colorectal cancer risk. Nat Genet, 46(6), 533–​542. PMCID: PMC4068797. Zhang X, Wu K, Cho E, et  al. 2013. Prospective cohort studies of bowel movement frequency and laxative use and colorectal cancer incidence in US women and men. Cancer Causes Control, 24(5), 1015–​1024. PMCID: PMC3633608. Zhao C, Ge Z, Wang Y, and Qian J. 2012. Meta-​analysis of observational studies on cholecystectomy and the risk of colorectal adenoma. Eur J Gastroenterol Hepatol, 24(4), 375–​381. PMID: 22410713. Ziegler RG, Devesa SS, and Fraumeni JF, Jr. 1986. Epidemiologic patterns of colorectal cancer. In:  DeVita VT, Hellman S, and Rosenberg SA (Eds.), Important Advances in Oncology (pp. 209–​232). Philadelphia: J. B. Lippincott. PMID: 3330534. Zorzi M, Fedeli U, Schievano E, et al. 2015. Impact on colorectal cancer mortality of screening programmes based on the faecal immunochemical test. Gut, 64(5), 784–​790. PMID: 25179811.

706

 70

37 Anal Cancer ANDREW E. GRULICH, FENGYI JIN, AND I. MARY POYNTEN

OVERVIEW Anal canal cancer is a generally uncommon cancer that has been increasing in incidence for several decades. In most geographic locations, squamous cell carcinoma (SCC) accounts for 70% or more of cases, and incidence is slightly higher in women than in men. The remaining cases are mainly adenocarcinoma, but the degree to which this represents misclassified rectal cancer is uncertain. In almost all cases, anal SCC is caused by persistent infection with high-​risk types of human papillomavirus (HPV), and HPV 16 accounts for 75% or more of all cases. Survival is highly stage-​dependent, and cure is usual if the cancer is diagnosed early. The main risk factor is anal exposure to HPV, and for this reason homosexual men are at particularly high risk. In women, risk is increased in those with higher numbers of sexual partners, and in those with a history of HPV-​related disease at genital sites. Anal HPV infection does not require a history of receptive anal intercourse. Tobacco smoking also increases risk of anal SCC. As a virus-​associated cancer, incidence is increased in people with immune deficiency, including in organ transplant recipients and in people with HIV. The highest incidence group is homosexual men with HIV, in whom anal cancer is one of the most common cancers, with an annual incidence exceeding 100/​100,000 in some studies. Condoms offer limited protection against HPV transmission, because the virus is transmitted through skin-​to-​skin contact. The HPV vaccine, administered before initiation of sexual activity, will eventually lead to reductions in incidence in immunized populations. A cytology-​based screening program, based on the cervical screening model, has been advocated by some in the field. However, population-​based implementation has been hindered by a lack of expertise and a lack of effective therapies for the anal cancer precursor, high-​grade squamous intra-​epithelial lesions.

CLASSIFICATION Anatomical Classification Anal cancer is a malignancy that arises from, or is predominantly located in, the anal canal. Proximally, the uppermost boundary of the anal canal is defined by the proximal portion of the anorectal ring (Welton et al., 2010). Distally, the anal canal ends at the anal verge, the muco-​cutaneous junction where it meets the perianal skin (American Joint Committee on Cancer, 2002). Cancers arising from the anal canal can be difficult to distinguish from cancers of the perianal skin arising from the anal margin (Mendenhall et  al., 1996). The fourth edition of the WHO Classification of Tumours of the Digestive System, published in 2010, proposed a new simplified classification to distinguish cancers arising from the anal canal from those of the perianal skin. In this classification, an anal canal tumor is defined as one that “cannot be seen in its entirety, or at all, when gentle traction is placed on the buttocks. A perianal cancer is found within a 5 cm radius of the anus and is seen completely when gentle traction is placed on the anus” (Welton et al., 2010). Topographically, invasive cancers arising from the anus and anal canal are classified as C21 under the International Classification of Disease for Oncology (ICD-​O, 3rd edition, 1st revision) (World Health Organization, 2013). Depending on the origin of the cancer, the topographical codes are further specified as cancer of the anus (C21.0),

the anal canal (C21.1), the cloacogenic zone (C21.2), and cancers that have originated from the overlapping zone of anus, anal canal, and rectum (C21.8). In some previously published analyses, anal cancer has included squamous cell carcinomas of the rectum (C20.9). The justification given for this is that the rectum is composed of glandular but not squamous cells, and that rectal squamous cell carcinomas are probably anal cancers miscoded as rectum, and can therefore be regarded as cases of anal cancer (Frisch, 2002; Joseph et al., 2008).

Histological Classification The anal canal consists of mucosa of three different histological types from proximal (internal) to distal (external) canal:  glandular, transitional, and squamous. The upper zone (colorectal zone) is lined with glandular mucosa indistinguishable from rectal mucosa. The middle zone (anal transitional zone) extends from the dentate line upward for up to 12  mm, where it gradually merges with the epithelium of the colorectal zone (Welton et al., 2010). The lower zone (squamous zone) extends from dentate line downward for about 1–​2 cm to the mucocutaneous junction, where it merges with perianal skin. It is covered with non-​keratinizing squamous epithelium devoid of skin appendages (hair follicles, apocrine glands, and sweat glands) (American Joint Committee on Cancer, 2002). Morphologically, anal cancers are classified as squamous cell carcinoma (8070/​3), verrucous carcinoma (8051/​3), adenocarcinoma (8140/​ 3), mucinous adenocarcinoma (8480/​3), and other carcinomas (8020/3) (World Health Organization, 2013). The terms “transitional cell” and “cloacogenic carcinoma” have been superseded, as they are now recognized as non-​keratinizing squamous cell carcinoma (American Joint Committee on Cancer, 2002). Squamous cell carcinoma includes keratinizing (8071) and non-​keratinizing (8072, 8073) subtypes, and often anal cancers show more than one subtype (Welton et al., 2010). Verrucous carcinoma is a form of well-​differentiated squamous cell carcinoma characterized by a papillary growth pattern. Some verrucous carcinomas contain HPV types 6 and 11. They are regarded as an intermediate state between condyloma and squamous cell carcinoma, and the clinical course is typically that of local destructive invasion without metastases (Welton et al., 2010). Globally, squamous cell carcinoma is the most common type of anal cancer, but this varies by region. Squamous cell carcinoma accounts for over 80% of anal cancers diagnosed in the United States, the United Kingdom, and the Netherlands, around 75% in Australia, and close to 70% in Canada (International Agency for Research on Cancer, 2013). In some regions, particularly in East and Southeast Asia, adenocarcinoma is the predominant histological type of anal cancer (59% in Hong Kong, 70% in Japan, and 73% in the Philippines) (International Agency for Research on Cancer, 2013). Whether the difference in the proportion of anal cancers that are adenocarcinomas represents a true difference or is in fact a misclassification of rectal adenocarcinoma is uncertain. Anal adenocarcinomas may have actually arisen from the rectum, and the association of these tumors with high-​risk types of HPV infection is not clear (Daling et al., 2004). Given that there are different types of tumors that can arise from the anal canal, it is important to clearly specify the anatomical and histological classification of anal cancer in research studies.

707

708

708

Part IV:  Cancers by Tissue of Origin

Variations in definition may introduce systematic bias when comparing incidence rates internationally or temporally. Combining data on adenocarcinoma and squamous cell carcinoma may lead to the masking of risk factors that are particular to only one of the two types.

TUMOR PROGRESSION MODEL Natural History of Anal HPV Infection and Anal Neoplasia Including Molecular Pathway Much of the current understanding of anal HPV infection and its natural history and progression to invasive cancer has been extrapolated from cervical cancer research. HPV is recognized as a necessary cause for cervical cancer, and this may also be the case in the anal canal (Bouvard et  al., 2009). HPV is transmitted by intimate skin-​to-​skin contact between sexual partners (Kjaer et  al., 2001; Rylander et  al., 1994). Most anogenital HPV infections are short-​lived and become undetectable within 1–​2  years (Ho et  al., 1998), but in some cases HPV infection may persist. In the case of oncogenic or high-​risk types of HPV (HR-​HPV) infection, persistent infection in the cervix can lead to the development of high-​grade precancerous cervical intraepithelial neoplasia (CIN) (Rodriguez et al., 2010). CIN can be divided into low-​ grade (CIN1) and high-​grade forms (CIN2 and CIN3) by microscopic appearance and their relation to risk of cancer development. CIN1 is less likely to be associated with HR-​HPV infection (Moscicki et al., 2008), and more likely to be the result of an acute infection with low-​ risk HPV types (LR-​HPV) that have an extremely low risk of progressing to cancer. In 2012, the College of American Pathologists and the American Society for Colposcopy and Cervical Pathology recommended a unified histopathologic nomenclature with a single set of diagnostic terms for all HPV-​associated pre-​invasive squamous lesions at all sites in the lower anogenital tract. It recommended a two-​tiered nomenclature of low-​grade and high-​grade squamous intra-​epithelial lesions (SIL), but that high-​grade lesions may be further subcategorized as IN2 or IN3, with specification of the anatomic site of origin (eg., CIN or AIN). The project concluded that IN2 at all anogenital sites represented a mixture of low-​grade and high-​grade SIL that cannot be reliably distinguished on hematoxylin and eosin staining (Darragh et al., 2012). In the cervix, the average time between HPV infection and CIN3 is thought to be 7–​15  years. Peak rates of HPV infection occur in the late teens or early twenties, and peak rates of CIN3 occur around 25–​30 years of age. However, prospective cohorts show that incident HPV infection can lead to CIN2/​3 within a few months (Moscicki et al., 2006). Women with screen-​detected invasive cancer tend to be more than 10 years older than women with CIN3, suggesting a long average time in the precancer state (Schiffman and Kjaer, 2003). In contrast to the cervix, there are few published longitudinal studies on the natural history of anal HPV infection. A number of studies have been performed in homosexual men, who are at high risk of anal cancer. Some striking differences in the epidemiology and natural history of anal HPV infection, compared with cervical HPV infection, have been described. First, the prevalence of anal HPV infection in homosexual men is much higher than that of cervical HPV in virtually any population of women. In a meta-​analysis of community-​based studies, anal HPV was detected in more than 90% of HIV-​positive and about two-​thirds of HIV-​negative homosexual men, and HR-​HPV was detected in about three-​quarters and over one-​third in the HIV-​positive and negative, respectively (Machalek et  al., 2012). Second, the age-​related pattern of anal HPV infection in homosexual men is clearly different from patterns observed in the cervix. In women the prevalence of cervical HPV infection peaks in the late teens and early twenties, soon after the onset of sexual activity (Burk et  al., 1996), whereas the prevalence of anal HPV in homosexual men remains high across all ages (Chin-​Hong et al., 2004). Many homosexual men continue to have multiple sexual partners past age 50, whereas in women, only among teenagers do more than a small minority report multiple recent sexual partners

(Poynten et al., 2016). These data suggest that the persistently high anal HPV prevalence across age groups may be related to patterns of sexual behavior. Third, homosexual men have a much higher prevalence of squamous intra-​epithelial lesions in the anal canal compared to the prevalence of squamous intra-​epithelial lesions in the cervix. Approximately a third of HIV-​positive and a quarter of HIV-​negative homosexual men have histologically confirmed high-​grade squamous intraepithelial lesions (HSIL) (Machalek et al., 2012). A cross-​ sectional comparison of anal HSIL prevalence with cancer incidence in homosexual men suggests that progression from HSIL may be less common in the anus than it is in the cervix (Machalek et al., 2012; McCredie et al., 2008). Persistent HPV infection and subsequent viral transformation result from the expression of certain HPV proteins. The HPV early proteins (E6 and E7) have oncogenic potential and are responsible for viral replication and transformation of the host cell (Huh, 2009). In the cervix, HPV can be found either in episomal or integrated forms or in mixed forms that contain both. Viral DNA integration of high-​risk HPV genotypes in the human genome is one of the key events associated with neoplastic progression of premalignant cervical lesions (zur Hausen, 1996). HPV E6/​E7 gene overexpression is consistently detected in cancers and is associated with dysregulation of the viral transcriptional repressor, E2, which is often lost during integration. The overexpression of E6 and E7 viral proteins bind and affect the normal functioning of tumor suppressors p53 and pRb, respectively. E6 binding to p53 results in degradation of p53, leading to dysregulation of repair of DNA damage and apoptotic pathways. E7 binding to pRb leads to dysregulation of cell cycle control and increase in cell proliferation and genomic instability (Dyson et al., 1989; Scheffner et  al., 1990). pRB-​E7 complex results in bypass of the negative feedback repression of the cyclin-​dependent kinase inhibitor, p16INK4a, and hence overexpression of p16 in conjunction with the cellular proliferation marker, Ki-​67, which can serve as a specific marker of high-​grade disease (McLaughlin-​Drubin and Munger, 2009).

SURVIVAL Survival from anal cancer is directly related to disease stage at diagnosis. Based on data from the US Surveillance, Epidemiology, and End Results (SEER) Program from 2005 to 2011, the overall 5-​year survival after a diagnosis of anal cancer was 65.7%. In those with localized disease (confined to primary site), 5-​year survival was 80.1%, compared with 58.6% among those patients whose cancer had spread to regional lymph nodes, and 30.7% for those whose disease had metastasized (SEER Cancer Statistics Factsheets, 2015). A small French retrospective series of 69 people treated between 1990 and 2000, with tumors of less than or equal to 1 cm at diagnosis, reported 5-​year overall survival of 94% and 5-​year disease-​free survival of 89% (Ortholan et al., 2005). Among 886 patients diagnosed with anal SSC between 2000 and 2007 in Sweden, Norway, and Denmark and treated according to Nordic guidelines, the 3-​year recurrence-​free survival ranged from 63% to 76%. Having a large primary tumor (greater than 5 cm) was independently associated with worse outcome. Other factors found to be associated with poorer prognosis included older age and male gender (Leon et al., 2014). A large US anal cancer trial investigated the impact of tumor node (TN) category of disease on overall and disease-​free survival in 620 patients. Patients with the poorest overall and disease-​free survival outcomes were those with the highest TN category of disease (Gunderson et al., 2013). In recent times, HIV infection does not appear to have an impact on anal cancer survival. Survival trends for 257 HIV-​positive patients with anal cancer were analyzed in the French Hospital Database on HIV. There was substantial improvement in survival between the pre-​ antiretroviral therapy (ART) era (1992–​ 1996) compared with 1997–​2000, but thereafter the 5-​year survival rate remained stable at 60%–​65% in the 2000–​2004 and 2005–​2009 periods. This was similar to survival after anal cancer diagnosis in the general population (Hleyhel et al., 2015).

 709

Anal Cancer

DESCRIPTIVE EPIDEMIOLOGY Demographic Variation in Incidence Anal cancer is an uncommon cancer in the general population globally. The annual age-​adjusted incidence rate of all anal cancer types combined is less than 2 per 100,000 in almost all countries that have published rates (Hartwig et al., 2012; SEER Cancer Statistics Factsheets, 2015; Welton et al., 2010). Anal cancer most commonly occurs in the sixth or seventh decade of life (Forman et  al., 2012). Incidence increases exponentially with age, with annual anal cancer rates greater than 5 per 100,000 reported in people aged over 65  years. In Australia, 48% of cases are diagnosed in people aged over 65  years (Jin et  al., 2011). The International Agency for Research on Cancer (IARC) reported that there were 27,000 new cases of anal cancer globally in 2008, with an estimated 88% attributable to HPV. There were slightly more cases among women (54%) than men (46%). The largest number occurred in the 50-​to 69-​year age group (43%), followed by those older than 70  years (36%), with only 22% of anal cancer cases occurring in those younger than 50 years. There was no difference in the number of new cases attributable to HPV infection between more-​developed and less-​developed regions of the world (de Martel et  al., 2012). In the United Kingdom, incident anal cancer rates increase rapidly with age, with very few cases of anal cancer diagnosed before 40 years of age. In England in 2012, two-​thirds of the 1043 registered anal cancer cases occurred in women, with the vast majority (88%) of these occurring in women aged over 50 years (Coffey et al., 2015). There is evidence of racial differences in anal cancer from the US SEER program data. In 2000, African-​American men had higher incidence rates than did other race-​specific and gender-​ specific groups (2.71 per 100,000) (Johnson et al., 2004). In 2010, African-​American men were statistically significantly younger by 5.5 years than non-​African-​American men at diagnosis of anal cancer (Robbins et al., 2015). Analyses comparing socioeconomic status and anal cancer incidence in the United States found that higher education, mid-​level income, and metropolitan residence were associated with increased age-​adjusted anal cancer incidence in men. Higher education and income were associated with increased age-​ adjusted anal incidence rates in women (Benard et al., 2008).

Temporal Trends Though an uncommon cancer, there is evidence that the incidence of anal cancer is increasing in both men and women. Recently, temporal trends in anal cancer were explored utilizing IARC’s Cancer Incidence in Five Continents. Anal cancer incidence increased since 1980 in the United States, Denmark, Japan, and Colombia, but declined in India (Mumbai) and was approximately constant in the Philippines (Forman et al., 2012). In the United States, anal cancer incidence rates have increased on average 2.2% each year over the last 10  years (SEER Cancer Statistics Factsheets, 2015). In most settings, anal cancer incidence is higher in women than in men, but in several locations, this difference in incidence between genders has decreased over recent years. In US SEER registries between 1973 and 1979, men had lower rates of anal cancer than women (1.06 per 100,000 compared with 1.39 per 100,000). From 1994 to 2000, the incidence had increased in both genders and had become approximately the same in men and women (2.04 per 100,000 and 2.06 per 100,000, respectively) (Johnson et  al., 2004). Similar increases in incidence have been recorded in cancer registries in Denmark, where the increase was first noticed in the 1960s and anal cancer incidence was much higher in women than men, and in residents of Copenhagen compared with other areas (Frisch et al., 1993; Nielsen et al., 2012). In Scotland during 1975 to 2002, incidence rates of anal cancer increased by more than 2-​fold in each sex (Brewster and Bhatti, 2006). In Australia, incidence rates of anal SCC increased by 3.4% per year in males and 1.9% in females in the period 1982 to 2005, and smaller increases of borderline significance were seen for anal adenocarcinoma (Jin

709

et al., 2011). In the United Kingdom, the incidence of all anal cancer combined increased in both men and women in the period 1960 to 2004. The increase was larger in women (Robinson et al., 2009). This increasing incidence of anal cancer in the general population may be related to changing sexual practices over time, such as earlier age of first sexual intercourse and higher number of lifetime sex partners (Grulich et al., 2012). In the United States, the proportion of all cases of anal cancer that occurred in people with HIV increased from 1.1% in males and 0% in females during 1980–​1984 to 28% and 1.2%, respectively, in the period 2001–​2005. Nevertheless, the annual increase in anal cancer incidence in people without HIV was still significant, at about 1.7% per year in males and 3.3% in females (Shiels et al., 2012).

ETIOLOGICAL FACTORS A small number of case-​control studies have examined risk factors for anal cancer. The first, published in 1987, and the most recent, a nested case-​control study within a cohort of HIV-​infected patients published in 2013, included all types of anal cancer without differentiation of type (Bertisch et al., 2013; Daling et al., 1987); three included only anal SCC and its variants (Frisch et al., 1997; Holly et al., 1989; Tseng et  al., 2003), and only one specifically included both SCC and adenocarcinoma (Daling et  al., 2004). The latter study presented odds ratios for anal SCC only, as numbers of adenocarcinoma cases were too small. For this reason, the results presented in the following can be interpreted as applying to anal SCC but, for the most part, not anal adenocarcinoma.

Human Papillomavirus Human papillomavirus (HPV) has been classified by IARC as a human carcinogen for several cancer types, including anal cancer (International Agency for Research on Cancer Working Group on the Evaluation of Carcinogenic Risk to Humans, 2012). A meta-​analysis published in 2009 summarized published data from 955 anal cancers from four continents. HPV was detected in 84% of anal SCC specimens (De Vuyst et al., 2009). The most common subtypes detected were HPV 16 (73%), HPV 18 (5.2%), and HPV 33 (4.8%). HPV prevalence was similar in the two main subtypes of anal SCC, basaloid and large cell SCC. A systematic review published in the same year (2009) reported HPV-​type distribution in 992 anal cancer cases (Hoots et  al., 2009). There was substantial overlap with the meta-​ analysis published in the same year (De Vuyst et al., 2009). Overall, HPV prevalence was 88%, HPV 16 prevalence was 66%, and HPV 18 prevalence was 5% (Hoots et al., 2009). More recently, data were reported on 366 anal cancer specimens from 16 French centers. Any HPV was found in 97% of cases, with HPV 16 being by far the most prevalent type (75%), followed by HPV 18, HPV 52, HPV 33, and HPV 51 (4%–​6%) (Abramowitz et  al., 2011). An Australian series of 112 anal cancer cases reported similar findings, with any HPV detected in 96% of specimens and HPV 16 in 75% of specimens (Hillman et al., 2014). A series of 146 anal cancer cases from seven US population-​based cancer registries reported that HPV DNA was detectable in 91% of anal cancer cases. HPV 16 was the most common type (77%), followed by HPV 33 (6.2%) and HPV 18 (3.4%) (Saraiya et  al., 2015). In 2015, new data on HPV prevalence and type distribution were reported from 496 anal cancer cases from 24 countries, based on retrospective testing of formalin-​fixed paraffin-​ embedded specimens from anal cancer cases diagnosed between 1986 and 2011. HPV was detected in 88% of cases. HPV 16 was the most frequent HPV type detected (81%), followed by HPV 18 (3.6%) (Alemany et  al., 2015). Taken together, these data demonstrate that HPV is found in around 85%–​95% of anal cancer cases, and that HPV 16 causes a larger proportion of cases than that demonstrated at the cervix (De Vuyst et al., 2009; Hoots et al., 2009). Given the limitation of the analytic techniques when performed on paraffin-​embedded tissues, it may be inferred that HPV is a necessary cause of anal SCC, as it is for cervical cancer.

710

710

Part IV:  Cancers by Tissue of Origin

There are limited data on HPV prevalence in anal adenocarcinoma, but two papers, based on inclusion of a total of only 27 cases, reported that around 40% of these cases were HPV-​positive (Daling et al., 2004; Hoots et al., 2009). In contrast, it was recently reported that only 3% of 62 anal adenocarcinomas were HPV-​positive (Alemany et al., 2015), supporting the presumption that these cases actually represented misclassified rectal adenocarcinoma. The presumed precursors of anal cancer, namely high-​grade squamous intra-​epithelial lesions, are also usually caused by HPV (Hoots et al., 2009; Moscicki et al., 2012). HPV 16 is more strongly related to higher-​than lower-​grade anal lesions (Hoots et al., 2009; Moscicki et al., 2012). This suggests that the risk of progression for HPV 16 –​ related lesions is higher than that for other HPV types.

Cofactors As anal infection with HPV is the underlying cause of most anal SCC, it logically follows that anal sexual exposure to HPV is a strong risk factor for anal SCC. In men, a history of homosexuality has been consistently described as being related to anal cancer. This was suggested by case series as early as the late 1970s (Daling et al., 1987) and has been confirmed in Danish marriage registry data, where men in homosexual partnerships were at elevated risk of anal SCC (RR = 31.2, 95% CI: 8.4, 79.8) (Frisch et al., 2003). Cancer registration data from the United Kingdom and the United States demonstrated 2-​fold higher incidence rates of anal cancer in never-​married as compared with married men (Melbye et al., 1994; Scholefield et al., 1990) and in case-​ control studies of anal SCC risk among men (Daling et al., 1987, 2004; Frisch et al., 1997; Holly et al., 1989; Tseng et al., 2003). In one US study, a population-​based case-​control study of anal cancer diagnosed between 1986 and 1998 reported that receptive anal intercourse by men who were not exclusively heterosexual was related strongly to the risk of anal cancer (OR 6.8) (Daling et al., 2004). A population-​based case-​control study conducted in Denmark and Sweden demonstrated significantly more men with anal cancer reported having had homosexual contact, compared with controls (p < 0.001) (Frisch et al., 1997). Among men, a history of receptive anal intercourse has been found to be a strong risk factor in all studies that have collected data on this behavior, increasing the risk of anal cancer by up to 50-​fold (Daling et al., 2004; Holly et al., 1989; Tseng et al., 2003). In the nested case-​ control study of HIV-​infected participants, 73% were men with a history of sex with men (Bertisch et al., 2013). In each of the five other case-​control studies that included men, fewer than 50% of anal cancer cases were in men who reported a history of homosexual contact (Daling et al., 1987, 2004; Frisch et al., 1997; Holly et al., 1989; Tseng et  al., 2003). Although it is likely that some men with anal SCC in these studies under-​reported a history of homosexual contact because of social approval bias, it is also highly likely that some men acquired anal HPV by means other than receptive anal intercourse. Having a higher number of lifetime sexual partners has been reported to be an independent risk factor among men and women (Daling et al., 2004; Frisch et al., 1997). The exact proportion of anal cancer that occurs in men with no history of homosexual anal sexual contact is unclear, and the association with homosexuality may be confounded by the association with impaired immune function caused by HIV infection. A meta-​analysis of anal HPV and associated disease in homosexual men reported there were only two cohort-​based estimates of anal cancer incidence in confirmed HIV-​negative homosexual men, with an incidence rate of 5 per 100,000 per year (Machalek et al., 2012). Though this estimate is higher than rates in the general population, and confidence intervals were relatively wide, the incidence is not as high as would be expected from the odds ratios reported in case-​control studies. In women, the association of anal cancer risk with receptive anal intercourse appears to be weaker than in men. Most case-​control studies have found significant (Frisch et  al., 1997; Hoots et  al., 2009) or nonsignificant (Daling et al., 1987; Holly et al., 1989) positive associations, with odds ratios of five or less. Only a minority of women with anal cancer report a history of receptive anal

intercourse. Most published case-​control studies have reported associations between anal SCC and indirect markers of increased likelihood of HPV exposure, such as a higher number of lifetime sexual partners and a history of a sexually transmitted infection (Daling et  al., 1982, 1987, 2004; Frisch et  al., 1997; Tseng et  al., 2003). In a study of anal HPV risk in women, anal HPV was common, and receptive anal intercourse was only weakly associated with risk (Goodman et al., 2008). Studies examining the association between tobacco exposure and cervical cancer risk in women who are infected with HPV have demonstrated that tobacco exposure is an independent risk factor with increasing cancer risk from longer duration and greater intensity of tobacco exposure (Appleby et al., 2006). Possible mechanisms include an exposure to carcinogens in cervical secretions and an immunosuppressive effect (Appleby et al., 2006). In all studies that have investigated the association between tobacco smoking and anal cancer, a risk has been detected in univariate analyses (Bertisch et al., 2013; Daling et al., 1987, 2004; Frisch et al., 1999; Holly et al., 1989; Richel et al., 2013; Tseng et al., 2003). Studies with careful control for confounders have reported an association at the multivariate level (Daling et  al., 2004; Frisch et al., 1999) suggesting that, as in cervical cancer, smoking may be an independent risk factor. In one of these studies, however, an increased risk was only found for premenopausal women who currently smoked (Frisch et al., 1999). In a large prospective study in the United Kingdom, tobacco smoking was an independent risk factor for anal cancer and the association with smoking was significantly greater for anal SCC than for adenocarcinoma of the anus (Coffey et al., 2015).

HIV People living with HIV have approximately 30-​fold increased rates of anal cancer compared to the general population (Grulich et al., 2007). Several studies have described an increase in anal cancer incidence in people living with HIV since 1996, when effective antiretroviral treatment for HIV was introduced. A  meta-​analysis of reports published before 2011 concluded that the annual anal cancer incidence in HIV-​positive men who have sex with men had increased from 22/​100,000 prior to 1996 to 78/​100,000 after that year (Machalek et al., 2012). In the US-​based Adult Spectrum of Disease study, the annual incidence of anal cancer among approximately 55,000 HIV-​ infected adults increased from 19 per 100,000 in 1992–​1995 to 78 per 100,000 in 2000–​2003 (Patel et al., 2008). In the US HIV/​AIDS cancer match study of over 275,000 people with HIV, anal cancer incidence increased by 3.8% per year between 1996 and 2010 (Robbins et al., 2014). In the North American AIDS Cohort Collaboration on Research and Design, anal cancer incidence in homosexual men increased from 90/​100,000 in 1996–​1999 to 131/​100,000 in 2004–​ 2007, and appears to have plateaued in recent years (Silverberg et al., 2012). During 1995–​2012 in the Netherlands, anal cancer incidence in people with HIV increased and peaked at 114 per 100,000 in 2005–​ 2006 and then declined to 72 per 100,000 in 2011–​2012 (Richel et al., 2015). In some settings where homosexual men comprise a majority of people with HIV, anal cancer has become the most common non-​ AIDS-​defining malignancy (van Leeuwen et  al., 2009). Incidence rates of over 100/​100,000 per year in some settings suggest that anal cancer has become one of the most common cancers in homosexual men with HIV. Though the risk of anal cancer is greatest in homosexual men with HIV, elevations in risk of more than 10-​fold compared with the general population are seen in other men with HIV, and in women with HIV (Chaturvedi et al., 2009).

HOST FACTORS Genetic Susceptibility Although the great majority of cases of anal cancer are associated with high-​risk HPV infection, most people who become infected with

 71

Anal Cancer high-​risk types of HPV do not develop anal cancer. It is estimated that about 90% of mucosal HPV infections clear within 2 years, and that this is mostly due to cell-​mediated immunity (Veldhuijzen et al., 2010). There has been substantial interest in whether genetic factors may underlie the reason that only a minority of HPV-​infected people develop cancer. Polymorphisms in a large number of candidate genes, including those regulating the cell cycle, DNA repair, apoptosis induction (including p53), and immune function genes have been studied as risk factors for HPV-​related cervical and head and neck cancers, but the results have not been consistent (Habbous et al., 2014). It is certainly plausible that genetic factors related to the control of cell-​ mediated immunity may be important risk factors for HPV-​related cancer (de Araujo Souza et al., 2009). Several studies have described associations between persistent cervical HPV infection (and cervical cancer risk) and polymorphisms of human leucocyte antigen (HLA) genes (de Araujo Souza et  al., 2009). Although some genome-​wide association studies of cervical cancer have also reported associations with HLA genes (Chen et al., 2013), they have not consistently confirmed the finding from hypothesis-​driven candidate gene studies (Johanneson et  al., 2014). No studies of genetic susceptibility that focus specifically on anal cancer, as opposed to other HPV-​related cancers, have been published.

Immune Function As described earlier, anal cancer occurs at substantially increased rates in people with HIV, and it also occurs at increased rates in other immune-​deficient populations, reflecting the role of immune function in controlling HPV infection and its consequences. Anal cancer incidence is raised about 5-​fold in solid organ transplant recipients who receive medical immune suppression (Grulich et al., 2007; Madeleine et  al., 2013). In people with HIV, the two risk factors of increased exposure to HPV and immune dysfunction are combined, and this population has the highest rates of anal cancer of any population. Anal cancer incidence is increased 10-​to 15-​fold in women with HIV, and in heterosexual men and injecting drug users with HIV, and is increased by around 50-​fold in gay and bisexual men with HIV (Chaturvedi et  al., 2009). Some large cohort studies in this population have reported anal cancer incidence of over 100/​100,000 per year (Grulich et al., 2012). In the HIV-​infected, anal cancer is not as strongly related to immune function as the AIDS-​defining cancers, Kaposi’s sarcoma and non-​ Hodgkin lymphoma (Grulich and Vajdic, 2015). However, a moderate association with declining CD4 count has been described in two recent cohort studies (Reekie et  al., 2010; Silverberg et  al., 2012). Severe past immune deficiency, as measured by low nadir CD4-​ positive lymphocyte count, has been associated with increased anal cancer risk (Bertisch et al., 2013; Grulich et al., 2012; Richel et al., 2013). Associations have been described with a current CD4 cell count of less than 200 cells/​μL (Silverberg et al., 2011). A CD4 count less than 200 cells/​µL 6–​7 years prior to diagnosis was most closely associated with anal cancer risk in the Swiss HIV Cohort study (Bertisch et al., 2013). A  large US cohort recently described a reduced risk of anal cancer in those who started ART early (with a CD4 count > 350/​μL) and in those with a long history of undetectable HIV viral load (Chiao et al., 2013). These data suggest that early initiation of HIV therapy to avoid severe immune deficiency may be associated with lower anal cancer risk.

HPV Disease at Other Sites In women, studies in Sweden, the United Kingdom, and the United States have demonstrated that anal cancer incidence is raised about 5-​ fold after the occurrence of cervical intra-​epithelial disease or invasive cervical cancer (Coffey et al., 2015; Edgren and Sparén, 2007; Saleem et al., 2011). It is raised by even more, about 20-​fold, after vulvar carcinoma (Saleem et al., 2011). This pattern likely reflects the physical proximity of the anus to the female genitalia, resulting in the spread of HPV infection.

711

PREVENTION Modification of Behavior HPV is the most prevalent of all sexually transmitted infections. This, coupled with the fact that it may be transmitted sexually on fingers (Nyitray, 2012)  and from anogenital skin, means that reduction in risky sexual behaviors is less successful in reducing incidence than it is for many other sexually transmitted infections, especially those which are predominantly transmitted via bodily fluids such as semen and cervical secretions. Condoms appear to provide partial protection against genital HPV infection, but estimates of preventive efficacy vary widely (Hariri and Warner, 2013). Receptive anal intercourse without protection of condom use increases the risk of anal HPV detection in homosexual men (Hu et al., 2013; Nyitray et al., 2011). Given the association of tobacco exposure with anal cancer risk noted earlier, it is plausible that smoking cessation may reduce the risk of anal cancer.

Vaccination HPV vaccination offers substantial efficacy in the prevention of anal high-​risk HPV infection and anal HSIL. The bivalent HPV vaccine, which protects against HPV 16 and HPV 18, reduced the rate of anal detection of these HPV types by 84% over 3  years of follow-​up in women without HPV 16 and HPV 18 infection at baseline (Kreimer et  al., 2011). In a per-​protocol analysis, the quadrivalent HPV vaccine, which protects against HPV 16, HPV 18, HPV 6, and HPV 11, reduced the risk of anal HSIL related to these HPV types in young men who have sex with men by 78% over 3 years of follow-​up (Palefsky et al., 2011). These data suggest that population-​based HPV vaccination should result in substantial declines in anal cancer incidence in young people who receive the HPV vaccine. Widespread use of HPV vaccines did not commence until 2007, and uptake has been geographically patchy. Countries with high HPV vaccine uptake in girls have seen enormous declines in the incidence of HPV-​associated genital warts in women (Ali et al., 2013), and in the prevalence of detectable cervical infection with high-​risk HPV types (Tabrizi et al., 2014). In addition, in settings with high uptake in young women, very large declines in genital warts have also been demonstrated in young heterosexual men, in the absence of male vaccination. These declines in unvaccinated males are believed to be due to herd immunity (Ali et al., 2013). However, declines in HPV-​related genital warts have not been seen in gay and bisexual men (Ali et al., 2013), suggesting that this population with highest rates of HPV-​related morbidity will only be protected if males are vaccinated. Currently, only a few countries recommend routine vaccination of boys, and only in Australia is there a universal offer of HPV vaccine to school-​aged boys (Brill, 2013). An alternative strategy to reduce HPV-​related morbidity in gay and bisexual men is a targeted approach in young adult men. The United Kingdom Department of Health Joint Committee on Vaccination and Immunisation recently released an interim statement recommending a targeted HPV vaccination program to vaccinate men who have sex with men aged 16 to 40 years attending genitourinary medicine clinics (United Kingdom Joint Committee on Vaccination and Immunisation Public Health England, 2014). In the United States, the Advisory Committee on Immunization Practices has recommended that men who have sex with men aged up to 26 should receive HPV vaccination (Markowitz et al., 2014). In Australia, in addition to publicly funded vaccination of school-​aged boys, targeted vaccination of adult gay and bisexual men is recommended (Australian Government Department of Health, 2014).

Screening and Treatment of Intra-​epithelial Neoplastic Precursors As described earlier, the natural history of anal HPV infection and anal HSIL is believed to resemble that of cervical HPV infection and cervical HSIL. Established high-​risk HPV infection leads to anal HSIL and, in a small proportion of cases, leads to invasive cancer over a period of

712

712

Part IV:  Cancers by Tissue of Origin

many years. The similarity in natural history has led some researchers to propose an anal swab-​based screening program in the highest risk population, gay and bisexual men with or without HIV (Palefsky et al., 1997). Population-​based implementation of cervical Pap screening in women has been associated with sustained and very substantial reductions in cervical cancer incidence. On the other hand, there is as yet no evidence that anal cancer screening leads to a decline in anal cancer incidence at an individual or population level. Rather, the impetus for screening for anal cancer has largely been based on reasoning by analogy (Arbyn et  al., 2012). There are currently no national guidelines that recommend anal cancer screening in any population. The analogy with cervical cancer may be misleading, as there are substantial differences in the natural history, screening, and treating of high-​risk HPV infection and HSIL of the cervix compared to that of the anus. In women, high-​risk HPV cervical infection rates peak in adolescence and rapidly decline after age 25 (Bruni et al., 2010). In gay and bisexual men, anal HPV is highly prevalent, and prevalence remains elevated at least until 50 years of age (Chin-​Hong et al., 2004; Vajdic et al., 2009). A lower prevalence of anal HPV is found in women, but has been reported to be over 20% in several studies (Nyitray, 2012). Based on limited data, it appears that progression rates from HSIL to cancer may be substantially lower for anal as compared to cervical HSIL (Grulich et al., 2012; McCredie et al., 2008). There is considerable debate about the accuracy of anal cancer screening tests. While the sensitivity of anal cytology in the detection of HSIL is believed to be similar to that of cervical cytology, the specificity is considerably lower (Roberts and Thurloe, 2012), and the diagnostic test, high-​resolution anoscopy, is technically more demanding than cervical colposcopy (Palefsky, 2012). In addition, treatment of anal HSIL is much more difficult than for cervical HSIL, because in general it is not possible to completely excise HSIL lesions. Recurrence rates are extremely high, and were close to 70% at 72 weeks in the largest randomized trial (Richel et al., 2013). For all of these reasons, the likely effectiveness of anal cancer screening is questionable. Given the much improved survival when anal cancer is diagnosed early, there has been substantial interest in the use of digital ano-​rectal examination as a screening test to diagnose anal cancer early. This technique is recommended as a screening test in European and US clinical guidelines for the management of people with HIV (Ong et al., 2014), but it is explicitly acknowledged that this is based only on “expert opinion.” There have been no studies of the test performance characteristics of digital ano-​rectal examination in the diagnosis of anal cancer, and no evidence that it will reduce cancer morbidity or mortality. Based on limited data, the test appears to be acceptable to gay and bisexual men living with HIV (Read et al., 2013).

FUTURE DIRECTIONS Although existing female HPV vaccination programs will lead to reductions in anal cancer incidence in years to come, this is unlikely to affect anal cancer rates in gay and bisexual men. The introduction of universal male HPV vaccination will eventually lead to reductions in anal cancer incidence in this population. Although the HPV vaccine is proven effective in gay and bisexual men aged up to 26, randomized trials are required investigate whether vaccination of older gay and bisexual men may reduce anal cancer risk in these older age groups. The design of screening programs for anal cancer has been hampered by the dearth of evidence on the natural history of anal HPV infection. Studies of the natural history of cervical HPV have greatly informed the design of screening programs in women, and similarly designed natural history studies are required in high-​risk populations, including gay and bisexual men, people with HIV, solid organ transplant recipients, and women with previous cervical HPV disease. In addition, trials of treatment of anal HSIL are required. A large multi-​ site randomized trial of topical or ablative treatment in preventing anal cancer in patients with HIV and anal HSIL (ANCHOR, clintrials.gov, identifier NCT02135419) is underway, and is scheduled to report in 2022. Studies investigating more effective HSIL treatments are also required.

References Abramowitz L, Jacquard A, Jaroud F, et al. 2011. Human papillomavirus genotype distribution in anal cancer in France:  the EDiTH V study. Int J Cancer, 129, 433–​439. Alemany L, Saunier M, Alvarado-​Cabrero I, et al. 2015. Human papillomavirus DNA prevalence and type distribution in anal carcinomas worldwide. Int J Cancer, 136, 98–​107. Ali H, Donovan B, Wand H, et  al. 2013. Genital warts in young Australians five years into national human papillomavirus vaccination programme: national surveillance data. BMJ, 346, f2032. PMID: 23599298. American Joint Committee on Cancer. 2002. Cancer staging manual. 6th ed. Chicago: Springer. Appleby P, Beral V, Berrington de González A, et  al. 2006. Carcinoma of the cervix and tobacco smoking:  collaborative reanalysis of individual data on 13,541 women with carcinoma of the cervix and 23,017 women without carcinoma of the cervix from 23 epidemiological studies. Int J Cancer, 118, 1481–​1495. Arbyn M, de Sanjose S, Saraiya M, et al. 2012. EUROGIN 2011 roadmap on prevention and treatment of HPV-​related disease. Int J Cancer, 131(9), 1969–​1982. PMCID: PMC3429628. Australian Government Department of Health. 2014. The Australian immunisation handbook, 10th ed. 2015 (updated June 2015). ISBN: 978-​7-​74241-​861-​2 Benard V, Johnson C, Thompson T, et  al. 2008. Examining the association between socioeconomic status and potential human papillomavirus-​ associated cancers. Cancer, 113, 2910–​2918. Bertisch B, Franceschi S, Lise M, et al. 2013. Risk factors for anal cancer in persons infected with HIV: a nested case-​control study in the Swiss HIV Cohort Study. Am J Epidemiol, 178, 877–​884. Bouvard V, Baan R, Straif K, et al. 2009. A review of human carcinogens—​Part B: biological agents. Lancet Oncol, 10(4), 321–​322. PMID: 19350698. Brewster D, and Bhatti L. 2006. Increasing incidence of squamous cell carcinoma of the anus in Scotland, 1975–​2002. Br J Cancer, 95, 87–​90. Brill D. 2013. Australia launches national scheme to vaccinate boys against HPV. BMJ, 346, f924. PMID: 23404164. Bruni L, Diaz M, Castellsague X, et al. 2010. Cervical human papillomavirus prevalence in 5 continents: meta-​analysis of 1 million women with normal cytological findings. J Infect Dis, 202(12), 1789–​1799. PMID: 21067372. Burk RD, Kelly P, Feldman J, et al. 1996. Declining prevalence of cervicovaginal human papillomavirus infection with age is independent of other risk factors. Sex Transm Dis, 23(4), 333–​341. PMID: 8836027. Chaturvedi A, Madeleine M, Biggar R, and Engels E. 2009. Risk of human papillomavirus-​ associated cancers among persons with AIDS. J Natl Cancer Inst, 101, 1120–​1130. Chen D, Juko-​Pecirep I, Hammer J, et al. 2013. Genome-​wide association study of susceptibility loci for cervical cancer. J Natl Cancer Inst, 105(9), 624–​ 633. PMID: 23482656. Chiao E, Hartman C, El-​Serag H, and Giordano T. 2013. The impact of HIV viral control on the incidence of HIV-​associated anal cancer. J Acquir Immune Defic Syndr, 63, 631–​638. PMCID: PMC3797186. Chin-​Hong PV, Vittinghoff E, Cranston RD, et  al. 2004. Age-​specific prevalence of anal human papillomavirus infection in HIV-​negative sexually active men who have sex with men: the EXPLORE study. J Infect Dis, 190(12), 2070–​2076. PMID: 15551204. Coffey K, Beral V, Green J, Reeves G, and Barnes I. 2015. Lifestyle and reproductive risk factors associated with anal cancer in women aged over 50 years. Br J Cancer, 112, 1568–​1574. Daling J, Weiss N, Hislop T, et al. 1987. Sexual practices, sexually transmitted diseases, and the incidence of anal cancer. N Engl J Med, 317, 973–​977. Daling J, Weiss N, Klopfenstein L, et  al. 1982. Correlates of homosexual behavior and the incidence of anal cancer. JAMA, 247, 1988–​1990. Daling JR, Madeleine MM, Johnson LG, et al. 2004. Human papillomavirus, smoking, and sexual practices in the etiology of anal cancer. Cancer, 101(2), 270–​280. PMID: 15241823. Darragh TM, Colgan TJ, Cox JT, et al. 2012. The Lower Anogenital Squamous Terminology Standardization Project for HPV-​Associated Lesions: background and consensus recommendations from the College of American Pathologists and the American Society for Colposcopy and Cervical Pathology. J Low Genit Tract Dis, 16(3), 205–​242. PMID: 22820980. de Araujo Souza PS, Sichero L, and Maciag PC. 2009. HPV variants and HLA polymorphisms:  the role of variability on the risk of cervical cancer. Future Oncol, 5(3), 359–​370. PMID: 19374542. de Martel C, Ferlay J, Franceschi S, et  al. 2012. Global burden of cancers attributable to infections in 2008: a review and synthetic analysis. Lancet Oncol, 12, 607–​615. PMID: 22575588. De Vuyst H, Clifford G, Nascimento M, Madeleine M, and Franceschi S. 2009. Prevalence and type distribution of human papillomavirus in carcinoma

 713

Anal Cancer and intraepithelial neoplasia of the vulva, vagina and anus: a metaanalysis. Int J Cancer, 124, 1626–​1636. Dyson N, Howley PM, Munger K, and Harlow E. 1989. The human papilloma virus-​16 E7 oncoprotein is able to bind to the retinoblastoma gene product. Science, 243(4893), 934–​937. PMID: 2537532. Edgren G, and Sparén P. 2007. Risk of anogenital cancer after diagnosis of cervical intraepithelial neoplasia:  a prospective population-​based study. Lancet Oncol, 8, 311–​316. PMID: 17395104 Forman D, de Martel C, Lacey C, et al. 2012. Global burden of human papillomavirus and related diseases. Vaccine, 30(Suppl 5), F12–​F23. Frisch M. 2002. On the etiology of anal squamous carcinoma. Dan Med Bull, 49(3), 194–​209. PMID: 12238281. Frisch M, Glimelius B, van den Brule A, et al. 1997. Sexually transmitted infection as a cause of anal cancer. N Engl J Med, 337, 1350–​1358. Frisch M, Glimelius B, Wohlfahrt J, Adami H, and Melbye M. 1999. Tobacco smoking as a risk factor in anal carcinoma: an antiestrogenic mechanism? J Natl Cancer Inst, 91, 708–​715. Frisch M, Melbye M, and Moller H. 1993. Trends in incidence of anal cancer in Denmark. Br Med J, 306, 419–​422. Frisch M, Smith E, Grulich A, and Johansen C. 2003. Cancer in a population-​ based cohort of men and women in registered homosexual partnerships. Am J Epidemiol 157, 966–​972. Goodman M, Shvetsov Y, McDuffie K, et al. 2008. Acquisition of anal human papillomavirus (HPV) infection in women:  the Hawaii HPV Cohort study. J Infect Dis, 197, 957–​966. Grulich A, van Leeuwen M, Falster MO, and Vajdic C. 2007. Incidence of cancers in people with HIV/​AIDS compared with immunosuppressed transplant recipients: a meta-​analysis. Lancet, 370, 59–​67. Grulich AE, Poynten M, Machalek DA, et al. 2012. The epidemiology of anal cancer. Sex Health, 9, 504–​508. PMID: 22958581. Grulich AE, and Vajdic CM. 2015. The epidemiology of cancers in human immunodeficiency virus infection and after organ transplantation. Semin Oncol, 42(2), 247–​257. PMID: 25843729. Gunderson L, Moughan J, Ajani J, et al. 2013. Anal Carcinoma: Impact of TN Category of Disease on Survival, Disease Relapse, and Colostomy Failure in US Gastrointestinal Intergroup RTOG 98-​11 Phase 3 Trial. Int J Radiat Oncol Biol Phys, 87, 638–​645. Habbous S, Pang V, Xu W, Amir E, and Liu G. 2014. Human papillomavirus and host genetic polymorphisms in carcinogenesis: a systematic review and meta-​analysis. J Clin Virol, 61(2), 220–​229. PMID: 25174543. Hariri S, and Warner L. 2013. Condom use and human papillomavirus in men. J Infect Dis, 208(3), 367–​369. PMID: 23644282. Hartwig S, Syrjanen S, Dominiak-​Felden G, Brotons M, and Castellsague X. 2012. Estimation of the epidemiological burden of human papillomavirus-​ related cancers and non-​malignant diseases in men in Europe: a review. BMC Cancer, 12, 30. Hillman RJ, Garland SM, Gunathilake MP, et al. 2014. Human papillomavirus (HPV) genotypes in an Australian sample of anal cancers. Int J Cancer, 135(4), 996–​1001. PMID: 24497322. Hleyhel M, Belot A, Bouvier A, et al. 2015. Trends in survival after cancer diagnosis among HIV-​infected individuals between 1992 and 2009. Results from the FHDH-​ANRS CO4 cohort. Int J Cancer, 137(10), 2443–​2453. Ho GY, Bierman R, Beardsley L, Chang CJ, and Burk RD. 1998. Natural history of cervicovaginal papillomavirus infection in young women. N Engl J Med, 338(7), 423–​428. PMID: 9459645. Holly E, Whittemore A, Aston D, et al. 1989. Anal cancer incidence: genital warts, anal fissure or fistula, hemorrhoids, and smoking. J Natl Cancer Inst, 81, 1726–​1731. Hoots B, Palefsky J, Pimenta J, and Smith J. 2009. Human papillomavirus type distribution in anal cancer and anal intraepithelial lesions. Int J Cancer, 124, 2375–​2383. Hu Y, Qian HZ, Sun J, et al. 2013. Anal human papillomavirus infection among HIV-​infected and uninfected men who have sex with men in Beijing, China. J Acquir Immune Defic Syndr, 64(1), 103–​114. PMCID: 3780393. Huh WK. 2009. Human papillomavirus infection: a concise review of natural history. Obstet Gynecol, 114(1), 139–​143. PMID: 19546771. International Agency for Research on Cancer (IARC). 2013. Cancer incidence in five continents. Lyon, France: IARC. International Agency for Research on Cancer Working Group on the Evaluation of Carcinogenic Risk to Humans. 2012. A review of human carcinogens. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans (Vol. Biological Agents, Volume 100B). Lyon, France: IARC. Jin F, Stein A, Conway E, et al. 2011. Trends in anal cancer in Australia, 1982–​ 2005. Vaccine, 29, 2322–​2327. Johanneson B, Chen D, Enroth S, Cui T, and Gyllensten U. 2014. Systematic validation of hypothesis-​driven candidate genes for cervical cancer in a genome-​wide association study. Carcinogenesis, 35(9), 2084–​2088. PMID: 24879636.

713

Johnson L, Madeleine M, Newcomer L, Schwartz S, and Daling J. 2004. Anal cancer incidence and survival:  the surveillance, epidemiology, and end results experience, 1973–​2000. Cancer, 101, 281–​288. Joseph DA, Miller JW, Wu X, et al. 2008. Understanding the burden of human papillomavirus-​associated anal cancers in the US. Cancer, 113(10 Suppl), 2892–​2900. PMCID: 2729501. Kjaer SK, Chackerian B, van den Brule AJ, et al. 2001. High-​risk human papillomavirus is sexually transmitted: evidence from a follow-​up study of virgins starting sexual activity (intercourse). Cancer Epidemiol Biomarkers Prev, 10(2), 101–​106. PMID: 11219765. Kreimer AR, Gonzalez P, Katki HA, et al. 2011. Efficacy of a bivalent HPV 16/​18 vaccine against anal HPV 16/​18 infection among young women: a nested analysis within the Costa Rica Vaccine Trial. Lancet Oncol, 12(9), 862–​870. PMCID: 3172992. Leon O, Guren M, Hagberg O, et al. 2014. Anal carcinoma: survival and recurrence in a large cohort of patients treated according to Nordic guidelines. Radiother Oncol, 113, 352–​358. Machalek DA, Poynten M, Jin F, et al. 2012. Anal human papillomavirus infection and associated neoplastic lesions in men who have sex with men: a systematic review and meta-​ analysis. Lancet Oncol, 13(5), 487–​ 500. PMID: 22445259. Madeleine MM, Finch JL, Lynch CF, Goodman MT, and Engels EA. 2013. HPV-​related cancers after solid organ transplantation in the United States. Am J Transplant, 13(12), 3202–​3209. PMCID: PMC4049182. Markowitz LE, Dunne EF, Saraiya M, et al. 2014. Human papillomavirus vaccination. MMWR Recomm Rep, 63(RR-​05), 1–​30. PMID: 25167164. McCredie MR, Sharples KJ, Paul C, et al. 2008. Natural history of cervical neoplasia and risk of invasive cancer in women with cervical intraepithelial neoplasia 3: a retrospective cohort study. Lancet Oncol, 9(5), 425–​434. PMID: 18407790. McLaughlin-​Drubin ME, and Munger K. 2009. Oncogenic activities of human papillomaviruses. Virus Res, 143(2), 195–​208. PMCID: PMC2730997. Melbye M, Rabkin C, Frisch M, and Biggar R. 1994. Changing patterns of anal cancer incidence in the United States, 1940–​1989. Am J Epidemiol, 139, 772–​780. Mendenhall WM, Zlotecki RA, Vauthey JN, and Copeland EM, 3rd. 1996. Squamous cell carcinoma of the anal margin. Oncology (Williston Park), 10(12), 1843–​1848; discussion 1848, 1853–​1844. PMID: 8985968. Moscicki A, Schiffman M, Burchell A et al. 2012. Updating the natural history of human papillomavirus and anogenital cancers. Vaccine, 30(Suppl 5), F24–​33. PMCID: PMC3700362. Moscicki AB, Ma Y, Wibbelsman C, et  al. 2008. Risks for cervical intraepithelial neoplasia 3 among adolescents and young women with abnormal cytology. Obstet Gynecol, 112(6), 1335–​1342. PMCID: PMC2735396. Moscicki AB, Schiffman M, Kjaer S, and Villa LL. 2006. Chapter 5: Updating the natural history of HPV and anogenital cancer. Vaccine, 24(Suppl 3), S3/​42–​51. PMID: 16950017. Nielsen A, Munk C, and Kjaer S. 2012. Trends in incidence of anal cancer and high-​grade anal intraepithelial neoplasia in Denmark, 1978–​2008. Int J Cancer, 130, 1168–​1173 Nyitray AG. 2012. The epidemiology of anal human papillomavirus infection among women and men having sex with women. Sex Health, 9(6), 538–​ 546. PMID: 22951119. Nyitray AG, Carvalho da Silva RJ, Baggio ML, et  al. 2011. Age-​specific prevalence of and risk factors for anal human papillomavirus (HPV) among men who have sex with women and men who have sex with men:  the HPV in men (HIM) study. J Infect Dis, 203(1), 49–​ 57. PMCID: 3086435. Ong JJ, Chen M, Grulich AE, and Fairley CK. 2014. Regional and national guideline recommendations for digital ano-​rectal examination as a means for anal cancer screening in HIV positive men who have sex with men: a systematic review. BMC Cancer, 14, 557. PMCID: PMC4137084. Ortholan C, Ramaioli A, Peiffert D, et al. 2005. Anal canal carcinoma: early-​ stage tumors < or =10 mm (T1 or Tis): therapeutic options and original pattern of local failure after radiotherapy. Int J Radiat Oncol Biol Phys, 62, 479–​485. Palefsky JM. 2012. Practising high-​resolution anoscopy. Sex Health, 9(6), 580–​ 586. PMID: 23380236. Palefsky JM, Giuliano AR, Goldstone S, et al. 2011. HPV vaccine against anal HPV infection and anal intraepithelial neoplasia. N Engl J Med, 365(17), 1576–​1585. PMID: 22029979. Palefsky JM, Holly EA, Hogeboom CJ, et al. 1997. Anal cytology as a screening tool for anal squamous intraepithelial lesions. J Acquir Immune Defic Syndr Hum Retrovirol, 14(5), 415–​422. PMID: 9170415. Patel P, Hanson DL, Sullivan PS, et  al. 2008. Incidence of types of cancer among HIV-​ infected persons compared with the general population in the United States, 1992–​2003. Ann Intern Med, 148(10), 728–​736. PMID: 18490686.

714

714

Part IV:  Cancers by Tissue of Origin

Poynten IM, Machalek D, Templeton D, et al. 2016. Comparison of age-​specific patterns of sexual behaviour and anal HPV prevalence in homosexual men with patterns in women. Sex Transm Infect, 92(3), 228–​231. PMID: 26306829. Read TR, Vodstrcil L, Grulich AE, et al. 2013. Acceptability of digital anal cancer screening examinations in HIV-​positive homosexual men. HIV Med, 14(8), 491–​496. PMID: 23590621. Reekie J, Kosa C, Engsig F, et  al. 2010. Relationship between current level of immunodeficiency and non-​acquired immunodeficiency syndrome-​ defining malignancies. Cancer, 116, 5306–​5315. Richel O, de Vries HJ, van Noesel CJ, Dijkgraaf MG, and Prins JM. 2013. Comparison of imiquimod, topical fluorouracil, and electrocautery for the treatment of anal intraepithelial neoplasia in HIV-​positive men who have sex with men: an open-​label, randomised controlled trial. Lancet Oncol, 14(4), 346–​353. PMID: 23499546. Richel O, Van der Zee R, Smith A, De Vries H, and Prins J. 2013. Gradually decreasing anal cancer incidence in the HIV+ population in the Netherlands after a decade of cART. Sex Health, 10, 586. Richel O, Van Der Zee RP, Smit C, De Vries HJ, and Prins JM. 2015. Brief report: anal cancer in the HIV-​positive population: slowly declining incidence after a decade of cART. J Acquir Immune Defic Syndr, 69(5), 602–​ 605. PMID: 26167621. Robbins H, Engels E, Pfeiffer R, and Shiels M. 2015. Age at cancer diagnosis for blacks compared with whites in the United States. J Natl Cancer Inst, 107,dju489. doi: 410.1093/​jnci/​dju1489. Robbins H, Shiels M, Pfeiffer R, and Engels E. 2014. Epidemiologic contributions to recent cancer trends among HIV-​infected people in the United States. AIDS, 28, 881–​890. Roberts JM, and Thurloe JK. 2012. Comparison of the performance of anal cytology and cervical cytology as screening tests. Sex Health, 9(6), 568–​ 573. PMID: 22951325. Robinson D, Coupland V, and Møller H. 2009. An analysis of temporal and generational trends in the incidence of anal and other HPV-​related cancers in Southeast England. Br J Cancer, 100, 527–​531. Rodriguez AC, Schiffman M, Herrero R, et  al. 2010. Longitudinal study of human papillomavirus persistence and cervical intraepithelial neoplasia grade 2/​3: critical role of duration of infection. J Natl Cancer Inst, 102(5), 315–​324. PMCID: 2831050. Rylander E, Ruusuvaara L, Almstromer MW, Evander M, and Wadell G. 1994. The absence of vaginal human papillomavirus 16 DNA in women who have not experienced sexual intercourse. Obstet Gynecol, 83(5 Pt 1), 735–​ 737. PMID: 8164934. Saleem A, Paulus J, Shapter A, et al. 2011. Risk of anal cancer in a cohort with human papillomavirus-​related gynecologic neoplasm. Obstet Gynecol, 17(3), 643–​649. Saraiya M, Unger ER, Thompson TD, et al. 2015. US assessment of HPV types in cancers:  implications for current and 9-​valent HPV vaccines. J Natl Cancer Inst, 107(6), djv086. PMID: 25925419. Scheffner M, Werness BA, Huibregtse JM, Levine AJ, and Howley PM. 1990. The E6 oncoprotein encoded by human papillomavirus types

16 and 18 promotes the degradation of p53. Cell, 63(6), 1129–​1136. PMID: 2175676. Schiffman M, and Kjaer SK. 2003. Chapter  2:  Natural history of anogenital human papillomavirus infection and neoplasia. J Natl Cancer Inst Monogr 31, 14–​19. PMID: 12807940. Scholefield J, Thornton Jones H, Cuzick J, and Northover J. 1990. Anal cancer and marital status. Br J Cancer, 62, 286–​288. SEER Cancer Statistics Factsheets. 2015. Anal cancer. Available from:  http://​seer.cancer.gov/​statfacts/​html/​anus.html Accessed March 20, 2015. Shiels MS, Pfeiffer RM, Chaturvedi AK, Kreimer AR, and Engels EA. 2012. Impact of the HIV epidemic on the incidence rates of anal cancer in the United States. J Natl Cancer Inst, 104(20), 1591–​1598. PMCID: PMC3611819. Silverberg M, Chao C, Leyden W, et  al. 2011. HIV infection, immunodeficiency, viral replication, and the risk of cancer. Cancer Epidemiol Biomarkers Prev, 20, 2551–​2559. Silverberg M, Lau B, Justice A, et al. 2012. Risk of anal cancer in HIV-​infected and HIV-​uninfected individuals in North America. Clin Infect Dis, 54, 1026–​1034. PMCID: PMC3297645. Tabrizi SN, Brotherton JM, Kaldor JM, et al. 2014. Assessment of herd immunity and cross-​protection after a human papillomavirus vaccination programme in Australia:  a repeat cross-​sectional study. Lancet Infect Dis, 14(10), 958–​966. PMID: 25107680. Tseng H, Morgenstern H, Mack T, and Peters R. 2003. Risk factors for anal cancer: results of a population-​based case-​control study. Cancer Causes Control, 14, 837–​846. PMID: 14682441. United Kingdom Joint Committee on Vaccination and Immunisation Public Health England. 2015. JCVI statement on HPV vaccination of men who have sex with men. Available from:  https://​www.gov.uk/​government/ ​ p ublications/ ​ j cvi- ​ s tatement- ​ o n- ​ h pv- ​ vaccination- ​ o f- ​ m en​who-​have-​sex-​with-​men. Vajdic CM, van Leeuwen MT, Jin F, et al. 2009. Anal human papillomavirus genotype diversity and co-​infection in a community-​based sample of homosexual men. Sex Transm Infect, 85(5), 330–​335. PMID: 19342375. van Leeuwen M, Vajdic C, Middleton M, et al. 2009. Continuing declines in some but not all HIV-​associated cancers in Australia after widespread use of antiretroviral therapy. AIDS, 23, 2183–​2190. Veldhuijzen NJ, Snijders PJ, Reiss P, Meijer CJ, and van de Wijgert JH. 2010. Factors affecting transmission of mucosal human papillomavirus. Lancet Infect Dis, 10(12), 862–​874. PMID: 21075056. Welton ML, Lambert R, and Bosman FT. 2010. Tumours of anal canal. In:  Bosman FT, Carneiro F, Hruban RH, and Theise ND (Eds.), WHO classification of tumours of the digestive system (4 ed., pp. 183–​193). Lyon, France: IARC. World Health Organization. 2013. International classification of disease for oncology (3rd ed., First revision). Geneva: WHO. zur Hausen H. 1996. Papillomavirus infections: a major cause of human cancers. Biochim Biophys Acta, 1288(2), F55–​78. PMID: 8876633.

 715

38 Leukemias MARTHA S. LINET, LINDSAY M. MORTON, SUSAN S. DEVESA, AND GRAÇA M. DORES

OVERVIEW The 2001 World Health Organization (WHO) classification of hematopoietic and lymphoid neoplasms categorized “the leukemias” into two major groupings—​myeloid and lymphoid neoplasms. Myeloid neoplasms, which are the primary focus of this chapter, include acute myeloid leukemia (AML), myelodysplastic syndromes (MDS), and myeloproliferative neoplasms (MPN). Lymphoid neoplasms are mostly reviewed as part of non-​Hodgkin lymphoma in Chapter 40 of this volume, although descriptive patterns and selected etiologic studies are briefly discussed in this chapter because of historical trends. Worldwide, leukemias are ranked 11th among all cancer types, comprising approximately 2.5% of all malignancies. For most leukemias, rates are higher in males than females, and age-​adjusted incidence rates show more limited international variation than most solid tumors, ranging from, ~3-​fold for acute lymphocytic leukemia (ALL), 4-​fold for chronic myeloid leukemia (CML), 10-​fold for AML, to 40-​ fold for chronic lymphocytic leukemia (CLL). Exposure to ionizing radiation and certain chemical carcinogens (e.g., cytotoxic chemotherapy, benzene, formaldehyde) are the most consistently associated risk factors for MDS and/​ or AML (MDS/​ AML). Radiation also has been linked with CML, and cigarette smoking with AML. Fewer risk factors have been identified for MPNs. Some evidence implicates increased risks of AML in rubber workers, farmers, and other agricultural workers. Several studies have reported increased risks of multiple myeloid neoplasms in patients with autoimmune diseases and in recipients of solid organ transplants. In addition, most myeloid neoplasms, with the possible exception of CML, demonstrate familial aggregation. Since established risk factors do not explain most of the occurrence of myeloid neoplasms, opportunities for prevention are currently limited, but would include reduction in exposure to radiation; limiting occupational and general population exposure to benzene, formaldehyde, and other chemicals; and cessation of and avoidance of smoking. Improvements in classification, ascertainment, diagnosis, and molecular characterization of myeloid neoplasms are critical to clarify etiology and to develop measures for prevention and effective treatment.

INTRODUCTION Most, if not all, acute and chronic leukemias appear to develop from a preleukemic state that progresses to overt leukemia over time (Shlush and Minden, 2015). Included among the preleukemic entities are myelodysplastic syndromes (MDS), myeloproliferative neoplasms (MPN), “overlap” disorders termed myelodysplastic/​myeloproliferative neoplasms (MDS/​MPN), and monoclonal B-​cell lymphocytosis. All of these entities are clonal stem cell disorders that can progress or transform into leukemia. Understanding the epidemiology of the leukemias and the preleukemic states has been complicated by changing classification schemes and by the fact that many preleukemic entities have not always been reportable to cancer registries, thereby often being excluded from population-​based cancer statistics. Worldwide in 2012, leukemias were ranked 11th among all cancer types, comprising approximately 2.5% of all malignancies and an estimated 352,000 incident cases (Ferlay et al., 2014). In the United

States, an estimated 60,140 cases will be diagnosed in 2016 (including 19,950 AML, 8220 CML, 6590 ALL, 18,960 CLL, and 6420 other leukemias), and the number of deaths from leukemia is estimated as 24,400 (including 10,430 AML, 1070 CML, 1430 ALL, 4660 CLL, and 6810 other leukemias) (Siegel et al., 2016). Leukemias are estimated to comprise 4% and 3% of all incident cancers among US males and females, respectively, and 4% of all cancer deaths in both males and females (Siegel et al., 2016). In the United States, all MDS and MPN became reportable to the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) Program in 2001 (Fritz et al., 2000). In 2012, 3981 and 3291 cases of MDS and MPN, respectively, were diagnosed in 18 cancer registry areas representing 26% of the US population, including 2304 and 1677 cases of MDS and 1672 and 1619 cases of MPN among males and females, respectively (SEER-​18). In contrast, based on data from 64 European cancer registries (1995–​2002), 7460 cases of MDS and 15,269 cases of MPN were estimated to occur annually (Visser et al., 2012). Mortality data on MDS and MPN are sparse. In 2009 there were 6007 deaths in the United States with an underlying cause of death specified as MDS (Polednak, 2013), and in 2006, 3303 deaths were attributed to MPN (Polednak, 2011). The unifying feature of the leukemias is that they arise from an accumulation of multiple, stepwise genetic and epigenetic changes in the hematopoietic stem cell (HSC) and committed progenitors. A preleukemic cell contains only a subset of the genetic and epigenetic changes characterizing leukemic cells (Shlush and Minden, 2015). In the normal state, HSC differentiate into progenitor cells that give rise to myeloid and lymphoid progenitor cells and eventually all mature blood elements (Hoffman et al., 2013; Shizuru et al., 2005). Throughout this highly regulated, hierarchical differentiation and maturation process, lymphoid and myeloid cells acquire distinct phenotypes. Genetic mutations involving primitive stem cells or early myeloid-​committed progenitors result in clonal proliferation and accumulation of immature hematopoietic cells (e.g., blasts) of myeloid lineage (e.g., AML) in the bone marrow, peripheral blood, or other tissues (Lichtman, 2016; Swerdlow et al., 2008). When the affected pluripotent stem cell results in maturation arrest of more mature myeloid cells and subsequent accumulation of these more differentiated phenotypes, chronic leukemias then ensue. In CML the affected pluripotent stem cell is consistently associated with a BCR-​ ABL1 fusion gene located on the Philadelphia chromosome, resulting in the accumulation of more mature myeloid cells of erythroid, granulocytic, monocytic, dendritic, and megakaryocytic lineages (Lichtman, 2015). For many of the lymphoid neoplasms, the “cell of origin” represents the stage of differentiation of the tumor cells, rather than the cell in which the initial transforming event occurred (Jaffe et al., 2001). Genetic mutations involving B-​cell progenitors may result in the accumulation of phenotypically immature-​appearing lymphoid cells (blasts), as seen in ALL, or mature-​appearing lymphocytes, as in CLL. The MDS are a heterogeneous group of clonal HSC neoplasms characterized by dysplasia (disordered maturation) in one or more cell lines and ineffective hematopoiesis that may result in peripheral cytopenias of one or more cell lines (Swerdlow et al., 2008). In contrast, the MPN are clonal HSC neoplasms associated with the proliferation of one or more of the myeloid lineages and absence of dysplasia. The MDS/​MPN include both dysplastic and proliferative features.

715

716

716

PART IV:  Cancers by Tissue of Origin

HEMATOPOIETIC AND LYMPHOID CLASSIFICATION SCHEMES Evolution of Classification Earlier reviews provided a comprehensive summary of the history of leukemia classification (Linet et al., 2006). The landmark French-​ American-​British (FAB) classification (Bennett et  al., 1976, 1989, 1994, 2013)  achieved international consensus on morphologic criteria. Subsequent efforts to incorporate developmental and functional aspects of hematopoiesis according to lineage, as well as key aspects of pathogenesis and cytogenetic and immunophenotypic characteristics (Bennett, 2000; McKenna, 2000), culminated in the 2001 WHO Classification of Tumors of the Hematopoietic and Lymphoid Tissue (Jaffe et al., 2001). This classification included genetic data that were more predictive of disease behavior and outcome than morphology and also added new disease categories. Cytogenetic alterations have long been identified as hallmarks of many cases of hematopoietic and lymphoid tumors, but the advent of and dramatic technical developments in high-​resolution profiling have led to notable advances in clarifying the genetic basis of these disorders. Certain markers have been identified as clinically meaningful therapeutic targets or as helpful prognostic markers (Bochtler et al., 2015; Inaba et al., 2013). With this rapid evolution and emergence of new information, the WHO classification was updated in 2008 (Swerdlow et al., 2008). The 2008 WHO classification considered lineage-​ specific disease categories (myeloid, lymphoid, and histiocytic/​ dendritic cell), distinguished immature neoplasms (e.g., AML, lymphoblastic leukemia/​lymphoma) from more mature neoplasms (e.g., MDS, MPN, MDS/​MPN), introduced new disease-​defining criteria, and identified new disease entities. Multidisciplinary experts in international working groups (such as the International Working Group for Myelofibrosis Research and Treatment, the European Group for the Immunologic Classification of Leukemia, and the National Cancer Institute-​sponsored Working Group on CLL) continue to meet and provide recommendations to ensure that the classification and updates will be clinically useful. The 2001 WHO Classification of Tumors of the Hematopoietic and Lymphoid Tissue categorized the lymphoid neoplasms into three broad categories: B-​cell neoplasms, T-​and NK-​cell neoplasms, and Hodgkin lymphoma. Within the former two categories, the leukemias were classified with the lymphomas due to several of these entities having circulating (blood) and solid (tissue) phases that represent different manifestations of the same disease (e.g., CLL [blood phase] and small lymphocytic lymphoma [tissue phase]; lymphoblastic leukemia [blood phase] and lymphoblastic lymphoma [tissue phase]) (Jaffe et al., 2001). Therefore, to be consistent with the WHO classification, this “leukemia” review will focus on the characteristics, descriptive epidemiology, and known and suspected risk factors of the myeloid neoplasms occurring in adults. Detailed findings from more recent epidemiologic studies of ALL and CLL will be found in Chapter 40 of this volume on non-​Hodgkin lymphoma, and the epidemiology of myeloid neoplasms of childhood will be found in Chapter 59 on childhood cancers. However, because earlier descriptions of leukemia incidence and mortality often focused on all forms of leukemia combined (e.g., AML, CML, ALL, CLL; hereafter designated total leukemia) and most epidemiologic studies prior to the last decade or so considered lymphoid leukemias in conjunction with myeloid leukemias, some material on lymphoid leukemias is included in this chapter in the sections on descriptive and analytical epidemiologic studies and in Chapter 40 of this book. The International Classification of Diseases (ICD) for Oncology (ICD-​O) classification, primarily used for coding tumor topography and morphology in cancer registries, has similarly evolved over time, and the 2001 WHO classification incorporated codes from the third edition of ICD-​O (ICD-​O-​3) (Fritz et al., 2000). The 2008 WHO classification included ICD-​O-​3 morphology codes and also proposed provisional codes for the forthcoming edition of ICD-​O-​4 that remain subject to change. The complex, continuing evolution of the international classification of hematopoietic and lymphoid neoplasms has led

some population-​based cancer registries to develop special measures to improve our understanding and interpretation of information in pathology and clinical records and thereby allow more accurate coding of these neoplasms (Ruhl et al., 2015).

MYELOID NEOPLASMS AND THE WHO CLASSIFICATION In the WHO classification, the term myeloid includes all cells that belong to granulocytic (neutrophil, eosinophil, basophil), monocytic/​macrophage, erythroid, megakaryocytic and mast cell lineages (Vardiman et al., 2009). Utilizing the WHO criteria, the diagnoses of myeloid neoplasms utilize morphologic, cytochemical, immunophenotypic, and cytogenetic characteristics to determine the lineage and maturation of the neoplastic cells obtained from peripheral blood and bone marrow upon initial clinical presentation, prior to initiation of treatment.

World Health Organization 2001 Classification Acute Myeloid Leukemia and Related Precursor Neoplasms

The 2001 WHO classification of myeloid neoplasms categorized AML arising de novo separately from AML evolving from antecedent MDS or MDS/​MPN in order to better reflect the postulated distinct underlying leukemogenic mechanisms and prognoses (Jaffe et  al., 2001). Whereas the latter (AML with multilineage dysplasia) is often associated with unfavorable cytogenetics, poor response to treatment, unfavorable prognosis, and genetic insults occurring over a lifetime (indicated by the increasing incidence with age), de novo AML typically is not associated with multilineage dysplasia, has a constant incidence throughout life, and is often associated with more favorable cytogenetic abnormalities, response to treatment, and prognosis. To better reflect the distinct clinical and biologic features of AML than the preceding morphology-​based FAB classification, the 2001 WHO classification considered four major disease subgroups: (1) AML with recurrent genetic abnormalities; (2) AML with multilineage dysplasia; (3) AML and MDS, therapy-​related; and (4) AML, not otherwise specified (NOS). Other significant changes in the 2001 WHO classification included a decrease in the blast percentage in the bone marrow or blood required to establish a diagnosis of AML from 30% to 20%. Furthermore, the presence of recurrent genetic abnormalities (t(8;21)(q22;q22), t(15;17)(q22;q12), and inv(16) (p13q22) or t(16;16) (p13;qi22)) was deemed diagnostic of AML irrespective of the percentage of blasts (Jaffe et al., 2001; Vardiman et al., 2002). The subsequent 2008 WHO classification added three new entities within the category of AML with recurrent genetic abnormalities: AML with t(6;9)(p23;q34); AML with inv(3)(q21q26.2) or t(3;3) (q21;q26.2); AML (megakaryoblastic) with t(1;22)(p13;q13); and two provisional entities (AML with mutated NPM1 and AML with CEBPA) (Swerdlow et  al., 2008; Vardiman et  al., 2009). Additional diagnostic refinements were further specified for acute promyelocytic leukemia with t(15; 17)(q22; q12) and AML with 11q23 (MLL). Other changes included renaming AML with multilineage dysplasia to AML with myelodysplasia-​related changes to include AML cases with an antecedent MDS or MPN/​MDS, myelodysplasia-​related cytogenetic abnormality, or with 50% or more dysplastic changes in two or more myeloid cell lines. The therapy-​related MDS/​AML category was renamed as therapy-​related myeloid neoplasms (t-​MDS/​AML) and the subcategories of alkylating agent/​radiation-​related and topoisomerase II inhibitor-​related AML were eliminated. Two additional new AML categories were added: (1) myeloid proliferations related to Down syndrome to include Down syndrome–​related transient abnormal myelopoiesis, MDS, and AML; and (2) blastic plasmacytic dendritic cell neoplasms. The complete list of AML and related myeloid precursors included in the 2008 WHO classification with an associated ICD-​O-​3 code are specified in Table 38–1.

 71

Table 38–1.  Age-​Adjusted Incidence of Acute Myeloid Leukemia and Related Neoplasms, Myelodysplastic Syndromes, Myeloproliferative Neoplasms, and Myelodysplastic/Myeloproliferative Neoplasms Diagnosed Among Adults ≥ 20 Years of Age in 18 Cancer Registry Areas of the Surveillance, Epidemiology and End Results Program According to Disease Subtype, Sex, and Race, 2001–​2012a All Races                

Total ICD-​O-​3 Code

All Sexesb

Males

No.

IR

4,024

 

Females

No.

IR

 

No.

0.56  

2,108

0.63  

1,916

0.51

WNH

IR

 

WH

Incidence Rate Ratios Blacks

 

No.

APIs

IR

  No.

  M:F   IRR

IR

WH: B: WNH WNH IRR IRR

API: WNH IRR

No.

IR

  No.

IR

 

2,578

0.56

  687

0.62  

383

0.49  

326

0.48  

1.22*

1.10*

0.87*

0.86*

AML and related neoplasms AML with recurrent genetic abnormalities

9866, 9871, 9896, 9897

AML with t(8;21) (q22;q22) AML with inv(16) (p13.1;q22) or t(16;16)(p13.1;q22) AML with t(15;17) (q22;q12) AML with t(9;11) (q22;q23) AML with myelodysplasia-​ related changes Therapy-​related myeloid neoplasms

9896

565

0.08  

313

0.10  

252

0.07

 

366

0.08

 

69

0.07  

57

0.07  

65

0.10  

1.44*

0.90

0.94

1.25

9871

410

0.06  

246

0.07  

164

0.04

 

297

0.07

 

58

0.06  

16

0.02  

30

0.04  

1.69*

0.86

0.34*

0.66*

9866

2,812

0.39  

1,438

0.42  

1,374

0.37

 

1,746

0.38

  531

0.47  

295

0.38  

207

0.30  

1.15*

1.21*

0.98

0.79*

9897

237

0.03  

111

0.04  

126

0.03

 

169

0.04

 

29

0.03  

15

 

24

0.04  

1.06

0.85

~

1.08

9895, 9984

3,086

0.45  

1,855

0.63  

1,231

0.32

 

2,401

0.49

  240

0.34  

205

0.34  

223

0.38  

1.97*

0.70*

0.69*

0.79*

9920, 9987

1,308

0.18  

610

0.19  

698

0.18

 

1,042

0.21

 

88

0.11  

101

0.14  

63

0.10  

1.06

0.52*

0.67*

0.48*

AML without recurrent genetic abnormalitiesc

9840, 9861, 9867, 9870, 9872-​9874, 9891, 9910, 9931

29,863

4.29  

16,204

5.36  

13,659

3.52

 

22,157

4.52

 2,863

3.71  

2,437

3.66   2,153

3.59  

1.52*

0.82*

0.81*

0.80*

AML, NOS AML with minimal differentiation AML without maturation AML with maturation Acute myelomonocytic leukemia Acute monoblastic and monocytic leukemia Acute erythroid leukemia Acute megakaryoblastic leukemia Acute basophilic leukemia Acute panmyelosis with myelofibrosis Myeloid sarcoma Blastic plasmacytoid dendritic cell neoplasm

9861 9872

19,011 1,059

2.74   0.15  

10,263 573

3.43   0.19  

8,748 486

2.24 0.13

   

14,096 771

2.86 0.16

 1,715   118

2.32   0.15  

1,675 93

2.53   1,349 0.14   70

2.30   0.11  

1.53* 1.52*

0.81* 0.96

0.89* 0.90

0.80* 0.72*

9873 9874 9867

1,530 1,764 3,029

0.22   0.25   0.43  

804 934 1,665

0.26   0.31   0.54  

726 830 1,364

0.19 0.22 0.35

     

1,090 1,290 2,240

0.23 0.27 0.46

  183   199   338

0.21   0.25   0.40  

99 117 219

0.14   0.17   0.32  

146 144 205

0.23   0.23   0.33  

1.34* 1.41* 1.52*

0.92 0.93 0.87*

0.64* 0.66* 0.69*

1.02 0.86 0.72*

9891

2,317

0.33  

1,262

0.41  

1,055

0.28

 

1,789

0.37

  223

0.28  

140

0.20  

157

0.26  

1.48*

0.75*

0.54*

0.70*

9840

654

0.09  

416

0.14  

238

0.06

 

499

0.10

 

49

0.06  

50

0.08  

50

0.09  

2.19*

0.59*

0.83

0.85

9910

180

0.03  

113

0.04  

67

0.02

 

130

0.03

 

19

0.02  

15

~

 

15

~

 

2.06*

0.91

~

~

9870

6

 

4

~

 

5

~

 

0

 

0

~

 

1

~

 

~

~

~

9931 9930 9727

~

~

~

~

 

2

~

313

0.05  

172

0.05  

141

0.04

 

247

0.05

 

19

0.02  

29

0.05  

16

0.03  

1.47*

0.47*

0.92

0.51*

379 259

0.05   0.04  

208 168

0.07   0.05  

171 91

0.05 0.02

   

269 174

0.06 0.04

   

44 38

0.04   0.04  

40 33

0.06   0.04  

19 12

0.03   ~  

1.44* 2.08*

0.79 0.99

1.01 1.04

0.52* ~

(continued)

718

Table 38–​1. Continued All Races                

Total ICD-​O-​3 Code

All Sexesb

Males  

No.

Females

No.

IR

IR

 

No.

43,517

6.39  

24,257

8.72  

19,260

4.85

WNH

IR

 

WH

Blacks

No.

IR

  No.

IR

 

34,005

6.73

 2,996

4.96  

3,237

Myelodysplastic syndromes

9980, 9982, 9983, 9985, 9986, 9989

Refractory anemia Refractory anemia with sideroblasts Refractory anemia with excess blasts Refractory cytopenia with multilineage dysplasia Myelodysplastic syndrome associated with isolated del(5q) Myelodysplastic syndrome, unclassifiable or NOS

9980

5,084

0.75  

2,634

0.95  

2,450

0.63

 

3,917

0.78

  318

0.52  

9982

3,649

0.54  

2,034

0.73  

1,615

0.41

 

2,936

0.59

  191

9983

5,875

0.86  

3,532

1.23  

2,343

0.60

 

4,638

0.93

9985

2,556

0.38  

1,660

0.59  

896

0.23

 

1,979

9986

1,076

0.16  

424

0.15  

652

0.16

 

9989

25,277

3.71  

13,973

5.08  

11,304

2.82

9740-​9742, 9875, 9950, 9960-​9964, 9975

31,505

4.48  

16,123

5.12  

15,382

9875

3,146

0.44  

1,753

0.52  

9963 9950

62 10,714

0.01   1.51  

42 6,177

9961 9962

2,984 9,276

0.43   1.33  

9964 9740-​9742

355 370

9960, 9975

4,598

Myeloproliferative neoplasms Chronic myelogenous leukemia, BCR/​ABL1 positived Chronic neutrophilic leukemia Polycythemia vera Primary myelofibrosis/​ myelosclerosis with myeloid metaplasia Essential thrombocythemia Chronic eosinophilic leukemia (hypereosinophilic syndrome), NOS Mastocytosis Chronic myeloproliferative disease, NOSe

Incidence Rate Ratios

 

No.

IR

APIs   No.

IR

  M:F   IRR

WH: B: WNH WNH IRR IRR

API: WNH IRR

5.44   2,651

4.95  

1.80*

0.74*

0.81*

0.74*

462

0.78  

284

0.53  

1.52*

0.67*

1.00

0.68*

0.32  

250

0.42  

188

0.35  

1.79*

0.55*

0.73*

0.61*

  412

0.63  

392

0.64  

377

0.68  

2.04*

0.68*

0.69*

0.73*

0.40

  199

0.31  

168

0.27  

170

0.31  

2.56*

0.77*

0.69*

0.79*

898

0.18

 

66

0.11  

55

0.09  

45

0.08  

0.93

0.64*

0.53*

0.47*

 

19,637

3.87

 1,810

3.07  

1,910

3.23   1,587

3.00  

1.80*

0.79*

0.83*

0.77*

3.98

 

23,228

4.76

 2,454

3.16  

2,962

4.30   2,027

3.29  

1.29*

0.66*

0.90*

0.69*

1,393

0.37

 

2,093

0.45

  412

0.41  

323

0.41  

205

0.30  

1.41*

0.91

0.92

0.68*

0.01   1.90  

20 4,537

0.01 1.16

   

45 8,352

0.01 1.71

  5   759

~   1.00  

7 724

~   1.05  

4 656

~   1.05  

2.80* 1.64*

~ 0.59*

~ 0.61*

~ 0.61*

1,763 3,613

0.58   1.18  

1,221 5,663

0.32 1.47

   

2,283 6,584

0.46 1.35

  207   689

0.32   0.88  

226 1,091

0.33   1.60  

206 625

0.34   1.03  

1.83* 0.80*

0.69* 0.66*

0.73* 1.19*

0.73* 0.77*

0.05   0.05  

215 187

0.06   0.06  

140 183

0.04 0.05

   

219 318

0.05 0.07

   

36 22

0.04   0.03  

59 16

0.08   0.02  

35 7

0.05   ~  

1.72* 1.16

0.74 0.40*

1.67* 0.35*

1.13 ~

0.67  

2,373

0.81  

2,225

0.57

 

3,334

0.67

  324

0.48  

516

0.80  

289

0.50  

1.42*

0.72*

1.19*

0.75*

 719

Myelodysplastic/​ myeloproliferative neoplasms Chronic myelomonocytic leukemia, NOS Atypical chronic myeloid leukemia, BCR/​ABL1 negatived Juvenile myelomoncytic leukemia Myelodysplastic/​ myeloproliferative neoplasm, unclassifiablee

9876, 9945, 9946, 9960, 9975

8,601

1.26  

4,861

1.69  

3,740

0.95

 

6,557

1.31

  586

0.92  

776

1.25  

503

0.89  

1.77*

0.70*

0.95

0.68*

9945

3,870

0.57  

2,407

0.85  

1,463

0.37

 

3,128

0.62

  250

0.42  

248

0.43  

204

0.38  

2.31*

0.67*

0.69*

0.60*

9876

133

0.02  

81

0.03  

52

0.01

 

95

0.02

 

12

~

 

12

~

 

10

~

 

9946

0

 

0

~

 

0

~

 

0

~

 

0

~

 

0

~

 

9960, 9975

Chronic myeloid leukemiac 9863, 9875, 9876 Chronic myeloid leukemia, NOSc

9863

~

~

2.05* ~

~

~

~

~

~

~

 

0

4,598

0.67  

2,373

0.81  

2,225

0.57

 

3,334

0.67

  324

0.48  

516

0.80  

289

0.50  

1.42*

0.72*

1.19*

0.75*

11,999

1.69  

6,771

2.09  

5,228

1.37

 

8,167

1.72

 1,413

1.53  

1,320

1.78  

763

1.16  

1.53*

0.89*

1.04

0.68*

8,720

1.24  

4,937

1.54  

3,783

0.98

 

5,979

1.25

  989

1.10  

985

1.35  

548

0.84  

1.57*

0.88*

1.08*

0.67*

Abbreviations: AML, acute myeloid leukemia; API, Asians/​Pacific Islanders; B, blacks; F, female; HW, Hispanic whites; ICD-​O-​3, International Clcassification of Diseases for Oncology; IR, incidence rate; IRR, incidence rate ratio; M, male; NHW, non-​Hispanic whites; NOS, not otherwise specified; No., number of cases. * IRR is statistically significant from 1.00 (P 0.005 Gy was 59% for the entire period (23% for ALL and 36% for CML). However, these two subtypes accounted for 80% (27% for ALL and 53% for CML) of the excess leukemias during 1950–​1955, but only 22% (10% for ALL and 12% for CML) during 1991–​2001. Hsu et al. (2013) speculated that CML and ALL rates may have been even higher than 80% within the first 5 years after the bombings (1945–​1949), prior to the availability of systematically ascertained data. In contrast, the fraction of AML attributable to radiation among cohort members with > 0.005 Gy was 38% for the entire period studied. AML accounted for only 20% of the excess leukemias during 1950–​1955, but accounted for 80% of the excess leukemias during 1991–​2001. This temporal pattern for AML is markedly different from that observed subsequent to cytotoxic chemotherapy, where most of the excess t-​AML occurs within 10 years of treatment (see further discussion later in this chapter). Richardson and colleagues were the first to evaluate leukemia mortality risks in atomic bomb survivors by subtype (2009). The

728

728

PART IV:  Cancers by Tissue of Origin

investigators followed up mortality among 86,611 survivors during 1950–​2000 and identified 310 deaths from all forms of leukemia. For AML mortality, the dose response pattern was best described by a quadratic dose–​response function that peaked approximately 10 years after exposure, while CML and ALL demonstrated a linear dose–​ response that did not vary with time since exposure. Excess leukemia mortality risk persisted for more than 5 decades. In the most recent decade evaluated (1991–​2000), 34% of leukemia deaths among those with radiation dose > 0.005 Gy were estimated to be attributable to radiation from the bombings. MDS was first linked with radiation exposure in the late 1980s, following a detailed histopathological review of myeloid malignancies among atomic bomb survivors (Matsuo et al., 1988). The first analysis, which was based on only 13 MDS cases, revealed a significant dose–​response for MDS mortality; the excess relative risk was several times greater than that seen for all solid cancers combined (Shimizu et al., 1999). An assessment of MDS diagnosed during 1984–​2004 in two populations of survivors in Nagasaki found a notably increased ERR per Gy of 4.3 (95% CI  =  1.6–​9.5) using a linear model based on 47 cases in the 22,245 survivors in the Life Span Study (compared with ERR per Gy  =  1.11; 95% CI  =  0.53–​2.08 for AML using the preferred quadratic model) and a significant excess based on 151 MDS cases in 64,026 survivors with known distance from the bomb hypocenter (Iwanaga et al., 2011). While the latency period to development of MDS cannot be accurately determined prior to the mid-​1980s, a 40-​year latency period was observed among survivors after the mid-​ 1980s. This time frame is similar to that of de novo MDS, but differs from the median peak latency of 4–​6 years observed for t-​MDS/​ AML (Bhatia, 2013). However, molecular characteristics of MDS in the atomic bomb survivors more closely resemble those of patients treated with alkylating agents, with 6 of 13 atomic bomb survivors with MDS having AML1 gene mutations (Harada et al., 2003). Further study is needed to clarify the molecular features of MDS associated with different exposures.

Radiation Workers

Historically, the major categories of workers exposed to ionizing radiation include medical radiation workers (radiologists and radiologic technologists), nuclear industry workers, radium dial workers, miners (uranium and tin), flight crew, and military servicemen exposed to above-​ground nuclear tests (Wakeford, 2009). Studies quantifying leukemia and other cancer risks in workers are important because the radiation exposures in these populations are protracted over decades, and these populations are monitored. Small radiation exposures are cumulative and, if associated with leukemia, could translate into meaningful numbers of patients with leukemia since there are millions of radiation workers. Results from radiation worker studies also contribute important information toward defining radiation protection measures (see Chapters 13 and 16 in this volume). Furthermore, these studies are useful because findings can be extrapolated to the general population experiencing low-​level protracted exposures to natural background radiation and repeated radiation exposures from diagnostic radiologic examinations. However, overall, interpretation of risks for cancer and other serious diseases in long-​term studies of medical radiation workers is complicated by the dramatically declining radiation doses to workers over time (Linet et  al., 2010). As with most occupational epidemiologic studies, information on potential confounders (e.g., workers’ personal diagnostic radiological imaging tests, radiotherapy, smoking, and genetic characteristics) is lacking, and few studies of medical radiation workers include women (see Chapter 16). A large excess mortality risk (approximately 10-​fold) of leukemia was initially reported among US radiologists in 1950 (March et  al., 1950). Eight major cohorts of radiologists and radation technologists have been actively followed up for leukemia, other cancers, and chronic diseases (reviewed in Linet et al., 2010; Yoshinaga et al., 2004). Collectively, the eight retrospective cohort investigations have studied radiologists or radiologic technologists who first began working over a

period spanning more than 80 years, including small numbers who first began working in the earliest years of the professions (e.g., between 1897 and 1926). Radiologists and X-​ray technicians employed in the first half of the twentieth century experienced notably elevated leukemia mortality (no subtype information provided), with increased risks ranging from 6-​to 8.8-​fold among those first joining professional societies (a proxy for first working) before 1940. Significantly elevated risks of incident non-​CLL leukemias were seen in US radiologic technologists who worked 5 or more years before 1950. Incidence of total leukemia was significantly elevated in Chinese X-​ray workers employed during 1950–​1980. Among British radiologists entering the profession after 1921 and US radiologists entering the workforce in 1940 or later, leukemia risks declined notably over time. There were no significant excesses in US radiologic technologists who first worked after 1950 (Linet et al., 2010; Yoshinaga et al., 2004). Accurate estimation of risk per unit of radiation has been limited due to absence of comprehensive historical dose reconstruction and, particularly, absence of recorded individual badge doses in the earliest years when exposures would have been greatest. A recent comprehensive historical reconstruction of individual occupational radiation doses for the US radiologic technologists cohort (Simon et al., 2014) provides a useful basis for estimating risks per unit dose for hematologic malignancies and other cancers, circulatory diseases, and cataracts. Results from earlier studies of nuclear workers can be found elsewhere (Boice, 2006; Polychronakis et al., 2013). Because radiation exposures of nuclear workers are mostly quite low and myeloid neoplasms are rare, pooled studies including large numbers of workers have been the most informative. A 15-​country study examining the relation between estimated cumulative occupational radiation dose and mortality risk of non-​CLL leukemia in a population of 407,391 nuclear workers using a 2-​year lag found an excess risk per Sievert (Sv) of 1.93 (90% CI  =  < 0–​7.14) based on 196 leukemia cases (Cardis et  al., 2005). The mean cumulative dose was estimated to be 19.4 mSv. In a subsequent study of 308,297 monitored nuclear workers from three countries employed for at least 1 year and followed up during the period 1944–​2005, risk of all non-​ CLL leukemias was significantly elevated using a 2-​year lag (ERR per Gy = 2.96; 90% CI = 1.17–​5.21) based on 531 leukemia cases (Leuraud et al., 2015). The mean cumulative occupational radiation dose across the three cohorts was estimated to be 15.9 mGy (range 0.0–​1,217.5 mGy). The excess risk was primarily due to a significant increase of CML (ERR per Gy = 10.45; 90% CI = 4.48–​19.65 based on 100 cases), whereas positive risk estimates for AML (ERR per Gy = 1.29; 90% CI = –​0.82–​4.28 based on 254 cases) and for ALL (ERR per Gy  =  5.80; 90% CI  =  not evaluable lower bound–​ 31.57, based on 30 cases) did not contribute notably to the overall risk. In contrast to the low-​level radiation exposures of most nuclear workers, external radiation exposures were high (mean cumulative dose of 800 mGy) for workers at the Mayak plutonium production facility in the Russian Federation during the early years of operation (1948–​1958). An elevated risk of non-​CLL leukemias (ERR per Gy = 0.99; 95% CI = 0.45–​2.12) was associated with external radiation exposures using a 2-​year lag. Risk from doses received 3–​5  years prior to diagnosis of non-​CLL leukemia was more than 10 times higher than the risk from doses received more than 5 years before diagnosis (Shilnikova et  al., 2003). There was no evidence of an association of plutonium exposure with non-​CLL leukemia in this population. Following the Chernobyl nuclear accident in 1986, clean-​up operations were carried out for years after the accident. The early clean-​up workers (also known as liquidators) experienced higher doses (mean cumulative radiation dose of 92 mGy) than most other nuclear workers (mean cumulative dose of 20 mGy). In a nested case-​control study of leukemia in a cohort of 110,645 Chernobyl clean-​ up workers from Ukraine, a significant linear dose-​response was observed for all leukemias based on 117 cases (ERR per Gy = 2.38; 95% CI = 0.49–​5.87) (Zablotska et al., 2013). Unexpectedly, in this study risks were significantly elevated for CLL (ERR per Gy = 2.58; 95% CI = 0.02–​8.43) as well as for non-​ CLL leukemias (ERR per Gy = 2.21; 95% CI = 0.05–​7.61); 16% of the leukemias diagnosed in this population (18% of CLL and 15%

 729

Leukemias of non-​CLL leukemias) were attributed to radiation exposure. The CLL cases in this study did not have unique features. MDS cases have been described in Chernobyl clean-​up workers, but radiation-​ related risks have not been reported (Gluzman et al., 2015). Although leukemia mortality risk was increased within10 years of first exposure (Darby et al., 1995), epidemiologic studies of uranium miners (Darby et al., 1995; McLaughlin, 2012; Tomasek et al., 1993; Zablotska et  al., 2014) have shown no overall association of cumulative radiation dose with leukemia mortality. Studies of radium dial painters who experienced excess risks of osteosarcomas and cancers of the nasal sinuses have found no clear evidence of excess risk of myeloid neoplasms (Boice, 2006), although in males occupationally exposed to radium, there is some evidence of an excess of leukemia, particularly of the same subtype(s) as seen in patients who received Thorotrast (Stebbings, 1998). Cancer risks have been evaluated in flight personnel. Excess risks of AML were described in Canadian airline pilots (Band et al., 1996) and in Danish cockpit crew flying more than 5000 hours (Gundestrup and Storm, 1999), but pooled analysis of airline crew cohorts from 10 countries found no evidence of elevated myeloid leukemia risk (Hammer et al., 2014). A critical question that is difficult to address in a single epidemiologic study is the risk of non-​CLL leukemia following low-​dose, protracted radiation exposure. In a meta-​analysis addressing this question, Daniels and Schubauer-​Berigan modeled results from 10 studies that were cohort or nested case-​control in design, reported quantitative estimates of exposure, were screened to reduce information overlap, and analyzed data using relative or excess relative risk per unit of radiation exposure (Daniels and Schubauer-​Berigan, 2011). These investigators estimated an excess relative risk at 100 mGy of 0.19 (95% CI = 0.07–​0.32) after adjusting for publication bias. They found no evidence of between-​study variance. The excess relative risk estimate was in agreement with the non-​CLL leukemia risk from the Life Span Study of the atomic bomb survivors.

Military Workers Exposed to Nuclear Weapons Tests and to Depleted Uranium

Military participating in maneuvers during nuclear weapons testing have been evaluated in a series of epidemiologic studies (reviewed in Boice, 2006). An excess of leukemia mortality (based on 10 cases), but not total cancer mortality, was reported among approximately 3000 military participants during a 1957 nuclear test in the United States (Caldwell et al., 1980). Among approximately 70,000 US military personnel who participated in one of five nuclear tests during the 1950s, a nonsignificant increase in risk for leukemia mortality was observed (Institute of Medicne [IOM], 2000). A study of 21,357 UK military participants reported an increased relative risk of non-​CLL leukemia mortality, but noted that this may have reflected a reduced risk in controls (Muirhead et  al., 2003). None of these studies, or others, included estimated radiation doses. Using recently available digital records, Till and colleagues have undertaken dose reconstruction for a planned case-​cohort study of leukemia and male breast cancer in a cohort of 115,000 US military participating in eight nuclear test series. The investigators have reported estimated median radiation doses ranging from 9.5 to 24 mGy, and further study is underway (Till et al., 2014). Based on concerns raised about a possible association of leukemia among military exposed to ammunition reinforced by depleted uranium, Storm and colleagues studied 13,552 men and 460 women deployed to the Balkans during 1992–​2001 and followed up through 2002. These investigators found no excess of leukemia (Storm et al., 2006), but the size of the population may have been too small to detect modest to moderately elevated risks.

Environmental Radiation

Few studies have examined radon or natural background radiation and risk of leukemia. While an ecological study suggested a correlation between indoor radon exposure and myeloid leukemia (Henshaw et  al., 1990), a subsequent comprehensive and critical review concluded that there was little evidence of a link (Laurier et  al., 2001).

729

An intriguing investigation that evaluated approximately 80,000 stable residents residing in underground dwellings in China who experienced about three-​fold higher cumulative radiation levels (6.4 mSv) than the average population worldwide found no evidence of an increase in leukemia (Wei and Sugahara, 2000). Among the limited numbers of studies of radon or of natural background radiation and leukemia, most have examined pediatric leukemia (see Chapter 59). Most of the studies of adult leukemias have been ecologic in design, few have included individual measurements, and those with measurements have been underpowered given the low radiation-​exposure levels (Boice, 2006). Data are also limited on cancer risks associated with environmental exposures to man-​made sources of radiation. A population of approximately 30,000 persons residing in villages next to the Techa River was exposed to chronic external and internal radiation during 1950–​1960 from releases by the Mayak nuclear weapons plutonium production plant in the Russian Federation. The median cumulative red bone marrow dose was 0.2 Gy, but doses ranged up to 2 Gy. In a follow-​up during 1953–​2005, a significant dose–​response relationship was seen for non-​CLL leukemia, with an estimated excess relative risk of 4.9 (95% CI = 1.6–​14.3) based on 70 cases. No excess risk or dose–​response relationship was observed for CLL (Krestinina et al., 2010). Spurred by a report from the United Kingdom of increased risk of leukemia and lymphoma occurring among young persons residing in proximity to nuclear plants, many ecologic studies and a few analytic epidemiologic studies have been conducted. Most of the studies have focused on childhood leukemia (Laurier et  al., 2008). A  large ecologic investigation examined total and specific forms of adult cancers, including leukemia, but found no association (Jablon et  al., 1991). Limitations acknowledged by the authors included lack of radiation dose measurements, absence of information about potential confounders, and the likely underpowered nature of the study, despite including a large population base of more than 900,000 cancer deaths from 1950 through 1984 (Jablon et al., 1991). A borderline significant increase in risk of non-​CLL leukemia and of ALL was observed in a case-​control study of more than 1000 leukemia deaths among persons living in southwest Utah in proximity to the Nevada Test Site (Stevens et al., 1990), with significantly elevated risks for those exposed to fallout under 20 years of age.

Non-​Ionizing Radiation: Extremely Low-​Frequency Magnetic Fields and Radiofrequency Exposures and Biological Effects from Extremely Low-​Frequency Magnetic Fields

Electromagnetic fields are produced by a growing number of sources that are ubiquitous worldwide. Extremely low-​frequency magnetic fields are produced from the generation, transmission, and use of electricity. Microwaves are generated by radio and television transmission, microwave ovens, mobile telephones and base stations, wireless local area networks, and smart meters (SCENIHR, 2015) (see Chapter 15 for more details). The primary known biological effect of electromagnetic fields is tissue heating. Electromagnetic fields generate energy that is proportional to the frequencies emitted (measured in hertz), and the energy they produce is too weak to break chemical bonds or to cause translocations in DNA. To date, laboratory studies have failed to demonstrate consistent, reproducible evidence of carcinogenicity, with the possible exception of a co-​carcinogenic effect of radiofrequency fields and a chemotherapy agent (see Chapter 15). Exposures largely have been studied in residential settings, in which most studies have assessed risks of pediatric leukemia and brain tumors (see Chapters 15, 56, and 59), or in occupational settings.

Residential Exposures to Extremely Low-​Frequency Magnetic Fields

Residential investigations of extremely low-​frequency magnetic field exposures in Nordic countries based on calculated historic exposures have shown no evidence of a significant increase in risk of leukemia in adults in Finland (Verkasalo et  al., 1996)  or Norway (Tynes and Haldorsen, 2003), but a borderline increase in risk of AML and CML

730

730

PART IV:  Cancers by Tissue of Origin

among persons residing in homes exposed to the highest estimated fields in Sweden (Feychting and Ahlbom, 1994). Risk of AML was not increased in adults residing in homes with measured exposures to extremely low-​frequency magnetic field levels in western Washington State (Severson et al., 1988), nor was risk of total leukemia elevated in persons living in close proximity to high power lines in the United Kingdom (Elliott et al., 2013).

Occupational Exposures to Extremely Low-​Frequency Magnetic Fields

Most epidemiologic studies of myeloid leukemia (or brain tumors) in workers considered to have high exposure to extremely low-​frequency magnetic fields (e.g., power linemen, utilities workers, and electronics workers) have used job titles and/​or a job exposure matrix as proxy measures of exposure (see Chapter  16 in this volume). A  few studies that incorporated measurements, as reviewed earlier, reported inconsistent findings for AML (Linet et  al., 2006). A  meta-​analysis published in 1997, which described an overall 40% increase in risk of AML among workers in jobs with high extremely low-​frequency magnetic field exposures (Kheifets et al., 1997), was updated a decade later and reported a lower pooled estimate for AML (pooled relative risk [RR]  =  1.09; 95% CI  =  0.98–​1.21) in this population (Kheifets et al., 2008). Risk for CML was also lower in the more recent meta-​ analysis (RR = 1.11; 95% CI = 0.94–​1.31) compared with the earlier one (RR = 1.24; 95% CI = 0.98–​1.57). Based on these results, Kheifets and colleagues concluded that the lack of a clear pattern of extremely low-​frequency magnetic field exposures and risks for AML, CML, and other leukemia subtypes did not support the hypothesis that these exposures were responsible for the observed excess risks. Subsequent to the 2008 meta-​analysis, an update of a cohort study of Danish utility workers found no evidence of increased risk of total leukemia (Johansen et al., 2007), while a population-​based cohort study in the Netherlands that utilized a job exposure matrix found a dose–​response relationship with estimated low and high exposure to extremely low-​ frequency magnetic fields and AML (Koeman, 2014).

Radiofrequency Exposures

There is little evidence that myeloid neoplasms are increased among people using mobile telephones or living in proximity to base stations (IARC, 2013). There are few epidemiologic studies of workers exposed to radiofrequency fields, and some of the results are difficult to interpret. In general, risks of leukemia overall and myeloid leukemia were not increased (IARC, 2013). An exception was an elevated risk of AML mortality among aviation electronics technicians (RR = 2.60; 95% CI = 1.53–​4.43, based on 23 deaths) (Groves et al., 2002). In this same cohort, a non-​significant increase in AML (RR = 1.87; 95% CI = 0.98–​3.58) was observed among 20,109 US Navy personnel who served on ships during the Korean War and were characterized as having high radiofrequency exposure based on expert assessment.

CHEMICAL EXPOSURES: MANUFACTURING, FARMING, MEDICATIONS Manufacturing Benzene

Benzene has been used for more than a century as a key component in the manufacturing of shoes, leather, rubber goods, paints, dyes, inks, lubricants, detergents, pesticides, and pharmaceuticals, and more recently in the production of styrene, polymers, latexes, hydroquinone, benzene hexachloride, plastics, resins, and insecticides (IARC, 2012). Jobs in crude oil refining and in sea and land transport of crude oil and gasoline also involve exposure to benzene, as do jobs in auto repair and bus garages. Surveys have led to estimates of more than 2.1 million benzene-​exposed manufacturing workers worldwide. Exposure sources to benzene in the general population include motor vehicle exhaust, tobacco smoke, contaminated water and foods, gasoline at pumping stations, and leaking underground gasoline storage tanks.

In 1982 the IARC concluded that there was sufficient evidence linking benzene with leukemia, particularly AML. The updated assessment by IARC noted that cohort studies in multiple industries and different countries demonstrated a dose-​response pattern for AML (IARC, 2012). Myeloid and lymphoid neoplasms, as well as many other types of cancer, also have been described in mice and rats following benzene exposure (IARC, 2012). A systematic review and meta-​analysis of four studies focusing on cumulative exposure to benzene in humans found evidence of a dose–​response pattern, with 3.2-​fold relative risk of AML for benzene exposure levels > 100 ppm-​years, although the trend was not statistically significant (Khalade et al., 2010). Data on AML risk at low levels of benzene exposure have been relatively limited due to small numbers of cases available for study. Among Chinese benzene-​exposed workers with cumulative exposures less than 40 ppm-​years, risks of AML (RR = 1.9; 95% CI = 0.5–​7.0, based on 5 cases) and the combined category of MDS/​AML (RR = 2.7; 95% CI = 0.8–​9.5, based on 7 cases) were non-​significantly increased (Hayes et  al., 1997). In a pooled and updated analysis of three nested case-​control studies carried out among cohorts of petroleum distribution workers with low levels of benzene exposure from Australia, Canada, and the United Kingdom, AML risks rose modestly, but not significantly, with increasing cumulative level of benzene exposure (< 0.348 ppm-​years: odds ratios [OR] = 1.00 [referent], based on 20 cases; 0.348–​2.93 ppm-​ years: OR = 1.04, 95% CI = 0.50–​2.19, based on 19 cases; > 2.93 ppm-​years: OR = 1.39, 95% CI = 0.68–​2.85, based on 21 cases). A similar relationship was observed for duration of exposure and peak exposure, but there was no clear pattern for average or maximum intensity of exposure. Risks were non-​significantly increased (OR ranged from 1.35 to 1.90) in the top quartiles of average and maximum exposure, and increased in those with peak exposure < 3 ppm (OR = 1.50; 95% CI = 0.82–​2.75). Significantly elevated risks of AML were seen among those who had ever been tanker drivers (OR = 2.02; 95% CI = 1.08–​3.28) (Rushton et al., 2014; Schnatter et  al., 2012). MDS/​AML was associated with recent (less than 10 years before diagnosis), but not distant (10 or more years before diagnosis) benzene exposure among Chinese benzene-​ exposed workers (Hayes et  al., 1997). Data from the long-​term follow-​up of a cohort of US Pliofilm workers suggest that the excess risk of leukemia diminished with time since exposure (Rinsky et al., 2002). In addition to AML, benzene also causes hematotoxicity at very low levels in benzene-​exposed workers, even for exposures below 1 ppm in air (Lan et al., 2004). Some (Raaschou-​Nielsen et al., 2016; Talbott et  al., 2011), but not all (Wilkinson et  al., 1999), studies have reported modestly increased risks of AML among community members exposed to gasoline vapors or traffic-​related air pollution and among those who reside in proximity to oil refineries. Efforts are underway to understand the mechanisms underlying benzene-​ associated leukemogenesis by identifying the critical genes and pathways that are involved in inducing genetic, chromosomal, and epigenetic abnormalities and genomic instability in HSCs; altered proliferation and differentiation of the HSCs; and dysregulation of stromal cells (McHale et al., 2012). These effects are likely modulated by benzene-​induced oxidative stress, reduced immunosurveillance, and aryl hydrocarbon dysregulation. Data assessing an association between benzene and MDS are limited. Risks of mortality from MDS were significantly increased among benzene-​exposed (7 cases) compared with unexposed workers (0 cases) among Chinese benzene-​exposed workers followed up during 1972–​ 1999 (Linet et  al., 2015). Cumulative benzene exposure demonstrated a monotonic dose–​ response relationship and significant trend with increasing benzene dose and MDS in the pooled three-​country study of petroleum distribution workers (OR  =  4.33; 95% CI  =  1.31–​14.3 at a cumulative exposure > 2.93 ppm-​years, based on a total of 29 cases with all levels of cumulative exposure) (Schnatter et al., 2012). Increased risks for MDS were observed among workers employed at terminal facilities (OR = 5.04; 95% CI = 1.58–​16.1) and among tanker drivers (OR = 2.16; 95% CI = 0.79–​5.88). Similar, albeit non-​significant or borderline significant, dose–​response patterns were observed for average exposure, maximum exposure, and peak exposures (> 3 ppm) to benzene and MDS.

 731

731

Leukemias Based on relatively few studies, CML has not been consistently linked with benzene exposure (IARC, 2012; Khalade et  al., 2010); however, a meta-​analysis found a moderate increase in risk of CML in studies of benzene workers that commenced follow-​ up after 1970 (OR  =  1.67; 95% CI  =  1.02–​2.74) (Vlaanderen et  al., 2012). The broader category of MPN and benzene exposure has only been studied in the three-​country investigation (Glass et al., 2014). Dose–​ response trends for CML (based on 28 cases) and MPN (excluding CML) (based on 30 cases) were not statistically significant for cumulative, average, or maximum benzene exposure, although risks rose with increasing cumulative dose experienced 2–​20 years before diagnosis. Among workers with cumulative exposure > 2.93 ppm-​years, risks were elevated for CML (OR = 12.6; 95% CI = 1.06–​150) and for MPN excluding CML (OR = 4.22; 95% CI = 0.95–​18.7) (Glass et al., 2014).

Formaldehyde

Concerns have arisen about health effects, including leukemia, of formaldehyde associated with workplace (health-​care, embalming, and manufacturing workers) and general population exposures, the latter from formaldehyde levels in new homes. In follow-​up of 25,619 manufacturing workers in 10 plants during 1966–​2004, risk of myeloid leukemia was associated with peak formaldehyde exposures (Beane Freeman et al., 2009; Hauptmann et al., 2003). Myeloid leukemia risk was non-​ significantly elevated for highest vs lowest (≥ 4 ppm vs. < 2 ppm) peak exposure for the period of follow-​up before 1966 through 2004 (RR = 1.78; 95% CI = 0.87–​3.64, p-​trend = 0.13). Risks were highest before 1980, but trend tests attained statistical significance only in 1990 when sufficient deaths had occurred. Risks were highest in the first 25 years following exposure and declined with continued follow-​ up. This pattern is similar to that observed for some other chemical exposures and AML risk (Linet et al., 2006). Funeral industry workers as well as anatomists and pathologists may also be exposed to formaldehyde. Based on a follow-​up during 1960–​1986 of a cohort of embalmers identified from funeral directors’ associations and licensing boards, a nested case-​control study of leukemia and selected other categories of cancer deaths was carried out. Compared with workers who performed fewer than 500 embalmings, those who performed more than 3068 embalmings experienced an elevated risk (OR = 3.0; 95% CI = 1.0–​9.2), as did those who performed embalmings for more than 34 years (OR = 3.9; 95% CI = 1.2–​12.5). Risk of myeloid leukemia rose significantly with increasing number of embalmings performed, number of years of embalming, estimated lifetime formaldehyde exposure in ppm-​years, and peak formaldehyde levels (Hauptmann et al., 2009). A study of 43 formaldehyde-​exposed and 51 unexposed workers in China demonstrated numerical chromosomal aberrations in myeloid progenitor cells (including chromosome 7 monosomy and chromosome 8 trisomy) of the exposed workers consistent with MDS/​AML; other changes were observed in peripheral blood reflecting bone marrow effects (Zhang et al., 2010). Although meta-​analyses published in 2010 reported conflicting findings on the relationship between formaldehyde and myeloid leukemia (Bachand et  al., 2010; Schwilk et  al., 2010), more recent studies, along with evidence of a biologically plausible mechanism, led an IARC working group to conclude that evidence was sufficient to designate formaldehyde as causal for leukemia, particularly myeloid leukemia (IARC, 2012).

Styrene and Butadiene Rubber Manufacturing

Workers in the styrene and butadiene rubber manufacturing industry have been repeatedly found to have excess mortality of total leukemia, more recently determined to be mostly due to CML, CLL, and, to a lesser extent, myeloid neoplasms (IARC, 2008). Although the most recent follow-​up data show only moderate excess risk of total leukemia, notable leukemia excesses were apparent in workers hired in the 1950s, those who were 20–​29 years since hire, and those who had worked > 10 years. Risks were increased in areas of the plants with higher exposures and in hourly workers, especially those hired in the 1950s, when exposures were higher. There were no measurements available before the 1970s. Average levels of butadiene from the

1970s through much of the 1980s ranged from 8 to 20 mg/​m , while those from the 1990s to the present have generally been < 2 mg/​m3. Dose–​response relationships have been observed for butadiene with CML and with myeloid neoplasms (for cumulative exposures > 425 ppm-​years, with CML: RR = 7.2, 95% CI = 1.1–​47.6; and total myeloid neoplasms including CML: RR = 2.4, 95% CI = 0.9–​6.8) (Delzell et  al., 2006), but estimates are imprecise due to small numbers of workers with specific leukemia subtypes. Evidence of carcinogenicity was considered to be sufficient for leukemia in workers in the styrene and butadiene rubber manufacturing and for butadiene (IARC, 2008). An IARC Working Group reaffirmed this conclusion as part of the comprehensive re-​evaluation published in the IARC monograph 100 (Pt F) (IARC, 2012). 3

Farming, Agricultural, and Related Exposures

As described previously, some studies of farmers and farm workers have shown modest excess risks of AML and virtually all other subtypes of leukemia (risks ranging from 1.1-​to 1.4-​fold), while others have shown no increase in risk of AML (Linet et al., 2006). International variation in risks may reflect differences in agriculture-​related exposures such as pesticides (particularly animal insecticides and herbicides), fertilizers, diesel fuel and exhaust, or infectious agents (Blair and Zahm, 1995). Few earlier studies that reported increased risk of AML among those living on a farm (Sinner et  al., 2005; Wong et  al., 2009) evaluated specific pesticide exposures in relation to AML. In the Agricultural Health Workers cohort, excess risk of leukemia was associated with use of chlordane and heptachlor (Purdue et al., 2007), alachlor (Lee et al., 2004), and the organophosphates fonofos (Mahajan et al., 2006) and diazinon (Beane Freeman et  al., 2005). Myeloid leukemia was increased among 20,000 persons aged 0–19 years residing in Seveso within 10 years after an industrial accident caused contamination of the region with 2,3,7,8-​tetrachlorobibenzo-​p-​dioxin (Pesatori et  al., 1993). Recent review of the evidence for dioxin does not support a strong association with myeloid leukemia (IARC, 2012). Biomarkers are needed that provide information about long-​term exposure to pesticides and that assess their chronic effects. Few studies have evaluated farming or agricultural work and risk of MDS or MPN, and findings, to date, are inconsistent (Anderson et al., 2012).

Medications Cytotoxic Chemotherapy Overview.  Relative risks of

developing t-​MDS/​AML following cytotoxic treatments are substantial (e.g., ≥ 3-​fold increased) and lifetime cumulative risks of t-​MDS/​AML range from < 1% to 10% (Candelaria and Duanes-​Gonzalez, 2015; Leone et al., 2010). Some t-​AML are preceded by MDS, while others are not, and latency periods, although generally fairly short, differ according to the type of cytotoxic chemotherapy. Although data are limited on the changing occurrence of t-​MDS/​AML over calendar time, a 34-​year assessment (1975–​2008) of 426,068 adults treated with chemotherapy for first primary cancers in the population-​based SEER Program demonstrated that there have been substantial changes in t-​AML risks over time that are consistent with known changes in treatment practices (Morton et  al., 2013). That study identified 801 cases of t-​MDS/​AML, a rate 4.7-​fold higher than expected in the general population, with nearly half the cases occurring after breast cancer or NHL. Over the study period, t-​MDS/​AML risks following treatment of NHL rose steadily, but declined following treatment of ovarian cancer and multiple myeloma. t-​MDS/​AML rates were highest during 1975–​ 1978 after treatment of primary breast cancer and Hodgkin lymphoma, then declined during the 1980s, followed by modest increases in the 1990s. Risks for t-​MDS/​AML were highest among those treated with chemotherapy for primary cancers at younger ages, although elevated risks were apparent regardless of age at treatment. Excess absolute risks of t-​MDS/​AML were highest following treatment of Hodgkin lymphoma and multiple myeloma, intermediate following treatment of lung and ovarian cancers and NHL, and lowest after treatment of breast cancer. Combination chemotherapy and radiotherapy was associated with non-​ significantly increased risks of t-​ MDS/​ AML

732

732

PART IV:  Cancers by Tissue of Origin

after treatment of cancers of the lung, breast, and ovary, but not after treatment of lymphoproliferative malignancies. The 2008 WHO classification removed the distinction between alkylating agent-​related and topoisomerase II inhibitor-​related t-​MDS/​AML, but for etiologic purposes the literature on t-​MDS/​AML risks associated with these two categories of chemotherapy is discussed separately.

Alkylating Agents. t-​AML associated with alkylating agents

generally occurs as a result of damage to DNA by methylation of DNA inter-​strand crosslink formation. The main methylating forms of alkylating agents include procarbazine, dacarbazine, and temozolomide (Leone et  al., 2010). Nitrosoureas and procarbazine are associated with a high risks of t-​MDS/​AML. For example, patients with Hodgkin lymphoma treated with MOPP (nitrogen mustard, vincristine, procarbazine, and prednisone) had an absolute risk of developing t-​MDS/​AML of 3.4%, compared with 1.3% for Hodgkin lymphoma patients treated with ABVD (doxorubicin, bleomycin, vinblastine, and dacarbazine), a regimen that does not include procarbazine or nitrogen mustard. Busulfan and melphalan are linked with higher risk of t-​AML than cyclophosphamide. As a result, cyclophosphamide has replaced busulfan and melphalan in several chemotherapy regimens. The pathogenesis of t-​MDS/​AML is frequently characterized by a preleukemic phase, trilineage dysplasia, and cytogenetic abnormalities involving monosomy of chromosome 5 or deletion of 5q and/​or monosomy of chromosome 7 or deletion of 7q. t-​MDS or t-​AML cases with monosomy of chromosome 17 or deletion of 17p, dicentric chromosomes, duplication or amplification of chromosome band 11q23, and other karyotypic abnormalities, but without abnormalities of chromosome 5, often have methylation of the CDKN4B gene promoter and somatic mutations of RUNX1 (Leone et  al., 2010). t-​MDS and t-​AML have been reported subsequent to Hodgkin lymphoma, NHL, multiple myeloma, polycythemia vera, and breast, ovarian, and testicular cancers treated with alkylating agents. Typically, t-​AML occurs 5–​7 years following treatment, and risk is related to cumulative alkylating drug dose.

Topoisomerase Inhibitors. Topoisomerase II inhibitors that

bind to the enzyme/​DNA complex at the strand cleavage stage of the topoisomerase reaction have been linked with elevated risk of t-​AML (Nitiss, 2009). As a result of blockage of the enzyme reaction, topoisomerase II inhibitors may leave DNA with a permanent DNA strand break. The resulting treatment-​ related leukemias may be myeloid (with the partner gene of MLL being chromosome 9)  or lymphoid (with the partner gene being chromosome 4) in lineage, and studies of gene expression profiles suggest that the leukemia originates within an undifferentiated hematopoietic stem cell. Both the antineoplastic effect and the leukemogenic effect of topoisomerase II inhibitors are due to chromosome translocation. Drugs that interact with topoisomerase II include epipodophyllotoxins that intercalate (e.g., doxorubicin) and those that do not intercalate (e.g., etoposide, teniposide). With the more recent widespread clinical use of topoisomerase II inhibitors, including anthracyclines (e.g., doxorubicin, daunarubicin) and anthracenedione (e.g., mitoxantrone), elevated risks of t-​AML have been observed, and risks appear to be higher among those treated at younger ages. The resultant t-​AML is generally not preceded by MDS, develops after a shorter latency period (median latency typically 2–​3 years), and has different cytogenetic abnormalities than alkylating agent-​associated t-​AML. Another contrast with alkylating agent-​associated t-​AML is the association of topoisomerase II inhibitors with balanced translocations involving the MLL gene on chromosome band 11q23 (Cowell et  al., 2012). Most MLL rearrangements are reciprocal translocations with many different partner genes including t(9;11) or t(4:11) but also include internal duplications, deletions, or inversions. Questions about the relationship between cumulative dose and frequency of treatment with epipodophyllotoxins with risk of t-​AML remain, as some previous studies were limited by the use of chemotherapeutic regimens that include both topoisomerase II inhibitors and alkylating agents.

Other Chemotherapy Agents.  Platinum agents (e.g., cis-

platin, carboplatin, oxaliplatin) are similar to alkylating agents in their mechanisms of action and resistance, with the exception that they do not alkylate DNA but rather form covalent metal adducts with DNA (Chabner et  al., 2011). Increasing doses of platinum-​ based chemotherapy for ovarian (Travis et al., 1999) and testicular cancers (Howard et  al., 2008)  have been quantitatively associated with increasing risks for t-​AML. A  10-​fold higher risk of t-​MDS/​ AML has been observed in breast cancer patients treated with mitoxantrone and methotrexate (an antimetabolite) or methotrexate and mitomycin C (an antibiotic that inhibits DNA synthesis) (Saso et al., 2000). Antimetabolites, including methotrexate, azathioprine, 6-​thio­guanine, and fludarabine (a nucleoside analogue), are used for some cancer treatments, as immunosuppressants in autoimmune diseases, or in recipients of organ transplants, the latter often including combination treatment with cyclosporine A and steroids. The antimetabolites share structural similarities with nucleotides and can be incorporated into DNA or RNA, thus causing inhibition of cell proliferation. Increased risks of AML have been reported in patients treated with azathioprine after organ transplantation or for autoimmune disease (Yenson et al., 2008). Leukemia risks may be higher in patients with low thiopurine S-​methyltransferase activity, and mechanisms may include aberrant mismatch repair and microsatellite instability (Karran, 2006). t-​MDS/​AML is increased among patients treated with nucleoside analogs (e.g., fludarabine, cladribine), alone or in combination with other agents (e.g., with chlorambucil or cyclophosphamide) often used to treat patients with CLL and other lymphoproliferative neoplasms (Leone et al., 2010). t-​MDS/​AML has also been associated with pretransplantation chemotherapy (e.g., mechlorethamine, chlorambucil) and/​or transplantation conditioning treatments that include total body irradiation, particularly at doses > 12 Gy (Metayer et al., 2003).

Transformation of MPN to t-​AML: The Role of Single versus Multiple Treatments or Other Factors Is Unclear.  MPN, including polycythemia vera, essential throm-

bocythemia, and primary myelofibrosis, can “transform” to AML (Abdulkarim et  al., 2009). There is variability in the risks of developing AML after different forms of MPN (Barbui, 2004; Mesa et al., 2005; Passamonti et  al., 2008). Mechanisms involved in leukemic transformation are not well understood, particularly because of the rarity of the event, the difficulty of disentangling the role of one or more treatments, and the likelihood that the causes of transformation are multifactorial. For example, in a nationwide Swedish cohort of 11,039 MPN patients diagnosed during 1958–​2005, 292 patients developed AML (n = 271) or MDS (n = 21) (Bjorkholm et al., 2011). A significantly increased risk of developing MDS/​AML was observed among MPN patients who received ≥ 1,000 MBq 32P, but it was unclear whether those receiving this high dose of 32P had also received alkylating agents and/​or hydroxyurea.

Other Medications Chloramphenicol.  Use of chloramphenicol has long been

linked with bone marrow suppression and risk of aplastic anemia (Fraumeni, 1967). Some patients with this hematological disorder have developed AML (Cohen and Creger, 1967), although risk of AML following use of chloramphenicol is unclear because rigorous epidemiological data are lacking (Fraumeni, 1967). A  dose–​ response relationship was observed between use of chloramphenicol and risk of childhood AML and ALL in Shanghai (Shu et al., 1987), but studies in adults using medical or pharmacy records have shown no association (Doody et al., 1996) or a non-​significant excess risk of acute leukemia (Traversa et  al., 1998). Use of topical chloramphenicol was not significantly associated with risk of acute leukemia or AML, based on data abstracted from general practitioner medical records in a large case-​control study in the United Kingdom. In this study, risk was non-​significantly increased if topical chloramphenicol was used three or more times, but there was no significant dose–​ response relationship (Smith et al., 2000).

 73

Leukemias

Non-​Steroidal Anti-​Inflammatory Drugs.  Clinical reports

described leukemia following treatment with phenylbutazone, but high-​quality epidemiologic studies are limited. In a prospective study of hematopoietic neoplasms in a health maintenance organization for which pharmacy records were the basis of information, there was no evidence of a significant association or a relationship with duration or cumulative amount of use of phenylbutazone and myelocytic leukemia (Friedman, 1982). Limited data have linked use of non-​steroidal anti-​inflammatory drugs (NSAIDs) with risk of AML (Kasum et al., 2003), but findings generally show a modestly reduced risk. A case-​control study in Los Angeles reported that use of NSAIDs was associated with a decreased risk of AML (Pogoda et  al., 2005). Based on pharmacy records from the National Health System in the population-​based case-​control study in the Province of Rome, use of very high doses of NSAIDs was linked with a modest, non-​significant reduction in risk of acute leukemia (Traversa et al., 1998). In case-​control, interview-​ based studies in Buffalo (Weiss et al., 2006) and Minnesota (Ross et al., 2011), aspirin was associated with a decreased risk of AML, and acetaminophen was linked with an increased risk. NSAIDs are characterized by anti-​inflammatory, anti-​pyretic, and analgesic properties and target the COX enzymes, thus inhibiting prostaglandin synthesis, oxidative cell damage, angiogenesis, and potentially signal transduction pathways that are believed to influence risk of malignancy. The specific mechanism(s) that may affect risk of hematopoietic malignancies after NSAID exposure is unknown (Bernard et al., 2008).

LIFESTYLE FACTORS Smoking Cigarette smokers are exposed to more than 70 chemicals that have been linked with cancer, including known leukemogens (e.g., benzene, formaldehyde, and polonium-​210) (CDC Surgeon General’s Report 2010). Since the early 1990s, a substantial number of studies have reported associations of cigarette smoking with AML. A recent meta-​analysis of 23 studies that included 7746 AML cases reported significantly elevated risks for current smokers (RR  =  1.40; 95% CI  =  1.22–​1.60) and ever smokers (RR  =  1.25; 95% CI  =  1.15–​ 1.36) (Fircanis et  al., 2014). Risks were notably higher for those who had smoked for ≥ 20 versus < 20 years and rose significantly with increasing number of cigarettes smoked per day and increasing number of pack-​years smoked. A growing number of studies have evaluated cigarette smoking and risk of MDS. A meta-​analysis of 14 studies that assessed 2588 MDS cases found significantly elevated risks among current (RR  =  1.81; 95% CI  =  1.24–​2.66) and ever (RR = 1.45; 95% CI = 1.25–​1.68) smokers, along with higher risks among those who smoked for ≥ 20 versus < 20  years, those who smoked ≥ 20 versus < 20 cigarettes per day, and those with higher number of pack-​years of smoking (Tong et  al., 2013). Combining AML and MDS in a meta-​analysis of 25 studies with 8074 cases of myeloid neoplasms that overlapped with the previously described meta-​analyses, investigators found similar results for MDS/​AML as for AML alone (Wang et al., 2015). Risks for MDS/​AML were significantly increased for current (RR  =  1.45; 95% CI  =  1.30–​1.62) and ever (RR = 1.23; 95% CI = 1.15–​1.32) smokers and were higher for those who smoked ≥ 20 versus < 20  years, ≥ 20 versus < 20 cigarettes per day, and a greater number of pack-​years. The mechanisms by which cigarette smoking increases risk of MDS/​AML are unknown. However, smoking has been shown to decrease circulating CD34 progenitor cells in healthy persons (Ludwig et al., 2010), as well as to decrease the number of erythrocyte and granulocyte colony-​forming units, to upregulate toll-​like receptor expression, to increase NF-​kB, AKT, and ERK expression, and to induce IL-​8 and TGF-​b1 production (Zhou et al., 2011). There have been fewer studies of cigarette smoking and risk of CML with some (Kinlen and Rogot, 1988; Kabat et al., 2013; Musselman et al., 2013), but not others (Bjork et al., 2001; Fernberg et al., 2007;

733

Richardson et  al., 2008; Strom et  al., 2009), finding an association. More recently, investigators examined the relationship of smoking with subtypes of MPN and found an association with polycythemia vera, but not essential thrombocythemia (Leal et al., 2014). Leal and colleagues postulated etiologic differences, with smoking-​related carcinogenic pathways linked with polycythemia vera, whereas obesity-​ related inflammatory pathways (see later in this chapter) appeared to be more important for essential thrombocythemia. In a population-​based case-​control study of myeloid leukemia, the elevated risk of AML associated with cigarette smoking declined with increasing number of years since quitting, while the risk reduction was more gradual for CML (Musselman et al., 2013).

Diet Overall, there have been a few exploratory studies assessing the possible role of diet in AML, and even fewer in MDS and MPN. Two case-​control studies (Li et al., 2006; Yamamura et al., 2013) and one cohort study (Ma et al., 2010) found that consumption of beef or meat, in general, increased risk of AML (Li et  al., 2006; Ma et  al., 2010; Yamamura et al., 2013), but there was no evidence of an association with level of doneness or meat mutagens (Ma et al., 2010). Findings are inconsistent, however, as to whether those who consume high levels of vegetables or fruits experience reduced risks of AML (Ma et al., 2010; Yamamura et al., 2013). Notably, higher dietary intake of isoflavones was associated with reduced risk of MDS in a hospital-​based case contol study in China (Liu, 2015).

Alcohol Some types of alcoholic drinks contain substances that have anti-​ carcinogenic properties, particularly polyphenols (Arranz et al., 2012). Studies evaluating alcohol consumption and AML risk have generally not found a relationship (Blackwelder et al., 1980; Carstensen et al., 1990; Hinds et al., 1980; Williams and Horm, 1977), demonstrated a modest inverse association (Brown et  al., 1992), or showed divergent results, with reduced risks for light or moderate beer intake and increased risks for moderate to heavy red wine consumption (Rauscher et al., 2004a). Hospital-​ based, case-​ control studies assessing alcohol consumption and MDS have shown conflicting results (Ido et  al., 1996; Liu et  al., 2016), while a large cohort study found no association (Ma et al., 2010). A meta-​analysis including 745 cases of MDS from five studies found a non-​significant increase in MDS (RR = 1.31; 95% CI = 0.79–​2.18) with higher versus lower alcohol consumption (Du et al., 2010). No association was observed between alcohol consumption and MPN in two cohort studies of women (Kroll et al., 2012; Leal et al., 2014).

Body Mass Index A meta-​ analysis including seven studies found an association of increasing body mass index (BMI) with increased risk of AML, which was estimated as a 3.1% increase in risk of AML per kg/​m2 (Larson and Wolk, 2008). Five studies included in a meta-​analysis of CML revealed an increase in relative risk among obese, but not overweight persons, but there was no evidence of a linear trend (Castillo et al., 2012). A large cohort study found increasing risk of MDS with increasing level of BMI (Ma et al., 2009). A case-​control study described obesity (both in early life and later in life) to be strongly associated with risk of AML (Poynter et al., 2016). The Iowa Women’s cohort study found that increased BMI was associated with increased risk of essential thrombocythemia, but not polycythemia vera (Leal et  al., 2014). Although the biological mechanism(s) that may be responsible for risk associated with BMI have not been clearly identified, possibilities include alterations in levels of insulin-​like growth factor-​1 (increased in response to the insulin resistance associated with obesity and demonstrates mitogenic activity in myeloid and lymphoid leukemia cell lines), leptin levels (increased in obesity and affects proliferation and differentiation of myeloid leukemia cell lines), or other hormones (e.g., adiponectin, insulin, or sex steroids)

734

734

PART IV:  Cancers by Tissue of Origin

or changes in the bone marrow environment (Karmali et  al., 2015; Poynter et al., 2016).

MEDICAL CONDITONS Autoimmune Disorders

Hair Dye Use Hair dyes, particularly in past decades, contained some carcinogenic chemicals, but the chemical constituents have changed over time (IARC, 2010). Hair dye use, particularly dark hair dyes for longer duration, has been weakly linked with adult AML in a large case-​control study (Rauscher et al., 2004b), but not in a large cohort study (Altekruse et al., 1999). Most studies have found no relationship for total or specific types of leukemias (Correa et  al., 2000; IARC, 2010; Zhang et  al., 2012). Weak associations have been reported in some case-​control studies of MDS (Ido et  al., 1996), CML (Cantor et  al., 1988), and essential thrombocythemia (Mele et al., 1996) based on small numbers of exposed cases and limited exposure data.

INFECTIOUS AGENTS Few studies have assessed the potential role of infectious agents in the etiology of myeloid neoplasms, and findings have not been consistent (Doody et al., 1992; Zheng et al., 1993). An intriguing finding was a positive association between early age at onset of childhood viral infections and risk of AML (Cooper et  al., 1996). Excess risk of myeloid leukemia has also been reported in patients with acquired immunodeficiency syndrome (AIDS) (Shiels and Engels, 2012), but the reasons are unknown. In a large study in Sweden that linked hospital discharge data with population-​based cancer registry data, the investigators found modest 30% statistically significant excess risks of adult AML and MDS associated with a history of any prior infectious disease. These findings were based on elevated risks of AML in relation to prior pneumonia, tuberculosis, intestinal infections, septicemia, hepatitis C, pyelonephritis, sinusitis, nasopharyngitis, upper respiratory infections, meningitis, cytomegalovirus, and cellulitis, with individual risk estimates ranging from 1.2 to 5.6, but the majority were < 2-​fold elevated (Kristinsson et  al., 2011). The elevated risks were still apparent when infections occurring less than 3 years before diagnosis were excluded. Increased risk of MDS was associated with prior history of pneumonia and cellulitis, and the elevated risks also were apparent when a latency of 3 or more years was considered. The Swedish linked registry analysis was based on infections treated on an inpatient basis, without specific validation of the diagnoses of infection, and without available treatment data and other potentially confounding information. These associations need to be confirmed in other populations. Given the indolent nature of many MDS cases and the often lengthy period between onset and clinical diagnosis, future studies should consider longer latency periods. Identification and validation of relevant infectious agents in animal and human studies would be helpful, particularly if sufficient numbers of MDS cases could be evaluated in a large cohort pooling project to determine the likelihood that the associations are etiological in nature.

REPRODUCTIVE FACTORS Data are limited and inconsistent on the relationship between exogenous hormone use and myeloid neoplasms. A  small population-​ based case-​control study of acute leukemia in the Province of Rome described an increased risk of AML among women who took oral contraceptives based on drugs received through the National Health Service in Italy at least 12 months before diagnosis (Traversa et al., 1998). In contrast, a small case-​control study in Minnesota reported a protective effect between longer duration of oral contraceptive use and adult acute leukemia (Poynter et al., 2013). The Minnesota study found little evidence for an association between other reproductive factors or use of hormone replacement therapy and risk of myeloid leukemia overall or, specifically, AML or CML.

Linkage of medical record or medical insurance information with population-​ based cancer registry incidence data has been used to assess the relationship between autoimmune disorders and myeloid neoplasms. Among 9468 AML cases, Kristinsson and colleagues considered 25 autoimmune disorders combined and found a 1.7-​fold increased risk of AML based on 359 patients with AML and any prior autoimmune disease (Kristinsson et  al., 2011). The excess risk was somewhat lower, albeit still significantly 1.4-​fold increased, if autoimmune disorders diagnosed within 3 years of diagnosis of AML were excluded. Combining the same 25 autoimmune disorders, Kristinsson and colleagues reported that risk of MDS was increased 2.1-​fold based on 133 patients with any autoimmune disorder among 1662 MDS cases; excluding patients whose autoimmune disorders were diagnosed within 3 years of MDS diagnosis, a lower 1.7-​fold excess risk was apparent (Kristinsson et  al., 2011). Using a subset of the same Swedish linked registry data, Hemminki and colleagues reported a similar significantly elevated risk of AML associated with a combined grouping of 33 autoimmune diseases (standardized incidence ratio [SIR] = 1.85), and risks of similar magnitude for CML (SIR = 1.68), other myeloid leukemia (SIR = 2.20), but somewhat lower risk for myelofibrosis (SIR = 1.36) (Hemminki et al., 2013). Based on an evaluation of 27 autoimmune diseases and risk of myeloid neoplasms among patients aged 66–​99 years old, results from a linkage of the US population-​based SEER cancer registries with Medicare data revealed increased risks of AML (OR = 1.29, based on 973 cases with any autoimmune disorder among 7824 individuals with AML) and MDS (OR = 1.50, based on 574 cases with any autoimmune disorder among 2471 individuals with MDS) (Anderson et al., 2009). Based on few cases, AML was associated with rheumatoid arthritis, systemic lupus erythematosus, polymyalgia rheumatic, autoimmune hemolytic anemia, systemic vasculitis, ulcerative colitis, and pernicious anemia, while MDS was associated with ulcerative colitis and pernicious anemia. Risks were not increased for CML or MPN. Anderson and colleagues suggested that the associations could be due to medications used to treat the autoimmune disorders (including alkylating agents and azathiaprine), shared genetic predispositions between autoimmune disorders and myeloid neoplasms, or involvement of the bone marrow by the autoimmune disorders (Anderson et  al., 2009). Several case-​control studies have found inconsistent associations of rheumatoid arthritis and myeloid neoplasms (Cartwright et al., 1988; Cooper et al., 1996; Severson et al., 1989; Zheng et al., 1993).

Organ Transplant Patients Solid organ transplant recipients were found to have significantly elevated risk of all myeloid neoplasms (SIR  =  4.6), AML (SIR  =  2.7), MDS (SIR = 2.7), CML (SIR = 2.3), and polycythemia vera (SIR = 7.2) (Morton et  al., 2014). Risks were highest among the patients who were youngest at the time of transplantation and declined notably with increasing age at transplantation for all disease subtypes. There was some variability in risk by time since transplantation, although the patterns generally were not consistent, except for a high rate in the first year for polycythemia vera that the investigators considered to be potentially spurious. Risks varied somewhat by type of organ transplant, with particularly high risk for lung recipients, which was consistent with increased AML and MDS risk associated with use of azathioprine and other antimetabolites (Yenson et al., 2008). However, the relatively small numbers of any given type of transplant and subsequent myeloid neoplasm made it difficult to estimate risks precisely. The finding of increased risks of myeloid neoplasms following organ transplantation, which were similar to the elevated risk of myeloid neoplasms among patients with human immunodeficiency virus (HIV)/​ AIDS and those with autoimmune diseases, suggests that immune dysfunction may be important in the etiology of myeloid neoplasms. Since use of azathioprine in solid organ recipients has declined over time due to the availability of alternate agents, it is possible that risks

 735

Leukemias of myeloid neoplasms may decrease with follow-​up of more recent organ recipients.

Allergic Disorders Overall, the evidence is weak for a relationship of allergic disorders and myeloid neoplasms. Some (Severson et  al., 1989), but not all (Cartwright et al., 1988; Cooper et al., 1996; Doody et al., 1992; Zheng et  al., 1993), case-​control studies reported a reduced risk of all or specific allergic conditions and AML. A US cohort study found little evidence of a protective effect, but included small numbers of all types of leukemia cases combined (Mills et al., 1992). Another US cohort study described no association of all or specific allergic conditions with myeloid neoplasms (Shadman et  al., 2013). A  Swedish cohort investigation reported that those with asthma and those with hives had an increased incidence of non-​CLL leukemia based on small numbers of exposed cases (Soderberg et al., 2004).

GENETIC RISK FACTORS Diverse lines of evidence provide strong support for a heritable component in the development of leukemia in general, and CLL in particular, but relatively few specific genes have been identified. Although individuals with a family history of leukemia have an elevated risk of developing leukemia themselves, CLL is well-​known to cluster in families, but familial occurrence of myeloid neoplasms is rare. Stronger evidence for the genetic basis of leukemia derives from studies of patients with rare genetic syndromes, but these cases account for only a small proportion of leukemias. Substantial advances in molecular techniques in the last decade have enabled increasingly broad investigations of the potential role of common genetic variation in leukemogenesis. Such studies hold great promise for further advancing our understanding of genetic susceptibility to leukemia, particularly as they consider specific leukemia subtypes and account for potential interactions of genetic susceptibility with other leukemia risk factors.

Familial Aggregation and Rare Genetic Syndromes The few studies of familial AML pedigrees, which generally include individuals diagnosed across a broad age range and with different types of MDS/​ AML, have identified several predisposing rare germline mutations with varying penetrance. The most established of these are in RUNX1, CEBPA, and GATA2, transcription factors that are thought to regulate myeloid differentiation (Babushok et  al., 2015; Churpek et al., 2013; Hahn et al., 2011; Nickels et al., 2013; Owen et al., 2008; Pan et al., 2015; Smith et al., 2004; Song et al., 1999). Familial platelet disorder with propensity to develop AML is an autosomal dominant syndrome that occurs due to mutations in RUNX1 (Mangan and Speck, 2011). Key observations among individuals with this disorder include variable rates of hematologic abnormalities, such as thrombocytopenia and abnormal platelet aggregation, and occurrence of a range of MDS/​ AML subtypes. Families with GATA2 mutations also frequently demonstrate hematologic abnormalities, including MonoMac syndrome, which is associated with reduced numbers of monocytes, natural killer cells, and B cells and increased risk of infections; and Emberger syndrome, which is characterized by lymphedema. In these patients, risk of MDS/​AML is strongly elevated, but penetrance is incomplete. In contrast, nearly all individuals in families with mutations in CEBPA eventually develop AML, most commonly AML with minimal maturation or AML with maturation, without preceding hematologic abnormalities. Because of the rare nature of pure familial MDS/​AML pedigrees, insight into the genetic basis of myeloid neoplasms is more likely to derive from studies of individuals with inherited genetic syndromes with diverse associated phenotypes, or from investigation of common genetic variants for specific myeloid neoplasms, as discussed further in the following. However, several new susceptibility genes have been proposed recently based on familial aggregation. TGM6

735

was identified as one such AML-​susceptibility gene based on linkage analysis and next-​generation sequencing of a multigenerational family with 11 AML cases inherited in an autosomal dominant fashion, without a clear pattern of preceding hematologic abnormalities (Pan et al., 2015). In another study based on four families, germline duplication of ATG2B and GSKIP on 14q32.2 was recently reported in association with megakaryopoiesis, with frequent progression to AML but also the occurrence of CMML and CML (Saliba et al., 2015). That study represents one of the few familial studies of myeloid neoplasms other than MDS/​AML. Data from large-​scale, multigenerational cancer registries show strong familial aggregation for MPN (Landgren et al., 2008) but not for CML (Bjorkholm et al., 2013), supporting the need for further familial studies of myeloid neoplasms. A number of rare, inherited genetic syndromes are characterized by elevated risk for developing leukemia, although they have a diverse set of presenting features. The best studied of these are the bone marrow failure syndromes, including Fanconi anemia, dyskeratosis congenita, congenital neutropenia, and Shwachman-​Diamond syndrome, which increase risk for both AML and MDS (Alter et  al., 2010; Babushok et al., 2015; Rommens et al., 2008). The magnitude of the risks can be difficult to quantify precisely because most studies include small numbers of patients. However, it is clear that risks of MDS/​AML are striking in these patients. For example, in a cohort of patients with dyskeratosis congenita, risk for AML was increased approximately 200-​ fold, and risk for MDS was increased over 2000-​ fold (Alter et  al., 2009). The mechanisms underlying these elevated risks are incompletely understood but are thought to relate to abnormal telomere maintenance, defective DNA repair, and abnormal hematopoietic differentiation and proliferation. Recent progress in understanding the genetic basis of these disorders holds promise for elucidating the mechanisms that confer such striking leukemia risks (Khincha and Savage, 2013). For example, the number of known susceptibility genes for dyskeratosis congenita has moved substantially beyond TERT and TERC to include a range of other genes (Federman and Sakamoto, 2005; Savage et al., 2008). Other hereditary conditions associated with increased risk of MDS/​AML include Li-​Fraumeni syndrome, ataxia-​ telangiectasia, and Bloom syndrome (Arora et  al., 2014; Ballinger et al., 2015; Olsen et al., 2001; Varley et al., 1997). Although the precise mechanism of leukemogenesis is not known, it is likely related to underlying genomic instability and defects in DNA repair. Familial monosomy 7 with inherited partial or complete monosomy 7 (a common cytogenetic abnormality in MDS/​AML) is associated with increased risk of developing MDS/​AML and is also associated with neurologic abnormalities, such as cerebellar ataxia or atrophy (Gaitonde et al., 2010). As described further in Chapter 59, children with Down syndrome (trisomy 21) also have very high risk of developing AML (Bjørge et  al., 2008; Xavier et  al., 2010). Whereas the genetic syndromes described in the preceding generally increase risk for MDS/​AML, neurofibromatosis 1 is associated with elevated risks for juvenile myelomonocytic leukemia (JMML), AML, and CML (Rosenbaum and Wimmer, 2014; Seminog and Goldacre, 2013), and Noonan syndrome is associated with JMML (Strullu et al., 2014). The associations observed with both of these syndromes may be related to RAS activation. Twin studies have provided insight into the genetic basis and natural history of leukemia, most notably demonstrating that concordant occurrences of leukemia in monozygotic twin pairs have a common clonal origin (Greaves et  al., 2003; Greaves and Wiemels, 2003). Twins with concordant leukemia most frequently develop ALL (Alpar et al., 2015; Couto et al., 2005), although some cases of AML have been reported (Debeljak et  al., 2013; Ng et  al., 1999; Udayakumar et al., 2014). However, these studies predominantly focused on childhood rather than adult leukemias.

Common Genetic Variation Early studies of common genetic variation in germline DNA and leukemia risk yielded modest insights into leukemogenesis. Most studies focused on genes related to DNA repair, carcinogen metabolism, and folate metabolism because of the importance of ionizing radiation

736

736

PART IV:  Cancers by Tissue of Origin

and chemical exposures in the etiology of leukemia (Bolufer et  al., 2006; Vijayakrishnan and Houlston, 2010). However, results from these studies were often inconsistent, possibly due to limited statistical power, broad case definitions, and investigation of relatively few genetic variants. With the advent of microarray technology for identifying large numbers of common genetic variants, agnostic interrogation of susceptibility variants across the entire genome is now possible. Although a number of genome-​wide association studies (GWAS) have been performed for childhood ALL (described in Chapter  59), only four GWAS have investigated susceptibility loci for any type of myeloid neoplasm. A study of 671 CML cases of Korean and European descent identified CML susceptibility loci at 6q25.1 and 17p11.1 (Kim et al., 2011). Two GWAS with over 3500 cases of myeloproliferative neoplasms identified a number of susceptibility loci, including 3q26.2, 3p24.2, 5p15.33, 6q23.3, and 9p24.1 (Kilpivaara et al., 2009; Tapper et al., 2015). The associations for some loci differed by JAK2 mutation status, supporting the importance of uniform case definitions in such studies. Finally, a study of 150 cases of t-​AML of European descent identified a locus at 17q12, albeit not at genome-​wide significance, that appeared to have a greater effect when the case population was restricted to t-​AML cases with abnormalities in chromosomes 5 and/​ or 7, which is highly correlated with antecedent alkylating agent exposure (Knight et  al., 2009). That study supports the intriguing possibility that certain susceptibility variants may only confer risk in the presence of a particular leukemogenic exposure. To maximize the discovery potential for identifying germline susceptibility to myeloid neoplasms, future investigations should account for heterogeneity in both exposures and disease subtypes. Attention is increasingly turning to the potential role of common genetic variation in relation to therapeutic response and disease prognosis. Results of initial studies that have focused on candidate genes that may confer drug resistance (e.g., drug metabolizing enzymes) have been conflicting, and expanded pharmacogenetic studies and GWAS are underway (Choi et al., 2013; Drenberg et al., 2016). For myeloid neoplasms, the investigation of germline variants related to therapeutic response and disease prognosis is in its infancy, in stark contrast to the extensive understanding of the prognostic as well as diagnostic importance of certain somatic changes. Additional research is needed to understand the mechanisms by which putative loci may contribute to leukemia development or prognosis, and to identify other associated genetic loci. Important directions for future research will be to compare risks for different myeloid neoplasms, and to consider whether underlying susceptibility variants may interact with other leukemia risk factors (Knight et al., 2009).

OPPORTUNITIES FOR PREVENTION Advances in molecular biology, genetics, and mechanistic aspects of pathogenesis of myeloid neoplasms have been notable, but progress has been slower in identifying etiologic factors, which limits opportunities for prevention. Nevertheless, efforts should be made to reduce medical, occupational, and environmental exposures to radiation, and to balance risks versus benefits of cytotoxic and other medications and chemicals implicated in the etiology of myeloid neoplasms. Radiologists, radiologic technologists, medical physicists, and manufacturers of X-​ray equipment should continuously seek to optimize diagnostic radiologic procedures to provide clinically important information at the lowest possible doses achievable and to provide estimates of radiation exposure from such procedures to physicians and patients. Reducing radiation exposure from diagnostic procedures is a shared responsibility of the referring medical practitioner and the radiologist, with increasing input from professional societies (e.g., the American College of Radiology) in the form of evidence-​and/​or consensus-​based guidelines produced by panels of experts with respect to the most appropriate examinations or modalities and the frequency of examinations recommended for clinical evaluations (American College of Radiology, 2015; Einstein et al., 2014). Unnecessary medical diagnostic radiologic procedures should be avoided. Radiation

and medical oncologists should continue to weigh risks of leukemia, second cancers, and other serious adverse sequelae from therapeutic radiation and cytotoxic therapy against the clinical benefits. Newer radiation therapy techniques and modalities have been designed to limit radiation scatter to normal tissue from therapeutic radiation. Research is needed to ascertain whether t-​MDS/​AML and other second cancer risks are lower with newer radiotherapy techniques or modalities than with conventional radiotherapy. Similarly, research is needed to better understand the risks associated with newer cytotoxic agents and other systemic therapies (e.g., taxanes, immunotherapy). Radiation workers should utilize all feasible radiation-​protective safety measures. Medical radiation workers should actively seek maximal use of all personal protective measures (including full coverage lead aprons, thyroid shields, and lead goggles), other protective measures (room shields), maintenance of the longest distance feasible from radiation sources that is still consistent with good patient care, and use of all badge monitoring devices as recommended by radiation safety officers. Radiation safety officers, radiologists, and hospital administration should select X-​ray equipment that provides the best visualization for the lowest possible exposures to patients and radiation workers, the greatest flexibility to incorporate adjustments for body size, and meaningful estimates of radiation exposure associated with radiological examinations. Nuclear workers should similarly use all recommended badge monitoring devices, and radiation safety officers should seek to reduce all unnecessary exposures. Workers in industries that utilize benzene and formaldehyde in manufacturing, transport, or storage and those in the rubber-​manufacturing industry should regularly employ all personal protective (respirators, clothing, and use of equipment to limit dermal exposures) and workplace safety measures (improve ventilation, minimize levels of relevant chemicals to the lowest achievable levels), wear monitoring devices, participate in formal safety training, and undergo regular evaluation to minimize their exposures to these chemicals. Manufacturers should reduce the amount of benzene in gasoline, paints, solvents, and other sources. Formaldehyde levels should be reduced and alternative, less toxic agents should be sought. Although further clarification is needed of the relation between specific pesticides and myeloid neoplasms, protective measures should be utilized in jobs and industries using such agents in an effort to limit exposures. For the general population, the most important personal measure to reduce risk of myeloid neoplasms is to avoid use of tobacco, and gasoline service stations should continue to upgrade devices to minimize exposure to benzene by those who pump gasoline. Ideally, individuals should consider using vehicles with sources of “fuel” that do not contain benzene. While the association of myeloid neoplasm risk with BMI remains to be elucidated, reducing overweight or obesity to normal weight has numerous other health benefits, including prevention of non-​hematologic cancers.

FUTURE DIRECTIONS AML has long been reported to population-​based cancer registries, but other major categories and subtypes of myeloid neoplasms have not been included until recently in some countries, or at all in others. With the rapid evolution of molecular markers, it is important for clinicians to incorporate the most recent guidelines into diagnostic evaluations, both for precision in disease classification and selection of appropriate treatment. Incomplete disease registration or inadequate diagnostic evaluation of myeloid neoplasms in the general population is a major obstacle for characterizing patterns and trends. Efforts are needed worldwide to completely ascertain, correctly diagnose, and employ state-​of-​the-​art molecular characterization of subtypes of AML, MDS, MPN, and MDS/​MPN to facilitate progress in identifying etiologic factors, preventive measures, and appropriate therapies. Epidemiologic studies have focused on AML and, to a lesser extent, on MDS, t-​MDS/​AML, and CML, but there has been little evaluation of risk factors by subtype of AML, MDS, MPN, or MDS/​MPN. Gaps in understanding the pathogenesis of myeloid neoplasms remain, even for well-​established risk factors. Epidemiologic studies of radiation

 73

Leukemias and myeloid neoplasms have generally not evaluated subtypes of AML, MDS, MPN, or MDS/​MPN. The intriguing association of radiation dose and risk of MDS in the atomic bomb survivors requires further follow-​ up in that population and comprehensive dose–​response assessment in other radiation-​exposed populations. The association of low-​level benzene exposure with MDS and absence of associations with AML and MPN in the pooled Canadian, United Kingdom, and Australian study should be examined in other populations with varying levels of benzene exposure and in relation to time since first exposure and other temporal characteristics. Incidence of myeloid neoplasms by major category and subtypes should be evaluated in exposed occupational populations in relation to dose–​response and temporal characteristics. Epidemiologic studies of cytotoxic agents have often focused on a single drug and have not adequately considered the joint effects of multiple agents or confounding exposures. Finally, most studies of well-​established risk factors have not assessed genetic or familial factors in these associations. Promising leads for further study include clarification of the association of myeloid neoplasms with BMI and the intriguing associations with infectious and inflammatory conditions. Although statistical associations of selected autoimmune conditions and specific myeloid neoplasms have been reproduced in Swedish and US populations, it will be clinically important to clarify whether these associations are related to medications used to treat these conditions, the conditions themselves, and/​or genetic characteristics. Further large-​scale studies are needed to identify genetic predisposition to myeloid neoplasms, considering both rare high-​penetrant and more common low-​risk variants. Such studies should consider both commonalities and differences in genetic susceptibility among the myeloid neoplasms. Intensive efforts are underway to identify host-​related genetic variables that influence risk of developing t-​AML (Knight et al., 2009). Few epidemiologic studies to date have been designed specifically to assess risk factors for MPN. Most of the reports on MPN have derived from studies designed to focus on other outcomes or exposures not specifically implicated in the occurrence of MPN. Data are limited on identification of determinants of transformation of MDS or MPN to t-​AML. Clinically, it would be valuable to clarify what exposures or genetic characteristics facilitate or protect against such transformation. Linked registry studies of transformation of MPN to t-​AML have limited information on confounders and have not disentangled main effects under study from known or suspected confounding exposures. The ultimate goal of epidemiologic research on myeloid neoplasms is to identify risk factors and thereby prevent occurrence of preleukemic and leukemic entities. The uncommon incidence of the major categories of myeloid neoplasms, changing classification schemes, and the rarity of subtypes within these major categories complicate efforts to undertake epidemiologic studies. Standard, single-​center case-​control studies have limited power to identify statistical associations that are modest to moderate in level, even if such risk factors (e.g., cigarette smoking, BMI, or diagnostic radiological procedures) involve millions of persons and are of public health importance. Lack of reproducibility of findings and delays between reports from the few existing case-​control studies have hampered progress in identifying risk factors. A multicenter, interdisciplinary approach (e.g., epidemiologists working together with expert clinicians and experienced cancer registry staff) is needed. Strategies might include (1) establishment of large cohorts of newly diagnosed patients with specific preleukemic disorders for long-​term follow-​up to identify risk factors that increase risk of transformation and to identify determinants that prevent transformation through cohort and nested case-​cohort studies; (2)  identification of cohorts of families with multigeneration occurrence of specific or mixtures of myeloid neoplasms for detailed genetic studies and investigations to assess gene–​environment interaction; and, if feasible, (3) large population-​based case-​control studies of groups of patients with specific preleukemic disorders and of myeloid leukemias to evaluate and compare risk factors more generally. The WHO classification of hematopoietic and lymphoid malignancies has transformed our understanding of pathogenesis, clinical, and prognostic aspects of these disorders and, with continued updates, offers new opportunities for epidemiologic research. Epidemiologic studies of

737

myeloid neoplasms require a shift in standard methods as well as international, multidisciplinary efforts to clarify risk factor differences and similarities across subtypes and to identify preventive measures.

ACKNOWLEDGMENTS The authors are indebted to Mr. David Check for preparation of figures and Table 38–1 and to Mr. Wayne Liu for technical assistance with manuscript preparation. Support for the work on this project was provided by the Intramural Research Program of the National Institutes of Health, the National Cancer Institute, and the US Public Health Service of the Department of Health and Human Services. References Abdulkarim K, Girodon F, Johansson P, et al. 2009. AML transformation in 56 patients with Ph-​MPD in two well defined populations. Eur J Haematol, 82(2), 106–​111. PMID: 19134023. Ades L, Itzykson R, and Fenaux P. 2014. Myelodysplastic syndromes. Lancet, 383(9936), 2239–​2252. PMID: 24656536. Allemani C, Weir HK, Carreira H, et al. 2015. Global surveillance of cancer survival 1995-​2009:  analysis of individual data for 25,676,887 patients from 279 population-​based registries in 67 countries (CONCORD-​2). Lancet, 385(9972), 977–​1010. PMID: 25467588. Alpar D, Wren D, Ermini L, et al. 2015. Clonal origins of ETV6-​RUNX1(+) acute lymphoblastic leukemia: studies in monozygotic twins. Leukemia, 29(4), 839–​846. PMID: 25388957. Altekruse SF, Henley SJ, and Thun MJ. 1999. Deaths from hematopoietic and other cancers in relation to permanent hair dye use in a large prospective study (United States). Cancer Causes Control, 10(6), 617–​625. PMID: 10616830. Alter BP, Giri N, Savage SA, et al. 2010. Malignancies and survival patterns in the National Cancer Institute inherited bone marrow failure syndromes cohort study. Br J Haematology, 150(2), 179–​188. Alter BP, Giri N, Savage SA, and Rosenberg PS. 2009. Cancer in dyskeratosis congenita. Blood, 113(26), 6549–​6557. PMCID: PMC2710915. American College of Radiology. 2015. Diagnostic Radiology:  Computed Tomography (CT) Practice Parameters and Technical Standards. Available from:  http://​www.acr.org/​quality-​safety/​standards-​guidelines/​ practice-​guidelines-​by-​modality/​ct. American College of Radiology. 2015. Radiation Safety. Available from: http://​ www.acr.org/​quality-​safety/​radiology-​safety/​radiation-​safety. Anderson LA, Duncombe AS, Hughes M, et al. 2012. Environmental, lifestyle, and familial/​ethnic factors associated with myeloproliferative neoplasms. Am J Hematol, 87(2), 175–​182. PMID: 22076943. Anderson LA, Pfeiffer RM, Landgren O, et al. 2009. Risks of myeloid malignancies in patients with autoimmune conditions. Br J Cancer, 100(5), 822–​828. PMCID: PMC2653768. Arora H, Chacon AH, Choudhary S, et  al. 2014. Bloom syndrome. Int J Dermatol, 53(7), 798–​802. PMID: 24602044. Arranz S, Chiva-​Blanch G, Valderas-​Martinez P, et al. 2012. Wine, beer, alcohol and polyphenols on cardiovascular disease and cancer. Nutrients, 4(7), 759–​781. PMCID: PMC3407993. Babushok DV and Bessler M. 2015. Genetic predisposition syndromes: when should they be considered in the work-​up of MDS? Best Pract Res Clin Haematol, 28(1), 55–​68. PMCID: PMC4323616. Bachand AM, Mundt KA, Mundt DJ, and Montgomery RR. 2010. Epidemiological studies of formaldehyde exposure and risk of leukemia and nasopharyngeal cancer: a meta-​analysis. Crit Rev Toxicol, 40(2), 85–​ 100. PMID: 20085478. Ballinger ML, Mitchell G, and Thomas DM. 2015. Surveillance recommendations for patients with germline TP53 mutations. Curr Opin Oncol, 27(4), 332–​337. PMID: 26049273. Band PR, Le ND, Fang R, et al. 1996. Cohort study of Air Canada pilots: mortality, cancer incidence, and leukemia risk. Am J Epidemiol, 143(2), 137–​ 143. PMID: 8546114. Barbui T. 2004. The leukemia controversy in myeloproliferative disorders: is it a natural progression of disease, a secondary sequela of therapy, or a combination of both? Semin Hematol, 41(2 Suppl 3), 15–​17. PMID: 15190518. Barbui T, Thiele J, Passamonti F et al. 2011. Survival and disease progression in essential thrombocythemia are significantly influenced by accurate morphologic diagnosis: an international study. J Clin Oncol, 29(23), 3179–​ 3184. PMID: 21747083. Beane Freeman LE, Blair A, Lubin JH et al. 2009. Mortality from lymphohematopoietic malignancies among workers in formaldehyde industries: the

738

738

PART IV:  Cancers by Tissue of Origin

National Cancer Institute Cohort. J Natl Cancer Inst, 101(10), 751–​761. PMCID: PMC2684555. Beane Freeman LE, Bonner MR, Blair A, et  al. 2005. Cancer incidence among male pesticide applicators in the Agricultural Health Study cohort exposed to diazinon. Am J Epidemiol, 162(11), 1070–​ 1079. PMID: 16236997. Bennett JM. 2000. World Health Organization classification of the acute leukemias and myelodysplastic syndrome. Int J Hematol, 72(2), 131–​133. PMID: 11039659. Bennett JM. 2013. The myelodysplastic syndromes:  a personal recollection of four decades of classification and prognostic scoring systems. Leuk Lymphoma, 54(12), 2588–​2591. PMID: 23547832. Bennett JM, Catovsky D, Daniel MT, et al. 1994. The chronic myeloid leukaemias: guidelines for distinguishing chronic granulocytic, atypical chronic myeloid, and chronic myelomonocytic leukaemia. Proposals by the French-​American-​British Cooperative Leukaemia Group. Br J Haematol, 87(4), 746–​754. PMID: 7986717. Bennett JM, Catovsky D, Daniel MT, et  al. 1976. Proposals for the classification of the acute leukaemias. French-​ American-​ British (FAB) co-​ operative group. Br J Haematol, 33(4), 451–​458. PMID: 188440. Bennett JM, Catovsky D, Daniel MT, et  al. 1989. Proposals for the classification of chronic (mature) B and T lymphoid leukaemias. French-​ American-​ British (FAB) Cooperative Group. J Clin Pathol, 42(6), 567–​584. PMCID: PMC1141984. Bernard MP, Bancos S, Sime PJ, and Phipps RP. 2008. Targeting cyclooxygenase-​2 in hematological malignancies: rationale and promise. Curr Pharm Des, 14(21), 2051–​2060. PMCID: PMC2745246. Bhatia S. 2013. Therapy-​related myelodysplasia and acute myeloid leukemia. Semin Oncol, 40(6), 666–​675. PMCID: PMC3867743. Bjørge T, Cnattingius S, Lie RT, Tretli S, and Engeland A. 2008. Cancer risk in children with birth defects and in their families:  a population based cohort study of 5.2 million children from Norway and Sweden. Cancer Epidemiol Biomarkers Prev, 17(3), 500–​506. Bjork J, Albin M, Welinder H, et al. 2001. Are occupational, hobby, or lifestyle exposures associated with Philadelphia chromosome positive chronic myeloid leukaemia? Occup Environ Med, 58(11), 722–​727. PMCID: PMC1740065. Bjorkholm M, Derolf AR, Hultcrantz M, et  al. 2011. Treatment-​related risk factors for transformation to acute myeloid leukemia and myelodysplastic syndromes in myeloproliferative neoplasms. J Clin Oncol, 29(17), 2410–​ 2415. PMCID: PMC3107755. Bjorkholm M, Kristinsson SY, Landgren O, and Goldin LR. 2013. No familial aggregation in chronic myeloid leukemia. Blood, 122(3), 460–​461. PMCID: PMC3716206. Blackwelder WC, Yano K, Rhoads GG, et al. 1980. Alcohol and mortality: the Honolulu Heart Study. Am J Med, 68(2), 164–​169. PMID: 7355889. Blair A, and Zahm SH. 1995. Agricultural exposures and cancer. Environ Health Perspect, 103 (Suppl 8), 205–​208. PMCID: PMC1518967. Bochtler T, Frohling S, and Kramer A. 2015. Role of chromosomal aberrations in clonal diversity and progression of acute myeloid leukemia. Leukemia, 29(6), 1243–​1252. PMID: 25673237. Boice JD, Jr. 2006. Ionizing radiation. In:  Schottenfeld D, and Fraumeni JF (Eds.), Cancer epidemiology and prevention (3rd ed., pp. 259–​ 293). New York: Oxford University Press. Boice JD, Jr., Blettner M, Kleinerman RA, et al. 1987. Radiation dose and leukemia risk in patients treated for cancer of the cervix. J Natl Cancer Inst, 79(6), 1295–​1311. PMID: 3480381. Boice JD, Jr., Morin MM, Glass AG, et al. 1991. Diagnostic x-​ray procedures and risk of leukemia, lymphoma, and multiple myeloma. JAMA, 265(10), 1290–​1294. PMID: 2053936. Bolufer P, Barragan E, Collado M, et al. 2006. Influence of genetic polymorphisms on the risk of developing leukemia and on disease progression. Leuk Res, 30(12), 1471–​1491. Borthakur G, Lin E, Jain N, Estey EE, Cortes JE, O’Brien S, Faderl S, Ravandi F, Pierce S, and Kantarjian H. (2009) Survival is poorer in patients with secondary core-​binding factor acute myelogenous leukemia compared with de novo core-​binding factor leukemia. Cancer, 115, 3217–​3221. Brown LM, Gibson R, Burmeister LF, et al. 1992. Alcohol consumption and risk of leukemia, non-​Hodgkin’s lymphoma, and multiple myeloma. Leuk Res, 16(10), 979–​984. PMID: 1405712. Caldwell GG, Kelley DB, and Heath CW, Jr. 1980. Leukemia among participants in military maneuvers at a nuclear bomb test: a preliminary report. JAMA, 244(14), 1575–​1578. PMID: 6932516. Campbell PJ, and Green AR. 2006. The myeloproliferative disorders. N Engl J Med, 355(23), 2452–​2466. PMID: 17151367. Cancer Genome Atlas Research Network. 2013. Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N Engl J Med, 368(22), 2059–​2074. PMID: 23634996.

Candelaria M, and Duenas-​Gonzalez A. 2015. Therapy-​related myelodysplastic syndrome. Expert Opin Drug Saf, 14(5), 655–​665. PMID: 25675961. Cantor KP, Blair A, Everett G, et al. 1988. Hair dye use and risk of leukemia and lymphoma. Am J Public Health, 78(5), 570–​571. PMCID: PMC1349342. Cardis E, Vrijheid M, Blettner M, et al. 2005. Risk of cancer after low doses of ionising radiation:  retrospective cohort study in 15 countries. BMJ, 331(7508), 77. PMCID: PMC558612. Carstensen JM, Bygren LO, and Hatschek T. 1990. Cancer incidence among Swedish brewery workers. Int J Cancer, 45(3), 393–​396. PMID: 2407667. Cartwright RA, Darwin C, McKinney PA, et al. 1988. Acute myeloid leukemia in adults: a case-​control study in Yorkshire. Leukemia, 2(10), 687–​690. PMID: 3172845. Castillo JJ, Reagan JL, Ingham RR, et  al. 2012. Obesity but not overweight increases the incidence and mortality of leukemia in adults:  a meta-​ analysis of prospective cohort studies. Leuk Res, 36(7), 868–​ 875. PMID: 22285508. Chabner B, Bertino J, Cleary J, et al. 2011. Cytotoxic agents. In: Brunton L, Chabner B, and Knollmann B (Eds.), Goodman & Gilman’s The pharmacological basis of therapeutics (12th ed.). New York: McGraw-​Hill. Choi H, Jung C, Sohn SK, et al. 2013. Genome-​wide genotype-​based risk model for survival in acute myeloid leukaemia patients with normal karyotype. Br J Haematol, 163(1), 62–​71. PMID: 24159609. Churpek JE, Lorenz R, Nedumgottil S, et  al. 2013. Proposal for the clinical detection and management of patients and their family members with familial myelodysplastic syndrome/​acute leukemia predisposition syndromes. Leuk Lymphoma, 54(1), 28–​35. PMID: 22691122. Cogle CR. 2015. Incidence and burden of the myelodysplastic syndromes. Curr Hematol Malig Rep, 10(3), 272–​281. PMID: 26134527. Cohen T, and Creger WP. 1967. Acute myeloid leukemia following seven years of aplastic anemia induced by chloramphenicol. Am J Med, 43(5), 762–​ 770. PMID: 5234001. Cooper GS, Kamel F, Sandler DP, Davey FR, and Bloomfield CD. 1996. Risk of adult acute leukemia in relation to prior immune-​related conditions. Cancer Epidemiol Biomarkers Prev, 5(11), 867–​ 872. PMID: 8922293. Correa A, Mohan A, Jackson L, Perry H, and Helzlsouer K. 2000. Use of hair dyes, hematopoietic neoplasms, and lymphomas:  a literature review. I. Leukemias and myelodysplastic syndromes. Cancer Invest, 18(4), 366–​ 380. PMID: 10808373. Couto E, Chen B, and Hemminki K. 2005. Association of childhood acute lymphoblastic leukaemia with cancers in family members. Br J Cancer, 93(11), 1307–​1309. Cowell IG, Sondka Z, Smith K, et al. 2012. Model for MLL translocations in therapy-​related leukemia involving topoisomerase IIbeta-​mediated DNA strand breaks and gene proximity. Proc Natl Acad Sci U S A, 109(23), 8989–​8994. PMCID: PMC3384169. Craig BM, Rollison DE, List AF, and Cogle CR. 2012. Underreporting of myeloid malignancies by United States cancer registries. Cancer Epidemiol Biomarkers Prev, 21(3), 474–​481. PMCID: 3662977. Dameshek W. 1951. Some speculations on the myeloproliferative syndromes. Blood, 6(4), 372–​375. PMID: 14820991. Daniels RD, and Schubauer-​Berigan MK. 2011. A meta-​analysis of leukaemia risk from protracted exposure to low-​dose gamma radiation. Occup Environ Med, 68(6), 457–​464. PMCID: PMC3095477. Darby SC, Whitley E, Howe GR, et al. 1995. Radon and cancers other than lung cancer in underground miners:  a collaborative analysis of 11 studies. J Natl Cancer Inst, 87(5), 378–​384. PMID: 7853419. Debeljak M, Kitanovski L, Pajic T, and Jazbec J. 2013. Concordant acute myeloblastic leukemia in monozygotic twins with germline and shared somatic mutations in the gene for CCAAT-​ enhancer-​ binding protein alpha with 13 years difference at onset. Haematologica, 98(7), e73–​74. PMCID: PMC3696596. Delzell E, Sathiakumar N, Graff J, Macaluso M, Maldonado G, Matthews R. Health Effects Institute. 2006. An updated study of mortality among North American synthetic rubber industry workers. Res Rep Health Eff Inst, 132, 1–​63. Dohner H, Weisdorf DJ, and Bloomfield CD. 2015. Acute myeloid leukemia. N Engl J Med, 373(12), 1136–​1152. PMID: 26376137. Doody MM, Linet MS, Glass AG, et al. 1996. Risks of non-​Hodgkin’s lymphoma, multiple myeloma, and leukemia associated with common medications. Epidemiology, 7(2), 131–​139. PMID: 8834551. Doody MM, Linet MS, Glass AG, et al. 1992. Leukemia, lymphoma, and multiple myeloma following selected medical conditions. Cancer Causes Control, 3(5), 449–​456. PMID: 1525326. Dores GM, Devesa SS, Curtis RE, Linet MS, and Morton LM. 2012. Acute leukemia incidence and patient survival among children and adults in the United States, 2001–​2007. Blood, 119(1), 34–​43. PMID: 22086414.

 739

Leukemias Drenberg CD, Paugh SW, Pounds SB, et  al. 2016. Inherited variation in OATP1B1 is associated with treatment outcome in acute myeloid leukemia. Clin Pharmacol Ther, 99(6), 651–​660. PMID: 26663398. Du Y, Fryzek J, Sekeres MA, and Taioli E. 2010. Smoking and alcohol intake as risk factors for myelodysplastic syndromes (MDS). Leuk Res, 34(1), 1–​5. PMID: 19747728. Einstein AJ, Berman DS, Min JK, et  al. 2014. Patient-​ centered imaging:  shared decision making for cardiac imaging procedures with exposure to ionizing radiation. J Am Coll Cardiol, 63(15), 1480–​1489. PMCID: PMC3994983. Elliott P, Shaddick G, Douglass M, et al. 2013. Adult cancers near high-​voltage overhead power lines. Epidemiology, 24(2), 184–​190. PMID: 23337237. Evans DG, Birch JM, Ramsden RT, Sharif S, and Baser ME. 2006. Malignant transformation and new primary tumours after therapeutic radiation for benign disease:  substantial risks in certain tumour prone syndromes. J Med Genet, 43(4), 289–​294. PMCID: PMC2563223. Federman N, and Sakamoto KM. 2005. The genetic basis of bone marrow failure syndromes in children. Mol Genet Metab, 86(1–​2), 100–​109. Ferlay J, Soerjomataram I, Ervik M, et al. 2014. Cancer Incidence and Mortality Worldwide. Available from: http://​www.wcrf.org/​int/​cancer-​facts-​figures/​ worldwide-​data. Accessed February 29, 2016. Fernberg P, Odenbro A, Bellocco R, et al. 2007. Tobacco use, body mass index, and the risk of leukemia and multiple myeloma:  a nationwide cohort study in Sweden. Cancer Res, 67(12), 5983–​5986. PMID: 17575169. Ferrara F, and Schiffer CA. 2013. Acute myeloid leukaemia in adults. Lancet, 381(9865), 484–​495. PMID: 23399072. Feychting M, and Ahlbom A. 1994. Magnetic fields, leukemia, and central nervous system tumors in Swedish adults residing near high-​voltage power lines. Epidemiology, 5(5), 501–​509. PMID: 7986864. Fircanis S, Merriam P, Khan N, and Castillo JJ. 2014. The relation between cigarette smoking and risk of acute myeloid leukemia: an updated meta-​ analysis of epidemiological studies. Am J Hematol, 89(8), E125–​132. PMID: 24753145. Fleenor CJ, Higa K, Weil MM, and DeGregori J. 2015. Evolved cellular mechanisms to respond to genotoxic insults: implications for radiation-​ induced hematologic malignancies. Radiat Res, 184(4), 341–​ 351. PMCID: PMC4617554. Fraumeni JF, Jr. 1967. Bone marrow depression induced by chloramphenicol or phenylbutazone: leukemia and other sequelae. JAMA, 201(11), 828–​834. PMID: 6071896. Friedman GD. 1982. Phenylbutazone, musculoskeletal disease, and leukemia. J Chronic Dis, 35(4), 233–​243. PMID: 7037810. Fritz A, Percy C, Jack A, et  al (Eds.). 2000. International classification of diseases for oncology (3rd ed.). Geneva, Switzerland:  World Health Organization. Gaitonde S, Boumendjel R, Angeles R, and Rondelli D. 2010. Familial childhood monosomy 7 and associated myelodysplasia. J Pediatr Hematol Oncol, 32(6), e236–​237. PMID: 20661156. Gangatharan SA, Grove CS, P’ng S, et al. 2013. Acute myeloid leukaemia in Western Australia 1991-​2005: a retrospective population-​based study of 898 patients regarding epidemiology, cytogenetics, treatment and outcome. Intern Med J, 43(8), 903–​911. PMID: 23611681. Glass DC, Schnatter AR, Tang G, Irons RD, and Rushton L. 2014. Risk of myeloproliferative disease and chronic myeloid leukaemia following exposure to low-​level benzene in a nested case-​control study of petroleum workers. Occup Environ Med, 71(4), 266–​274. PMID: 24532441. Gluzman DF, Sklyarenko LM, Koval SV, et  al. 2015. Myelodysplastic syndromes in Chernobyl clean-​up workers. Ann Hematol, 94(10), 1639–​ 1643. PMID: 26208666. Godley LA, and Larson RA. 2008. Therapy-​related myeloid leukemia. Semin Oncol, 35(4), 418–​429. PMCID: PMC2600445. Goldberg SL, Chen E, Corral M, et al. 2010. Incidence and clinical complications of myelodysplastic syndromes among United States Medicare beneficiaries. J Clin Oncol, 28(17), 2847–​2852. PMID: 20421543. Granfeldt Ostgard LS, Medeiros BC, Sengelov H, Norgaard M, Andersen MK, Dufva IH, Friis LS, Kjeldsen E, Marcher CW, Preiss B, Severinsen M, and Norgaard JM. 2015. Epidemiology and clinical significance of secondary and therapy-​related acute myeloid leukemia: a national population-​based cohort study. J Clin Oncol, 33, 3641–​3649. Greaves MF, Maia AT, Wiemels JL, and Ford AM. 2003. Leukemia in twins: lessons in natural history. Blood, 102(7), 2321–​2333. PMID: 12791663. Greaves MF, and Wiemels J. 2003. Origins of chromosome translocations in childhood leukaemia. Nat Rev Cancer, 3(9), 639–​ 649. PMID: 12951583. Greenberg P, Cox C, LeBeau MM, et al. 1997. International scoring system for evaluating prognosis in myelodysplastic syndromes. Blood, 89(6), 2079–​ 2088. PMID: 9058730.

739

Greenberg PL, Tuechler H, Schanz J, et al. 2012. Revised international prognostic scoring system for myelodysplastic syndromes. Blood, 120(12), 2454–​2465. PMID: 22740453. Griem ML, Kleinerman RA, Boice JD, Jr. et  al. 1994. Cancer following radiotherapy for peptic ulcer. J Natl Cancer Inst, 86(11), 842–​ 849. PMID: 8182765. Grimwade D. 2001. The clinical significance of cytogenetic abnormalities in acute myeloid leukaemia. Best Pract Res Clin Haematol, 14(3), 497–​529. PMID: 11640867. Grimwade D, Hills RK, Moorman AV, et al. 2010. Refinement of cytogenetic classification in acute myeloid leukemia:  determination of prognostic significance of rare recurring chromosomal abnormalities among 5876 younger adult patients treated in the United Kingdom Medical Research Council trials. Blood, 116(3), 354–​365. PMID: 20385793. Groves FD, Page WF, Gridley G et al. 2002. Cancer in Korean war navy technicians: mortality survey after 40 years. Am J Epidemiol, 155(9), 810–​818. PMID: 11978584. Gruppo Italiano Studio Policitemia. 1995. Polycythemia vera: the natural history of 1213 patients followed for 20 years. Ann Intern Med, 123(9), 656–​ 664. PMID: 7574220. Gundestrup M, and Storm HH. 1999. Radiation-​induced acute myeloid leukaemia and other cancers in commercial jet cockpit crew: a population-​based cohort study. Lancet, 354(9195), 2029–​2031. PMID: 10636367. Hahn CN, Chong CE, Carmichael CL et al. 2011. Heritable GATA2 mutations associated with familial myelodysplastic syndrome and acute myeloid leukemia. Nat Genet, 43(10), 1012–​1017. PMCID: PMC3184204. Hammer GP, Auvinen A, De Stavola BL, et al. 2014. Mortality from cancer and other causes in commercial airline crews: a joint analysis of cohorts from 10 countries. Occup Environ Med, 71(5), 313–​322. PMID: 24389960. Harada H, Harada Y, Tanaka H, Kimura A, and Inaba T. 2003. Implications of somatic mutations in the AML1 gene in radiation-​associated and therapy-​ related myelodysplastic syndrome/​ acute myeloid leukemia. Blood, 101(2), 673–​680. PMID: 12393679. Hauptmann M, Lubin JH, Stewart PA, Hayes RB, and Blair A. 2003. Mortality from lymphohematopoietic malignancies among workers in formaldehyde industries. J Natl Cancer Inst, 95(21), 1615–​1623. PMID: 14600094. Hauptmann M, Stewart PA, Lubin JH, et  al. 2009. Mortality from lymphohematopoietic malignancies and brain cancer among embalmers exposed to formaldehyde. J Natl Cancer Inst, 101(24), 1696–​ 1708. PMCID: PMC2794303. Hayes RB, Yin SN, Dosemeci M, et  al. 1997. Benzene and the dose-​related incidence of hematologic neoplasms in China. Chinese Academy of Preventive Medicine-​-​National Cancer Institute Benzene Study Group. J Natl Cancer Inst, 89(14), 1065–​1071. PMID: 9230889. Hemminki K, Liu X, Forsti A, et al. 2013. Subsequent leukaemia in autoimmune disease patients. Br J Haematol, 161(5), 677–​687. PMID: 23565673. Henshaw DL, Eatough JP, and Richardson RB. 1990. Radon as a causative factor in induction of myeloid leukaemia and other cancers. Lancet, 335(8696), 1008–​1012. PMID: 1970069. Hinds MW, Kolonel LN, Lee J, and Hirohata T. 1980. Associations between cancer incidence and alcohol/​cigarette consumption among five ethnic groups in Hawaii. Br J Cancer, 41(6), 929–​940. PMCID: PMC2010370. Hoffman R, Benz EJ, Jr., Silberstein LE, et al. 2013. Hematology: basic principles and practice, 6th ed. Philadelphia: Saunders/​Elsevier. Howard R, Gilbert E, Lynch CF, et al. 2008. Risk of leukemia among survivors of testicular cancer: a population-​based study of 42,722 patients. Ann Epidemiol, 18(5), 416–​421. PMCID: PMC4034178. Hsu WL, Preston DL, Soda M, et al. 2013. The incidence of leukemia, lymphoma and multiple myeloma among atomic bomb survivors:  1950–​ 2001. Radiat Res, 179(3), 361–​382. PMCID: PMC3875218. IARC. 2001. Ionizing radiation, part  2:  some internally deposited radionuclides. IARC Monogr Eval Carcinog Risks Hum, 78, 595. IARC. 2008. 1,3-​butadiene, ethylene oxide and vinyl halides (vinyl fluoride, vinyl chloride and vinyl bromide). IARC Monogr Eval Carcinog Risks Hum, 97, 510. IARC. 2010. Some aromatic amines, organic dyes and related exposures. IARC Monogr Eval Carcinog Risks Hum, 99. [Personal use of hair dyes: pp. 558–​ 658.] Lyon, France: International Agency for Research on Cancer. IARC. 2012. Chemical and related occupations. IARC Monogr Eval Carcinog Risks Hum 100 (Pt F). Benzene:  pp.  249–​294; Butadiene:  pp.  309–​ 338; Dioxins and related:  pp.  339–​378; Formaldehyde:  pp.  401–​435. Occupational exposure in the rubber manufacturing industry:  pp.  541–​ 562. Lyon, France: International Agency for Research on Cancer. IARC. 2013. Non-​ionizing radiation, part  2:  radiofrequency electromagnetic fields. IARC Monogr Eval Carcinog Risks Hum, 102, 460. IARC. 2014. Cancer incidence in five continents, Vol. X. Scientific Publication 164. Lyon, France: International Agency for Research on Cancer.

740

740

PART IV:  Cancers by Tissue of Origin

Ido M, Nagata C, Kawakami N et al. 1996. A case-​control study of myelodysplastic syndromes among Japanese men and women. Leuk Res, 20(9), 727–​731. PMID: 8947581. Inaba H, Greaves M, and Mullighan CG. 2013. Acute lymphoblastic leukaemia. Lancet, 381(9881), 1943–​1955. PMCID: PMC3816716. IOM (Institute of Medicine). 2000. The Five Series Study: Mortality of military participants in U.S.  nuclear weapons tests. Washington, DC:  National Academy Press. Iwanaga M, Hsu WL, Soda M et  al. 2011. Risk of myelodysplastic syndromes in people exposed to ionizing radiation:  a retrospective cohort study of Nagasaki atomic bomb survivors. J Clin Oncol, 29(4), 428–​434. PMID: 21149671. Jablon S, Hrubec Z, and Boice JD, Jr. 1991. Cancer in populations living near nuclear facilities: a survey of mortality nationwide and incidence in two states. JAMA, 265(11), 1403–​1408. PMID: 1999880. Jaffe ES, Harris NL, Stein H, and Vardiman JW (Eds.). 2001. World Health Organization classification of tumours: pathology and genetics of haematopoietic and lymphoid tissues. Lyon: IARC Press. James C, Ugo V, Le Couedic JP, et al. 2005. A unique clonal JAK2 mutation leading to constitutive signalling causes polycythaemia vera. Nature, 434(7037), 1144–​1148. PMID: 15793561. Johansen C, Raaschou Nielsen O, Olsen JH, and Schuz J. 2007. Risk for leukaemia and brain and breast cancer among Danish utility workers:  a second follow-​ up. Occup Environ Med, 64(11), 782–​ 784. PMCID: PMC2078420. Kabat GC, Wu JW, Moore SC, et  al. 2013. Lifestyle and dietary factors in relation to risk of chronic myeloid leukemia in the NIH-​AARP Diet and Health Study. Cancer Epidemiol Biomarkers Prev, 22(5), 848–​854. PMCID: PMC3849026. Kadia TM, Ravandi F, Cortes J, and Kantarjian H. 2015. toward individualized therapy in acute myeloid leukemia: a contemporary review. JAMA Oncol, 1(6), 820–​828. PMID: 26181162. Karmali R, Dalovisio A, Borgia JA, et al. 2015. All in the family: Clueing into the link between metabolic syndrome and hematologic malignancies. Blood Rev, 29(2), 71–​80. PMID: 25433571. Karran P. 2006. Thiopurines, DNA damage, DNA repair and therapy-​related cancer. Br Med Bull, 79–​80, 153–​170. PMID: 17277075. Kasum CM, Blair CK, Folsom AR, Ross JA. 2003. Non-​ steroidal anti-​ inflammatory drug use and risk of adult leukemia. Cancer Epidemiol Biomark Prev, 12, 534–​7. Kaye FJ. 2009. Mutation-​associated fusion cancer genes in solid tumors. Mol Cancer Ther, 8(6), 1399–​1408. PMID: 19509239. Khalade A, Jaakkola MS, Pukkala E, and Jaakkola JJ. 2010. Exposure to benzene at work and the risk of leukemia:  a systematic review and meta-​ analysis. Environ Health, 9, 31. PMCID: PMC2903550. Kheifets L, Monroe J, Vergara X, Mezei G, and Afifi AA. 2008. Occupational electromagnetic fields and leukemia and brain cancer:  an update to two meta-​ analyses. J Occup Environ Med, 50(6), 677–​ 688. PMID: 18545095. Kheifets LI, Afifi AA, Buffler PA, Zhang ZW, and Matkin CC. 1997. Occupational electric and magnetic field exposure and leukemia: a meta-​ analysis. J Occup Environ Med, 39(11), 1074–​1091. PMID: 9383718. Khincha PP, and Savage SA. 2013. Genomic characterization of the inherited bone marrow failure syndromes. Semin Hematol, 50(4), 333–​347. PMCID: PMC3835370. Kilpivaara O, Mukherjee S, Schram AM, et al. 2009. A germline JAK2 SNP is associated with predisposition to the development of JAK2(V617F)-​ positive myeloproliferative neoplasms. Nat Genet, 41(4), 455–​ 459. PMCID: PMC3676425. Kim DH, Lee ST, Won HH, et al. 2011. A genome-​wide association study identifies novel loci associated with susceptibility to chronic myeloid leukemia. Blood, 117(25), 6906–​6911. PMID: 21540461. Kinlen LJ, and Rogot E. 1988. Leukaemia and smoking habits among United States veterans. BMJ, 297(6649), 657–​659. PMCID: PMC1834341. Knight JA, Skol AD, Shinde A, et al. 2009. Genome-​wide association study to identify novel loci associated with therapy-​related myeloid leukemia susceptibility. Blood, 113(22), 5575–​5582. PMCID: PMC2689055. Koeman T, van den Brandt PA, Slottje P, et al. 2014. Occupational extremely low-​frequency magnetic field exposure and selected cancer outcomes in a prospective Dutch cohort. Cancer Causes Control, 25(2), 203–​214. PMID: 24241907. Kralovics R, Passamonti F, Buser AS, et al. 2005. A gain-​of-​function mutation of JAK2 in myeloproliferative disorders. N Engl J Med, 352(17), 1779–​ 1790. PMID: 15858187. Krestinina L, Preston DL, Davis FG, et al. 2010. Leukemia incidence among people exposed to chronic radiation from the contaminated Techa River, 1953–​2005. Radiat Environ Biophys, 49(2), 195–​201. PMID: 20012750.

Kristinsson SY, Bjorkholm M, Hultcrantz M, et  al. 2011. Chronic immune stimulation might act as a trigger for the development of acute myeloid leukemia or myelodysplastic syndromes. J Clin Oncol, 29(21), 2897–​ 2903. PMCID: PMC3138717. Kroll ME, Murphy F, Pirie K, et al. 2012. Alcohol drinking, tobacco smoking and subtypes of haematological malignancy in the UK Million Women Study. Br J Cancer, 107(5), 879–​887. PMCID: PMC3425977. Lan Q, Zhang L, Li G, et al. 2004. Hematotoxicity in workers exposed to low levels of benzene. Science, 306(5702), 1774–​1776. PMCID: PMC1256034. Landgren O, Goldin LR, Kristinsson SY, et  al. 2008. Increased risks of polycythemia vera, essential thrombocythemia, and myelofibrosis among 24,577 first-​degree relatives of 11,039 patients with myeloproliferative neoplasms in Sweden. Blood, 112(6), 2199–​2204. PMCID: PMC2532797. Larsson SC, and Wolk A. 2008. Overweight and obesity and incidence of leukemia: a meta-​analysis of cohort studies. Int J Cancer, 122, 1418–​21. Laurier D, Jacob S, Bernier MO, et al. 2008. Epidemiological studies of leukaemia in children and young adults around nuclear facilities: a critical review. Radiat Prot Dosimetry, 132(2), 182–​190. PMID: 18922823. Laurier D, Valenty M, and Tirmarche M. 2001. Radon exposure and the risk of leukemia: a review of epidemiological studies. Health Phys, 81(3), 272–​ 288. PMID: 11513461. Leal AD, Thompson CA, Wang AH, et al. 2014. Anthropometric, medical history and lifestyle risk factors for myeloproliferative neoplasms in the Iowa Women’s Health Study cohort. Int J Cancer, 134(7), 1741–​1750. PMCID: PMC3947165. Lee WJ, Hoppin JA, Blair A, et  al. 2004. Cancer incidence among pesticide applicators exposed to alachlor in the Agricultural Health Study. Am J Epidemiol, 159(4), 373–​380. PMID: 14769641. Lehmann S, Ravn A, Carlsson L, et al. 2011. Continuing high early death rate in acute promyelocytic leukemia:  a population-​based report from the Swedish Adult Acute Leukemia Registry. Leukemia, 25(7), 1128–​1134. PMID: 21502956. Leone G, Fianchi L, Pagano L, and Voso MT. 2010. Incidence and susceptibility to therapy-​related myeloid neoplasms. Chem Biol Interact, 184(1–​2), 39–​45. PMID: 20026017. Leuraud K, Richardson DB, Cardis E, et al. 2015. Ionising radiation and risk of death from leukaemia and lymphoma in radiation-​monitored workers (INWORKS): an international cohort study. Lancet Haematol, 2(7), e276–​e281. PMCID: PMC4587986. Li Y, Moysich KB, Baer MR, et al. 2006. Intakes of selected food groups and beverages and adult acute myeloid leukemia. Leuk Res, 30(12), 1507–​ 1515. PMID: 16678899. Lichtman MA. 2016. Classification and clinical manifestations of the clonal myeloid disorders. Kaushansky K, Lichtman MA, Prchal JT, et al (eds). Williams Hematology, 9e. (pp. 1275–​1290). New  York:  McGraw Hill Medical. Linet MS. 1985. The leukemias: epidemiological aspects. New York: Oxford University Press. Linet MS, and Cartwright RA. 1988. Chronic lymphocytic leukemia: epidemiology and etiologic findings. Nouv Rev Fr Hematol, 30(5–​6), 353–​357. PMID: 3222144. Linet MS, Devesa SS, and Morgan GJ. 2006. The leukemias. In: Schottenfeld D, and Fraumeni JF (Eds.), Cancer epidemiology and prevention (3rd ed., pp. 841–​871). New York: Oxford University Press. Linet MS, Kim KP, Miller DL, et al. 2010. Historical review of occupational exposures and cancer risks in medical radiation workers. Radiat Res, 174(6), 793–​808. PMCID: PMC4098897. Linet MS, Slovis TL, Miller DL, et al. 2012. Cancer risks associated with external radiation from diagnostic imaging procedures. CA Cancer J Clin, 62(2), 75–​100. PMCID: PMC3548988. Linet MS, Yin SN, Gilbert ES, et  al. 2015. A retrospective cohort study of cause-​specific mortality and incidence of hematopoietic malignancies in Chinese benzene-​exposed workers. Int J Cancer, 137(9), 2184–​2197. PMID: 25944549. Liu P, Holman CD, Jin J, and Zhang M. 2015. Dietary isoflavone intake is associated with a reduced risk of myelodysplastic syndromes. Br J Nutr, 114(12), 2110–​2115. PMID: 26458988. Liu P, Holman CD, Jin J, and Zhang M. 2016. Alcohol consumption and risk of myelodysplastic syndromes:  a case-​ control study. Cancer Causes Control, 27(2), 209–​216. PMID: 26590915. Ludwig A, Jochmann N, Kertesz A, et al. 2010. Smoking decreases the level of circulating CD34 + progenitor cells in young healthy women: a pilot study. BMC Womens Health, 10, 20. PMCID: PMC2891626. Ma X, Lim U, Park Y, et al. 2009. Obesity, lifestyle factors, and risk of myelodysplastic syndromes in a large US cohort. Am J Epidemiol, 169(12), 1492–​1499. PMCID: PMC2727203.

 741

Leukemias Ma X, Park Y, Mayne ST, et al. 2010. Diet, lifestyle, and acute myeloid leukemia in the NIH-​AARP cohort. Am J Epidemiol, 171(3), 312–​322. PMCID: PMC2842202. Mabuchi K, Fugiwara S, Preston DL, Nakamura N, and Shore RE. 2011. Atomic bomb survivors: long-​term health effects of radiation. In: Shrieve DS, and Loeffler JS (Eds.), Human radiation injury. Philadelphia: Wolters Kuwer/​ Lippincott Williams and Wilkins. Mahajan R, Blair A, Lynch CF, et al. 2006. Fonofos exposure and cancer incidence in the agricultural health study. Environ Health Perspect, 114(12), 1838–​1842. PMCID: PMC1764168. Mangan JK, and Speck NA. 2011. RUNX1 mutations in clonal myeloid disorders:  from conventional cytogenetics to next generation sequencing, a story 40  years in the making. Crit Rev Oncog, 16(1–​2), 77–​91. PMCID: PMC3248792. March HC. 1950. Leukemia in radiologists in a 20 year period. Am J Med Sci, 220(3), 282–​286. PMID: 14771053. Matsuo T, Tomonaga M, Bennett JM, et al. 1988. Reclassification of leukemia among A-​ bomb survivors in Nagasaki using French-​ American-​ British (FAB) classification for acute leukemia. Jpn J Clin Oncol, 18(2), 91–​96. PMID: 2898029. Maynadie M, Girodon F, Manivet-​Janoray I, et al. 2011. Twenty-​five years of epidemiological recording on myeloid malignancies: data from the specialized registry of hematologic malignancies of Cote d’Or (Burgundy, France). Haematologica, 96(1), 55–​61. PMID: 20971817. McHale CM, Zhang L, and Smith MT. 2012. Current understanding of the mechanism of benzene-​induced leukemia in humans: implications for risk assessment. Carcinogenesis, 33(2), 240–​252. PMCID: PMC3271273. McKenna RW. 2000. Multifaceted approach to the diagnosis and classification of acute leukemias. Clin Chem, 46(8 Pt 2), 1252–​ 1259. PMID: 10926919. McKeown SR, Hatfield P, Prestwich RJ, Shaffer RE, and Taylor RE. 2015. Radiotherapy for benign disease; assessing the risk of radiation-​induced cancer following exposure to intermediate dose radiation. Br J Radiol, 88(1056), 20150405. PMID: 26462717. McLaughlin J. 2012. An historical overview of radon and its progeny: applications and health effects. Radiat Prot Dosimetry, 152(1–​3), 2–​8. PMID: 22914338. McNally RJ, Roman E, and Cartwright RA. 1999. Leukemias and lymphomas: time trends in the UK, 1984–​93. Cancer Causes Control, 10(1), 35–​ 42. PMID: 10334640. McQuilten ZK, Wood EM, Polizzotto MN, et  al. 2014. Underestimation of myelodysplastic syndrome incidence by cancer registries:  results from a population-​ based data linkage study. Cancer, 120(11), 1686–​ 1694. PMID: 24643720. Mele A, Visani G, Pulsoni A, et al. 1996. Risk factors for essential thrombocythemia: A case-​control study. Italian Leukemia Study Group. Cancer, 77(10), 2157–​2161. PMID: 8640685. Mesa RA, Li CY, Ketterling RP, et al. 2005. Leukemic transformation in myelofibrosis with myeloid metaplasia: a single-​institution experience with 91 cases. Blood, 105(3), 973–​977. PMID: 15388582. Mesa RA, Loegering D, Powell HL et al. 2005. Heat shock protein 90 inhibition sensitizes acute myelogenous leukemia cells to cytarabine. Blood, 106(1), 318–​327. PMCID: PMC1895127. Metayer C, Curtis RE, Vose J, et al. 2003. Myelodysplastic syndrome and acute myeloid leukemia after autotransplantation for lymphoma: a multicenter case-​control study. Blood, 101(5), 2015–​2023. PMID: 12393427. Mettler FAJr., Huda W, Yoshizumi TT, Mahesh M. 2008. Effective doses in radiology and diagnostic nuclear medicine:  a catalog. Radiology, 248, 254–​264. Meyer SC, and Levine RL. 2014. Translational implications of somatic genomics in acute myeloid leukaemia. Lancet Oncol, 15(9), e382–​394. PMID: 25079101. Mills PK, Beeson WL, Fraser GE, and Phillips RL. 1992. Allergy and cancer:  organ site-​specific results from the Adventist Health Study. Am J Epidemiol, 136(3), 287–​295. PMID: 1415150. Morton LM, Dores GM, Tucker MA, et  al. 2013. Evolving risk of therapy-​ related acute myeloid leukemia following cancer chemotherapy among adults in the United States, 1975–​2008. Blood, 121(15), 2996–​3004. PMCID: PMC3624944. Morton LM, Gibson TM, Clarke CA, et  al. 2014. Risk of myeloid neoplasms after solid organ transplantation. Leukemia, 28(12), 2317–​2323. PMCID: PMC4197126. Muirhead CR, Bingham D, Haylock RG, et  al. 2003. Follow up of mortality and incidence of cancer 1952–​98 in men from the UK who participated in the UK’s atmospheric nuclear weapon tests and experimental programmes. Occup Environ Med, 60(3), 165–​172. PMCID: PMC1740497.

741

Muirhead CR, O’Hagan JA, Haylock RG, et  al. 2009. Mortality and cancer incidence following occupational radiation exposure: third analysis of the National Registry for Radiation Workers. Br J Cancer, 100(1), 206–​212. PMCID: PMC2634664. Musselman JR, Blair CK, Cerhan JR, et  al. 2013. Risk of adult acute and chronic myeloid leukemia with cigarette smoking and cessation. Cancer Epidemiol, 37(4), 410–​416. PMCID: PMC3819424. Nardi V, Winkfield KM, Ok CY, et  al. 2012. Acute myeloid leukemia and myelodysplastic syndromes after radiation therapy are similar to de novo disease and differ from other therapy-​related myeloid neoplasms. J Clin Oncol, 30(19), 2340–​2347. PMID: 22585703. NCRP (National Council on Radiation Protection and Measurements). 2009. Ionizing radiation exposure of the population of the United States. Bethesda, MD: NCRP. NCRP (National Council on Radiation Protection and Measurements). 2011. Second primary cancers and cardiovascular disease after radiation therapy. Bethesda, MD: NCRP. Ng KC, Tan AM, Chong YY, Lau LC, and Lou J. 1999. Congenital acute megakaryoblastic leukemia (M7) with chromosomal t(1;22)(p13;q13) translocation in a set of identical twins. J Pediatr Hematol Oncol, 21(5), 428–​430. PMID: 10524459. Nickels EM, Soodalter J, Churpek JE, and Godley LA. 2013. Recognizing familial myeloid leukemia in adults. Ther Adv Hematol, 4(4), 254–​269. PMCID: PMC3734901. Nitiss JL. 2009. Targeting DNA topoisomerase II in cancer chemotherapy. Nat Rev Cancer, 9(5), 338–​350. PMCID: PMC2748742. NRC (National Research Council). 2006. Health risks from exposure to low levels of ionizing radiation: BEIR VII Phase 2. Washington, DC: National Academy of Sciences. Olsen JH, Hahnemann JM, Borresen-​Dale AL, et al. 2001. Cancer in patients with ataxia-​telangiectasia and in their relatives in the nordic countries. J Natl Cancer Inst, 93(2), 121–​127. PMID: 11208881. Owen C, Barnett M, and Fitzgibbon J. 2008. Familial myelodysplasia and acute myeloid leukaemia:  a review. Br J Haematol, 140(2), 123–​132. PMID: 18173751. Pan LL, Huang YM, Wang M, et  al. 2015. Positional cloning and next-​ generation sequencing identified a TGM6 mutation in a large Chinese pedigree with acute myeloid leukaemia. Eur J Hum Genet, 23(2), 218–​223. PMCID: PMC4297898. Park JH, Qiao B, Panageas KS, et al. 2011. Early death rate in acute promyelocytic leukemia remains high despite all-​trans retinoic acid. Blood, 118(5), 1248–​1254. PMID: 21653939. Passamonti F, Rumi E, Arcaini L, et al. 2008. Prognostic factors for thrombosis, myelofibrosis, and leukemia in essential thrombocythemia: a study of 605 patients. Haematologica, 93(11), 1645–​1651. PMID: 18790799. Passamonti F, Rumi E, Pungolino E, et al. 2004. Life expectancy and prognostic factors for survival in patients with polycythemia vera and essential thrombocythemia. Am J Med, 117(10), 755–​ 761. PMID: 15541325. Patel JP, Gonen M, Figueroa ME, et al. 2012. Prognostic relevance of integrated genetic profiling in acute myeloid leukemia. N Engl J Med, 366(12), 1079–​1089. PMID: 22417203. Pesatori AC, Consonni D, Tironi A, et  al. 1993. Cancer in a young population in a dioxin-​contaminated area. Int J Epidemiol, 22(6), 1010–​1013. PMID: 8144281. Pogoda JM, Katz J, McKean-​Cowdin R, et  al. 2005. Prescription drug use and risk of acute myeloid leukemia by French-​American-​British subtype: results from a Los Angeles County case-​control study. Int J Cancer, 114(4), 634–​638. PMID: 15609330. Pogoda JM, Nichols PW, Ross RK, et al. 2011. Diagnostic radiography and adult acute myeloid leukaemia: an interview and medical chart review study. Br J Cancer, 104(9), 1482–​1486. PMCID: PMC3101931. Polednak AP. 2011. US death rates from myeloproliferative neoplasms, and implications for cancer surveillance. J Registry Manag, 38(2), 87–​92. PMID: 22096879. Polednak AP. 2013. Trend (1999–​2009) in U.S. death rates from myelodysplastic syndromes: utility of multiple causes of death in surveillance. Cancer Epidemiol, 37(5), 569–​574. PMID: 23773300. Polychronakis I, Dounias G, Makropoulos V, Riza E, and Linos A. 2013. Work-​ related leukemia:  a systematic review. J Occup Med Toxicol, 8(1), 14. PMCID: PMC3668148. Poynter JN, Fonstad R, Blair CK, et al. 2013. Exogenous hormone use, reproductive history and risk of adult myeloid leukaemia. Br J Cancer, 109(7), 1895–​1898. PMCID: PMC3790163. Poynter JN, Richardson M, Blair CK, et al. 2016. Obesity over the life course and risk of acute myeloid leukemia and myelodysplastic syndromes. Cancer Epidemiol, 40, 134–​140. PMCID: PMC4738058.

742

742

PART IV:  Cancers by Tissue of Origin

Preston-​Martin S, and Pogoda JM. 2003. Estimation of radiographic doses in a case-​control study of acute myelogenous leukemia. Health Phys, 84(2), 245–​259. PMID: 12553655. Preston-​Martin S, Thomas DC, Yu MC, and Henderson BE. 1989. Diagnostic radiography as a risk factor for chronic myeloid and monocytic leukaemia (CML). Br J Cancer, 59(4), 639–​644. PMCID: PMC2247133. Purdue MP, Hoppin JA, Blair A, Dosemeci M, and Alavanja MC. 2007. Occupational exposure to organochlorine insecticides and cancer incidence in the Agricultural Health Study. Int J Cancer, 120(3), 642–​649. PMCID: PMC1971137. Raaschou-​Nielsen O, Ketzel M, Harbo Poulsen A, and Sorensen M. 2016. Traffic-​related air pollution and risk for leukaemia of an adult population. Int J Cancer, 138(5), 1111–​1117. PMID: 26415047. Rauscher GH, Shore D, and Sandler DP. 2004a. Alcohol intake and incidence of de novo adult acute leukemia. Leuk Res, 28(12), 1263–​1265. PMID: 15475066. Rauscher GH, Shore D, and Sandler DP. 2004b. Hair dye use and risk of adult acute leukemia. Am J Epidemiol, 160(1), 19–​25. PMID: 15229113. Richardson D, Sugiyama H, Nishi N, et al. 2009. Ionizing radiation and leukemia mortality among Japanese atomic bomb survivors, 1950–​2000. Radiat Res, 172(3), 368–​382. PMID: 19708786. Richardson DB, Terschuren C, Pohlabeln H, Jockel KH, and Hoffmann W. 2008. Temporal patterns of association between cigarette smoking and leukemia risk. Cancer Causes Control, 19(1), 43–​ 50. PMID: 17906957. Rinsky RA, Hornung RW, Silver SR, and Tseng CY. 2002. Benzene exposure and hematopoietic mortality: a long-​term epidemiologic risk assessment. Am J Ind Med, 42(6), 474–​480. PMID: 12439870. Rivina L, Davoren M, and Schiestl RH. 2014. Radiation-​induced myeloid leukemia in murine models. Hum Genomics, 8, 13. PMCID: PMC4128013. Rollig C, Bornhauser M, Thiede C, et al. 2011. Long-​term prognosis of acute myeloid leukemia according to the new genetic risk classification of the European LeukemiaNet recommendations:  evaluation of the proposed reporting system. J Clin Oncol, 29(20), 2758–​2765. PMID: 21632498. Roman E, Smith A, Appleton S, et al. 2016. Myeloid malignancies in the real world:  Occurrence, progression and survival in the UK’s population-​ based Haematological Malignancies Research Network 2004–​15. Cancer Epidemiol, Jun 42, 186–​198. PMID: 27090942. Rommens J, and Durie P. 2008. Shwachman-​Diamond syndrome. In:  Pagon R, Bird T, Dolan C, and Stephens K (Eds.), GeneReviews [Internet]. Seattle: University of Washington. Rosenbaum T, and Wimmer K. 2014. Neurofibromatosis type 1 (NF1) and associated tumors. Klin Padiatr, 226(6–​7), 309–​315. PMID: 25062113. Ross JA, Blair CK, Cerhan JR, et  al. 2011. Nonsteroidal anti-​inflammatory drug and acetaminophen use and risk of adult myeloid leukemia. Cancer Epidemiol Biomarkers Prev, 20(8), 1741–​1750. PMCID: PMC3153558. Ruhl J, Adamo M, and Dickie L. 2015. Hematopoietic and lymphoid neoplasm coding manual. Bethesda, MD: National Cancer Institute. Rushton L, Schnatter AR, Tang G, and Glass DC. 2014. Acute myeloid and chronic lymphoid leukaemias and exposure to low-​ level benzene among petroleum workers. Br J Cancer, 110(3), 783–​ 787. PMCID: PMC3915135. Sakata R, Kleinerman RA, Mabuchi K, et al. 2012. Cancer mortality following radiotherapy for benign gynecologic disorders. Radiat Res, 178(4), 266–​ 279. PMCID: PMC3471655. Saliba J, Saint-​Martin C, Di Stefano A, et  al. 2015. Germline duplication of ATG2B and GSKIP predisposes to familial myeloid malignancies. Nat Genet, 47(10), 1131–​1140. PMID: 26280900. Sant M, Allemani C, Tereanu C, et al. 2010. Incidence of hematologic malignancies in Europe by morphologic subtype: results of the HAEMACARE project. Blood, 116(19), 3724–​3734. PMID: 20664057. Saso R, Kulkarni S, Mitchell P, et  al. 2000. Secondary myelodysplastic syndrome/​ acute myeloid leukaemia following mitoxantrone-​ based therapy for breast carcinoma. Br J Cancer, 83(1), 91–​94. PMCID: PMC2374543. Savage SA, Giri N, Baerlocher GM, et al. 2008. TINF2, a component of the shelterin telomere protection complex, is mutated in dyskeratosis congenita. Am J Hum Genet, 82(2), 501–​509. PMCID: PMC2427222. SCENIHR (Scientific Committee on Emerging and Newly Identified Health Risks). 2015. Potential health effects of exposure to electromagnetic fields (EMF): http://​ec.europa.eu/​health/​scientific_​committees/​emerging/​ docs/​scenihr_​o_​041.pdf. Schlenk RF, Dohner K, Krauter J, et  al. 2008. Mutations and treatment outcome in cytogenetically normal acute myeloid leukemia. N Engl J Med, 358(18), 1909–​1918. PMID: 18450602. Schnatter AR, Glass DC, Tang G, Irons RD, and Rushton L. 2012. Myelodysplastic syndrome and benzene exposure among petroleum

workers:  an international pooled analysis. J Natl Cancer Inst, 104(22), 1724–​1737. PMCID: PMC3502195. Schwilk E, Zhang L, Smith MT, Smith AH, and Steinmaus C. 2010. Formaldehyde and leukemia: an updated meta-​analysis and evaluation of bias. J Occup Environ Med, 52(9), 878–​886. PMID: 20798648. SEER. 2015a. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database:  Incidence—​SEER 9 Regs Research Data, Nov 2014 Sub (1973–​ 2012) Katrina/​ Rita Population Adjustment—​ Linked to County Attributes—​ Total U.S., 1969–​ 2013 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2015, based on the November 2014 submission. SEER. 2015b. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Incidence—​SEER 13 Regs Research Data, Nov 2014 Sub (1992–​ 2012) Katrina/​ Rita Population Adjustment—​ Linked to County Attributes—​ Total U.S., 1969–​ 2013 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2015, based on the November 2014 submission. SEER. 2015c. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Incidence—​SEER 18 Regs Research Data + Hurricane Katrina Impacted Louisiana Cases, Nov 2014 Sub (2000–​2012) Katrina/​Rita Population Adjustment—​Linked to County Attributes—​Total U.S., 1969–​2013 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2015, based on the November 2014 submission. Seminog OO, and Goldacre MJ. 2013. Risk of benign tumours of nervous system, and of malignant neoplasms, in people with neurofibromatosis:  population-​based record-​linkage study. Br J Cancer, 108(1), 193–​ 198. PMCID: PMC3553528. Severson RK, Davis S, Thomas DB, et  al. 1989. Acute myelocytic leukemia and prior allergies. J Clin Epidemiol, 42(10), 995–​1001. PMID: 2809658. Severson RK, Stevens RG, Kaune WT, et al. 1988. Acute nonlymphocytic leukemia and residential exposure to power frequency magnetic fields. Am J Epidemiol, 128(1), 10–​20. PMID: 3381818. Shadman M, White E, De Roos AJ, and Walter RB. 2013. Associations between allergies and risk of hematologic malignancies: results from the VITamins and lifestyle cohort study. Am J Hematol, 88(12), 1050–​1054. PMCID: PMC4001851. Shiels MS, and Engels EA. 2012. Increased risk of histologically defined cancer subtypes in human immunodeficiency virus-​infected individuals: clues for possible immunosuppression-​related or infectious etiology. Cancer, 118(19), 4869–​4876. PMCID: PMC3366173. Shilnikova NS, Preston DL, Ron E, et al. 2003. Cancer mortality risk among workers at the Mayak nuclear complex. Radiat Res, 159(6), 787–​798. PMID: 12751962. Shimizu Y, Pierce DA, Preston DL, and Mabuchi K. 1999. Studies of the mortality of atomic bomb survivors. Report 12, part II. Noncancer mortality: 1950–​1990. Radiat Res, 152(4), 374–​389. PMID: 10477914. Shizuru JA, Negrin RS, and Weissman IL. 2005. Hematopoietic stem and progenitor cells: clinical and preclinical regeneration of the hematolymphoid system. Annu Rev Med, 56, 509–​538. PMID: 15660525. Shlush LI, and Minden MD. 2015. Preleukemia:  the normal side of cancer. Curr Opin Hematol, 22(2), 77–​84. PMID: 25575035. Shore RE, Moseson M, Harley N, and Pasternack BS. 2003. Tumors and other diseases following childhood x-​ray treatment for ringworm of the scalp (Tinea capitis). Health Phys, 85(4), 404–​408. PMID: 13678280. Shu XO, Gao YT, Linet MS, et al. 1987. Chloramphenicol use and childhood leukaemia in Shanghai. Lancet, 2(8565), 934–​937. PMID: 2889862. Siegel RL, Miller KD, and Jemal A. 2016. Cancer statistics, 2016. CA Cancer J Clin, 66(1), 7–​30. PMID: 26742998. Simon SL, Preston DL, Linet MS, et al. 2014. Radiation organ doses received in a nationwide cohort of U.S. radiologic technologists: methods and findings. Radiat Res, 182(5), 507–​528. PMCID: PMC4406479. Sinner PJ, Cerhan JR, Folsom AR, and Ross JA. 2005. Positive association of farm or rural residence with acute myeloid leukemia incidence in a cohort of older women. Cancer Epidemiol Biomarkers Prev, 14(10), 2446–​2448. PMID: 16214930. Smith A, Howell D, Patmore R, Jack A, and Roman E. 2011. Incidence of haematological malignancy by sub-​type: a report from the Haematological Malignancy Research Network. Br J Cancer, 105(11), 1684–​ 1692. PMCID: 3242607. Smith A, Roman E, Howell D, et  al. 2009. The Haematological Malignancy Research Network (HMRN): a new information strategy for population based epidemiology and health service research. Br J Haematol, 148(5), 739–​753. PMCID: 3066245.

 743

Leukemias Smith AG, Dovey GJ, and Cartwright RA. 2000. Topical chloramphenicol and the risk of acute leukaemia in adults. Pharmacoepidemiol Drug Saf, 9(3), 215–​219. PMID: 19025822. Smith ML, Cavenagh JD, Lister TA, and Fitzgibbon J. 2004. Mutation of CEBPA in familial acute myeloid leukemia. N Engl J Med, 351(23), 2403–​2407. PMID: 15575056. Soderberg KC, Hagmar L, Schwartzbaum J, and Feychting M. 2004. Allergic conditions and risk of hematological malignancies in adults:  a cohort study. BMC Public Health, 4, 51. PMCID: PMC534807. Sokal JE, Cox EB, Baccarani M, et  al. 1984. Prognostic discrimination in “good-​ risk” chronic granulocytic leukemia. Blood, 63(4), 789–​ 799. PMID: 6584184. Song WJ, Sullivan MG, Legare RD, et al. 1999. Haploinsufficiency of CBFA2 causes familial thrombocytopenia with propensity to develop acute myelogenous leukaemia. Nat Genet, 23(2), 166–​175. PMID: 10508512. Srour SA, Devesa SS, Morton LM, et al. 2016. Incidence and patient survival of myeloproliferative neoplasms and myelodysplastic/​myeloproliferative neoplasms in the United States, 2001–​2012. Br J Haematol, 174(3), 382–​ 396. PMID: 27061824. Stebbings JH. 1998. Dose-​response analyses of osteonecrosis in New Jersey radium workers point to roles for other alpha emitters. Health Phys, 74(5), 602–​607. PMID: 9570164. Stevens W, Thomas DC, Lyon JL, et al. 1990. Leukemia in Utah and radioactive fallout from the Nevada test site: a case-​control study. JAMA, 264(5), 585–​591. PMID: 2366297. Storm HH, Jorgensen HO, Kejs AM, and Engholm G. 2006. Depleted uranium and cancer in Danish Balkan veterans deployed 1992–​2001. Eur J Cancer, 42(14), 2355–​2358. PMID: 16857358. Strom SS, Yamamura Y, Kantarijian HM, and Cortes-​Franco JE. 2009. Obesity, weight gain, and risk of chronic myeloid leukemia. Cancer Epidemiol Biomarkers Prev, 18(5), 1501–​1506. PMCID: PMC2918285. Strullu M, Caye A, Lachenaud J, et  al. 2014. Juvenile myelomonocytic leukaemia and Noonan syndrome. J Med Genet, 51(10), 689–​ 697. PMID: 25097206. Swerdlow SH, Campo E, Harris NL, et  al (Eds.). 2008. WHO classification of tumours of haematopoietic and lymphoid tissues (4th ed.). Lyon: International Agency for Research on Cancer. Talbott EO, Xu X, Youk AO, et al. 2011. Risk of leukemia as a result of community exposure to gasoline vapors: a follow-​up study. Environ Res, 111(4), 597–​602. PMID: 21453914. Tapper W, Jones AV, Kralovics R, et al. 2015. Genetic variation at MECOM, TERT, JAK2 and HBS1L-​MYB predisposes to myeloproliferative neoplasms. Nat Commun, 6, 6691. PMCID: PMC4396373. Tefferi A, and Pardanani A. 2015. Myeloproliferative neoplasms: a contemporary review. JAMA Oncol, 1(1), 97–​105. PMID: 26182311. Thariat J, Marcy PY, Lacout A, et al. 2012. Radiotherapy and radiology: joint efforts for modern radiation planning and practice. Diagn Interv Imaging, 93(5), 342–​350. PMID: 22483122. Thygesen LC, Nielsen OJ, and Johansen C. 2009. Trends in adult leukemia incidence and survival in Denmark, 1943–​2003. Cancer Causes Control, 20(9), 1671–​1680. PMID: 19672681. Till JE, Beck HL, Aanenson JW, et  al. 2014. Military participants at U.S. Atmospheric nuclear weapons testing: methodology for estimating dose and uncertainty. Radiat Res, 181(5), 471–​484. PMCID: PMC4082990. Tomasek L, Darby SC, Swerdlow AJ, Placek V, and Kunz E. 1993. Radon exposure and cancers other than lung cancer among uranium miners in West Bohemia. Lancet, 341(8850), 919–​923. PMID: 8096265. Tong H, Hu C, Yin X, et al. 2013. A meta-​analysis of the relationship between cigarette smoking and incidence of myelodysplastic syndromes. PLoS One, 8(6), e67537. PMCID: PMC3689714. Traversa G, Menniti-​Ippolito F, Da Cas R, et al. 1998. Drug use and acute leukemia. Pharmacoepidemiol Drug Saf, 7(2), 113–​ 123. PMID: 15073735. Travis LB, Hauptmann M, Gaul LK, et al. 2003. Site-​specific cancer incidence and mortality after cerebral angiography with radioactive thorotrast. Radiat Res, 160(6), 691–​706. PMID: 14640794. Travis LB, Holowaty EJ, Bergfeldt K, et  al. 1999. Risk of leukemia after platinum-​based chemotherapy for ovarian cancer. N Engl J Med, 340(5), 351–​357. PMID: 9929525. Tucker JD. 2010. Chromosome translocations and assessing human exposure to adverse environmental agents. Environ Mol Mutagen, 51(8–​9), 815–​824. PMID: 20213842. Tynes T, and Haldorsen T. 2003. Residential and occupational exposure to 50 Hz magnetic fields and hematological cancers in Norway. Cancer Causes Control, 14(8), 715–​720. PMID: 14674735. Udayakumar AM, Surekhatony M, and Wali YA. 2014. Myelodysplasia progressing to acute myeloid leukemia in monozygotic twins with

743

monosomy 7 as sole abnormality. Leuk Lymphoma, 55(11), 2672–​2674. PMID: 24598016. U.S. Department of Health and Human Services. How Tobacco Smoke Causes Disease:  The Biology and Behavioral Basis for Smoking-​Attributable Diseases. A  Report of the Surgeon General. Atlanta, GA:  U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, 2010. Vardiman JW, Harris NL, and Brunning RD. 2002. The World Health Organization (WHO) classification of the myeloid neoplasms. Blood, 100(7), 2292–​2302. PMID: 12239137. Vardiman JW, Thiele J, Arber DA, et al. 2009. The 2008 revision of the World Health Organization (WHO) classification of myeloid neoplasms and acute leukemia:  rationale and important changes. Blood, 114(5), 937–​ 951. PMID: 19357394. Varley J, Evans D, and Birch J. 1997. Li-​Fraumeni syndrome: a molecular and clinical review. Br J Cancer, 76(1), 1–​14. Verkasalo PK, Pukkala E, Kaprio J, Heikkila KV, and Koskenvuo M. 1996. Magnetic fields of high voltage power lines and risk of cancer in Finnish adults:  nationwide cohort study. BMJ, 313(7064), 1047–​1051. PMCID: PMC2352388. Vijayakrishnan J, and Houlston R. 2010. Candidate gene association studies and risk of childhood acute lymphoblastic leukemia:  a systematic review and meta-​ analysis. Haematologica, 95(8), 1405–​ 1414. PMID: 20511665. Visser O, Trama A, Maynadie M, et al. 2012. Incidence, survival and prevalence of myeloid malignancies in Europe. Eur J Cancer, 48, 2357–​66. PMID 22720878 Vlaanderen J, Lan Q, Kromhout H, Rothman N, and Vermeulen R. 2012. Occupational benzene exposure and the risk of chronic myeloid leukemia: a meta-​analysis of cohort studies incorporating study quality dimensions. Am J Ind Med, 55(9), 779–​785. PMID: 22729623. Voso MT, Fenu S, Latagliata R, et al. 2013. Revised International Prognostic Scoring System (IPSS) predicts survival and leukemic evolution of myelodysplastic syndromes significantly better than IPSS and WHO Prognostic Scoring System:  validation by the Gruppo Romano Mielodisplasie Italian Regional Database. J Clin Oncol, 31(21), 2671–​ 2677. PMID: 23796988. Wakeford R. 2009. Radiation in the workplace-​a review of studies of the risks of occupational exposure to ionising radiation. J Radiol Prot, 29(2A), A61–​79. PMID: 19454806. Wang P, Liu H, Jiang T, and Yang J. 2015. Cigarette smoking and the risk of adult myeloid disease:  a meta-​analysis. PLoS One, 10(9), e0137300. PMCID: PMC4560392. Wei L, and Sugahara T. 2000. An introductory overview of the epidemiological study on the population at the high background radiation areas in Yangjiang, China. J Radiat Res, 41 Suppl, 1–​7. PMID: 11142208. Weiss HA, Darby SC, Fearn T, and Doll R. 1995. Leukemia mortality after X-​ray treatment for ankylosing spondylitis. Radiat Res, 142(1), 1–​11. PMID: 7899552. Weiss JR, Baker JA, Baer MR, et al. 2006. Opposing effects of aspirin and acetaminophen use on risk of adult acute leukemia. Leuk Res, 30(2), 164–​169. PMID: 16099041. Wheatley K, Brookes CL, Howman AJ, et al. 2009. Prognostic factor analysis of the survival of elderly patients with AML in the MRC AML11 and LRF AML14 trials. Br J Haematol, 145(5), 598–​605. PMID: 19344426. Wick RR, Nekolla EA, Gaubitz M, and Schulte TL. 2008. Increased risk of myeloid leukaemia in patients with ankylosing spondylitis following treatment with radium-​ 224. Rheumatology (Oxford), 47(6), 855–​ 859. PMID: 18390588. Wilkinson P, Thakrar B, Walls P, et al. 1999. Lymphohaematopoietic malignancy around all industrial complexes that include major oil refineries in Great Britain. Occup Environ Med, 56(9), 577–​580. PMCID: PMC1757785. Williams RR, and Horm JW. 1977. Association of cancer sites with tobacco and alcohol consumption and socioeconomic status of patients: interview study from the Third National Cancer Survey. J Natl Cancer Inst, 58(3), 525–​547. PMID: 557114. Wong O, Harris F, Yiying W, and Hua F. 2009. A hospital-​based case-​control study of acute myeloid leukemia in Shanghai:  analysis of personal characteristics, lifestyle and environmental risk factors by subtypes of the WHO classification. Regul Toxicol Pharmacol, 55(3), 340–​ 352. PMID: 19703505. Wright JD, St Clair CM, Deutsch I, et al. 2010. Pelvic radiotherapy and the risk of secondary leukemia and multiple myeloma. Cancer, 116(10), 2486–​ 2492. PMID: 20209618. Xavier AC, and Taub JW. 2010. Acute leukemia in children with Down syndrome. Haematologica, 95(7), 1043–​1045.

74

744

PART IV:  Cancers by Tissue of Origin

Yamamura Y, Oum R, Gbito KY, Garcia-​Manero G, and Strom SS. 2013. Dietary intake of vegetables, fruits, and meats/​beans as potential risk factors of acute myeloid leukemia: a Texas case-​control study. Nutr Cancer, 65(8), 1132–​1140. PMID: 24168094. Yenson PR, Forrest D, Schmiegelow K, and Dalal BI. 2008. Azathioprine-​associated acute myeloid leukemia in a patient with Crohn’s disease and thiopurine S-​ methyltransferase deficiency. Am J Hematol, 83(1), 80–​83. PMID: 17696202. Yoshinaga S, Mabuchi K, Sigurdson AJ, Doody MM, and Ron E. 2004. Cancer risks among radiologists and radiologic technologists:  review of epidemiologic studies. Radiology, 233(2), 313–​321. PMID: 15375227. Zablotska LB, Bazyka D, Lubin JH, et al. 2013. Radiation and the risk of chronic lymphocytic and other leukemias among Chornobyl cleanup workers. Environ Health Perspect, 121(1), 59–​65. PMCID: PMC3553431. Zablotska LB, Lane RS, Frost SE, and Thompson PA. 2014. Leukemia, lymphoma and multiple myeloma mortality (1950–​ 1999) and incidence

(1969–​1999) in the Eldorado uranium workers cohort. Environ Res, 130, 43–​50. PMCID: PMC4002578. Zhang L, Tang X, Rothman N, et al. 2010. Occupational exposure to formaldehyde, hematotoxicity, and leukemia-​specific chromosome changes in cultured myeloid progenitor cells. Cancer Epidemiol Biomarkers Prev, 19(1), 80–​88. PMCID: PMC2974570. Zhang Y, Kim C, and Zheng T. 2012. Hair dye use and risk of human cancer. Front Biosci (Elite Ed), 4, 516–​528. PMID: 22201892. Zheng W, Linet MS, Shu XO, et  al. 1993. Prior medical conditions and the risk of adult leukemia in Shanghai, People’s Republic of China. Cancer Causes Control, 4(4), 361–​368. PMID: 8347786. Zhou J, Eksioglu EA, Fortenbery NR, et al. 2011. Bone marrow mononuclear cells up-​regulate toll-​like receptor expression and produce inflammatory mediators in response to cigarette smoke extract. PLoS One, 6(6), e21173. PMCID: PMC312.

 745

39

Hodgkin Lymphoma HENRIK HJALGRIM, ELLEN T. CHANG, AND SALLY L. GLASER

OVERVIEW Hodgkin lymphoma (HL) is a malignant neoplasm of the lymphatic system. The malignant cell clone derives from germinal center B lymphocytes in ~98% of cases, the rest being of T-​lymphocyte origin. Each year, HL is diagnosed in roughly 66,000 individuals worldwide. HL is curable with modern therapy in the vast majority of patients, with 5-​ year survival rates exceeding 90% for early-​stage disease. However, so far this excellent prognosis has been achieved at the expense of a high incidence of severe long-​term treatment complications such as secondary malignancies, and endocrine and cardiovascular diseases. In affluent Western countries, HL occurrence follows a distinctive and unusual bimodal age distribution, with one incidence peak among adolescents and younger adults and another in older adults. In these populations, the early peak renders HL among the most common malignancies in adolescents and younger adults. In socioeconomically less affluent populations, in contrast, the adolescent and younger-​adult incidence peak is less pronounced, whereas incidence of HL in young boys may be higher than in affluent populations. It is believed that this difference in age-​specific incidence patterns reflects that HL in children and in adolescents and younger adults is influenced by socioeconomic environment in childhood. HL incidence has increased among adolescents and younger adults periodically in recent decades, while it has been stable or even decreased in older adults. The causes of HL remain incompletely characterized, but infection with the Epstein-​Barr virus (EBV) is considered causally associated with the roughly one-​third of all HL cases that harbor the virus in the malignant cells. There are epidemiological differences between EBV-​positive and EBV-​negative HL, and in particular risk of EBV-​positive HL is influenced by immunological factors. No infectious agent has been associated convincingly with the large EBV-​negative HL subset, and no specific risk factors have been identified consistently for this group.

TUMOR CLASSIFICATION Clinical Presentation HL may occur at all ages, but typical patients are young adults presenting with painless lymphadenopathy, most commonly of the neck (Stein, 2008; Stein and Morgan, 2004). The patients may have nonspecific symptoms, such as fever, night sweats, itching of the skin, alcohol intolerance, weight loss, and/​or fatigue. The prevalence of such symptoms varies by disease stage, from less than 10% in patients with stage I to two-​thirds in patients with stage IV disease (Stein and Morgan, 2004). Definitive diagnosis is made by tumor biopsy with subsequent histological examination (Stein and Morgan, 2004).

Histology and Cell of Origin Histologically, HL lesions are exceptional in that the malignant cells typically make up less than 1% of the tumor lesion. These cells are most often characteristic large, multinucleated cells with a slightly basophilic cytoplasm, so-​called Reed-​Sternberg cells, or mononuclear variants thereof, referred to as Hodgkin, mummified,

lymphocyte-​predominant, or popcorn cells, depending on their morphology (Poppema et al., 2008; Stein et al., 2008). Instead of proliferating malignant cells, the tumor is dominated by an abundant admixture of non-​neoplastic inflammatory and accessory cells (i.e., B and T lymphocytes, plasma cells, eosinophils, neutrophils, histiocytes, and fibroblasts) (Stein et al., 2008). Often, CD4-​positive T lymphocytes can be seen surrounding the malignant cells to form rosette-​like structures (Stein, 2008). Because the Reed-​Sternberg cells are scarce, their origin was challenging to resolve. However, it is now understood that these malignant cells in most instances are outgrowths from mature germinal center B-​lymphocyte clones (Kanzler et al., 1996), and in rare cases (~2%) of T-​lymphocyte origin (Muschen et al., 2000). Consequently, the historical name Hodgkin’s disease was substituted by the name HL around the turn of the twenty-​first century (Jaffe et al., 1998).

WHO Classification For HL, the current WHO classification has evolved from the 1966 Lukes-​Butler/​Rye classification (Lukes et al., 1966). A most important modification between the previous and the current classifications is that whereas the Rye classification considered all variants of HL as phenotypes of the same disease, WHO has since 2001 recognized two broad forms of HL. These are classical HL (cHL), which makes up ~95% of all cases, and nodular lymphocyte predominant HL (Table 39–1) (Poppema et al., 2008; Stein et al., 2008). The two forms of HL differ in clinical behavior and with respect to tumor lesion microenvironment, tumor cell morphology and, importantly from a diagnostic viewpoint, tumor cell immune phenotype (Stein, 2008). Under the assumption that classical and nodular lymphocyte predominant HL also have different etiologies, they are now often considered separately in epidemiological investigations. In accordance with this distinction, the present chapter will consider the two separately when possible and will mostly address cHL. Where the two subtypes are not considered independently, data can be assumed to apply mainly to cHL, since it represents the vast majority of all HL.

Subtypes of cHL

Based on the composition of the tumor microenvironment and the morphology of the malignant cells, four variants of cHL are recognized currently. These are referred to as nodular sclerosis, mixed cellularity, lymphocyte-​rich, and lymphocyte-​depleted cHL (Table 39–1). Although archetypical cases of the different cHL variants may be distinguished relatively easily, studies have demonstrated that histological subtyping of cHL is prone to considerable inter-​observer variation (Glaser et al., 2001; Glaser and Swartz, 1990; Jarrett et al., 2003). This variability is important because the histological subtype of cHL historically plays an important role in the effort to understand the epidemiology of the lymphoma.

Clinical Relevance of Hodgkin Lymphoma Classification To the vast majority of patients, the sub-​classification of HL is of limited value to treatment allocation. Rather, the choice of treatment first and foremost rests on clinical parameters, which are used to classify

745

746

746

PART IV:  Cancers by Tissue of Origin

Table 39–1. Historical and Current Classification of Hodgkin Lymphoma Jackson and Parker, 1947 Granuloma Sarcoma Paragranuloma

Rye Conference, 1966 Nodular sclerosis Mixed cellularity Lymphocytic depletion Lymphocytic predominance

WHO, 2008 Nodular sclerosis Mixed cellularity Lymphocyte-​depleted Lymphocyte-​rich Nodular lymphocyte predominant

Proportion ~ 70% ~ 20% < 1% ~ 5% ~ 5%

Note: Congruence between subtypes in different classifications is not perfect. Source: Reproduced from Hjalgrim (2012).

patients as having early-​stage or advanced-​stage disease according to the Ann Arbor staging system, and for the group of patients with advanced stage disease further by the International Prognostic Score (Derenzini and Younes, 2011; Rathore and Kadin, 2010). Situations where HL histological classification is currently considered clinically relevant include recommendation of more lenient treatment for patients with early-​stage nodular lymphocyte predominant HL and the use of cHL histology to distinguish between patients with favorable and unfavorable early-​stage disease in some protocols (Derenzini and Younes, 2011; Rathore and Kadin, 2010).

INCIDENCE HL is diagnosed in ~3,800 females and ~4,800 males in the United States annually, yielding age standardized (world) rates of 2.2 per 100,000 in women and 2.8 per 100,000 in men. HL has a favorable prognosis with modern treatment, and 5-​year survival rates are 80% or above for the patient group as a whole. The high survival is reflected in the corresponding age-​standardized (world) mortality rates of 0.2 per 100,000 in women and 0.3 per 100,000 in men in the United States (Ferlay et al., 2013). Worldwide, an estimated 27,500 females and 38,500 males were diagnosed with HL in 2012, with HL making up ~0.5% of all malignancies (Forman et al., 2013). These figures correspond to annual age-​ standardized (world) incidence rates of 0.7 per 100,000 in women and 1.1 per 100,000 in men (Ferlay et al., 2013). The incidence of HL varies with level of socioeconomic development between and within populations and is 3–​4-​fold higher (age-​ standardized [world] 2.1 per 100,000 for both sexes combined) in more developed regions of the world than in less developed regions (0.6 per 100,000 for both sexes combined) (Figure 39–1) (Ferlay et al., 2013). HL incidence, moreover, varies by race and ethnicity beyond what is attributable to socioeconomic differences. For instance, incidence rates are markedly lower in Japan (age standardized [world] 0.5 per 100,000 both sexes combined) and Singapore (0.8 per 100,000 both sexes) than in other comparably developed regions outside Asia, for example the European Union (2.2 per 100,000 both sexes combined) and North America (2.4 per 100,000 both sexes combined) (Ferlay et al., 2013). Worldwide, the approximate number of deaths attributed to HL in 2012 was 25,500. This total translates into age-​standardized mortality rates of 0.3 per 100,000 per year in women (10,000 deaths) and 0.4 per 100,000 in men (15,500 deaths) (Figure 39–1) (Ferlay et al., 2013). It is noteworthy that, despite incidence differences, HL mortality rates are similar or even higher in less developed regions than in more developed regions (Figure 39–1). The low mortality rate in more developed regions underscores the good prognosis that HL generally has, provided access to modern treatment. At the same time, the similar mortality rate in less developed regions, despite lower incidence rates, highlights the inadequate access to diagnosis, treatment, and follow-​up care in these populations (Chatenoud et al., 2013). The potentially good prognosis of HL should not lead to the conclusion that further research into the etiology of HL is unwarranted. Once a nearly invariably fatal disease, HL is today cured in 90% or

more of those with early-​stage disease (Armitage, 2010). However, it has become increasingly clear that the high cure rates in HL have been achieved at the price of a high frequency of serious treatment side effects, including cardiovascular, endocrine, and pulmonary diseases, and secondary malignancies, in addition to psychological disorders related to a cancer diagnosis per se early in life (e.g., Glimelius et al., 2015; Matasar et al., 2015). This burden supports the utility of disease prevention and, thus, the continued need for better understanding of the causes of HL.

Age-​Specific Incidence Patterns: HL Overall Patterns of age-​specific HL incidence vary markedly both between and within populations. The characterization of this variation has been pivotal to the formulation of epidemiological models for HL and has led to the tradition of considering HL in children (age 0–​14 years), adolescents and younger adults (age 15–​34 years), and older adults (age 50+ years) as separate entities (see later discussion) (Correa and O’Conor, 1971; Gutensohn and Cole, 1977; MacMahon, 1957; 1966). In most affluent, predominantly Caucasian populations, a hallmark of HL epidemiology is the conspicuous bimodal distribution of age-​specific incidence rates, with two distinct incidence peaks, one among adolescents and younger adults and another among older adults (Figure 39–2). HL incidence before the age of 10 years is generally low in these settings. However, analyses of childhood HL incidence in the Nordic countries in the period 1978–​2010 (Hjalgrim et al., 2016) and across 19 European countries in the period 1978–​1997 (Clavel et al., 2006) have pointed out a small well-​defined childhood incidence peak, culminating around the age of 5 years and particularly prominent among boys. Statistical modeling of the Nordic data suggested that HL diagnosed before the age of 8 years and HL diagnosed after the age of 8 years are epidemiologically distinct, and that HL in children older than 8 years is likely to be epidemiologically similar to HL in adolescents and younger adults (Hjalgrim et al., 2016). Incidence patterns differ slightly between males and females. In affluent populations, HL incidence is higher among males in all age groups except in adolescence and younger adulthood, when it may be highest among females (Figure 39–2). It has been known since the 1970s that HL incidence rates among adolescents and younger adults are positively correlated with population-​level socioeconomic development (Correa and O’Conor, 1971)  (see discussion in Mueller and Grufferman, 2006). The presence and persistence over time of this socioeconomic pattern in adolescents and younger adults were most convincingly demonstrated by analyses of neighborhood socioeconomic status of 3794 HL patients diagnosed in California in the years 1988–​1992. Here, Clarke and colleagues observed incidence rate ratios of 1.22 (95% confidence interval [CI] 1.05–​1.42) in young-​adult males and of 1.44 (95% CI 1.22–​1.70) in young-​adult females residing in the highest compared with the lowest tertiles of socioeconomic status at diagnosis (Clarke et al., 2005) (Figure 39–3). Of note, regardless of the variation in absolute incidence rates, the bimodal age-​incidence distribution was apparent in each of the population strata (Figure 39–3). Conversely, among children, especially boys, HL incidence was reported in the 1970s to be higher in socioeconomically deprived populations than in affluent populations (Correa and O’Conor, 1971). The subsequent evidence for socioeconomic variation in childhood HL incidence is not overwhelming. This may in part reflect that HL is rare in childhood and that regional cancer registration infrastructure depends on level of socioeconomic development. However, even with these reservations, higher HL incidence rates among boys aged less than 10 years are seen in less developed countries (e.g., in Latin American and Indian cancer registries) than in the United States (Figure 39–2) (Forman et al., 2013). For HL among older adults, no strong association with socioeconomic status has been established, despite original observations suggesting a relationship to affluence (MacMahon, 1966).

 74



747

Hodgkin Lymphoma Hodgkin lymphoma ASR (W) per 100,000, all ages Male

Female

Northern America Western Europe Australia/New Zealand Southern Europe Northern Europe More developed regions Western Asia Central and Eastern Europe Northern Africa Central America South America Caribbean World Southern Africa Eastern Africa South-Central Asia Less developed regions Western Africa Middle Africa Micronesia Melanesia South-Eastern Asia Eastern Asia Polynesia 3

2

1

0 Incidence

1 Mortality

As already mentioned, HL incidence overall also varies by ethnicity and/​or race. This variation is apparent in age-​specific incidence rates, for example in the United States, where HL incidence rates are generally highest among non-​Hispanic whites and lowest in Asian Americans across all age groups (Figure 39–4). Again, the bimodal age-​distribution is visible in all subpopulations (Figure 39–4). The race-​specific variation in HL incidence is likely to reflect, to some extent, differences in socioeconomic status. However, in the previously mentioned California survey, racial/​ethnic variations in HL incidence were also apparent within strata of socioeconomic levels, suggesting that other factors also contribute (Clarke et al., 2005). In this regard, Glaser and colleagues recently reported that HL incidence rates among Hispanics and Asians/​ Pacific Islanders in California varied significantly by neighborhood measures of acculturation, independently of neighborhood socioeconomic status (Glaser et al., 2015a). This finding indicates that at least for these two racial/​ ethnic groups, certain lifestyle characteristics that are not fully captured by traditional measures of socioeconomic status influence HL occurrence.

Age-​Specific Incidence Patterns: cHL Variants

The HL age-​incidence peaks and their respective associations with sex and level of socioeconomic development mirror the distributions of the nodular sclerosis and mixed cellularity cHL variants. In affluent populations, the incidence of nodular sclerosis cHL manifests a bimodal age pattern; however, unlike in cHL overall, the incidence peak among adolescents and younger adults is far more prominent than the peak in

2

3

Figure 39–1.  Age-​standardized incidence and mortality rates for Hodgkin lymphoma by sex and geographic region and regional development. Source: Ferlay et al. (2013).

older adults (Figure 39–5). Accordingly, nodular sclerosis is the most common variant of cHL in affluent populations, especially among adolescents and younger adults. By inference from the geographic variation for HL overall, nodular sclerosis cHL incidence in younger adults is associated with higher socioeconomic status, as demonstrated in data from California (Clarke et  al., 2005). Nodular sclerosis cHL is more common in females than in males in adolescence and young adulthood, but not in older adulthood. Mixed cellularity cHL incidence in affluent countries increases gradually with age, with a small incidence peak in younger adulthood (Figure 39–5). It may be the most common of the cHL variants only in the youngest children (Clavel et  al., 2006; Englund et  al., 2015). Some evidence (Cozen et al., 1992), but not all (Clarke et al., 2005), suggests that risk of mixed cellularity cHL correlates inversely with socioeconomic status. Across all age groups, including children, mixed cellularity cHL is more common in males than in females. Lymphocyte-​rich and lymphocyte-​depleted cHL incidence rates also tend to increase with age, but as seen in Table 39–1 and in Figure 39–5, these subtypes constitute only a very small proportion of all cases and, therefore, are not well studied.

Age-​Specific Incidence Patterns: cHL Epstein-​Barr Virus Status

Clonal EBV is present in the malignant cells in a proportion of cHL, and in epidemiological studies tumor EBV status may serve as an alternative to the traditional morphological classification of cHL. Based on

748

748

PART IV:  Cancers by Tissue of Origin 7.00

6.00

Rate/100,000/year

5.00

4.00

3.00

2.00

1.00

0.00

0

10

20

30

40

50

60

70

80

90

100

Age (years) at diagnosis Indian males

Indian females

US males

US females

Figure 39–2.  Age-​specific incidence rates of Hodgkin lymphoma in US (black lines) and Indian (grey lines) males (full lines) and females (broken lines) in the period 2003–​2007. Source: Forman et al. (2013).

distributions of EBV positivity identified in analyses of tumor series and on known risk factors, the age-​specific incidence of EBV-​positive cHL is believed to have three peaks: one in early childhood (primarily among boys), one in adolescents and younger adults, and one in older adults. In contrast, the age-​specific incidence of EBV-​negative cHL is expected to have a more or less unimodal distribution, peaking in younger adults (Jarrett, 2002).

2

In a population-​based investigation, Jarrett et al. determined the tumor EBV status of 537 cases of cHL diagnosed in patients age 16–​74 years in Scotland and Newcastle, United Kingdom, between 1993 and 1997 and used this information to estimate age-​and sex-​ specific incidence rates for EBV-​positive and EBV-​negative cHL (Figure 39–6). As predicted (Jarrett, 2002), the age-​ incidence curve for EBV-​negative HL was largely unimodal and peaked in younger adults, whereas the distribution of EBV-​positive cHL featured separate peaks in younger adults and in older adults (Jarrett et al., 2003). Of note, EBV-​positivity is more common in mixed cellularity than in nodular sclerosis cHL (Glaser et al., 1997; Lee et al., 2014). This pattern is often construed as evidence that EBV-​positive cHL most often is of the mixed cellularity subtype. However, because nodular sclerosis cHL is considerably more common than the other HL subtypes, it makes up half or more of all EBV-​positive cHL cases (Chang et al., 2004b; Glaser et al., 2008; Hjalgrim et al., 2007; Huang et al., 2012b; Jarrett et al., 2003). Thus, although the age-​incidence curve for EBV-​positive cHL would be expected to resemble that for mixed cellularity cHL, that for EBV-​negative cHL may not track as closely with the distribution of nodular sclerosis cHL.

1

Incidence Trends

7 6

Rate/100,000

5 4 3

0 0–4 10–14 20–24 30–34 40–44 50–54 60–64 70–74 80–84 Age at diagnosis High

Medium

Low

Figure  39–3. Age-​ specific incidence rates of Hodgkin lymphoma by tertile of neighborhood socioeconomic status, California, 1988–​ 1992. Source: Reproduced with permission from Clarke et al. (2005).

There has been no general trend in overall HL incidence in recent decades, although closer examination reveals some subgroup-​specific trends, as reviewed in Hjalgrim (2012). In the Nordic countries between 1990 and 2013, age-​adjusted incidence has increased slightly among women, but has remained stable among men (Engholm et al., 2010). In the United States over the period 1975–​2012, trends have differed by gender and race/​ethnicity: incidence has decreased among men and increased slightly among black women. In the period 1992–​ 2012, US rates increased among black men and Asian women and men (Howlader et al., 2014).

 749



749

Hodgkin Lymphoma 7

Incidence per 100,000 per year

6 5 4 3 2 1 0

00–09 yr 10–19 yr 20–29 yr 30–39 yr 40–49 yr 50–59 yr 60–69 yr 70–79 yr

80+ yr

Age group Whites

Blacks

Hispanics

Asian/Pacific Islanders

Figure 39–4.  Age-​specific incidence rates for Hodgkin lymphoma overall in both sexes combined by race/​ethnicity in the United States (2008–​2012); US SEER 18. Source: Howlader et al. (2014).

Reported HL incidence trends have also varied considerably by age within populations. Thus, in various periods since the 1960s, HL incidence has been reported to have increased among adolescents and younger adults in several regions worldwide, including Europe, North America, Israel, and Singapore (reviewed in Hjalgrim, 2012). Typically, rates of increase have been higher among females than among males, and often have been particularly pronounced among adolescents as compared with younger adults. In the Nordic countries, for instance, between 1978 and 1997, annual percentage

increases in incidence rates were 5.9% (95% CI 4.1%–​7.7%) and 1.8% (95% CI 0.6%–​3.0%) among females age 10–​19 years and 20–​ 29 years, respectively. In Nordic males, the corresponding figures were 3.6% (95% CI 2.0%–​5.1%) and 1.1% (95% CI 0.0%–​2.1%) (Hjalgrim et al., 2001). Among older adults, in contrast, incidence rates have decreased in the same period in many regions (Hjalgrim, 2012). The decreasing incidence among older adults to some extent reflects improved diagnostic precision, as it is well documented that misclassification of

3.5

Incidence per 100,000 per year

3 2.5 2 1.5 1 0.5 0

00–09 yr

10–19 yr

20–29 yr 30–39 yr 40–49 yr 50–59 yr 60–69 yr 70–79 yr

80+ yr

Age group Nodular sclerosis

Mixed cellularity

Lymphocyte rich

Lymphocyte depleted

Not otherwise specified

Figure 39–5.  Age-​specific incidence rates for classical Hodgkin lymphoma in both sexes combined by histologic subtype in the United States (2000–​ 2011); US SEER 18. Source: Howlader et al. (2014).

750

750

PART IV:  Cancers by Tissue of Origin (a)

6

Rate/100,000

5 4 3 2 1 0

(b)

20

30

40 50 Age (years)

60

70

2.5

Rate/100,000

2.0 1.5 1.0

and older adults (50+ years) are etiologically heterogeneous conditions (MacMahon, 1957, 1966). Thus, theoretically the remarkable age-​specific incidence distribution in Western industrialized populations essentially represents the superimposition of distributions for the younger-​adult and older-​adult disease entities. In its original form, this model held that HL in adolescents and younger adults is of an infectious origin—​ likely a chronic granulomatous inflammation—​ whereas HL in older adults is a more traditional neoplastic disease (MacMahon, 1966). The late infection or polio model centers on HL in children and in adolescents/​younger adults and elaborates on the suspicion of an infectious cause of HL. Hence, the model speculates that HL, like paralytic polio, may be a rare manifestation of a common childhood infection, and that the risk of HL increases with age at infection (Correa and O’Conor, 1971; Gutensohn and Cole, 1977; Newell, 1970). Assuming that exposure to such an agent in early childhood—​ as infectious disease exposure in general—​ differs between socioeconomically deprived (higher exposure) and affluent (lower exposure) settings, the late infection model would explain the affluence-​dependent variation in age-​specific HL incidence patterns.

Definition of HL Subgroups in Epidemiological Research

0.5

non-​Hodgkin lymphomas as HL previously inflated reported HL incidence rates, especially among older adults (Glaser et al., 2001; Glaser and Swartz, 1990; Jarrett et al., 2003). Presumably, however, correction of this error cannot entirely explain the decreasing incidence in this age group. When information on histological subtype has been available, the increase in overall HL incidence among adolescents and younger adults appears to reflect increasing incidence mainly of nodular sclerosis cHL (reviewed in Hjalgrim, 2012). The incidence of other and unspecified histological variants of HL has mostly remained stable or has decreased. However, as pointed out previously, the interpretation of the diverging trends should be made with caution because of difficulties and changes over time in cHL sub-​classification (Glaser et al., 2001; Glaser and Swartz, 1990; Jarrett et al., 2003), with changes in diagnostic procedures, e.g., the use of excisional biopsies securing sufficient material for sub-​classification, adding another layer of complexity to the interpretation of incidence data, as discussed in Glaser et al. (2015b).

In terms of epidemiology, the suggested etiologic heterogeneity in HL may be detrimental to efforts to identify risk factors if it is ignored (Cole et al., 1968). However, almost constituting a circular argument, the division of HL into etiologically distinct entities amenable for risk factor studies requires some form of classification that reflects their respective etiologies with reasonable accuracy. It was in the absence of such a classification that MacMahon first proposed the three age groups at diagnosis, mentioned previously, to approximate the supposedly etiologically separate entities (MacMahon, 1957, 1966). Although congruence is not 100% between previous and current classification schemes, the histological variants of cHL that are recognized today were essentially described nearly 50  years ago (Table 39–1) (Lukes et al., 1966). As previously discussed, the cHL subtypes are unevenly distributed among age groups and populations of different socioeconomic status. Consequently, the histological subtypes, specifically nodular sclerosis and mixed cellularity HL, have also been suggested to have separate etiologies (Cozen et  al., 1992; MacMahon et  al., 1971; Morton et  al., 2007; Newell et al., 1970). More recently, HL EBV status has been proposed as a marker of etiological heterogeneity (Armstrong et al., 1998; Jarrett et al., 1991). As already mentioned and described further later in the chapter, the proportion of HL that is positive for EBV in a particular population depends on factors such as age, sex, race/​ethnicity, socioeconomic status, and histological subtype (Glaser et al., 1997; Glaser et al., 2008). Thus, analogous to histology-​specific variation explaining some of the epidemiological variation in age-​specific HL occurrence, differences in the epidemiology of EBV-​positive and EBV-​negative HL might explain epidemiological variation between the different histological variants of cHL.

EPIDEMIOLOGICAL MODELS AND GROUPINGS

EBV INFECTION AND HODGKIN LYMPHOMA

Historically, the variation in age-​specific HL incidence between and within populations of different socioeconomic development fostered two disease models for HL: the multiple diseases and the late infection models. These models have dominated epidemiological research on HL for decades and continue to constitute its interpretational frame of reference. Although the models are not mutually exclusive, they are presented separately here in keeping with tradition. Briefly, the multiple diseases model rests on the observed bimodal age distribution of HL in affluent countries and posits that HL in children (0–​14  years), adolescents and younger adults (15–​34  years),

Because the evidence for a causal association between EBV infection and HL makes up a coherent body of scientific literature, it is presented here separately from the discussion of other HL risk factors. The EBV is a ubiquitous lymphotrophic herpesvirus that chronically infects the majority of the world’s adult populations and which has been associated with several different types of cancer (see Francheschi et al., Chapter 24 of this volume). Several aspects of the epidemiology of EBV infection are compatible with the scenario laid out by the late infection model of HL. There is a strong correlation between level of socioeconomic development and age at primary infection with the

0.0

20

30

40

50

60

70

Age (years)

Figure 39–6.  Top panel: Age-​specific incidence rates for classical Hodgkin lymphoma overall (dotted line) and for EBV-​negative (squares) and EBV-​ positive (triangles) classical Hodgkin lymphoma. Lower panel:  Age-​ specific incidence rates for EBV-​positive Hodgkin lymphoma in males (squares) and females (triangles). Source:  Reproduced with permission from Jarrett et al. (2003).

 751



751

Hodgkin Lymphoma

virus, which generally occurs at younger ages in socioeconomically deprived populations than in affluent populations (see Chapter 24). Moreover, primary EBV infection typically carries no or only mild symptoms when it occurs in childhood. In contrast, with primary EBV infection in adolescence or later, upwards of 50% or more of infected individuals will experience so-​called infectious mononucleosis (IM) (Balfour et al., 2013; Crawford et al., 2006). EBV infection has been implicated in HL etiology by molecular-​ biological, serological, and classical epidemiological lines of evidence.

Molecular-​Biological Evidence The strongest evidence that EBV causes cHL development is the demonstration of the virus in all the malignant cells in a subset of patients (Pallesen et al., 1991; Poppema et al., 1985; Uccini et al., 1989; Weiss et al., 1987). The EBV episome in the Reed-​ Sternberg cells is monoclonal, meaning that the infection occurred prior to the malignant transformation that gave rise to the cells. In the Reed-​Sternberg cells, the virus expresses three proteins that are produced during the latent phase of EBV infection: EBV nuclear antigen (EBNA) 1, latent membrane protein (LMP) 1, and LMP 2A (Kuppers, 2009). This pattern of EBV antigen expression is known as EBV latency pattern type II, which is also seen in nasopharyngeal carcinoma (see Chang and Hildesheim, Chapter 26 of this volume). These three viral products are known to affect the infected cell in ways that plausibly could explain the tumor’s development (Kuppers, 2009). The Reed-​Sternberg cells are, as mentioned earlier, believed to be outgrowths from germinal center B cells, as evidenced by clonal immunoglobulin (Ig) gene rearrangements. However, the cells lack B-​cell receptors, which in roughly 25% of cases can be attributed to mutations that interfere with the normal coding capacity of functional Ig genes (Kuppers, 2009; Vockerodt et al., 2015). Under normal circumstances, these mutations would result in apoptosis, but in in vitro studies, EBV infection can prevent apoptosis by expression of LMP2, which evokes signals mimicking an active B-​ cell receptor. The expression of LMP1 by EBV may also help the crippled B cell survive. Accordingly, expression of LMP1 mimics an activated CD40 receptor, which in turn activates the NF-​κB, JAK/​ STAT, and phosphatidylinositol-​3-​kinase/​AKT signaling pathways, all of which are constitutively activated in Reed-​Sternberg cells and that may contribute to B-​cell immortalization, as well as to abnormal cell differentiation and proliferation (Kuppers, 2009; Vockerodt et al., 2015). As summarized by Farrell and Jarrett (2011), a number of molecular-​ biological differences between EBV-​ positive and EBV-​ negative cHL further support a causal role for EBV in EBV-​positive cHL. Accordingly, EBV-​negative cHL displays mutations of genes encoding inhibitors of NF-​κB, in particular TNFAIP3 (A20), and also differs from EBV-​positive cHL regarding the expression of multiple receptor tyrosine kinase pathways (Farrell and Jarrett, 2011).

EBV Prevalence in HL The prevalence of EBV in HL variants has been assessed in more than 120 studies (Lee et  al., 2014). Collectively, the many studies have cemented the observation that the virus is present only in a subset of HL and that its prevalence varies non-​randomly by demographic and tumor characteristics (Glaser et  al., 1997, 2008; Lee et al., 2014). Data on HL EBV status for more than 13,000 patients reported in the literature as of 2012 were recently summarized. Overall, 48% of the investigated tumors were EBV-​positive. In univariate analyses, HL EBV positivity was higher in males (44%) than in females (27%); higher in patients younger than 15 years (70%) than in older patients (41%); higher in mixed cellularity (66%), lymphocyte-​ depleted (52%) and lymphocyte-​rich (47%) than in nodular sclerosis cHL (29%); and, finally, higher in patients in studies from Africa (74%), Central and South America (61%), and Asia (56%) than in

patients in studies from Europe (36%) and North America (32%) (Lee et al., 2014). The preceding findings are not immediately translatable as predictors of HL EBV-​positivity, partly because of the selective nature of several of the studied patient populations, and partly because the different patient and tumor characteristics are mutually related (see section in this chapter on incidence). For instance, mixed cellularity cHL is more common in men than in women and more common in deprived than in affluent settings. However, the observed variation in HL EBV prevalence among strata of patients and the associations implied by it nevertheless correspond fairly well with the current understanding of demographic and tumor-​related determinants of HL EBV positivity (Glaser et al., 1997, 2008). Accordingly, in a seminal study of HL tumor EBV positivity including 1546 patients from 14 studies, Glaser and colleagues limited the impact of confounding by carrying out multivariate analysis (Glaser et  al., 1997). The analyses showed that associations with ethnicity and with age, sex, histological subtype, and regional economic level remained after mutual adjustment, indicating that they are independently associated with HL EBV positivity (Tables 39–2 and 39–3). Of particular interest, the study by Glaser et  al. showed that the prevalence of EBV in HL was the lowest in younger adults (15–​ 39  years) regardless of race/​ethnicity or regional level of socioeconomic development, even though both of these other factors were also determinants of HL EBV status (Glaser et al., 1997). EBV positivity was particularly high among the youngest children (i.e., those less than 10 years of age at diagnosis). The recent review also considered the association between clinical stage and HL-​EBV status (Lee et  al., 2014). In 31 studies together including 5336 cases, the proportion of EBV-​positive tumors was higher in stage III or IV disease (37%) than in stage I or II disease (32%), corresponding to an odds ratio (OR) of 1.21 (95% CI 1.07–​1.37).

Serological Evidence Prior to the discovery of EBV in Reed-​Sternberg cells, case-​control studies demonstrating aberrant patterns of anti-​EBV antibodies in HL patients provided the first tangible evidence that the virus might contribute to HL development. Accordingly, investigations from the 1970s and 1980s together including more than 2000 patients and controls showed that while the prevalence of IgG antibodies against the (lytic) EBV viral capsid antigen (VCA) was essentially similar in patients and controls, patients typically displayed higher titers (IARC, 1997). In addition, patients displayed a higher prevalence and higher titers of antibodies against the (lytic) early antigen (EA) complex than controls. In a decisive investigation, Mueller et al. (1989) demonstrated that the serologic anomalies preceded HL diagnosis. Specifically, accessing Table 39–2.  Odds Ratios (OR) and 95% Confidence Intervals (CI) for EBV-​Positive Hodgkin Lymphoma Associated with Histologic Subtype, Regional Economic Development and Sex Age Stratum (years) Adjusted OR

Variable

Comparison

Histologic subtype

MC vs. NS

0–​14 15–​49 50+

7.3 13.4 4.9

3.8–1​ 4.2 9.0–1​ 9.9 2.8–​8.7

Regional economic development

Less vs. more1

0–​14 15–​49 50+

6.0 0.9 0.8

2.0–1​ 8.0 0.4–​2.3 0.2–​3.0

Sex

Female vs. male1

0–​14 15–​49 50+

0.6 0.4 1.2

0.3–​1.1 0.3–​0.6 0.7–​2.0

1

Analyses controlled for country, clinical stage, and component case series. MC = mixed cellularity; NS = nodular sclerosis. 1 represents reference group. Source: Reproduced with permission from Glaser et al. (1997).

95% CI

752

752

PART IV:  Cancers by Tissue of Origin Table 39–3. Odds Ratios (OR) and 95% Confidence Intervals (CI) for EBV-​Positive Hodgkin Lymphoma Associated with Age Strata Variable

Comparison

Age

0–​14 vs. 15–​491 years

Regional Development

Histologic Subtype

Less developed

NS MC NS MC NS MC NS MC

More developed 50 vs. 15–​491 years

Males

Less developed More developed

Adjusted OR 10.0 5.5 1.6 0.9 1.2 0.4 1.3 0.5

Females 95% CI

Adjusted OR

95% CI

2.6–​39.7 1.3–​23.3 0.8–​3.0 0.4–​1.7 0.4–​3.9 0.1–​1.4 0.7–​2.4 0.3–​0.8

14.3 7.9 2.2 1.2 3.6 1.3 3.9 1.4

3.4–​59.9 1.7–​36.0 1.0–​4.7 0.5–​2.7 1.1–​11.3 0.4–​4.3 2.2–​7.1 0.8–​2.8

Analyses controlled for country, clinical stage, and component cases series. MC= mixed cellularity; NS=nodular sclerosis. 1 represents reference group. Source: Reproduced with permission from Glaser et al. (1997).

four serum repositories, the authors identified and measured anti-​EBV antibodies in blood samples from 43 individuals who developed HL on average more than 4 years after blood sampling and from 96 matched controls. In univariate analyses, HL risk was associated with elevated IgG titers against both VCA (OR 2.6, 90% CI 1.1–​6.1) and the diffuse form of EA (OR 2.6, 90% CI 1.1–​6.1). In addition, the presence of IgG antibodies against EBV nuclear antigen complex (EBNA) (OR = 4.0; 90% CI 1.4–​11.4) and elevated titers of IgA antibodies against VCA (OR = 3.7; 90% CI 1.4–​9.3) also carried an increased risk of HL. After mutual adjustment, statistically significant associations were seen for elevated IgA (OR = 4.1; 90% CI 1.3–​12.9) and presence of IgM antibodies (OR = 0.07; 90% CI 0.01–​ 0.53) against VCA, and for elevated IgG antibodies against EBNA (OR = 6.7; 90% CI 1.8–​24.5) (Mueller et al., 1989). The association with EBNA antibody patterns attracted particular attention because it implies an inadequate immune response to EBV in HL patients. Briefly, among six different EBNA variants coded by EBV, EBNA2 is highly expressed in the early stages of primary EBV infection, whereas EBNA1 expression dominates as the acute infection is resolved and latent infection is established. During the normal course of primary EBV infection, corresponding changes are seen in the humoral immune response, as the ratio of anti-​EBNA1 to anti-​ EBNA2 antibody levels gradually increases from values ≤ 1 to values > 1; failure to do so signals an abnormal immune response to EBV (Henle et al., 1987; Miller et al., 1987). Mueller and colleagues pursued their original observation in three ensuing investigations to demonstrate that an EBNA1:EBNA2 ratio ≤ 1 is also associated with an increased risk of EBV-​positive HL. First, using post-​treatment serum samples from participants in a previous case-​control investigation (Gutensohn and Cole, 1981), they compared anti-​ EBNA1 and anti-​ EBNA2 antibody levels between HL cases and sibling controls with and without self-​reported IM (Mueller et al., 2012). In analyses adjusted for age, sex, self-​reported IM history, and elevated titres of IgG antibodies against VCA and EA, anti-​ EBNA1:anti-​EBNA2 antibody level ratios ≤ 1 were associated with a 2.43-​fold (95% CI 1.05–​5.65) increased risk of HL. The increased HL risk was observed irrespective of self-​reported history of IM, suggesting that anti-​EBNA1:anti-​EBNA2 antibody level ratios ≤ 1 do not merely mediate an association between IM and HL (Mueller et al., 2012). The group next examined antibody profiles in 88 and 282 (largely post-​ treatment) patients with EBV-​ positive and EBV-​ negative cHL, respectively, enrolled in a case-​ control investigation carried out between 1997 and 2001 in the northeastern United States. This investigation showed that all patients with EBV-​positive tumors also were EBV sero-​positive (as indicated by detectable anti-​VCA IgG), whereas this was case for only 94% of patients with EBV-​negative tumors (Chang et  al., 2004b). Thus, all 18 EBV-​sero-​negative cHL patients with known tumor EBV status also had EBV-​negative tumors.

In analyses adjusted for age, sex, educational level, smoking status, and other EBV antibodies, risk of EBV-​positive cHL was associated with elevated titres of IgG antibodies against VCA (OR = 3.6; 95% CI 1.4–​8.7) and with anti-​EBNA1:anti-​EBNA2 antibody level ratios ≤ 1 (OR = 3.2; 95% CI 1.1–​9.0). Finally, Mueller and colleagues took advantage of the US Department of Defense Serum Repository to assess differences in EBV antibody patterns in prediagnostic samples from 40 and 88 individuals who developed EBV-​positive and EBV-​negative HL, respectively, an average of 30 months subsequent to blood sampling, and from 368 matched controls (Levin et al., 2012). Analyses showed that prediagnostic samples were EBV sero-​positive for all patients with EBV-​positive cHL, but only for 89.8% of the patients with EBV-​negative cHL and for 92.4% of the controls. In crude analyses, patients with EBV-​positive tumors were statistically significantly more likely than their matched controls to have elevated levels of IgG antibodies against VCA, to have antibodies against EA and EBNA2 (both RR= 2.5 (95% CI 1.1-5.8)), and to have an anti-​EBNA1:anti-​EBNA2 antibody level ratios ≤ 1 (Table 39–4). In contrast, there were no differences in anti-​EBV antibody profiles between patients with EBV-​negative tumors and their matched controls. In multivariate case-series analyses, only anti-​EBNA1:anti-​ EBNA2 antibody level ratios ≤ 1 were statistically significantly associated with HL EBV positivity (Table 39–4) (Levin et  al., 2012).

Circulating Cell-​Free EBV DNA In an extension of research detecting EBV genome products in tumor tissue, Gallagher and colleagues in 1999 demonstrated cell-​free EBV DNA in serum samples from 30 of 33 patients with virus-​positive HL, compared with only 6 of 26 patients with virus-​negative cHL. Moreover, only 5 of 12 patients with IM, and none of 15 healthy controls, were positive for circulating free EBV DNA (Gallagher et al., 1999). It is clear from this and other studies that circulating free EBV DNA closely correlates with tumor EBV status, and that it is a marker of poor prognosis (Kanakry et al., 2013). However, in the context of a causal association, arguably the most important study is an extension of the prospective nested case-​control study of EBV serology preceding HL diagnosis (Levin et  al., 2012), also using samples from the US Department of Defense Serum Repository (Yang et  al., 2002). Using the same study design, the authors compared the presence of circulating free EBV DNA in 41 and 115 individuals who developed EBV-​positive and EBV-​negative cHL, respectively, subsequent to blood sampling, along with matched controls. In patients with EBV-​positive disease, circulating free EBV DNA could be detected even in samples predating cHL diagnosis by 7 years, and at 2 years before diagnosis the future patients were 21-​fold (95% CI 2.3–​192.7) more likely than their matched controls

 753



753

Hodgkin Lymphoma

Table 39–4.  Association of Elevated EBV Antibody Titer and Low Antibody Ratio of EBNA-​1/​EBNA-​2 with Hodgkin Lymphoma Risk Overall and EBV+ and EBV-​Hodgkin Lymphoma

Antibody

Referent

VCA (IgG ≥ 1:2560)† EA complex (≥ 1: 40)† EBNA complex (≥ 1:1280)† EBNA-​1 (≥ 1:1280)† EBNA-​2 (≥ 1: 80)† EBNA-​1/​EBNA-​2 ratio (≤ 1.0)

< 1:2560 < 1:40 < 1:1280 < 1:1280 < 1:80 > 1.0

All Cases vs. Matched Controls (N = 128 sets) 2.1 (1.2–​3.7) 1.7 (1.0–​2.9) 1.2 (0.7–​2.1) 1.2 (0.6–​2.3) 1.6 (0.9–​2.8) 1.4 (0.7–​2.7)

EBV+HL cases vs. Matched Controls (N = 40 sets) 3.1 (1.1–​8.7) 2.4 (0.9–​6.3) 1.7 (0.6–​4.7) 1.4 (0.4–​4.6) 1.3 (0.5–​3.3) 4.7 (1.6–​13.8)

EBV-​HL cases vs. Matched Controls (N = 88 sets) 1.7 (0.9–​3.5) 1.4 (0.7–​2.8) 1.0 (0.5–​2.0) 1.1 (0.5–​2.5) 1.8 (0.9–​3.5) 0.4 (0.1–​1.3)

EBV+ HL cases vs. EBV-​HL Cases (N = 40 vs 88)* 1.4 (0.5–​3.8) 1.5 (0.5–​4.2) 1.7 (0.5–​6.3) 0.93 (0.2–​4.8) 0.81 (0.3–​2.5) 14.0 (2.7–​72.5)

Data are RR (95% CI). * Unconditional logistic regression adjusted for age, sex, race, year of serum collection, and histology. † Elevated titer levels. Source: Reproduced with permission from Levin et al. (2012).

to display circulating free DNA. Also, levels of circulating free EBV DNA increased as diagnosis was approached. In contrast, no difference was seen between future patients with EBV-​negative HL and their matched controls (Mantel-​Haenszel OR = 0.5; 95% CI 0.11–​2.19).

Infectious Mononucleosis Even before the first description of EBV (Epstein et al., 1964), co-​occurrence of IM and HL had been noted in case reports (see summary in Kaplan, 1972). More than 20 studies have since investigated this association, with most reporting increased HL risk after IM (for a detailed review, see Hjalgrim, 2012). In register-​based cohort studies based on serologically confirmed or hospital-​ diagnosed IM, reported relative risks have ranged from nil in one small study of 1234 Finnish patients hospitalized for IM (Lumio and Karjalainen, 1993) to six-​fold increased among nearly 2800 UK patients hospitalized with IM (Goldacre et al., 2008). In case-​control studies, mostly resting on self-​reported IM history, reported ORs have varied from slightly less than unity (Chang et al., 2004c) to 7.5 less than five years after IM (Levine et al., 1998). The increased risk appears to be specific to HL, and not part of a general cancer predisposition among IM patients. For instance, in a large Scandinavian cohort study assessing cancer risk in more than 38,000 IM patients, the relative risk of cancer overall was 1.03 (95% CI 0.98–​1.09) (Hjalgrim et al., 2000). With few exceptions (Becker et al., 2009; Serraino et al., 1991), studies suggest that the increased risk of HL after IM is a transient phenomenon (Bernard et al., 1987; Levine et al., 1998; Munoz et al., 1978; Rosdahl et  al., 1974; Hjalgrim et  al., 2000, 2007). Because IM most commonly occurs in adolescents, the temporal risk variation implies that most HLs potentially attributable to IM present in adolescents and younger adults. Meanwhile, there is no evidence to suggest that the increased HL risk is particular to any specific age at IM (Hjalgrim et al., 2000). Although data are limited, increased risks of HL after IM appear to apply to both men and women (see Hjalgrim, 2012). Only a few studies have considered HL EBV status with respect to the IM association, with conflicting results. In two Scandinavian investigations, IM was associated exclusively with increased EBV-​positive HL risk. One was a cohort study of 38,555 individuals with serologically confirmed or hospital-​diagnosed IM (relative risk [RR] EBV-​ positive HL = 4.0; 95% CI 3.4–​4.5); the other was an interview-​based case-​control study of 586 incident cHL patients and 3,187 population controls (OR EBV-​positive HL = 3.23; 95% CI 1.89–​5.55) (Hjalgrim et al., 2003, 2007). In both investigations, the increased risk of EBV-​ positive cHL was found to vary by time since IM diagnosis. A preferential association with EBV-​positive cHL was also observed in a British population-​based case-​control investigation among persons age 16–​24 years (Alexander et  al., 2000), in which self-​reported IM was associated with nine-​fold increased EBV-​positive cHL risk (OR = 9.16;

95% CI 1.07–​78.31), but not with risk of EBV-​negative cHL (OR = 1.60; 95% CI 0.63–​4.07). In another British population-​ based case-​ control investigation, self-​reported IM was associated with increased risks of both EBV-​ positive cHL (OR = 2.59; 95% CI 1.24–​5.43) and EBV-​negative cHL (OR  =  2.11; 95% CI 1.14–​3.90) (Alexander et  al., 2003). A  slight variation in age-​specific relative risks by cHL EBV status suggested shorter intervals between IM and EBV-​positive HL than between IM and EBV-​ negative cHL, compatible with the Scandinavian observations. The European findings contrast with the results of three American investigations, none of which observed associations between self-​ reported IM and HL risk, whether overall or by EBV-​status (Chang et al., 2004b, 2004c; Glaser et al., 2005; Sleckman et al., 1998). Even if uncontrolled confounding by risk factors shared by IM and HL remains theoretically possible, the rather solid evidence of an association between serologically confirmed IM and HL, along with its specificity to EBV-​positive cHL, is consistent with a causal relationship (see Hjalgrim, 2012). Thus, the most likely explanation for the contrasting American observations remains a mixture of misclassification of IM based on self-​report and participation bias. Under the assumption of such causality, statistical modeling in the two Scandinavian investigations suggested that the median incubation period of IM-​induced EBV-​positive cHL is on the order of 3–​4 years (Figure 39–7) (see Hjalgrim, 2012).

FAMILY HISTORY In 1959, when Razis et al. first formally investigated familial aggregation of HL, this tendency had long been suspected from numerous published reports and case series (Razis et al., 1959). In their investigation, Razis and colleagues compared first-​degree family history of HL based on medical chart review among 1102 HL patients compared with four other groups of patients treated at a single institution, and reported an approximately three-​fold increased risk among HL patients’ relatives (Razis et al., 1959). Despite the methodological limitations of this first study, its results still stand. Thus, several subsequent investigations have reported between approximately two-​ fold and nine-​ fold increased risks of HL among first-​degree relatives of HL patients (Crump et al., 2012; Friedman et  al., 2005; Grufferman et  al., 1976; Haim et  al., 1982; Kharazmi et al., 2015; McDuffie et al., 2009; Paltiel et al., 2000; Pang et al., 2008; Rudant et al., 2007), or in the combined group of first-​and second-​degree relatives (Kerber and O’Brien, 2005; Kerzin-​Storrar et al., 1983). Accordingly, it has been estimated that in the Swedish population the heritability of HL is 28.4% (Shugart et al., 2000). Results from twin studies are mixed. A study of more than 13,000 twins from North America indicated that HL risk was 99 (95% CI 48–​182) times higher in identical than in fraternal co-​twins of cases

754

754

PART IV:  Cancers by Tissue of Origin

Relative risk

20 15 10 5 0 0

5

10

15

20

Time (years) since mononucleosis Overall

EBV pos

EBV neg

Figure  39–7. Temporal variation in relative risk of classical Hodgkin lymphoma overall and by EBV status by time since infectious mononucleosis as predicted by statistical modeling based on the age, gender, and country-​specific distribution of infectious mononuleosis among population controls in the SCALE study. Source:  Reproduced from Hjalgrim et al. (2007).

(Mack et al., 1995). By contrast, no concordance was observed among 116 HL patients (57 monozygotic and 69 dizygotic twins) in a Nordic cohort study of 203,691 twins (Mucci et al., 2016). The many familial studies provide little information to distinguish between genetic and environmental explanations of familial associations. Broadly, studies suggest that familial HL cases tend to occur at earlier ages than non-​familial cases (Shugart et al., 2000), that familial clustering applies to HL at all ages (i.e., including childhood) (Crump et al., 2012; Friedman et al., 2005; Rudant et al., 2007), and that HL risk is more increased by sibling than parental/​offspring HL history (Kharazmi et al., 2015; Paltiel et al., 2000). Some investigations have suggested that being of the same sex as an affected sibling may increase HL risk more than being of the opposite sex. In their survey in Greater Boston, 1959–​1973, Grufferman and colleagues identified 13 pairs of siblings both diagnosed with HL, 12 of which were same-​sex (p  =  0.01) (Grufferman et  al., 1977). In a large register-​based study of 57,475 first-​degree relatives of 13,922 HL patients in the Nordic countries, HL risk in sisters of female HL patients was 9.4-​fold (95% CI 5.9-​to 14-​fold) increased relative to the general population, compared with 4.5-​fold (95% CI 2.9-​to 6.7-​fold) increased in brothers of male HL patients and 5.9-​fold (95% CI 4.3-​to 8.1-​fold) increased in opposite sex siblings of HL patients (Kharazmi et al., 2015). Together, these findings suggest that interaction between genetic constitution and childhood environment is important to HL risk. Within families, HL also clusters with other lymphomas (Chang et al., 2005b; Goldin et al., 2009; Wang et al., 2007). Whether this clustering rests on shared genetic predisposition, common environmental risk factors, or a combination remains to be elucidated.

GENETIC PREDISPOSITION HLA Based on the suspicion of a viral etiology, HL was among the first diseases scrutinized for associations with the human leukocyte antigen (HLA) complex, and in 1967 the first to be associated with an HLA antigen, specifically Payne’s 4c antigen in the contemporary nomenclature (Amiel, 1967). By 1979, data for 2670 HL patients studied across more than 25 investigations indicated that HLA serotypes A1, B5, B8, and B18 were overrepresented among HL patients,

with A1 in particular among patients with mixed cellularity HL (Hors and Dausset, 1983). The understanding of constitutional susceptibility to HL has advanced considerably with the development of genotyping techniques that provide greater detail than serological HLA typing (e.g., by enabling HLA class II allele typing), and with the inclusion of HL EBV status in a number of investigations. Recent extensive reviews of the literature have been summarized in the following (Diepstra et al., 2005a; Kushekhar et al., 2014; McAulay and Jarrett, 2015; Sud et al., 2015). One of the most important observations in recent years is the demonstration of variation in HLA associations between EBV-​positive and EBV-​negative HL. This was first reported in landmark papers by Diepstra, Niens, and colleagues, who demonstrated that HLA-​ A*01 and HLA-​A*02 are statistically significantly over-​and underrepresented, respectively, among patients with EBV-​positive cHL compared with both healthy controls and patients with EBV-​negative cHL (Diepstra et  al., 2005b; Niens et  al., 2006, 2007). The associations have been confirmed in other patient populations (Dutch, English/​Scottish, and Scandinavian) (Hjalgrim et  al., 2010; Huang et  al., 2012b), and in one investigation were found to be mutually independent, such that there was a 9.45-​fold (95% CI 4.60–​19.4) difference in risk of HL EBV positivity between HLA-​A*01 and HLA-​ A*02 homozygotes (Hjalgrim et al., 2010). The associations between HLA-​A*01 and HLA-​A*02 and EBV-​ positive HL were corroborated by a genome-​ wide association study (GWAS) of HL in a European (Caucasian) patient population (Urayama et  al., 2012), which partially overlapped with the previous studies (Hjalgrim et al., 2010; Niens et al., 2007). Here, associations with two single nucleotide polymorphisms (SNPs), rs2734986 and rs6904029, were found to be explained by linkage disequilibrium with HLA-​A*01 and HLA-​A*02, respectively (Figure 39–8). EBV-​positive HL has been associated less consistently with other HLA types. Most promising in this regard are HLA-​B37 (and HLA-​ B*37:01), HLA-​DPB1*01:01, HLA-​DRB1*15:01, and -​DR10 (Huang et al., 2012b; Johnson et al., 2015). In contrast, EBV-​negative HL has been associated most strongly with HLA class II alleles (Kushekhar et  al., 2014; McAulay and Jarrett, 2015). GWAS of Caucasian HL patients have unanimously reported strong associations between rs6903608, located in the HLA class II region, and EBV-​negative HL (Figure 39–8) (Enciso-​Mora et al., 2010; Urayama et al., 2012) or nodular sclerosis cHL in younger adults, the majority of which would be EBV-​negative (Cozen et  al., 2012; Frampton et al., 2013). The linkage disequilibrium of rs6903608 with HLA-​DRB1*15:01 and -​DQB1*06:02 may in part explain the reported overrepresentation of these HLA alleles in EBV-​negative HL (Huang et al., 2012b; Johnson et al., 2015) and in young-​adult nodular sclerosis HL (Harty et al., 2002; Moutsianas et al., 2011). Accordingly, inclusion of rs6903608 in multivariate models either markedly reduces or completely abrogates the HLA alleles’ associations with EBV-​ negative HL (Johnson et  al., 2015) or with nodular sclerosis cHL (Moutsianas et al., 2011). Two of the GWAS also reported putative HL susceptibility loci near the MICB gene on the border between HLA class I and class III regions, and in HLA-​DRA (Cozen et al., 2014; Urayama et al., 2012). However, using a Bayesian variable selection modeling method, their associations with HL were found most likely to be explained by HLA alleles (Johnson et al., 2015). Other HLA alleles suggested to be associated with EBV-​negative HL are HLA-​DQB1*03:03, HLA-​DRB1*03:01, and HLA-​B*15:01 (Johnson et  al., 2015), as well as HLA-​ DQA1*02:01 and HLA-​ DPB1*03:01 (Moutsianas et al., 2011). Despite differences in HL incidence among racial/​ethnic groups, little is known about genetic predisposition to HL in non-​Caucasians, although one investigation in a Chinese study population suggested HLA-​ A*02:07 to be associated with an increased risk of EBV-​ positive cHL and a decreased risk of EBV-​negative cHL (Huang et al., 2012a).

 75



755

Hodgkin Lymphoma Class I

Extended Class I

Class III

Class II

HLA-DRA: rs6903608

30

HLA-DRA: rs6903608 25

All classical HL

HLA-DRA: rs2395185

–Log10 (P value)

EBV-positive HL EBV-negative HL

20

HLA-A: rs2734986 MICB : rs2248462

15 HLA-A: rs6904029 10

5

0 27

28

29

30

31

32

33

Position in MHC Region (MB)

Figure 39–8.  Association between genetic variants and total classical Hodgkin lymphoma (cHL [grey]), Epstein-​Barr virus (EBV)-​positive cHL (dark grey) and EBV-​negative cHL (light grey) within an approximately 6.5 Mb region of the extended major histocompatibility complex located at 6p21. Multiple logistic regression analysis was performed assuming a log-​additive genetic model and adjusting for sex (male or female), country (up to eight indicator variables after excluding one country as the reference, depending on the analysis: France, Germany, Spain, Czech Republic, Ireland, United Kingdom, Denmark, Sweden, and the Netherlands), and eight principal components analysis eigenvectors. All statistical tests were two-​sided. Arrows indicate cHL-​associated regions indexed by five SNPs. Source: Reproduced with permission from Urayama et al. (2012).

Non-​HLA Associations

Interpretation of Genetic Findings

Associations with HL risk have been assessed for more than 80 gene loci outside the HLA region in targeted analyses, and statistically significant associations have been reported for variants in more than 20 genes, albeit in many instances only in single studies as yet without independent replication (see reviews by Kushekhar et  al., 2014; Sud et al., 2015). Sud and colleagues recently performed a meta-​analysis of data from studies of individual candidate SNPs and of related information derived from two HL GWAS (Enciso-​Mora et  al., 2010; Frampton et  al., 2013). Nominally significant associations were observed for nine variants in IL-​1A, IL-​1B, IL-​4A, and IL-​10RA, genes involved in immune function and response, in CYP2C9 (two variants), which is involved in carcinogen metabolism, in XPG/​ERCC5 and ERCC1, which are involved in DNA repair, and in NFKB1A, which is involved in signal transduction and inflammation (Sud et al., 2015). However, as pointed out by the authors, false-​positive report probabilities were high in all instances, and analyses were not corrected for multiple comparisons, meaning that caution in the interpretation is warranted. The GWAS of HL have so far identified six risk loci outside the HLA region, none of which is among previously reported candidate SNPs (Kushekhar et al., 2014; Sud et al., 2015). The candidate genes associated with these loci are the REL gene, a member of the NFκB transcription factor family, at 2p16.1; the eomesodermin (EOMES) involved in cytotoxic (CD8+) T-​lymphocyte (CTL) differentiation at 3p24.1; IL13 at 5q31; Hsp70 subfamily B suppressor1-​like protein and/​or avian myeloblastosis viral oncogene homolog (HBS1L-​MYB) located in an area of the genome important to hematopoiesis at 6q23.3; the PVT1 (non-​protein coding Pvt1 oncogene) involved in expression of multiple microRNAs at 8q24.2; the GATA3 important to Th2 T-​cell differentiation at 10p14; and, finally, a locus close to the transcription factor 3 (TCF3) important to both B-​and T-​cell development at 19p13.3 (see review by Kushekhar et al., 2014). Among these associations, REL, IL3, and PVT1 may be the strongest for EBV-​negative HL (Kushekhar et al., 2014). These findings offer promising targets for deep sequencing and functional studies.

The results of the genetic analyses support a constitutional basis for the observed familial accumulation of HL and, possibly, for some aspects for the ethnicity-​dependent variation in HL risk. From a practical point of view for etiologic research, the studies also highlight the importance of stratifying HL, ideally directly on tumor EBV status and histology, or, at a minimum, on combinations of histology, age, and sex to approximate HL EBV status. Given the strong evidence of environmental risk factors for HL, such stratification would be prudent both in genetic and in environmental epidemiologic studies. Accordingly, absent such information, studies may not be comparable or even informative. The genetic studies may also provide insight into the pathogenesis of HL. Accordingly, the association with HLA class I alleles indicates that development of EBV-​positive HL is in some way influenced by cytotoxic CTLs. HLA-​A*02:01 and other HLA-​A*02 variants (with the notable exception of HLA-​A*02:07) are capable of presenting a wide range of EBV-​derived latent and lytic peptides, including LMP1 and LMP2a expressed in the Reed-​Sternberg cells (Hislop et al., 2007). The reduced EBV-​positive HL risk seen with HLA-​A*02 therefore could simply reflect an efficient presentation of EBV-​derived peptides to CTLs, resulting in a correspondingly efficient immune response regulating either the number of infected B lymphocytes at risk of malignant transformation, or targeting Reed-​Sternberg precursor cells (McAulay and Jarrett, 2015). Such a mechanism also would explain the increased risk of EBV-​positive HL associated with HLA-​A*02:07, which is unable to present LMP1 and LMP2a, observed in Chinese HL patients (Huang et al., 2012a). In contrast to HLA-​A*02, there are no confirmed EBV-​derived peptides that are restricted to HLA-​ A*01 (Hislop et  al., 2007). Theoretically, one mechanism explaining the association of HLA-​ A*01 with EBV-​positive HL could be a deficient cellular immune response toward EBV (Hjalgrim et al., 2010). However, as pointed out by McAulay and Jarrett (2015), an increased risk of EBV-​positive HL is also observed in HLA-​A*01/​HLA-​A*02 heterozygotes. An alternative hypothesis, therefore, could be that HLA-​A*01 in some way

756

756

PART IV:  Cancers by Tissue of Origin

influences (abrogates) responses restricted through other HLA alleles (McAulay and Jarrett, 2015). Conversely, to the extent that it is not entirely explained by linkage disequilibrium with rs6903608 (as suggested by Johnson et  al., 2015; McAulay and Jarrett, 2015; Moutsianas et al., 2011), the association of EBV-​negative HL with HLA class II alleles indicates that T-​helper lymphocytes (CD4+ T-​lymphocytes) play an important role in the susceptibility to this HL variant (Huang et al., 2012b; Kushekhar et al., 2014). Kushekhar and colleagues have developed more elaborate models for constitutional HL susceptibility that include HLA and some of the other related genes, several of which have been shown to be aberrantly expressed in tumor cells (Kushekhar et al., 2014) (Figure 39–9). Broadly, the proposed models involve (1) facilitation or promotion of Reed-​Sternberg (precursor) cell proliferation and survival, (2) the ability or inability to mount efficient immune responses to Reed-​Sternberg precursor cells, and (3) an influence on T-​cell functionality and modulation of the micro-​environment in the direction of a permissive Th2 milieu (Kushekhar et al., 2014).

IL 13

IL4RA

576R IL4RA (+) 110Q IL13 (+) STAT6 (↑)

IL4RA Pre-HRS

TCF3 (↓) REL (NFKBIA) (↑) STAT3 (↑) DNA repair (–)

HRS

(b) CTLs

EBV

A*02

A*02

GCB

Pre-HRS

A*01

FAMILY STRUCTURE AND CHILDHOOD SOCIOECONOMIC ENVIRONMENT The late infection model holds that HL in children and in adolescents and younger adults has an infectious etiology, and that the later in adolescence and young adulthood the infection with the suspected agent occurs, the greater is the risk of HL. Inspired by international variation in HL incidence patterns, this model has also been explored in studies at the level of the individual. Over time, evidence for risk factors identified by such investigations seems to have weakened, but nevertheless remains important to the understanding of HL epidemiology. Gutensohn and colleagues compared the number of siblings among prevalent adolescent and young adult HL patients in Boston in 1972 with the number expected from other population series and observed that HL risk decreased with an increasing number of siblings (Gutensohn et al., 1975). Gutensohn and Cole then conducted a case-​control study that showed that various measures associated with increasing infectious disease pressure in childhood (e.g., having more siblings and more playmates and living in multifamily housing units in childhood as well as having a less educated mother) were associated with reduced HL risk at ages 15–​39 years (Gutensohn and Cole, 1981). The inverse association with number of siblings, older ones in particular, was largely corroborated by later case-​control (Bernard et al., 1987; Bonelli et al., 1990; Chatenoud et al., 2005) and register-​based investigations. Westergaard et  al. (1997) ascertained information on family structure in the Danish Population Register for 2.1  million Danes and followed these individuals for HL diagnosis in the Danish Cancer Register from 1968 to 1992. Their analyses suggested that among adolescents and younger adults, HL risk was inversely associated with sibship size (relative risk per child  =  0.91; 95% CI 0.81–​1.03) and birth order (relative risk per position = 0.85; 95% CI 0.71–​1.01) (Westergaard et al., 1997). Analogously, Chang et al. identified 2140 HL patients registered in the Swedish Cancer Registry in 1958–​1998 and 10,024 population controls. Based on information on family structure from the Swedish Multi Generation Register, their analyses showed that HL risk at age 15–​39 years decreased with number of older siblings (ptrend = 0.01), but did not vary by total number of siblings, number of younger siblings, or paternal occupation (Chang et al., 2004a). However, more recent case-​control investigations generally have been unable to reproduce the original associations between childhood environment and HL risk in adolescence and younger adulthood. In a case-​control study of 202 younger-​adult women diagnosed with HL in the years 1988–​1994 and 254 controls in California, Glaser and colleagues reproduced none of the associations originally reported by Gutensohn and Cole (Glaser et al., 2002). Meanwhile, other measures of childhood socioeconomic affluence, such as living in family-​owned

IL 13

(a)

R131 FCGR2A A*01 (–)

A*01

HRS

No CTLs

(c)

576R IL4RA (+), 110Q IL13 (+) S159 IL10RA (+), STAT6 (↑)

IL4RA

Th2 Th2

Th2

IL4RA IL13

IL13 IL13

IL4RA

IL4RA

Pre-HRS HRS

Figure  39–9.  Putative susceptibility mechanisms of genetic associations in cHL. A:  Enhanced survival signals for precursor Hodgkin and Reed-​ Sternberg (pre-​HRS). The variants associated with hyperactivity (110Q IL13 and 576R IL4RA), decreased expression of B-​cell lineage commitment or differentiation factors (TCF3), increased transcriptional activity (REL [NFKBIA], STAT3, STAT6), and decreased DNA-​repair capacity, and provide enhanced growth and survival signals for the pre-​HRS cell. In addition, IL13 and IL4RA polymorphisms can contribute to growth and survival of pre-​HRS cells. B: EBV-​associated malignant transformation of pre-​HRS cells. HLA-​A*02 carriers can effectively induce EBV-​ specific CTLs and control the number of EBV+ pre-​HRS cells throughout the malignant transformation process. HLA-​A*01 carriers cannot mount EBV-​specific CTL responses to the latency type II infection pattern in pre-​ HRS cells. In addition, EBV+ B cells in R131 FCGR2A carriers are less effectively cleared by CTLs. C:  Th2-​based microenvironment and cHL susceptibility mechanisms. The autocrine growth signaling mediated by hyperactive 110Q IL13, 576R IL4RA, and S159 IL10RA, and increased expression of STAT6 can enhance Th2 differentiation and subsequent secretion of IL13, providing growth and stimulation signals throughout the transformation process of pre-​HRS cells. GCB = germinal center B cell. Source: Reproduced with permission from Kushekhar et al. (2014).

(as opposed to rented) homes and among US-​born women, not sharing bedrooms at age 11, carried increased HL risk (Glaser et al., 2002). When stratified according to HL EBV-​status, low birth order and having had a single bedroom at age 11 years were associated with decreased and increased EBV-​positive HL risk, respectively (Glaser et al., 2005).

 75



Hodgkin Lymphoma

Perhaps the most interesting observation in the study by Glaser and colleagues was the clear tendency for associations with presumed markers of childhood socioeconomic affluence to become stronger with increasing age at diagnosis, suggestive of a birth-​cohort phenomenon (Glaser et al., 2002). This led the authors to speculate that recent societal phenomena (e.g., preschool attendance, which had increased markedly since the 1950s) might have replaced the family environment as an important source of childhood infectious disease pressure (Glaser et al., 2002). Chang and colleagues similarly observed no association between sibship size, number of childhood playmates, housing density, housing ownership, and/​or maternal education and risk of HL at age 15–​54 years in their northeastern US case-​control study among 470 patients and 557 population controls (Chang et al., 2004c). Of note, however, having attended nursery school or day care for more than a year was associated with a decreased risk of HL in the adolescent and younger/​middle-​aged adult age group (OR = 0.64; 95% CI 0.45–​0.92), with no apparent variation by HL EBV status (Chang et al., 2004c). In a Danish-​Swedish case-​control study with 586 cHL patients diagnosed in the period 1999–​2002 at age 18–​74, and 3187 population controls, there also was no association between self-​reported number of older or younger siblings, housing density, or paternal education and risk of cHL at age 18–​44 years (Hjalgrim et  al., 2007). In stratified analyses, increasing number of siblings, whether younger or older, tended to be associated with decreased EBV-​positive cHL risk at age 18–​44 years, and number of older siblings with increased EBV-​positive cHL risk at age 45–​74 years (Hjalgrim et  al., 2007). In this investigation, higher maternal education was associated with cHL risk at age 18–​44 years in Sweden (OR for increasing category = 1.36; 95% CI 1.07–​1.73), with no heterogeneity by HL EBV status, whereas no such association was seen in Denmark. Of note, kindergarten attendance also was associated with a decreased HL risk, albeit not at a statistically significant level (OR = 0.78; 95% CI 0.56–​1.09). The predictions of the late infection model have also been explored for HL in childhood, for which associations opposite those for HL in adolescence and early adulthood are expected. In the Danish cohort study by Westergaard et  al. (1997), HL risk in childhood was positively associated with sibship size (relative risk per child = 1.28; 95% CI 1.00–​1.63) and with birth order (relative risk per position = 1.26; 95% CI 0.92–​1.73). This pattern was statistically significantly different from the pattern observed for the adolescent and younger adult age group discussed earlier (phomogeneity < 0.05). In contrast, no association with family structure was observed for HL risk before age 15 years in the Swedish register-​study by Chang and colleagues (2004a) or in a similar American register-​based investigation of 477 patients diagnosed in the period 1980–​2004 (von Behren et al., 2011). In the American data, analyses revealed an increased childhood HL risk for highest household educational attainment of less than 12 years compared with more than 17 years (OR = 2.17; 95% CI 1.07–​ 4.38) (Carozza et al., 2010). The largest case-​control study to date of HL diagnosed before the age of 15 years included 517 North American patients (80% participation rate) and 784 controls (65% participation rate). In this investigation, increasing sibship size was associated with increasing HL risk (ptrend = 0.04), while increasing maternal educational attainment (ptrend = 0.0001) and increasing household income at birth of the child (ptrend < 0.0001) were inversely associated with HL risk (Linabery et al., 2014). In a French study, maternal education beyond high school was similarly associated with a reduced childhood HL risk (OR = 0.7; 95% CI 0.4–​1.0) in a comparison of 128 HL patients age 5–​14 years and 848 population controls (Rudant et  al., 2011). Birth order and number of children in the household were not statistically significantly associated with HL risk in this study. For older adults, reported associations between risk of HL and family structure or socioeconomic status are not convincing (e.g., Chang et al., 2004a, 2004c; Gutensohn, 1982; Hjalgrim et al., 2007). If EBV-​ positive HL in older adulthood is related to reactivation of latent EBV infection, then other factors besides family structure (especially in

757

childhood) and socioeconomic status might be expected to be more important predictors of risk.

ANTHROPOMETRY Birth characteristics (i.e., birth weight and length and other measures of fetal growth) and adolescent and adult anthropometric measures may reflect childhood socioeconomic status. Hence, based on the late infection model, the expectation is that measures of fetal growth would be negatively associated with HL risk in childhood and positively associated with HL risk in adolescence and early adulthood. There is little to suggest that high birth weight and related birth characteristics are associated with childhood HL risk, whereas a positive association with risk of HL in adolescence and early adulthood has been reported in more investigations. In a large Swedish register-​ based investigation, Crump and colleagues studied 3,571,574 individuals born in the years 1973–​2008 and followed for HL through 2009. HL occurred in 943 cohort members, and in adjusted analyses, HL risk was positively associated with birth weight (relative risk per 1000-​g increase  =  1.24; 95% CI 1.09–​1.42), birth length (relative risk per 1-​cm increase = 1.03; 95% CI 1.01–​1.08), and fetal growth rate (relative risk per 1 standard-​deviation increase  =  1.09; 95% CI 1.03–​1.16). The association with fetal growth applied equally to risk of HL before (179 patients) and after (764 patients) the age of 15 years (phomogeneity = 0.87) and to both nodular sclerosis and mixed cellularity HL (Crump et al., 2012). In contrast, in a recent meta-​analysis of data from four independent investigations including a total of 844 cases, HL risk in childhood (before age 15 years in three studies and before age 18 in the fourth) was associated with neither reported birth weight less than 2.5 kg (fixed effects OR = 0.94; 95% CI 0.54–​1.65) nor birth weight of 4.0 kg or more (fixed effects OR = 0.94; 95% CI 0.64–​1.38) as compared with normal birth weight (Papadopoulou et al., 2012). Besides the Swedish investigation (Crump et al., 2012), two investigations support a positive association of birth weight with HL risk in adulthood. One is a Danish register-​based study reporting higher mean birth weights in 33 HL patients than in 99 matched controls (p < 0.01) (Isager and Andersen, 1978). The other investigation is a case-​control study of 312 California women and 325 matched population controls, in which self-​reported birth weight was positively associated with HL risk at age 19–​44 years (ptrend, tertiles = 0.07), while negatively associated with HL risk in older adults (age 45–​79 years; ptrend, tertiles = 0.05) (Keegan et al., 2006). Consistent with the supposition that childhood socioeconomic affluence underlies the suggested relationship with birth weight, stature in late childhood and/​or early adolescence has been positively associated with HL risk in adulthood (Isager and Andersen, 1978; Keegan et al., 2006). Data for a corresponding association with adult stature are inconsistent, with some investigations pointing to positive associations (Gutensohn and Cole, 1981; Keegan et al., 2006; Murphy et al., 2013), and others reporting no difference between HL patients and controls (La Vecchia et al., 1990; Lim et al., 2007; Paffenbarger et al., 1977; Willett and Roman, 2006). Of greater interest in terms of possibilities for the prevention of HL is the positive association between overweight and obesity and HL risk reported in some investigations. In a meta-​analysis of five prospective investigations published in the years 2005–​2009, which included a total of 1557 HL patients, Larsson and Wolk showed that obesity (body mass index ≥ 30 kg/​m2) carried an increased HL risk (pooled relative risk estimate = 1.41; 95% CI 1.14–​1.75), whereas overweight (25  < BMI ≤ 30  kg/​m2) did not (pooled relative risk estimate  =  0.97; 95% CI 0.85–​1.12). No information was provided regarding sex-​specific associations (Larsson and Wolk, 2011). In a subsequent analysis in the UK Million Women study, an increase of 10 kg/​m2 in BMI was associated with a relative risk of HL of 1.64 (95% CI 1.21–​2.21) (Murphy et al., 2013). The results of case-​control investigations are less consistent. No association between BMI or obesity and HL risk was reported in four

758

758

PART IV:  Cancers by Tissue of Origin

case-​control studies in Scandinavia (Chang et al., 2005a), Italy (Bosetti et al., 2005), California (Keegan et al., 2006), and Japan (Kanda et al., 2010b). In contrast, Willett and Roman found that self-​reported obesity was associated with an increased risk of HL among UK men, but not women (Willett and Roman, 2006), whereas Li and colleagues in an American case-​control investigation found that HL risk in younger-​ and older-​adult women increased and decreased, respectively, across rising quartiles of BMI, whereas BMI was not associated with HL risk in men (Li et al., 2013). As already mentioned, taller stature could plausibly be associated with HL risk as a marker of higher socioeconomic status, and therefore lower infectious disease pressure, in childhood. The excess cancer risk in those of taller stature (including men relative to women) also may be due to a greater number of susceptible cells in a specific organ (Walter et al., 2013), although it is unclear whether this explanation would apply to B lymphocytes. Obesity, meanwhile, could conceivably increase EBV-​positive HL risk by inducing a state of chronic inflammation and immune dysfunction (Exley et al., 2014; Mraz and Haluzik, 2014) that fosters tumor progression and loss of immunological control of EBV infection.

INFECTIONS OTHER THAN EBV Given that the association between HL risk and childhood socioeconomic affluence is hypothesized to be mediated primarily through infectious disease exposure in early life, several studies have assessed the direct association of HL risk with infectious diseases. Reduced risk of HL overall or of subsets of HL in adolescence and early adulthood with self-​reported history of one or more common childhood infectious diseases has been reported in several case-​control investigations (Monnereau et al., 2007; Montella et al., 2006), although not in all (Serraino et al., 1991). For example, Alexander et al. (2000) found self-​reported history of two or more childhood infectious diseases (measles, rubella, mumps, chicken pox, and pertussis) to be associated with decreased risk of EBV-​positive (OR = 0.18; 95% CI 0.03–​0.95) and EBV-​negative (OR = 0.43; 95% CI 0.21–​0.86) HL in a UK case-​control study of 118 HL patients age 15–​24 years (participation rate 90%) and 237 matched population controls (participation rate less than 50%). Likewise, self-​reported history of measles, mumps, or rubella was similarly associated with a reduced risk of EBV-​positive HL at age 19–​44 years (adjusted OR = 0.3; 95% CI 0.1–​1.0), but not with any other subgroup of HL in a population-​based case-​control study of California women (Glaser et al., 2005). Altogether, while a reduced HL risk with common childhood infectious diseases might reflect biological mechanisms, such as a skewing of the immune system in a Th1 direction, other methodological explanations for the observed associations, such as selection/​participation bias, seem more likely. The introduction of vaccinations for many childhood infectious diseases renders future exploration of the association with these conditions next to impossible.

Infections Other Than Childhood Infectious Diseases Other problems impede the interpretation of results of studies that investigate later-​life infections and HL risk. These include not least the shorter intervals between disease exposure and HL diagnosis, and the possibilities of confounding from an association between immune dysfunction and HL risk (see following discussion) and of reverse causality due to immune impairment with incipient HL. The problem of distinguishing between causal and non-​causal mechanisms was illustrated by a British register-​ based study. Newton and colleagues scrutinized general practitioners’ records for 214 HL patients age 16–​69 years and 214 matched controls to compare the frequency of medical visits in the period up to 15 years before diagnosis (or pseudo-​diagnosis in controls). Their analyses showed that an increasingly larger proportion of eventual HL

patients than of controls had visited their practitioner for infectious disease complaints in the decade before (pseudo)diagnosis, whereas no such difference was seen for visits for non-​ infectious complaints, except in the year immediately preceding (pseudo)diagnosis (Newton et al., 2007). Some reported associations between HL and specific infections can be considered as non-​causal with some degree of certainty. This includes, for instance, occasionally observed associations between elevated HL risk and self-​reported history of herpes zoster (Kristinsson et  al., 2015; Serraino et  al., 1991)  or warts (Becker et  al., 2005; Serraino et  al., 1991), both of which are known to be markers of immune dysfunction. In other instances, one cannot rule out contributions from biological mechanisms, for example, through chronic immune stimulation. Kristinsson et al. (2015) compared histories of infectious and inflammatory disease recorded in the population-​based Swedish inpatient register between 7414 HL patients and 29,240 matched controls. Disregarding the last year before (pseudo)diagnosis in HL patients and controls, history of any infection carried a 1.11-​fold increased HL risk, representing associations with such conditions as sinusitis (OR  =  1.81; 95% CI 1.06–​3.07), tuberculosis (OR  =  1.76; 95% CI 1.01–​3.07), encephalitis (OR = 7.88; 95% CI 1.97–​31.5), and the combined group of gastrointestinal infections (OR = 1.16; 95% CI 1.00–​ 1.35). Latency analyses suggested that the associations with sinusitis, tuberculosis, and herpes zoster were apparent for up to 10 years before HL diagnosis. The relation between infectious disease history and HL in childhood is similarly unclear. Based on parental recall of history of infections and immune disorders in their case-​control investigation of 517 patients younger than 15 years and 784 controls, Linabery and colleagues (2014) found that a reported history of any infection in the child (OR = 1.69; 95% CI 0.98–​2.91) or in a sibling (OR = 2.04; 95% CI 1.01–​4.14) was associated with an increased HL risk. In contrast, recurrent early common infections, as reported by subjects’ mothers, were associated with a decreased HL risk at age 5–​14 years (OR = 0.5; 95% CI 0.3–​0.9) in the French case-​control investigation including 128 and 848 population controls (Rudant et al., 2011).

IMMUNE FUNCTION Profound immune deficiency has been associated repeatedly with an increased HL risk. This is true for primary immune deficiencies and for acquired immune deficiencies as seen in conjunction with HIV/​ AIDS or with organ transplantation (see Chapter 25 in this volume for a review). HL occurring in the context of most though not all immune deficiency disorders is virtually always EBV-​positive (Gaulard et al., 2008; Raphael et  al., 2008; Swerdlow et  al., 2008; Van Krieken et al., 2008). Consistent with this pattern, a large follow-​up study of Americans AIDS patients showed that risk was strongly increased for mixed cellularity (relative risk = 18.3; 95% CI 15.9–​20.9) and lymphocyte-​ depleted HL (relative risk  =  35.3; 95% CI 24.7–​48.8), which are in turn associated with EBV positivity, and less strongly for nodular sclerosis HL (relative risk = 4.7; 95% CI 3.9–​5.5) (Frisch et al., 2001). A common underlying biological mechanism proposed for the pathogenesis of EBV-​positive HL in the context of these immune deficiency conditions is that EBV-​infected lymphocytes may proliferate and undergo malignant transformation, possibly promoted by chronic immune stimulation because of loss of immunological control (Swerdlow et al., 2008). Studies of patients with HIV/​AIDS have shown that HL risk correlates with degree of immune deficiency as measured by CD4+ T-​cell counts (e.g., Reekie et al., 2010; Silverberg et al., 2011). Interestingly, one investigation suggested that the association between HL risk and declining immune competence may not be linear, but rather that the risk may be higher in moderately immune-​deficient than in more or less severely immune-​deficient individuals (Biggar et al., 2006). The underlying explanation for this could be that a minimal immune response is required for the tumor lesion to become clinically

 759



Hodgkin Lymphoma

manifest (Biggar et al., 2006). This observation led to speculation as to whether the introduction of highly active antiretroviral treatment would be accompanied by an increased occurrence of HL among HIV-​infected persons. However, current evidence suggests the converse, that is, that prolonged antiretroviral treatment is accompanied by decreased HL risk (Kowalkowski et al., 2014).

COMORBIDITY

759

While these observations would suggest a positive association between atopic disease and HL risk, this is not borne out by the epidemiological literature. Even as statistically significantly elevated HL risks have been reported in association with eczema history in some investigations (Bernard et al., 1987; Cozen et al., 2009), statistically significantly decreased HL risks have been reported with asthma (Soderberg et al., 2006)  or hay fever (Becker et  al., 2005). Meanwhile, the majority of studies have reported null-​findings (Martinez-​Maza et al., 2010).

Autoimmune Diseases

TOBACCO

There has been considerable interest in the connection between HL risk and autoimmune disease, prompted in part by the hypothesis that non-​ malignant immunological diseases might increase the risk of malignancies of the immune system (i.e., lymphomas). HL occurrence was recently assessed among 878,000 Swedish patients registered with any of 33 autoimmune diseases in the Swedish Inpatient Register (Fallah et al., 2014). Overall, the standardized incidence ratio (SIR) for HL was 2.0 (95% CI 1.8–​2.2) in the patients with autoimmune disease, but varied from 5.2 (95% CI 4.2–​6.3) in the period < 1 year, over 2.0 (95% CI 1.7–​2.4) in the period 1–​4 years, to 1.5-​fold (95% CI 1.2–​1.7) in the period more than 5  years after the autoimmune disease diagnosis. When a lag period of 5 years between autoimmune and HL diagnoses was imposed, statistically significant increased risks were observed for rheumatoid arthritis (SIR = 2.8; 95% CI 2.1–​3.7), autoimmune hemolytic anemia (SIR 16.6; 95% CI 3.1–​ 49.2), Behcet’s disease (SIR 4.0; 95% CI 1.3–​9.3), and systemic lupus erythematosus (SIR 4.1; 95% CI 1.5–​9.0). Statistically non-​significant SIRs above 2 were observed for numerous other autoimmune diseases. The temporal variation in the relative risk of HL observed in the Swedish investigation is suggestive of non-​causal mechanisms, such as diagnostic misclassification and surveillance bias. These are likely to attenuate over time, but are hardly limited to only a 1-​year period after autoimmune disease diagnosis—​the lag period typically employed by other register-​based investigations—​and therefore may contribute to reported increased risks (Anderson et al., 2009; Kristinsson et al., 2009; Landgren et al., 2006). The association between autoimmune disease and HL also could reflect shared etiologies. This hypothesis has been assessed indirectly in Scandinavian register-​based investigations of HL risk by familial autoimmune disease history. In Sweden, Ekstrom et al. (2003) observed no unusual HL risk among first-​degree relatives of rheumatoid arthritis patients except children age 0–​14 years (SIR = 3.18; 95% CI 1.03–​7.42) (Ekstrom et al., 2003). No such association was reported in a similarly designed Danish investigation (Mellemkjaer et  al., 2000). Landgren and colleagues reported increased HL risk among first-​degree relatives of patients with sarcoidosis (OR = 1.8; 95% = 1.01–​3.1) and ulcerative colitis (OR = 1.6; 95% CI 1.02–​2.6) (Landgren et al., 2006). Epidemiological similarities between multiple sclerosis and HL in 1970 led Newell to suggest shared etiologic factors or pathways with HL (Newell, 1970). Evidence in support of this hypothesis has been mounting. Specifically, in partly overlapping register-​based investigations, multiple sclerosis was found to cluster with HL in younger adults, both within individuals (Montgomery et al., 2016) and between first-​degree relatives (Hjalgrim et  al., 2004; Landgren et  al., 2005). While this clustering of diseases may reflect shared environmental risk factors, such as infectious mononucleosis (Nielsen et al., 2007), a meta-​analysis of GWAS data also demonstrated substantial overlap in genetic predisposition to the two conditions (Khankhanian et  al., 2016).

Among its many adverse health effects on humans, tobacco smoking also interferes with the immune system. It is therefore conceivable that smoking would also be associated with an increased risk of HL, perhaps in particular of the EBV-​positive form. Findings regarding the association between tobacco smoking and HL risk recently have been subject to two meta-​analyses. Based on the same set of investigations, they both concluded that compared with never smoking, current smoking is associated with a 30%–​40% increased risk of HL overall and that the association exhibits a positive exposure-​response-​like pattern, which, however, most likely reflects the inclusion of never smokers as reference in the dose–​response analysis (Kamper-​Jorgensen et al., 2013). One of the meta-​analyses showed that the increased risk associated with current smoking applies to both nodular sclerosis HL (relative risk estimate  =  1.35; 95% CI 1.12–​1.63) and mixed cellularity HL (relative risk estimate  =  2.53; 95% CI 1.72–​3.72) (Sergentanis et al., 2013). The larger relative risk for mixed cellularity HL suggests that current smoking might be more strongly associated with EBV-​positive HL than EBV-​negative HL. This interpretation is consistent with results from the other meta-​analysis, which found a summary relative risk of EBV-​ positive HL of 2.26 (95% CI 1.69–​3.02) and summary relative risk of EBV-​negative HL of 1.40 (95% CI 1.08–​1.81) (Castillo et al., 2011).

Allergy A related but less studied question concerns the association between HL and allergic/​atopic diseases such as eczema, hay fever, and asthma. Adding to the interest here is that HL at diagnosis displays features indicative of a pro-​allergic immune response, such as elevated IgE levels and a decreased Th-​1 immune responses (reviewed by Martinez-​ Maza et al., 2010).

ALCOHOL Two prospective cohort studies have examined the association between alcohol intake and HL risk (Klatsky et al., 2009; Lim et al., 2007). Both found relative risk estimates that were less than unity for alcohol drinking at study entry, but no associations or exposure-​response trends were statistically significant. Power to detect an association in these studies was limited, with one study including only 57 HL patients (Lim et al., 2007) and the other including 43 patients (Klatsky et al., 2009). The suggestion of an inverse association between HL risk and alcohol consumption is on the whole supported by case-​control investigations, which consistently have reported decreased HL risk among alcohol users (Bernard et  al., 1987; Besson et  al., 2006; Monnereau et al., 2008; Nieters et al., 2006), even if not at a statistically significant level (Kanda et al., 2010a; Willett et al., 2007). The association has been reported for both younger adult and older adults (Besson et  al., 2006; Monnereau et  al., 2008), for both EBV-​ positive and -​negative HL (Besson et al., 2006), and in one study for both nodular sclerosis and mixed cellularity cHL (Monnereau et al., 2008), but seemingly only nodular sclerosis cHL in another (Besson et al., 2006). One study suggested that the reduced risk was restricted to nonsmokers (Gorini et al., 2007). The interpretation of these studies must consider that HL patients may experience discomfort upon intake of alcohol (Stein and Morgan, 2004). This is presumably just as true for incipient (undiagnosed) disease, and may result in reverse causality, that is, lower alcohol consumption due to HL.

UV RADIATION The potential protective effect of vitamin D on cancer risk has attracted considerable attention in recent decades, including as regards HL. The

760

760

PART IV:  Cancers by Tissue of Origin

body’s production of vitamin D is dependent on exposure of the skin to UV radiation, with diet being a secondary, exogenous source. An inverse association between UV radiation exposure and HL risk has been suggested by ecological investigations showing higher HL incidence rates in geographic regions with lower estimated levels of ambient UV radiation (Bowen et al., 2016; van Leeuwen et al., 2013). This association is corroborated by a recent pooled analysis including 1320 HL patients and 6381 controls showing that two measures of UV light exposure, sunburn (OR = 0.77; 95% CI 0.63–​0.95) and use of sun lamps (OR = 0.81; 95% CI 0.69–​0.96), along with estimated cumulative lifetime UV radiation exposure, were associated with decreased risks of HL (Monnereau et al., 2013). Interestingly, analyses indicated that the associations were more pronounced for EBV-​positive HL than for EBV-​negative HL. For estimated cumulative lifetime exposure, for instance, risk of EBV-​positive HL decreased with increasing cumulative lifetime exposure, with the OR for the most exposed compared with the least exposed being 0.56 (95% CI 0.35–​0.91) (ptrend, quartiles = 0.03) (Monnereau et al., 2013). Proposed biological mechanisms for the reported association involve induction of vitamin D3, an anti-​inflammatory effect of UV-​ radiation-​induced T-​cell regulation, or enhancement the DNA damage responses by effector T cells induced by the expression of EBV oncogenes (Monnereau et al., 2013).

MEDICATION No medications have been established as being associated with markedly increased risks of HL. Two studies, one an American case-​control investigation and the other a Danish register-​based study, have reported less than a two-​fold elevated HL risk among regular users of selective Cox-​2 inhibitors or acetaminophen (Chang et  al., 2011, 2004d). However, because the association was stronger with shorter interval to HL diagnosis, suspicion of reverse causality or possibly confounding by indication was raised. In contrast, both studies observed a decreased HL risk with long-​ term aspirin use and a meta-​analysis of the two studies identified an OR of 0.62 (95% CI 0.46–​0.82) (Chang et al., 2011). If the relationship is causal, the reduced HL risk induced by aspirin could be mediated through inhibition of the constitutively active NF-​κB pathway in HL.

OCCUPATION

FUTURE DIRECTIONS In 1958, Brian MacMahon inferred from epidemiological and clinical data that two etiologically heterogeneous variants of HL—​implicitly cHL by modern nosology—​exist. Today, more than half a century later, evidence is overwhelming that EBV-​positive cHL and EBV-​negative cHL define two such etiologically separate entities. The two cHL variants display both molecular-​biological and epidemiological differences, for example, with respect to patient HLA constitution and association with immune function as summarized in the so-​called four-​ disease model (Figure 39–10) (Jarrett, 2002). Future investigations can shed further light on the precise mechanisms by which EBV and HLA interact to cause EBV-​positive cHL. Whether EBV sequence variation affects risk of EBV-​positive cHL is another area for potentially fruitful study. In contrast, the causes of EBV-​negative cHL remain to be determined. Indeed, there are few good leads as to the etiology of this cHL variant, which comprises the majority of cases. Further heterogeneity of EBV-​negative cHL, for example, by histology, age group, or other factors, may be partly responsible for obscuring risk factor associations. Nodular lymphocyte dominant HL is another disease subgroup for which very little is known about risk factors, and that could be an intriguing topic for future research, provided that sufficient case numbers could be identified. Classifications are fundamental for the diagnosis, treatment, and investigation of diseases, including the lymphohematopoietic malignancies (Harris et al., 2008). Accordingly, the definition of a disease ideally incorporates information about etiology, pathological characteristics, and clinical behavior, and as new knowledge accumulates, it may be relevant to revise existing disease definitions. Unfortunately, however, the evidence of etiological heterogeneity between EBV-​positive and EBV-​negative cHL has yet to percolate to the clinical setting. Thus, although the current WHO classification of cHL devotes separate chapters for the histological subtypes of cHL, such distinctions currently have limited clinical implications. Moreover, presumably because of the highly effective therapy for HL offered today, there seems to be little incentive to include new markers of suspected etiologic relevance into the current HL classification. The potential health costs incurred by delays in the identification of preventable risk factors for cHL and possibly in introduction of etiology-​adapted therapy in cHL are difficult to estimate. Accordingly,

Incidence

Incidence

There is no convincing evidence of an association between HL risk and occupation, occupational exposures, or environmental chemical exposures, as noted in the previous edition of the present chapter (Mueller and Grufferman, 2006), and recently reiterated by a review of the entire literature since 1960 (Charbotel et  al., 2014). As possible exceptions to this general pattern, the review singled out possible

increased risks with pesticide exposures (e.g., Navaranjan et al., 2013), although no specific pesticide has been shown consistently to increase HL risk, and with occupational exposure to high-​molecular-​weight agents associated with asthma (Kogevinas et  al., 2004). However, the latter finding was not confirmed in two subsequent investigations, which if anything pointed to a reduced risk (Espinosa et  al., 2013; Mirabelli et al., 2009).

0

10 20 30 40 50 60 70 80 Age group (years)

0

10 20 30 40 50 60 70 80 Age group (years)

Figure 39–10.  Four-​disease model according to Jarrett (2002). Modeled age-​specific incidence rates of EBV-​positive classical Hodgkin lymphoma (left) and EBV-​negative classical Hodgkin lymphoma (right). For EBV-​positive classical Hodgkin lymphoma, incidence peaks occur around typical ages at primary EBV infection and with an age-​dependent increase in immune deficiency. EBV-​negative classical Hodgkin lymphoma follows a unimodal age distribution. Source: Reproduced from Hjalgrim (2012).

 761



Hodgkin Lymphoma

some investigations have suggested that HL prognosis varies among strata defined by combinations of age and EBV status (Clarke et al., 2001; Diepstra et al., 2009; Jarrett et al., 2005; Keegan et al., 2005). From an epidemiological point of view, international collaboration to ensure large studies with sufficient statistical power seems to be the most promising way forward to identify cHL risk factors of relevance for preventive efforts and to characterize the web of causes and consequences of the interplay among sex, age, histological subtype, and tumor EBV status. Optimally, such endeavors should involve clinicians from the outset to ensure ready identification and adaptation of observations relevant to the clinical setting. This transdisciplinary approach could help to overcome the boundaries that have traditionally separated etiologic research on cHL from clinical practice, and in this way have the potential to yield insights beneficial to both realms. References Alexander FE, Jarrett RF, Lawrence D, et al. 2000. Risk factors for Hodgkin’s disease by Epstein-​Barr virus (EBV) status: prior infection by EBV and other agents. Br. J. Cancer, 82(5), 1117–​1121. PMCID: PMC2374437. Alexander FE, Lawrence DJ, Freeland J et al. 2003. An epidemiologic study of index and family infectious mononucleosis and adult Hodgkin’s disease (HD): evidence for a specific association with EBV+ve HD in young adults. Int. J. Cancer, 107(2), 298–​302. PMID: 12949811. Amiel JL. 1967. Study of leucocyte phenotypes in Hodgkin’s disease:  histocompatibility testing (pp. 79–​81). Copenhagen: Munksgaard. Anderson LA, Gadalla S, Morton LM, et al. 2009. Population-​based study of autoimmune conditions and the risk of specific lymphoid malignancies. Int. J. Cancer, 125(2), 398–​405. PMCID: PMC2692814. Armitage JO. 2010. Early-​ stage Hodgkin’s lymphoma. N. Engl. J.  Med., 363(7), 653–​662. PMID: 20818856. Armstrong AA, Alexander FE, Cartwright RA, et al. 1998. Epstein-​Barr virus and Hodgkin’s disease: further evidence for the three disease hypothesis. Leukemia, 12, 1272–​1276. PMID: 9697883. Balfour HH, Jr., Odumade OA, Schmeling DO, et al. 2013. Behavioral, virologic, and immunologic factors associated with acquisition and severity of primary Epstein-​Barr virus infection in university students. J. Infect. Dis., 207(1), 80–​88. PMCID: PMC3523797. Becker N, Deeg E, Rudiger T, and Nieters A. 2005. Medical history and risk for lymphoma: results of a population-​based case-​control study in Germany. Eur. J. Cancer, 41(1), 133–​142. PMID: 15617998. Becker N, Fortuny J, Alvaro T, et al. 2009. Medical history and risk of lymphoma: results of a European case-​control study (EPILYMPH). J. Cancer Res. Clin. Oncol., 135(8), 1099–​1107. PMID: 19205736. Bernard SM, Cartwright RA, Darwin CM, et al. 1987. Hodgkin’s disease: case control epidemiological study in Yorkshire. Br. J. Cancer, 55(1), 85–​90. PMID: 3814482. Besson H, Brennan P, Becker N, et al. 2006. Tobacco smoking, alcohol drinking and Hodgkin’s lymphoma:  a European multi-​centre case-​control study (EPILYMPH). Br. J. Cancer, 95(3), 378–​384. PMCID: PMC2360649. Biggar RJ, Jaffe ES, Goedert JJ, et al. 2006. Hodgkin lymphoma and immunodeficiency in persons with HIV/​AIDS. Blood, 108(12), 3786–​3791. PMCID: PMC1895473. Bonelli L, Vitale V, Bistolfi F, Landucci M, and Bruzzi P. 1990. Hodgkin’s disease in adults:  association with social factors and age at tonsillectomy. A case-​control study. Int. J. Cancer, 45(3), 423–​427. PMID: 2307531. Bosetti C, Dal Maso L, Negri E, et al. 2005. Re: Body mass index and risk of malignant lymphoma in Scandinavian men and women. J. Natl. Cancer Inst., 97(11), 860–​861. PMID: 15928310. Bowen EM, Pfeiffer RM, Linet MS, et al. 2016. Relationship between ambient ultraviolet radiation and Hodgkin lymphoma subtypes in the United States. Br. J. Cancer, 114(7), 826–​831. PMCID: 4984855. Carozza SE, Puumala SE, Chow EJ, et al. 2010. Parental educational attainment as an indicator of socioeconomic status and risk of childhood cancers. Br. J. Cancer, 103(1), 136–​142. PMCID: PMC2905284. Castillo JJ, Dalia S, and Shum H. 2011. Meta-​analysis of the association between cigarette smoking and incidence of Hodgkin’s Lymphoma. J. Clin. Oncol., 29(29), 3900–​3906. PMID: 21911724. Chang ET, Froslev T, Sorensen HT, and Pedersen L. 2011. A nationwide study of aspirin, other non-​steroidal anti-​inflammatory drugs, and Hodgkin lymphoma risk in Denmark. Br. J.  Cancer, 105(11), 1776–​ 1782. PMCID: 3242601. Chang ET, Hjalgrim H, Smedby KE, et al. 2005a. Body mass index and risk of malignant lymphoma in Scandinavian men and women. J. Natl. Cancer Inst., 97(3), 210–​218. PMID: 15687364.

761

Chang ET, Montgomery SM, Richiardi L, et  al. 2004a. Number of siblings and risk of Hodgkin’s lymphoma. Cancer Epidemiol. Biomarkers Prev., 13(7), 1236–​1243. PMID: 15247136. Chang ET, Smedby KE, Hjalgrim H, et al. 2005b. Family history of hematopoietic malignancy and risk of lymphoma. J. Natl. Cancer Inst., 97(19), 1466–​1474. PMID: 16204696. Chang ET, Zheng T, Weir EG, et al. 2004c. Childhood social environment and Hodgkin’s lymphoma: new findings from a population-​based case-​control study. Cancer Epidemiol. Biomarkers Prev., 13(8), 1361–​ 1370. PMID: 15298959. Chang ET, Zheng T, Lennette ET, et  al. 2004b. Heterogeneity of risk factors and antibody profiles in epstein-​ barr virus genome-​ positive and -​negative hodgkin lymphoma. J. Infect. Dis., 189(12), 2271–​2281. PMID: 15181575. Chang ET, Zheng T, Weir EG, et al. 2004d. Aspirin and the risk of Hodgkin’s lymphoma in a population-​based case-​control study. J. Natl. Cancer Inst., 96(4), 305–​315. PMID: 14970279. Charbotel B, Fervers B, and Droz JP. 2014. Occupational exposures in rare cancers:  A  critical review of the literature. Crit. Rev. Oncol. Hematol., 90(2), 99–​134. PMID: 24387944. Chatenoud L, Bertuccio P, Bosetti C, et al. 2013. Hodgkin’s lymphoma mortality in the Americas, 1997–​2008: achievements and persistent inadequacies. Int. J. Cancer, 133(3), 687–​694. PMID: 23335127. Chatenoud L, Gallus S, Altieri A, et al. 2005. Number of siblings and risk of hodgkin’s and other lymphoid neoplasms. Cancer Epidemiol. Biomarkers Prev., 14(2), 552. PMID: 15734990. Clarke CA, Glaser SL, Dorfman RF, et al. 2001. Epstein-​Barr virus and survival after Hodgkin disease in a population-​based series of women. Cancer, 91(8), 1579–​1587. PMID: 11301409. Clarke CA, Glaser SL, Keegan THM, and Stroup A. 2005. Neighborhood socioeconomic status and Hodgkin’s lymphoma incidence in California. Cancer Epidemiol. Biomarkers Prev., 14(6), 1441–​ 1447. PMID: 15941953. Clavel J, Steliarova-​Foucher E, Berger C, Danon S, and Valerianova Z. 2006. Hodgkin’s disease incidence and survival in European children and adolescents (1978–​1997):  report from the Automated Cancer Information System project. Eur. J. Cancer, 42(13), 2037–​2049. PMID: 16919769. Cole P, MacMahon B, and Aisenberg A. 1968. Mortality from Hodgkin’s disease in the United States: evidence for the multiple-​aetiology hypothesis. Lancet, 2(7583), 1371–​1376. PMID: 4177934. Correa P, and O’Conor GT. 1971. Epidemiologic patterns of Hodgkin’s disease. Int. J. Cancer, 8(2), 192–​201. PMID: 5133848. Cozen W, Hamilton AS, Zhao P, et  al. 2009. A protective role for early oral exposures in the etiology of young adult Hodgkin lymphoma. Blood, 114(19), 4014–​4020. PMCID: PMC2774542. Cozen W, Katz J, and Mack TM. 1992. Risk patterns of Hodgkin’s disease in Los Angeles vary by cell type. Cancer Epidemiol. Biomarkers Prev., 1(4), 261–​268. PMID: 1303125. Cozen W, Li D, Best T, et  al. 2012. A genome-​wide meta-​analysis of nodular sclerosing Hodgkin lymphoma identifies risk loci at 6p21.32. Blood, 119(2), 469–​475. PMCID: PMC3257012. Cozen W, Timofeeva MN, Li D, et al. 2014. A meta-​analysis of Hodgkin lymphoma reveals 19p13.3 TCF3 as a novel susceptibility locus. Nature Commun., 5, 3856. PMCID: PMC4055950. Crawford DH, Macsween KF, Higgins CD, et al. 2006. A cohort study among university students:  identification of risk factors for Epstein-​Barr virus seroconversion and infectious mononucleosis. Clin. Infect. Dis., 43(3), 276–​282. PMID: 16804839. Crump C, Sundquist K, Sieh W, Winkleby MA, and Sundquist J. 2012. Perinatal and family risk factors for Hodgkin lymphoma in childhood through young adulthood. Am. J.  Epidemiol., 176(12), 1147–​ 1158. PMCID: PMC3571233. Derenzini E, and Younes A. 2011. Predicting treatment outcome in classical Hodgkin lymphoma:  genomic advances. Genome Med., 3(4), 26. PMCID: PMC3129642. Diepstra A, Niens M, te Meerman GJ, Poppema S, and van den Berg A. 2005a. Genetic susceptibility to Hodgkin’s lymphoma associated with the human leukocyte antigen region. Eur. J. Haematol. Suppl. 66, 34–​41. PMID: 16007866. Diepstra A, Niens M, Vellenga E, et al. 2005b. Association with HLA class I in Epstein-​ Barr-​ virus-​ positive and with HLA class  III in Epstein-​ Barr-​ virus-​ negative Hodgkin’s lymphoma. Lancet, 365(9478), 2216–​ 2224. PMID: 15978930. Diepstra A, van Imhoff GW, Schaapveld M, et al. 2009. Latent Epstein-​Barr virus infection of tumor cells in classical Hodgkin’s lymphoma predicts adverse outcome in older adult patients. J. Clin. Oncol., 27(23), 3815–​ 3821. PMID: 19470931.

762

762

PART IV:  Cancers by Tissue of Origin

Ekstrom K, Hjalgrim H, Brandt L, et al. 2003. Risk of malignant lymphomas in patients with rheumatoid arthritis and in their first-​degree relatives. Arthritis Rheum., 48(4), 963–​970. PMID: 12687538 Enciso-​Mora V, Broderick P, Ma Y, et  al. 2010. A genome-​wide association study of Hodgkin’s lymphoma identifies new susceptibility loci at 2p16.1 (REL), 8q24.21 and 10p14 (GATA3). Nat. Genet., 42(12), 1126–​1130. PMCID: PMC4268499. Engholm G, Ferlay J, Christensen N, et al. 2010. NORDCAN: a Nordic tool for cancer information, planning, quality control and research. Acta Oncol., 49(5), 725–​736. PMID: 20491528. Englund A, Hopstadius C, Enblad G, Gustafsson G, and Ljungman G. 2015. Hodgkin lymphoma:  a survey of children and adolescents treated in Sweden 1985–​2009. Acta Oncol., 54(1), 41–​48. PMCID: PMC4364274. Epstein MA, Achong BG, and Barr YM. 1964. Virus particles in cultured lymphoblasts from Burkitt’s lymphoma. Lancet 1, 702–​703. PMID 14107961. Espinosa A, Zock JP, Benavente Y, et  al. 2013. Occupational exposure to immunologically active agents and risk for lymphoma:  the European Epilymph case-​ control study. Cancer Epidemiol., 37(4), 378–​ 384. PMID: 23522435. Exley MA, Hand L, O’Shea D, and Lynch L. 2014. Interplay between the immune system and adipose tissue in obesity. J. Endocrinol., 223(2), R41–​R48. PMID: 25228503. Fallah M, Liu X, Ji J, et al. 2014. Hodgkin lymphoma after autoimmune diseases by age at diagnosis and histological subtype. Ann. Oncol., 25(7), 1397–​1404. PMID: 24718892. Farrell K, and Jarrett RF. 2011. The molecular pathogenesis of Hodgkin lymphoma. Histopathology, 58(1), 15–​25. PMID: 21261680. Ferlay J, Soerjomataram I, Ervik M, et al. 2013, 6/​1/​2015. GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality Worldwide: IARC CancerBase No. 11 [Internet]. Available from: http://​globocan.iarc.fr/​. Accessed July 20, 2015. Forman D, Brewster DH, Gombe-​Mbalawa C, et al. 2013, 6/​1/​2015. Cancer in Five Continents. http://​ci5/​. iarc. fr, X. Frampton M, da Silva Filho MI, Broderick P, et al. 2013. Variation at 3p24.1 and 6q23.3 influences the risk of Hodgkin’s lymphoma. Nature Commun., 4, 2549. PMID: 24149102 Friedman DL, Kadan-​Lottick NS, Whitton J, et  al. 2005. Increased risk of cancer among siblings of long-​term childhood cancer survivors: a report from the childhood cancer survivor study. Cancer Epidemiol. Biomarkers Prev., 14(8), 1922–​1927. PMID: 16103438. Frisch M, Biggar RJ, Engels EA, and Goedert JJ. 2001. Association of cancer with AIDS-​related immunosuppression in adults. JAMA, 285(13), 1736–​ 1745. PMID: 11277828. Gallagher A, Armstrong AA, MacKenzie J, et al. 1999. Detection of Epstein-​ Barr virus (EBV) genomes in the serum of patients with EBV-​associated Hodgkin’s disease. Int. J. Cancer, 84(4), 442–​448. PMID: 10404101. Gaulard P, Swerdlow SH, Harris NL, Jaffe ES, and Sundström C. 2008. Other iatrogenic immunodeficiency-​ associated disorders. In: Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H, Thiele J, and Vardiman JW (Eds.), Tumours of haematopoietic and lymphoid tissues (3rd ed., pp. 350–​351). Lyon: IARC. Glaser SL, Chang ET, Clarke CA et  al. 2015a. Hodgkin lymphoma incidence in ethnic enclaves in California. Leuk. Lymphoma, 56(12),1–​23. PMCID: PMC4801145. Glaser SL, Clarke CA, Keegan TH, Chang ET, and Weisenburger DD. 2015b. Time trends in rates of Hodgkin lymphoma histologic subtypes:  true incidence changes or evolving diagnostic practice? Cancer Epidemiol. Biomarkers Prev., 24(10), 1474–​1488. PMCID: PMC4592457. Glaser SL, Clarke CA, Nugent RA, Stearns CB, and Dorfman RF. 2002. Social class and risk of Hodgkin’s disease in young-​adult women in 1988–​94. Int. J. Cancer, 98(1), 110–​117. PMID: 11857394. Glaser SL, Dorfman RF, and Clarke CA. 2001. Expert review of the diagnosis and histologic classification of Hodgkin disease in a population-​based cancer registry: interobserver reliability and impact on incidence and survival rates. Cancer, 92(2), 218–​224. PMID: 11466672. Glaser SL, Gulley ML, Clarke CA, et al. 2008. Racial/​ethnic variation in EBV-​ positive classical Hodgkin lymphoma in California populations. Int. J. Cancer, 123(7), 1499–​1507. PMCID: PMC2775059. Glaser SL, Keegan TH, Clarke CA, et al. 2005. Exposure to childhood infections and risk of Epstein-​Barr virus-​defined Hodgkin’s lymphoma in women. Int. J. Cancer, 115(4), 599–​605. PMID: 15700307. Glaser SL, Lin RJ, Stewart SL, et  al. 1997. Epstein-​Barr virus-​associated Hodgkin’s disease:  epidemiologic characteristics in international data. Int. J. Cancer, 70, 375–​382. PMID: 9033642. Glaser SL, and Swartz WG. 1990. Time trends in Hodgkin’s disease incidence:  the role of diagnostic accuracy. Cancer, 66(10), 2196–​ 2204. PMID: 2224775.

Glimelius I, Ekberg S, Linderoth J, et al. 2015. Sick leave and disability pension in Hodgkin lymphoma survivors by stage, treatment, and follow-​up time-​a population-​based comparative study. J. Cancer Surviv, 9(4), 599–​ 609. PMID: 25689964. Goldacre MJ, Wotton CJ, and Yeates DG. 2009. Associations between infectious mononucleosis and cancer: record-​ linkage studies. Epidemiol. Infect., 137(5), 672–​680. PMID: 18840316. Goldin LR, Bjorkholm M, Kristinsson SY, Turesson I, and Landgren O. 2009. Highly increased familial risks for specific lymphoma subtypes. Br. J. Haematol., 146(1), 91–​94. PMCID: PMC2702464. Gorini G, Stagnaro E, Fontana V et al. 2007. Alcohol consumption and risk of Hodgkin’s lymphoma and multiple myeloma: a multicentre case-​control study. Ann. Oncol., 18(1), 143–​148. PMID: 17047000. Grufferman S, Cole P, Smith PG, and Lukes RJ. 1977. Hodgkin’s disease in siblings. N. Engl. J. Med., 296(5), 248–​250. PMID: 83110. Gutensohn N, and Cole P. 1977. Epidemiology of Hodgkin’s disease in the young. Int. J. Cancer, 19, 595–​604. PMID: 863541. Gutensohn N, and Cole P. 1981. Childhood social environment and Hodgkin’s disease. N. Engl. J. Med., 304(3), 135–​140. PMID: 6255329. Gutensohn N, Li FP, Johnson RE, and Cole P. 1975. Hodgkin’s disease, tonsillectomy and family size. N. Engl. J.  Med., 292(1), 22–​ 25. PMID: 1109186. Gutensohn NM. 1982. Social class and age at diagnosis of Hodgkin’s disease:  new epidemiological evidence for the “two-​disease hypothesis.” Cancer Treat. Rep., 66, 689–​695. PMID: 7074638. Haim N, Cohen Y, and Robinson E. 1982. Malignant lymphoma in first-​degree blood relatives. Cancer, 49, 2197–​2200. Harris NL, Campo E, Jaffe ES, et al. 2008. Introduction to the WHO classification of tumours of haematopoietic and lymphoid tissues. In: Jaffe ES, Harris NL, Stein H, and Vardiman JW (Eds.), Tumours of haematopoietic and lymphoid tissues (pp. 14–​15). Lyon: IARC. Harty LC, Lin AY, Goldstein AM, et  al. 2002. HLA-​ DR, HLA-​ DQ, and TAP genes in familial Hodgkin disease. Blood, 99(2), 690–​ 693. PMID: 11781255. Henle W, Henle G, Andersson J, et al. 1987. Antibody responses to Epstein-​ Barr virus-​determined nuclear antigen (EBNA)-​1 and EBNA-​2 in acute and chronic Epstein-​Barr virus infection. Proc. Natl. Acad. Sci., U. S. A., 84(2), 570–​574. PMCID: PMC304252. Hislop AD, Taylor GS, Sauce D, and Rickinson AB. 2007. Cellular responses to viral infection in humans: lessons from Epstein-​Barr virus. Ann. Rev. Immunol., 25, 587–​617. PMID: 17378764. Hjalgrim H. 2012. On the aetiology of Hodgkin lymphoma. Dan. Med. J., 59(7), B4485. PMID: 22759852. Hjalgrim H, Askling J, Pukkala E, et al. 2001. Incidence of Hodgkin’s disease in Nordic countries. Lancet, 358(9278), 297–​298. PMID: 11498219. Hjalgrim H, Askling J, Rostgaard K, et al. 2003. Characteristics of Hodgkin’s lymphoma after infectious mononucleosis. N. Engl. J.  Med., 349(14), 1324–​1332. PMID: 14523140. Hjalgrim H, Askling J, Sorensen P, et  al. 2000. Risk of Hodgkin’s disease and other cancers after infectious mononucleosis. J. Natl. Cancer Inst., 92(18), 1522–​1528. PMID: 10995808. Hjalgrim H, Rasmussen S, Rostgaard K, et  al. 2004. Familial clustering of Hodgkin lymphoma and multiple sclerosis. J. Natl. Cancer Inst., 96(10), 780–​784. PMID: 15150306. Hjalgrim H, Rostgaard K, Johnson PC, et al. 2010. HLA-​A alleles and infectious mononucleosis suggest a critical role for cytotoxic T-​cell response in EBV-​related Hodgkin lymphoma. Proc. Natl. Acad. Sci. U. S. A., 107(14), 6400–​6405. PMCID: PMC2851961. Hjalgrim H, Smedby KE, Rostgaard K, et al. 2007. Infectious mononucleosis, childhood social environment, and risk of Hodgkin lymphoma. Cancer Res., 67(5), 2382–​2388. PMID: 17332371. Hjalgrim LL, Rostgaard K, Engholm G, et al. 2016. Aetiologic heterogeneity in pediatric Hodgkin lymphoma? Evidence from the Nordic countries, 1978–​2010. Acta Oncol., 55(1), 85–​90. PMID: 26073450. Hors J, and Dausset J. 1983. HLA and susceptibility to Hodgkin’s disease. Immunol. Rev., 70, 167–​192. PMID: 6403456. Howlader N, Noone AM, Krapcho M, et al. 2014. Surveillance, Epidemiology, and End Results (SEER) Program (http://​www.seer.cancer.gov/​) SEER*Stat Database:  Incidence—​ SEER 18 Regs Research Data + Hurricane Katrina Impacted Louisiana Cases, Nov 2014 Sub (2000–​2012) Katrina/​ Rita Population Adjustment—​ Linked to County Attributes—​ Total U.S., 1969–​ 2013 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2015, based on the November 2014 submission. National Cancer Institute. Bethesda, MD. http://​seer.cancer.gov/​csr/​1975_​2012/​ Huang X, Hepkema B, Nolte I, et  al. 2012a. HLA-​ A*02:07 is a protective allele for EBV negative and a susceptibility allele for EBV

 763



Hodgkin Lymphoma

positive classical Hodgkin lymphoma in China. PLoS One, 7(2), e31865. PMCID: PMC3280205. Huang X, Kushekhar K, Nolte I, et  al. 2012b. HLA associations in classical Hodgkin lymphoma:  EBV status matters. PLoS One, 7(7), e39986. PMCID: PMC3393726. IARC. 1997. Epstein-​Barr virus and Kaposi’s sarcoma herpesvirus/​herpesvirus 8. IARC Monogr. Eval. Carcinog. Risks Hum., 70, 2075. Isager H, and Andersen E. 1978. Pre-​morbid factors in Hodgkin’s disease. I.  Birth weight and growth pattern from 8 to 14  years of age. Scand. J. Haematol., 21(3), 250–​255. PMID: 715377. Jaffe ES, Harris NL, Diebold J, and Müller-​Hermelink HK. 1998. World Health Organization Classification of lymphomas:  a work in progress. Ann. Oncol., 9 (Suppl 5), S25–​S30. PMID: 9926234. Jarrett RF. 2002. Viruses and Hodgkin’s lymphoma. Ann. Oncol., 13 (Suppl 1), 23–​29. PMID: 12078898. Jarrett RF, Gallagher A, Jones DB, et al. 1991. Detection of Epstein-​Barr virus genomes in Hodgkin’s disease: relation to age. J. Clin. Pathol., 44(10), 844–​848. PMCID: PMC496672. Jarrett RF, Krajewski AS, Angus B, et al. 2003. The Scotland and Newcastle epidemiological study of Hodgkin’s disease: impact of histopathological review and EBV status on incidence estimates. J. Clin. Pathol., 56(11), 811–​816. PMCID: PMC1770114. Jarrett RF, Stark GL, White J, et al. 2005. Impact of tumor Epstein-​Barr virus status on presenting features and outcome in age-​defined subgroups of patients with classic Hodgkin lymphoma:  a population-​ based study. Blood, 106(7), 2444–​2451. PMID: 15941916. Johnson PC, McAulay KA, Montgomery D et al. 2015. Modeling HLA associations with EBV-​positive and -​negative Hodgkin lymphoma suggests distinct mechanisms in disease pathogenesis, 137(5), 1066–​1075. Int. J. Cancer. PMCID: PMC4737225. Kamper-​Jorgensen M, Rostgaard K, Glaser SL, et al. 2013. Cigarette smoking and risk of Hodgkin lymphoma and its subtypes: a pooled analysis from the International Lymphoma Epidemiology Consortium (InterLymph). Ann. Oncol., 24(9), 2245–​2255. PMCID: PMC3755332. Kanakry JA, Li H, Gellert LL, et  al. 2013. Plasma Epstein-​Barr virus DNA predicts outcome in advanced Hodgkin lymphoma:  correlative analysis from a large North American cooperative group trial. Blood, 121(18), 3547–​3553. PMCID: PMC3643756. Kanda J, Matsuo K, Inoue M, et al. 2010a. Association of alcohol intake with the risk of malignant lymphoma and plasma cell myeloma in Japanese: a population-​ based cohort study (Japan Public Health Center-​ based Prospective Study). Cancer Epidemiol. Biomarkers Prev., 19(2), 429–​ 434. PMID: 20086115. Kanda J, Matsuo K, Suzuki T, et al. 2010b. Association between obesity and the risk of malignant lymphoma in Japanese:  a case-​control study. Int. J. Cancer, 126(10), 2416–​2425. PMID: 19821488. Kanzler H, Kuppers R, Hansmann ML, and Rajewsky K. 1996. Hodgkin and Reed-​Sternberg cells in Hodgkin’s disease represent the outgrowth of a dominant tumor clone derived from (crippled) germinal center B cells. J. Exp. Med., 184(4), 1495–​1505. PMCID: PMC2192840. Kaplan HS. 1972. Hodgkin’s disease, 1st ed. Cambrigde, MA:  Harvard University Press. Keegan TH, Glaser SL, Clarke CA, et al. 2005. Epstein-​Barr virus as a marker of survival after Hodgkin’s lymphoma: a population-​based study. J. Clin. Oncol., 23(30), 7604–​7613. PMID: 16186595. Keegan TH, Glaser SL, Clarke CA, et al. 2006. Body size, physical activity, and risk of Hodgkin’s lymphoma in women. Cancer Epidemiol. Biomarkers Prev., 15(6), 1095–​1101. PMID: 16775165 Kerber RA, and O’Brien E. 2005. A cohort study of cancer risk in relation to family histories of cancer in the Utah population database. Cancer, 103(9), 1906–​1915. PMID: 15779016 Kerzin-​Storrar L, Faed MJW, MacGillivray JB, and Smith PG. 1983. Incidence of familial Hodgkin’s disease. Br. J. Cancer, 47(5), 707–​712. PMCID: PMC2011375. Khankhanian P, Cozen W, Himmelstein DS, et  al. 2016. Meta-​analysis of genome-​ wide association studies reveals genetic overlap between Hodgkin lymphoma and multiple sclerosis. Int. J.  Epidemiol. PMCID: PMC5005944. Kharazmi E, Fallah M, Pukkala E, et  al. 2015. Risk of familial classical Hodgkin lymphoma by relationship, histology, age, and sex:  a joint study from five Nordic countries. Blood, 126(17), 1990–​1995. PMID: 26311361. Klatsky AL, Li Y, Baer D, et al. 2009. Alcohol consumption and risk of hematologic malignancies. Ann. Epidemiol., 19(10), 746–​753. PMID: 19394864. Kogevinas M, Zock JP, Alvaro T, et al. 2004. Occupational exposure to immunologically active agents and risk for lymphoma. Cancer Epidemiol. Biomarkers Prev., 13(11 Pt 1), 1814–​1818. PMID: 15533913.

763

Kowalkowski MA, Mims MA, Day RS, et al. 2014. Longer duration of combination antiretroviral therapy reduces the risk of Hodgkin lymphoma: a cohort study of HIV-​infected male veterans. Cancer Epidemiol., 38(4), 386–​392. PMCID: PMC4447432. Kristinsson SY, Gao Y, Bjorkholm M, et al. 2015. Hodgkin lymphoma risk following infectious and chronic inflammatory diseases: a large population-​ based case-​control study from Sweden. Int. J. Hematol., 101(6), 563–​568. PMID: 25758095. Kristinsson SY, Landgren O, Sjoberg J, et al. 2009. Autoimmunity and risk for Hodgkin’s lymphoma by subtype. Haematologica, 94(10), 1468–​1469. PMCID: PMC2754970. Kuppers R. 2009. The biology of Hodgkin’s lymphoma. Nature Rev. Cancer, 9(1), 15–​27. PMID: 19078975. Kushekhar K, van den Berg A, Nolte I, et  al. 2014. Genetic associations in classical hodgkin lymphoma: a systematic review and insights into susceptibility mechanisms. Cancer Epidemiol. Biomarkers Prev., 23(12), 2737–​2747. PMID: 25205514. La Vecchia C, Negri E, Parazzini F, et  al. 1990. Height and cancer risk in a network of case-​control studies from northern Italy. Int. J. Cancer, 45(2), 275–​279. PMID: 2303293. Landgren O, Engels EA, Pfeiffer RM, et al. 2006. Autoimmunity and susceptibility to Hodgkin lymphoma: a population-​based case-​control study in Scandinavia. J. Natl. Cancer Inst., 98(18), 1321–​1330. PMID: 16985251. Landgren O, Kerstann KF, Gridley G, et al. 2005. Re: Familial clustering of Hodgkin lymphoma and multiple sclerosis. J. Natl. Cancer Inst., 97(7), 543–​544. PMID: 15812083. Larsson SC, and Wolk A. 2011. Body mass index and risk of non-​Hodgkin’s and Hodgkin’s lymphoma:  a meta-​analysis of prospective studies. Eur. J. Cancer, 47(16), 2422–​2430. PMID: 21733676. Lee JH, Kim Y, Choi JW, and Kim YS. 2014. Prevalence and prognostic significance of Epstein-​Barr virus infection in classical Hodgkin’s lymphoma: a meta-​analysis. Arch. Med. Research, 45(5), 417–​431. PMID: 24937173. Levin LI, Chang ET, Ambinder RF, et al. 2012. Atypical prediagnosis Epstein-​ Barr virus serology restricted to EBV-​ positive Hodgkin lymphoma. Blood, 120(18), 3750–​3755. PMCID: PMC3488887. Levine R, Zhu K, Gu Y, et al. 1998. Self-​reported infectious mononucleosis and 6 cancers: a population-​based, case-​control study. Scand. J. Infect. Dis., 30(3), 211–​214. PMID: 9790125. Li Q, Chang ET, Bassig BA, et  al. 2013. Body size and risk of Hodgkin’s lymphoma by age and gender: a population-​based case-​control study in Connecticut and Massachusetts. Cancer Causes Control, 24(2), 287–​295. PMCID: PMC3557669. Lim U, Morton LM, Subar AF, et al. 2007. Alcohol, smoking, and body size in relation to incident Hodgkin’s and non-​Hodgkin’s lymphoma risk. Am. J. Epidemiol., 166(6), 697–​708. PMID: 17596266. Linabery AM, Erhardt EB, Fonstad RK, et  al. 2014. Infectious, autoimmune and allergic diseases and risk of Hodgkin lymphoma in children and adolescents: a Children’s Oncology Group study. Int. J. Cancer, 135(6), 1454–​1469. PMCID: PMC4107091. Lukes RJ, Craver LF, Hall TC, Rappaport H, and Ruben P. 1966. Report of the nomenclature committee. Cancer Res., 26(1311), 1311. Lumio J, and Karjalainen S. 1993. Patients treated in hospital for infectious mononucleosis and risk of cancer. Scand. J.  Infect. Dis., 25, 283–​288. PMID: 8362223. Mack TM, Cozen W, Shibata DK, et  al. 1995. Concordance for Hodgkin’s disease in identical twins suggesting genetic susceptibility to the young-​adult form of the disease. N. Engl. J.  Med., 332(7), 413–​418. PMID: 7824015. MacMahon B. 1957. Epidemiological evidence of the nature of Hodgkin’s disease. Cancer, 10(5), 1045–​1054. PMID: 13472655 MacMahon B. 1966. Epidemiology of Hodgkin’s disease. Cancer Res., 26, 1189–​1201. PMID: 5329907 MacMahon B, Cole P, and Newell GR. 1971. Hodgkin’s disease: one entity or two? Lancet, 1(7692), 240–​241. PMID: 4099902. Martinez-​Maza O, Moreno AD, and Cozen W. 2010. Epidemiological evidence:  IgE, allergies, and hematopoietic malignancies. In:  Penichet ML, and Jensen-​ Jarolim E (Eds.), Cancer and IgE (pp. 79–​ 136). New York: Springer. Matasar MJ, Ford JS, Riedel ER, et  al. 2015. Late morbidity and mortality in patients with Hodgkin’s lymphoma treated during adulthood. J. Natl. Cancer Inst., 107(4). PMID: 25717170. McAulay KA, and Jarrett RF. 2015. Human leukocyte antigens and genetic susceptibility to lymphoma. Tissue Antigens, 86(2), 98–​ 113. PMID: 26189878. McDuffie HH, Pahwa P, Karunanayake CP, Spinelli JJ, and Dosman JA. 2009. Clustering of cancer among families of cases with Hodgkin Lymphoma (HL), Multiple Myeloma (MM), Non-​ Hodgkin’s Lymphoma (NHL),

764

764

PART IV:  Cancers by Tissue of Origin

Soft Tissue Sarcoma (STS) and control subjects. BMC Cancer, 9, 70. PMCID: PMC2653543. Mellemkjaer L, Alexander F, and Olsen JH. 2000. Cancer among children of parents with autoimmune diseases. Br. J.  Cancer, 82(7), 1353–​1357. PMCID: PMC2374493. Miller G, Grogan E, Rowe D, et al. 1987. Selective lack of antibody to a component of EB nuclear antigen in patients with chronic active Epstein-​Barr virus infection. J. Infect. Dis., 156(1), 26–​35. PMID: 3036961. Mirabelli MC, Zock JP, D’Errico A, et  al. 2009. Occupational exposure to high molecular weight allergens and lymphoma risk among Italian adults. Cancer Epidemiol. Biomarkers Prev., 18(10), 2650–​ 2654. PMCID: PMC2762781. Monnereau A, Glaser SL, Schupp CW, et al. 2013. Exposure to UV radiation and risk of Hodgkin lymphoma: a pooled analysis. Blood, 122(20), 3492–​ 3499. PMCID: PMC3829118. Monnereau A, Orsi L, Troussard X, et al. 2007. History of infections and vaccinations and risk of lymphoid neoplasms: does influenza immunization reduce the risk? Leukemia, 21(9), 2075–​2079. PMID: 17508003. Monnereau A, Orsi L, Troussard X, et  al. 2008. Cigarette smoking, alcohol drinking, and risk of lymphoid neoplasms: results of a French case-​control study. Cancer Causes Control, 19(10), 1147–​1160. PMID: 18781390. Montella M, Maso LD, Crispo A, et  al. 2006. Do childhood diseases affect NHL and HL risk? A case-​control study from northern and southern Italy. Leuk. Res., 30(8), 917–​922. PMID: 16406019. Montgomery S, Hajiebrahimi M, Burkill S, et al. 2016. Multiple sclerosis and risk of young-​adult-​onset Hodgkin lymphoma. Neurol. Neuroimmunol. Neuroinflamm., 3(3), e227. PMCID: PMC4841639. Morton LM, Turner JJ, Cerhan JR, et al. 2007. Proposed classification of lymphoid neoplasms for epidemiologic research from the Pathology Working Group of the International Lymphoma Epidemiology Consortium (InterLymph). Blood, 110(2), 695–​708. PMCID: PMC1924473. Moutsianas L, Enciso-​ Mora V, Ma YP, et  al. 2011. Multiple Hodgkin lymphoma-​associated loci within the HLA region at chromosome 6p21.3. Blood, 118(3), 670–​674. PMID: 21596858. Mraz M, and Haluzik M. 2014. The role of adipose tissue immune cells in obesity and low-​grade inflammation. J. Endocrinol., 222(3), R113–​R127. PMID: 25006217 Mucci LA, Hjelmborg JB, Harris JR, et al. 2016. Familial risk and heritability of cancer among twins in Nordic countries. JAMA, 315(1), 68–​76. PMID: 26746459. Mueller N, Evans A, Harris NL, et al. 1989. Hodgkin’s disease and Epstein-​ Barr virus:  altered antibody pattern before diagnosis. N. Engl. J.  Med., 320(11), 689–​695. PMID: 2537928. Mueller NE, and Grufferman S. 2006. Hodgkin lymphoma. In: Schottenfeld D, and Fraumeni Jr JF (Eds.), Cancer epidemiology and prevention (3rd ed., pp. 872–​897). Oxford: Oxford University Press. Mueller NE, Lennette ET, Dupnik K, and Birmann BM. 2012. Antibody titers against EBNA1 and EBNA2 in relation to Hodgkin lymphoma and history of infectious mononucleosis. Int. J.  Cancer, 130(12), 2886–​2891. PMCID: PMC3899938. Munoz N, Davidson RJ, Witthoff B, Ericsson JE, and de The G. 1978. Infectious mononucleosis and Hodgkin’s disease. Int. J. Cancer, 22(1), 10–​13. PMID: 681023 Murphy F, Kroll ME, Pirie K, et al. 2013. Body size in relation to incidence of subtypes of haematological malignancy in the prospective Million Women Study. Br. J. Cancer, 108(11), 2390–​2398. PMCID: PMC3681016. Muschen M, Rajewsky K, Brauninger A, et al. 2000. Rare occurrence of classical Hodgkin’s disease as a T cell lymphoma. J. Exp. Med., 191(2), 387–​ 394. PMCID: PMC2195757. Navaranjan G, Hohenadel K, Blair A, et al. 2013. Exposures to multiple pesticides and the risk of Hodgkin lymphoma in Canadian men. Cancer Causes Control, 24(9), 1661–​1673. PMCID: PMC3734598. Newell GR. 1970. Etiology of multiple sclerosis and Hodgkin’s disease. Am. J. Epidemiol., 91, 119–​122. PMID: 5416245. Newell GR, Cole SR, Miettinen OS, and MacMahon B. 1970. Age differences in the histology of Hodgkin’s disease. J. Natl. Cancer Inst., 45, 311–​317. Newton R, Crouch S, Ansell P, et  al. 2007. Hodgkin’s lymphoma and infection: findings from a UK case-​control study. Br. J. Cancer, 97(9), 1310–​ 1314. PMCID: PMC2360469. Nielsen TR, Rostgaard K, Nielsen NM, et  al. 2007. Multiple sclerosis after infectious mononucleosis. Arch. Neurol., 64(1), 72–​75. PMID: 17210811. Niens M, Jarrett RF, Hepkema B, et al. 2007. HLA-​A*02 is associated with a reduced risk and HLA-​A*01 with an increased risk of developing EBV+ Hodgkin lymphoma. Blood, 110(9), 3310–​3315. PMID: 17630352. Niens M, van den Berg A, Diepstra A, et  al. 2006. The human leukocyte antigen class  I  region is associated with EBV-​positive Hodgkin’s lymphoma:  HLA-​ A and HLA complex group 9 are putative candidate

genes. Cancer Epidemiol. Biomarkers Prev., 15(11), 2280–​ 2284. PMID: 17119058. Nieters A, Deeg E, and Becker N. 2006. Tobacco and alcohol consumption and risk of lymphoma: results of a population-​based case-​control study in Germany. Int. J. Cancer, 118(2), 422–​430. PMID: 16080191. Paffenbarger RS, Jr., Wing AL, and Hyde RT. 1977. Characteristics in youth indicative of adult-​onset Hodgkin’s disease. J. Natl. Cancer Inst., 58, 1489–​1491. PMID: 857036. Pallesen G, Hamilton-​Dutoit SJ, Rowe M, and Young LS. 1991. Expression of Epstein-​Barr virus latent gene products in tumour cells of Hodgkin’s disease [see comments]. Lancet, 337(8737), 320–​322. PMID: 1671232. Paltiel O, Schmit T, Adler B, et al. 2000. The incidence of lymphoma in first-​ degree relatives of patients with Hodgkin disease and non-​Hodgkin lymphoma: results and limitations of a registry-​linked study. Cancer, 88(10), 2357–​2366. PMID: 10820359. Pang D, Alston RD, Eden TO, and Birch JM. 2008. Cancer risks among relatives of children with Hodgkin and non-​Hodgkin lymphoma. Int. J. Cancer, 123(6), 1407–​1410. PMID: 18561317. Papadopoulou C, Antonopoulos CN, Sergentanis TN, et al. 2012. Is birth weight associated with childhood lymphoma? A  meta-​analysis. Int. J.  Cancer, 130(1), 179–​189. PMID: 21351088. Poppema S, Delsol G, Pileri S, et al. 2008. Nodular lymphocyte predominant Hodgkin lymphoma. In: Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H, Thiele J, and Vardiman JW (Eds.), Tumours of haematopoietic and lymphoid tissues (3rd ed., pp. 323–​325). Lyon: IARC. Poppema S, van Imhoff, Torensma R, and Smit J. 1985. Lymphadenopathy morphologically consistent with Hodgkin’s disease associated with Epstein-​Barr virus infection. Am. J. Clin. Pathol., 84(3), 385–​390. PMID: 2994459. Raphael M, Said J, Borisch B, Cesarman E, and Harris NL. 2008. Lymphomas associated with HIV infection. In:  Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H, Thiele J, and Wardiman.J.W (Eds.), Tumours of haematopoietic and lymphoid tissues (3rd ed., pp. 340–​342). Lyon: IARC. Rathore B, and Kadin ME. 2010. Hodgkin’s lymphoma therapy:  past, present, and future. Expert Opin. Pharmacother., 11(17), 2891–​ 2906. PMCID: PMC2978071. Razis DV, Diamond HD, and Craver LF. 1959. Familial Hodgkin’s disease: its significance and implications. Ann. Intern. Med., 51, 933–971. PMID: 14436684. Reekie J, Kosa C, Engsig F, et  al. 2010. Relationship between current level of immunodeficiency and non-​ acquired immunodeficiency syndrome-​defining malignancies. Cancer, 116(22), 5306–​ 5315. PMID: 20661911. Rosdahl N, Olesen Larsen S, and Clemmesen J. 1974. Hodgkin’s disease in patients with previous infectious mononucleosis:  30  years’ experience. BMJ, 2, 253–​256. PMCID: PMC1610532. Rudant J, Menegaux F, Leverger G, et  al. 2007. Family history of cancer in children with acute leukemia, Hodgkin’s lymphoma or non-​Hodgkin’s lymphoma: the ESCALE study (SFCE). Int. J. Cancer, 121(1), 119–​126. PMID: 17330239. Rudant J, Orsi L, Monnereau A, et  al. 2011. Childhood Hodgkin’s lymphoma, non-​Hodgkin’s lymphoma and factors related to the immune system:  the Escale Study (SFCE). Int. J.  Cancer, 129(9), 2236–​2247. PMID: 21170962. Sergentanis TN, Kanavidis P, Michelakos T, and Petridou ET. 2013. Cigarette smoking and risk of lymphoma in adults: a comprehensive meta-​analysis on Hodgkin and non-​Hodgkin disease. Eur. J. Cancer Prev., 22(2), 131–​ 150. PMID: 22759975. Serraino D, Franceschi S, Talamini R, et al. 1991. Socio-​economic indicators, infectious disease and Hodgkin’s disease. Int. J.  Cancer, 47, 352–​357. PMID: 1993542. Shugart YY, Hemminki K, Vaittinen P, Kingman A, and Dong C. 2000. A genetic study of Hodgkin’s lymphoma:  an estimate of heritability and anticipation based on the familial cancer database in Sweden. Hum. Genet., 106(5), 553–​556. PMID: 10914686. Silverberg MJ, Chao C, Leyden WA, et  al. 2011. HIV infection, immunodeficiency, viral replication, and the risk of cancer. Cancer Epidemiol. Biomarkers Prev., 20(12), 2551–​2559. PMCID: PMC3237725. Sleckman BG, Mauch PM, Ambinder RF, et  al. 1998. Epstein-​Barr virus in Hodgkin’s disease: correlation of risk factors and disease characteristics with molecular evidence of viral infection. Cancer Epidemiol. Biomarkers Prev., 7, 1117–​1121. PMID: 9865430. Soderberg KC, Jonsson F, Winqvist O, Hagmar L, and Feychting M. 2006. Autoimmune diseases, asthma and risk of haematological malignancies: a nationwide case-​control study in Sweden. Eur. J. Cancer, 42(17), 3028–​ 3033. PMID: 16945522.

 765



Hodgkin Lymphoma

Stein H. 2008. Hodgkin lymphoma:  introduction. In:  Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H, Thiele J, and Vardiman JW (Eds.), Tumours of haematopoietic and lymphoid tissues (3rd ed., pp. 322–​322). Lyon: IARC. Stein H, Delsol G, Pileri S, et al. 2008. Classical Hodgkin lymphoma: introduction. In: Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H, Thiele J, and Vardiman JW (Eds.), Tumours of haematopoietic and lymphoid tissues (3rd ed., pp. 326–​329). Lyon: IARC. Stein RS, and Morgan DS. 2004. Hodgkin disease. In: Greer JP, Foester J, Lukens JN, Rodgers GM, Paraskevas F, and Glader B (Eds.), Wintrobe’s clinical hematology (Ed 11, Vol. 2, pp. 2521–​2554). Philadelphia: Lippincott Williams & Wilkins. Sud A, Hemminki K, and Houlston RS. 2015. Candidate gene association studies and risk of Hodgkin lymphoma:  a systematic review and meta-​ analysis. Hematol. Oncol. PMID: 26053036. Swerdlow SH, Webber SA, Chadbrun A, and Ferry JA. 2008. Post-​transplant lymphoproliferative disorders. In:  Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H, Thiele J, and Wardiman JW (Eds.), Tumours of haematopoietic and lymphoid tissues (3rd ed., pp. 343–​349). Lyon: IARC. Uccini S, Monardo F, Ruco LP et  al. 1989. High frequency of Epstein-​Barr virus genome in HIV-​positive patients with Hodgkin’s disease. Lancet, 1(8652), 1458. 4009736 Urayama KY, Jarrett RF, Hjalgrim H, et  al. 2012. Genome-​ wide association study of classical Hodgkin lymphoma and Epstein-​Barr virus status-​defined subgroups. J. Natl. Cancer Inst., 104(3), 240–​ 253. PMCID: PMC3274508. Van Krieken JH, Onciu M, Elenitoba-​Johnson KSJ, and Jaffe ES. 2008. Lymphoproliferative diseases associated with primary immune disorders. In:  Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H, Thiele J, and Wardiman JW (Eds.), Tumours of haematopoietic and lymphoid tissues (Vol. 4, pp. 336–​339). Lyon: IARC. van Leeuwen MT, Turner JJ, Falster MO et  al. 2013. Latitude gradients for lymphoid neoplasm subtypes in Australia support an association

765

with ultraviolet radiation exposure. Int. J.  Cancer, 133(4), 944–​951. PMID: 23382012. Vockerodt M, Yap LF, Shannon-​ Lowe C, et  al. 2015. The Epstein-​ Barr virus and the pathogenesis of lymphoma. J. Pathol., 235(2), 312–​322. PMID: 25294567. von Behren J, Spector LG, Mueller BA, et  al. 2011. Birth order and risk of childhood cancer: a pooled analysis from five US States. Int. J. Cancer, 128(11), 2709–​2716. PMCID: PMC3008504. Walter RB, Brasky TM, Buckley SA, Potter JD, and White E. 2013. Height as an explanatory factor for sex differences in human cancer. J. Natl. Cancer Inst., 105(12), 860–​868. PMCID: PMC3687370. Wang SS, Slager SL, Brennan P, et al. 2007. Family history of hematopoietic malignancies and risk of non-​Hodgkin lymphoma (NHL): a pooled analysis of 10 211 cases and 11 905 controls from the International Lymphoma Epidemiology Consortium (InterLymph). Blood, 109(8), 3479–​ 3488. PMCID: PMC1852242. Weiss LM, Strickler JG, Warnke RA, Purtilo DT, and Sklar J. 1987. Epstein-​ Barr viral DNA in tissues of Hodgkin’s disease. Am. J. Pathol., 129(1), 86–​91. PMCID: PMC1899692. Westergaard T, Melbye M, Pedersen JB, et  al. 1997. Birth order, sibship size and risk of Hodgkin’s disease in children and young adults: a population-​ based study of 31  million person-​years. Int. J.  Cancer, 72(6), 977–​981. PMID: 9378561. Willett EV, O’Connor S, Smith AG, and Roman E. 2007. Does smoking or alcohol modify the risk of Epstein-​Barr virus-​positive or -​negative Hodgkin lymphoma? Epidemiology, 18(1), 130–​136. PMID: 17099321. Willett EV, and Roman E. 2006. Obesity and the risk of Hodgkin lymphoma (United Kingdom). Cancer Causes Control, 17(8), 1103–​1106. PMID: 16933061. Yang J, Lin L, Levin L, et al. 2002. Cell-​free EBV is detected in patients with EBV(+) Hodgkin’s lymphoma (HL) years in advance of diagnosis. Oral presentation at the 10th International Symposium on Epstein-​Barr Virus & Associated Malignant Diseases. July 16–​21, 2002, Cairns, Australia.

76

 76

40

The Non-​Hodgkin Lymphomas JAMES R. CERHAN, CLAIRE M. VAJDIC, AND JOHN J. SPINELLI

OVERVIEW The non-​Hodgkin lymphomas (NHL) are a heterogeneous group of over 40 lymphoid neoplasms that have undergone a major redefinition over the last 25 years, in part due to major advances in immunology and genetics, as well as implementation of the World Health Organization (WHO) classification system. NHLs are considered clonal tumors of B-​ cells, T-​ cells, or natural killer (NK) cells arrested at various stages of differentiation, regardless of whether they present in the blood (lymphoid leukemia) or lymphoid tissues (lymphoma). In the United States, the age-​standardized NHL incidence rate (per 100,000) doubled from 1973 (10.2) to 2004 (21.4) and then stabilized, while 5-​year relative survival rates improved from 42% in 1973 to 70% in 2004. These changes have greatly impacted age-​adjusted NHL mortality rates (per 100,000), which were 5.7 in 1973, peaked at 8.9 in 1997, and have been declining since that time, most recently at 5.9 in 2012. The end of this NHL “epidemic” has also been observed in other Western countries. Globally, there is a large variation in NHL incidence rates and distribution of NHL subtypes. Established risk factors for NHL or specific NHL subtypes include infectious agents (HTLV-​1, HIV, EBV, HHV8, HCV, and H. pylori), lindane, immune dysregulation (primary immunodeficiency, transplantation, autoimmunity, and immunosuppressive drugs), family history of lymphoma, and common genetic variants identified by genome-​wide association studies (GWAS). There is growing evidence for a role of certain medical history (atopy, breast implants, prior chemotherapy) and lifestyle (hair dye, alcohol, sun exposure) factors in NHL etiology, while a role for other occupational, environmental (pesticides, solvents, PCBs) or lifestyle (diet, obesity, physical activity, smoking) factors have produced mixed results. There has been substantial progress on evaluating etiologic heterogeneity by NHL subtype, with GWAS hits largely (but not exclusively) unique to specific subtypes, while the InterLymph NHL Subtypes Project has identified medical history, lifestyle, family history, and occupational risk factors that are common among subtypes, as well as distinct to a single or a few subtypes. Cumulatively, the established risk factors do not explain the NHL epidemic, and beyond avoiding oncogenic infectious agents, lindane and immunosuppression, there are currently limited prevention strategies.

TUMOR MODEL AND CLASSIFICATION Molecular Pathogenesis Lymphomas are a heterogeneous group of over 40 neoplasms (Table 40–​1) that arise from lymphocytes and produce tumors in the lymph nodes, lymphatic organs (e.g., tonsils, spleen), and extranodal lymphatic tissue. The lymphatic system is part of the immune system, and advances in molecular and tumor biology and immunology in particular have greatly improved our understanding of lymphomagenesis. Current knowledge supports the concept that the many lymphoma subtypes are clonal tumors of immature or mature B-​cells, T-​cells, or natural killer (NK)-​cells, arrested at various stages of differentiation (Jaffe et al., 2008). B-​cell neoplasms appear to recapitulate the normal stages of B-​cell differentiation, and most (but not all) B-​cell

lymphomas can be linked to a normal cell counterpart (Figure 40–​1). T-​cells and NK-​cells share many immunophenotypic and functional properties and are generally considered together; their differentiation and relationship to normal counterparts is not as well characterized (Figure 40–​2). Current understanding supports the hypothesis that lymphomas are aberrant outcomes of normal physiologic mechanisms used by the adaptive immune system. In the bone marrow and thymus, V(D) J gene recombination, which involves DNA double-​strand breaks, is used by both immature B-​and T-​cells for the assembly of immunoglobulin heavy-​chain and light-​chain genes. Once B-​cells have matured from the bone marrow, they migrate to peripheral lymphoid organs, where they can go through the germinal center reaction, which includes clonal expansion, somatic hypermutation of IgG genes, class-​switch recombination, and selection/​apoptosis. While these physiologic mechanisms allow for the creation of immense antibody diversity and specificity, they come with a trade-​off of susceptibility to neoplasia, mainly by generating reciprocal chromosomal translocations (involving the Ig and other loci) and off-​target somatic hypermutation of proto-​oncogenes, leading to aberrant expression of oncogenes and/​or inactivation of tumor suppressor genes (Klein and Dalla-​Favera, 2008; Shaffer et al., 2002). Of note, while there are approximately equal numbers of B-​and T-​cells in the human body, in Western countries approximately 90% of the lymphoid malignancies are of B-​cell origin, highlighting the critical role of the germinal center reaction in lymphomagenesis (Kuppers, 2005). Chromosomal translocations are a hallmark of many lymphoma subtypes, including t(14;18) in most follicular lymphomas (FLs) and some diffuse large B-​cell lymphomas (DLBCLs) (leading to BCL-​ 2 overexpression); t(8;14) in Burkitt lymphoma (BL) (leading to c-​ MYC overexpression); t(11;18) in mantle cell lymphoma (leading to cyclin D1 overexpression); t(11;18) in mucosa-​associated lymphoid tissue (MALT) lymphoma (leading to the API2/​MALT1 fusion protein); and t(2;5) in anaplastic large cell lymphoma (ALCL) (resulting in ALK fusion proteins) (Kuppers, 2005). Copy number alterations (amplifications and deletions) and somatic mutations are additional drivers of lymphomagenesis. Key somatic alterations identified by next-​ generation sequencing include mutations impacting lymphoid signaling (CD79B), NF-​κB signaling (CARD11, MYD88) and histone modification (CREBBP, EZH2, MEF2B, MLL2/​MLL3) in DLBCL; histone modification (CREBBP, MEF2B, EP300, MLL2) in FL; NOTCH1, mRNA splicing (SF3B1), DNA damage/​repair (ATM, POT1), apoptosis (BIRC3) and innate immunity (MYD88, TLR2) in chronic lymphocytic leukemia (CLL); NOTCH2 in splenic marginal zone lymphoma (MZL); NOTCH1 in mantle-​cell lymphoma (MCL); TCF3/​ID3 and CCND in BL; and MYD88 in Waldenström macroglobulinemia (WM) (Kahl and Yang, 2016; Mullighan, 2013; Watson et al., 2013). Of particular note, in comparison to most solid tumors, which typically have substantial genomic instability, lymphomas generally have a more stable genome (Shankland et al., 2012). Finally, gene expression profiling has identified novel oncogenic pathways within apparently homogenous subtypes (Nogai et  al., 2011), most prominently the distinction between germinal center-​like DLBCL (which retains a germinal center expression program) and activated B-​cell-​like DLBCL (which uses the plasma-​cell expression program) (Lenz and Staudt, 2010).

767

768

Table 40–​1.  NHL WHO Subtypes from SEER18, United States, 2003–​2012: Overall, Race-​Specific, and Sex-​Specific Age-​Adjusted Incidence Rates per 100,00; Annual Percent Change in Incidence Rate (APC); Incidence Rate Ratios (IRR); and Overall and Race Specific 5-​Year Relative Survival Rates† Overall

Race-​Specific White

Black

Sex-​Specific Asian or Pacific Islander

WHO Subtype (Abbreviation)

Rate

APC

Rate

APC

Rate

APC

Non-​Hodgkin lymphoma, plasma cell neoplasms included Non-​Hodgkin lymphoma, plasma cell neoplasms exluded Non-​Hodgkin lymphoma, B-​cell (excludes plasma cell neoplasms) Precursor acute lymphoblastic leukemia/​lymphoma, B-​cell (B-​ALL/​LBL) Mature B-​cell neoplasms (excludes plasma cell neoplasms) Chronic/​Small lymphocytic leukemia/​lymphoma (CLL/​SLL) Prolymphocytic leukemia, B-​ cell (P-​BLL) Mantle-​cell lymphoma (MCL) Lymphoplasmacytic lymphoma/​Waldenstrom macroglobulinemia (LPL/​WM) Diffuse large B-​cell lymphoma, NOS (DLBCL-​NOS) Intravascular large B-​cell lymphoma (IVBCL) Primary effusion lymphoma (PEL) Mediastinal large B-​cell lymphoma (MLBCL) Burkitt lymphoma/​leukemia (BL) Marginal-​zone lymphoma (MZL) Splenic MZL (SMZL) Extranodal MZL, MALT type (EMZL) Nodal MZL (NMZL) Follicular lymphoma (FL) Hairy-​cell leukemia (HCL) Non-​Hodgkin lymphoma, B-​cell, NOS Non-​Hodgkin lymphoma, T-​cell

32.01

–​0.01

33.04

–​0.07

31.15

–​0.78

0.9

25.83

–​0.26

27.38

–​0.3

18.52

–​0.32

23.25

0.05

24.86

0.04

15.47

–​0.15

1.33

3.45*

1.48

3.67*

0.69

20.50

–​0.53*

21.91

–​0.55*

13.68

5.79

–​1.70*

6.22

–​1.86*

0.02

~

0.03

0.77 0.63

1.16 –​1.06*

6.91

Male

APC

IRR‡

Rate

APC

Rate

APC

IRR¶

White

Black

All Other

20.02

–​0.04

0.6

25.79

–​0.13

39.86

–​0.32

1.5

66.3%

67.5%

56.1%

61.6%

0.7

15.74

–​0.12

0.6

20.82

–​0.37

32.10

–​0.24

1.5

71.5%

72.3%

63.7%

66.1%

0.6

13.78

0.22

0.6

18.82

–​0.07

28.85

0.07

1.5

72.8%

73.4%

66.0%

67.7%

0.5

1.09

0.37

0.7

1.21

4.07*

1.45

3.02*

1.2

67.2%

67.0%

65.7%

70.1%

–​0.74

0.6

11.73

–​0.2

0.5

16.41

–​0.72*

25.67

–​0.47*

1.6

73.6%

74.3%

66.5%

67.8%

4.52

–​1.88*

0.7

1.45

–​1.03

0.2

4.15

–​1.88*

7.91

–​1.40*

1.9

79.5%

79.9%

70.9%

74.9%

~

˄

~

~

˄

~

~

0.02

~

0.04

~

2.0

38.4%

35.0%

˄

˄

0.86 0.68

1.18 –​1.32*

0.34 0.37

~ ~

0.4 0.5

0.34 0.35

~ ~

0.4 0.5

0.42 0.48

1.56 –​0.57

1.23 0.85

0.82 –​1.71*

2.9 1.8

54.8% 77.9%

55.1% 78.5%

56.8% 63.4%

48.9% 68.6%

0.2

7.20

0.34

4.79

–​0.98

0.7

5.92

0.16

0.8

5.68

–​0.32

8.42

0.60*

1.5

59.8%

60.4%

54.8%

57.2%

0.01

~

0.01

~

˄

~

~

˄

~

~

0.01

~

0.01

~

1.0

˄

˄

˄

˄

0.01

~

0.01

~

˄

~

~

0.00

~

~

˄

~

0.02

~

~

20.8%

18.8%

˄

˄

0.05

~

0.05

~

0.05

~

1.0

0.05

~

1.0

0.06

~

0.04

~

0.7

80.4%

85.7%

˄

˄

0.40

0.41

0.42

0.16

0.35

1.1

0.8

0.35

~

0.8

0.21

1.62

0.61

–​0.12

2.9

56.2%

57.6%

46.1%

53.6%

1.95

1.26*

2.02

1.28*

1.45

2.01

0.7

1.46

–​0.93

0.7

1.89

0.89

2.05

1.70*

1.1

89.3%

89.2%

89.1%

88.8%

0.16 1.16

–​0.58 1.44*

0.18 1.17

–​0.46 1.45*

0.08 0.93

~ 1.96

0.4 0.8

0.06 1.05

~ –​1.44

0.3 0.9

0.16 1.16

–​0.44 1.05

0.17 1.19

–​0.35 1.82*

1.1 1.0

84.0% 93.8%

83.7% 94.3%

˄ 91.6%

˄ 91.0%

0.62 3.65 0.30 1.42

1.39 –​1.53* –​0.87 5.49*

0.66 4.08 0.33 1.47

1.42 –​1.45* –​1.1 5.47*

0.43 1.66 0.11 1.10

~ –​1.22 ~ 4.50*

0.7 0.4 0.3 0.7

0.35 1.69 0.11 0.95

~ –​1.75 ~ 5.31*

0.5 0.4 0.3 0.6

0.58 3.39 0.11 1.19

0.87 –​1.46* –​2.2 4.96*

0.69 3.98 0.51 1.73

2.01 –​1.58* –​0.72 5.78*

1.2 1.2 4.6 1.5

81.8% 85.2% 93.9% 66.3%

81.3% 85.4% 94.4% 66.8%

82.3% 81.8% 80.6% 59.9%

81.7% 82.7% 91.5% 62.0%

2.13

0.42

2.03

0.07

2.69

1.26

1.3

1.74

–​0.2

0.9

1.62

0.47

2.72

0.27

1.7

61.9%

63.5%

52.7%

54.5%

4.51*

IRR‡ Rate

Female

5-​Year Relative Survival

Overall

 769

Precursor acute lymphoblastic leukemia/​lymphoma, T-​cell (T-​ALL/​LBL) Mature non-​Hodgkin lymphoma, T-​cell Mycosis fungoides/​Sezary syndrome (MF/​SS) Peripheral T-​cell lymphoma, NOS (PTCL-​NOS) Angioimmunoblastic T-​cell lymphoma (AITL) Subcutaneous panniculitis-​like T-​cell lymphoma (SPTCL) Anaplastic large cell lymph (ALCL) Hepatosplenic T-​cell lymphoma (HSTL) Enteropathy-​associated T-​cell lymphoma (EATL) Cutaneous T-​cell lymphoma, NOS (CTCL) Primary cutaneous anaplastic large cell lymphoma (PCALCL) Adult T-​cell leukemia/​ lymphoma (ATLL) NK/​T-​cell lymphoma, nasal type/​aggressive NK leukemia (NKTCL) T-​cell large granular lymphocytic leukemia (T-​LGL) T-​cell prolymphocytic leukemia (T-​PLL) Non-​Hodgkin lymphoma, T-​cell, NOS Non-​Hodgkin lymphoma, unknown lineage

0.14

~

0.15

~

0.15

~

1.0

0.10

~

0.7

0.09

~

0.19

~

2.1

62.6%

64.4%

51.9%

58.1%

1.97

1.39*

1.88

1.11*

2.51

2.12

1.3

1.62

0.71

0.9

1.52

1.31*

2.51

1.28*

1.7

62.1%

63.7%

53.2%

54.3%

0.49

1.31

0.44

1.11

0.66

1.56

1.5

0.32

~

0.7

0.40

1.47

0.60

0.88

1.5

89.1%

89.4%

80.5%

94.4%

0.42

1.03

0.39

0.6

0.65

2.75

1.7

0.36

~

0.9

0.30

–​0.22

0.55

1.93*

1.8

35.7%

37.9%

29.8%

24.3%

0.15

3.60*

0.15

3.55*

0.09

~

0.6

0.20

~

1.3

0.13

0.17

2.9

1.3

43.6%

44.8%

27.8%

44.0%

3.92*

0.01

~

0.01

~

0.03

~

3.0

˄

~

~

0.02

~

0.01

~

0.5

51.1%

40.8%

˄

˄

0.23

–​3.32*

0.24

–​3.02*

0.27

~

1.1

0.15

~

0.6

0.17

–​2.20*

0.30

–​3.88*

1.8

55.7%

57.3%

45.4%

49.8%

0.01

~

0.01

~

˄

~

~

˄

~

~

˄

~

0.01

~

~

21.2%

˄

˄

˄

0.01

~

0.01

~

˄

~

~

˄

~

~

0.01

~

0.02

~

2.0

12.2%

17.3%

˄

˄

0.22

0.17

0.22

–​0.39

0.30

~

1.4

0.08

~

0.4

0.17

0.75

0.29

–​0.26

1.7

79.6%

82.0%

69.7%

66.2%

0.10

0.03

0.10

–​0.33

0.10

~

1.0

0.05

~

0.5

0.08

–​0.07

0.14

–​0.24

1.8

90.3%

91.4%

76.0%

˄

0.19

10.26*

0.17

9.99*

0.28

~

1.6

0.20

~

1.2

0.12

0.25

10.66*

2.1

54.3%

58.4%

41.0%

47.4%

0.08

3.05*

0.07

3.64

0.03

~

0.4

0.17

~

2.4

0.06

~

0.11

1.6

1.8

38.7%

39.9%

˄

35.1%

9.35*

0.02

~

0.02

~

˄

~

~

˄

~

~

0.02

~

0.02

~

1.0

65.5%

63.5%

˄

˄

0.03

~

0.03

~

0.06

~

2.0

˄

~

~

0.03

~

0.04

~

1.3

14.6%

12.4%

˄

˄

0.01

~

0.01

~

˄

~

~

˄

~

~

0.01

~

0.02

~

2.0

38.5%

43.2%

˄

˄

0.45

–​14.97*

0.48

–​15.02*

0.36

~

0.8

0.22

~

0.5

0.38

~

0.53

~

1.4

57.8%

57.1%

55.6%

65.5%

† Rates are per 100,000 and age-​adjusted to the 2000 US Std Population (19 age groups) standard. Actuarial method is Ederer II method used for cumulative expected. ‡ IRR relative to white rate ¶ IRR relative to male rate ˄ Statistic not displayed due to fewer than 25 cases. ~ Statistic could not be calculated. * The APC is significantly different from zero (p < 0.05). Percent changes were calculated using 1 year for each end point and APCs were calculated using weighted least squares method.

70

770

PART IV:  Cancers by Tissue of Origin Central lymphoid

Peripheral Lymphoid Tissue

Precursor B-cells

Peripheral (mature) B-cells

Bone Marrow

B Lymphoblastic leukemia/ lymphoma

Hairy cell leukemia Prolymphocytic leukemia

DLBCL (ABC type) Primary Mediastinal B-cell lymphoma

Splenic marginal zone lymphoma

Memory B-cell

Marginal Zone Mantle Zone

B-CLL/SLL mutated V gene

Germinal Center

V-region gene recombination

Class switch recombination

Somatic hypermutation

MALT lymphoma

Clonal expansion T-cell

B-cell precursor Selection Naive B-cell

B-cell

Dark Zone

FDC

Plasma blast

Differentiation Light Zone

Multiple myeloma

Apoptosis No BCR

Lymphoplasmablastic lymphoma Primary effusion lymphoma

CD5+ B-cell Apoptosis

Plasma cell

B-CLL/SLL unmutated V gene Mantle cell lymphoma Follicular lymphoma Burkitt lymphoma DLBCL (GCB type) Lymphocyte – predominant Hodgkin lymphoma

Classical Hodgkin lymphoma Post-transplant lymphoma

Figure  40–​1.  Diagrammatic representation of B-​cell differentiation and the relationship to major B-​cell neoplasms. B-​cell neoplasms are assigned to their proposed normal B-​cell counterpart, which correspond to stages of B-​cell maturation. Precursor B-​cells that mature in the bone marrow can undergo apoptosis or develop into mature naïve B-​cells. Most lymphomas are derived from germinal center B-​cells (where somatic hypermutation and heavy chain class switching occur) or B-​cells that have passed through the germinal center, indicating the major role of the germinal center in B-​cell lymphomagenesis. Post germinal center cells include both long-​lived plasma cells and memory type B-​cells. B-​cell lymphomas that have not gone through the germinal center have unmutated V-​region genes. Solid arrows represent B-​cell differentiation steps and dashed arrows link the various lymphomas to their proposed normal counterparts. Abbreviations: BCR, B-​cell receptor; FDC, Follicular dendridic cells; GCB, germinal center B-​cell; ABC, activated B-​cell. Adapted from Kuppers 2005 and Jaffe et al. 2008.

The lymphoma microenvironment is diverse and includes non-​ malignant immune cells, stroma, blood vessels, and extracellular matrix. The percentage of malignant lymphoma cells within the microenvironment is highly variable, ranging from < 1% of cells in Hodgkin lymphoma (HL), a heterogeneous mix of tumor and background cells exemplified by FL, to nearly all tumor cells as in BL, suggesting different levels of dependence of lymphoma subtypes on the microenvironment (Scott and Gascoyne, 2014). Many lymphoma subtypes can only be cultured in vitro by including key components of the microenvironment (e.g., stromal cells, cytokine cocktails, etc.) (Scott and Gascoyne, 2014). Gene expression profiling of lymphoma tissue samples have shown that the signals from the non-​malignant cells are associated with prognosis (Dave et al., 2004; Glas et al., 2005; Steidl et al., 2010) and promising novel therapies are also targeting the microenvironment (Scott and Gascoyne, 2014). Integration of these observations leads to the hypothesis that lymphoma is a disease of functional B-​, T-​, and NK-​cells in which intrinsic (specific molecular alterations of the tumor) and extrinsic (microenvironment, including immunologic and other regulatory networks) factors interact to promote neoplastic growth (de Jong, 2005).

Classification The current WHO classification system (Swerdlow et  al., 2008b), which has been incorporated into the International Classification of Diseases—​Oncology (ICD-​O) (Fritz et  al., 2000), recognizes B-​cell neoplasms, T-​cell/​natural killer (NK)-​cell neoplasms, and HL; plasma cell tumors (e.g., multiple myeloma) are classified as B-​cell neoplasms (Jaffe et al., 2008). The B-​and T/​NK-​cell neoplasms are further stratified into immature (or precursor to a lineage of interest) and mature (peripheral) neoplasms. In this classification approach, lymphoid malignancies of the same lineage arising in the bone marrow or peripheral blood (leukemias) and lymphoid malignancies arising in the peripheral organs (lymphomas) are grouped together; for example, acute lymphoblastic leukemia (ALL) and lymphoblastic lymphoma (LBL) are grouped together, as are CLL and small lymphocytic leukemia (SLL). Subtypes within the major WHO categories are defined using a multiparameter approach that includes morphology, immunophenotype, genetics, and clinical features (Jaffe et al., 2008). A revision of the fourth edition of the WHO Classification was recently released, and although it did not allow for any new definitive entities,

 71



771

The Non-Hodgkin Lymphomas Central lymphoid tissue

Peripheral Lymphoid Tissue

Precursor T-cells

Peripheral (mature) T-cells NK-cell

Follicle

Spleen Mucosa Peripheral blood

Bone Marrow

γδ T-cell

FDC

Skin

Progenitor T-cell/ Prothymocyte

Effector T-cell Naïve T-cell

TFH

T-blast Memory T-cell

CD4 Subcapsular cortical thymocyte

Common thymocyte

CD4 Naïve T-cell

αβ CD4+ CD8+

Effector T-cell

AG

CD8

Thymus T lymphoblastic lymphoma/leukemia

CD8

T-blast Memory T-cell

Peripheral (mature) T-cell and NK-cell lymphomas/leukemias: Mycosis fungoides/Sezary Syndrome; Peripheral T-cell; Angioimmunoblastic; Subcutaneous panniculitis-like T-cell; Anaplastic large cell; Hepatosplenic T-cell; Enteropathy-associated T-cell; Primary cutaneous anaplastic large cell; Adult T-cell; NK/T-cell; nasal type/aggressive NK leukemia; T-cell large granular lymphocytic leukemia, T-cell prolymphocytic leukemia

Figure 40–​2.  Diagrammatic representation of T-​cell differentiation and the relationship to major T-​cell neoplasms. T-​cell neoplasms correspond to different stages of maturation, although this is not well understund. Lymphoid progenitors enter the thymus where precursor T-​cells developed in varied types of naïve T-​cells. The precise maturational process of natural killer cells and γδ T-​cells is not fully understood. The αβ T-​cells leave the thymus where upon exposure to antigen they can undergo blast transformation and develop further into CD4+ and CD8+ effector and memory T-​cells. T regulatory cells are the major type of CD4+ effector T-​cells. Another specific type of effector T-​cells is the follicular helper T-​cell that is found in germinal centers. Upon antigenic stimulation, T-​cell responses may occur independent of the germinal center or in the context of a germinal center reaction. Abbreviations: FDC, Follicular dendridic cells; TFH, Follicular Helper T-​cell; AG, antigen. Solid arrows represent T-​cell differentiation steps. The T-​cell lymphoma subtypes are listed along the bottom of the Figure. Adapted from Jaffe et al. 2008.

it incorporated new genetic/​molecular and clinical data into current disease entities and added a limited number of new provisional entities (Swerdlow et al., 2016). Current treatment and management approaches are predicated on the WHO subtypes (Shankland et al., 2012), which, depending on the subtype, have a wide variety of approaches (including observation, immunotherapy, antibiotics, radiation, and chemotherapy) and prognoses (including the ability to cure some of the subtypes, such as DLBCL). Rapid changes in our understanding and classification of lymphoid neoplasms have challenged both descriptive and analytic epidemiology, which have traditionally grouped hematologic cancers into leukemia, lymphoma, and multiple myeloma, with the lymphomas divided into HL and NHL, an approach that is increasingly unsustainable. In 2007, the InterLymph Consortium developed a nested classification system based on the WHO system and ICD-​O-​3 to facilitate analysis of lymphoid malignancies in epidemiologic studies (Morton et al., 2007); this was updated (Turner et  al., 2010) with the fourth edition of the WHO Classification (Swerdlow et  al., 2008b). This system was first incorporated into the SEER statistics in 2009 (Horner et al., 2009); the major WHO-​ defined subtypes are listed in Table 40–​1. For this edition of Cancer

Epidemiology and Prevention, we have maintained the traditional chapters for HL and multiple myeloma, but for leukemia, we have moved more recent descriptive and analytic epidemiology findings for lymphoid leukemias to this chapter. Many of the lymphoma subtypes were first formalized in the 1980s and 1990s, and some subtypes such as DLBCL are likely to be redefined based on new biologic knowledge. From an epidemiologic perspective, we review the literature for NHL both as a single entity and as having potentially etiologically distinct subtypes. A singular focus on NHL subtypes does not allow a full integration of historical trends or a discussion of much of the recent epidemiologic literature. Also, our ability to reliably distinguish subtypes is still evolving. Furthermore, NHL arises from a common precursor cell, and some NHLs can transform from one type to another (e.g., FL to DLBCL; CLL to DLBCL). There are also likely to be common etiologic agents across subtypes, either impacting the precursor cell or impacting lymphoid cells by similar mechanisms, just at different points in their maturation (Morton et al., 2007; Weisenburger, 1992). In other circumstances, factors unique to a specific subtype may be more relevant. These evolving concepts have been incorporated into this edition of Cancer Epidemiology and Prevention.

72

772

PART IV:  Cancers by Tissue of Origin

Precursors Monoclonal gammopathy of undetermined significance (MGUS), which is the precursor state to multiple myeloma, was first described in 1978 (see Chapter  41). In contrast, a precursor state to CLL was only first identified in the early 1990s, and was ultimately given the name “monoclonal B lymphocytosis” (MBL) (Marti et  al., 2005). MBL is defined as the presence of a clonal B-​cell population in the peripheral blood with fewer than 5 × 109/​L B-​cells and no other signs of a lymphoproliferative disorder. The clonal B-​cell population usually has a CLL phenotype (co-​expression of CD5, CD19, and CD23, and weak expression of CD20, CD79b, and surface immunoglobulin), although MBL with atypical CLL (CD5+, CD23–​, and CD20bright) or non-​CLL (CD5–​) phenotypes also occur (Strati and Shanafelt, 2015). In 2008, the International Workshop on CLL (IWCLL) incorporated MBL into the diagnostic criteria for CLL, which included changing the CLL definition from an absolute lymphocyte count to an absolute B-​lymphocyte count of 5 × 109/​L (Hallek et al., 2008). This change in the diagnostic criteria has led to a reduced incidence of CLL (and an increase in MBL), has shifted the distribution to a higher Rai stage at CLL diagnosis, and has shortened the median time to treatment (Call et al., 2014). CLL-​like MBL is also classified into low-​count (< 0.5 × 109/​L) versus high-​count (0.5 to 5 × 109/​L) based on the size of the B-​cell clone (Strati and Shanafelt, 2015). Low-​count MBL has a prevalence of ~5% in adults aged 40 and older but rarely progresses to CLL (Fazi et al., 2011), while high-​count MBL (often incidentally identified when evaluating a lymphocytosis) progresses to CLL, requiring therapy at a rate of 1%–​2% per year (Rawstron et  al., 2008; Rossi et al., 2009b; Shanafelt et al., 2009). While most MBL does not progress to CLL, a nested case-​control study of CLL found MBL in pre-​ diagnostic samples of virtually all of the CLL cases (Landgren et al., 2009), and family members of CLL cases have a higher prevalence of MBL (Marti et al., 2003; Rawstron et al., 2002), both of which further support MBL as a CLL precursor. Other precursors have also been identified. MBL with atypical-​CLL phenotypes and MBL with non-​CLL phenotypes have been suggested to be precursors to indolent MCL (Karube et al., 2014), MZL (Xochelli et  al., 2014), and DLBCL (Malecka et  al., 2015); these phenotypes also appear to be characterized by low potential to progress. Several potential precursors for FL have also been described (Mamessier et al., 2014), and the hallmark t(14;18) translocation in FL can be found in clonal circulating cells in 50%–​70% of otherwise healthy individuals (Limpens et  al., 1995), much higher than the incidence of FL, supporting this event as necessary but not sufficient for lymphomagenesis (Roulland et al., 2011). Using a nested case-​control design with banked blood specimens, clonal analysis of t(14;18) junctions in paired prediagnostic blood versus tumor samples demonstrated that progression to FL occurred from t(14;18)-​positive committed precursors, and that higher levels of circulating t(14;18) cells were associated with > 20-​fold higher risk of FL, even 15 years before diagnosis (Roulland et al., 2014).

DESCRIPTIVE EPIDEMIOLOGY International Comparisons Globally, NHL exhibits a large variation in incidence rates. Using estimated rates for 2012 from Globocan (Ferlay et al., 2013) and the classic definition of NHL to facilitate cross-​country comparisons, Figure 40–​3 highlights the range in incidence rates, with higher rates in more developed regions and lower rates in developing regions. The highest age-​standardized rates per 100,000 were observed in North America for both males (14.6) and females (10.2), while the lowest incidence rates were observed in South-​Central Asia for both males (3.3) and females (1.8), a 4.4-​fold difference for males and a 5.7-​fold difference for females. However, international variability in NHL mortality rates is smaller: 2.6-​fold for males and 3.1-​fold for females. A major explanation for the latter observation is that survival rates are much better in developed countries, based on a comparison of the ratio of incidence

to mortality rates, which is likely due to better treatment and management of aggressive lymphomas in these regions. While incidence rates are much higher in North America, Australia/​ New Zealand, and Europe, of the actual global estimated number of incident cases in 2012 (385,741), the number of cases from these regions (169,425) is similar to Asia (150,026) and proportionally, almost 50% of the estimated deaths due to NHL globally occur in Asia. Within the United States, there is also clear geographic variability, with the highest NHL mortality rates for 1995–​2004 occurring in the upper Midwest and Northeastern United States, which has suggested several etiologic hypotheses discussed later in the chapter.

Migrant Data Based on 2002 Globocan data, the age-​adjusted NHL rate per 100,000 in South Asian males residing in India (3.2) was similar to those residing in Singapore (3.1), but lower than those residing in the United Kingdom (11.3); females showed similar patterns (Rastogi et  al., 2008). During 1980–​1997, the age-​standardized NHL incidence rate per 100,000 for Hong Kong migrants to British Columbia, Canada, (7.11) was comparable to those of Hong Kong (7.91), but was much lower than the British Columbia background population rate (11.8) (Au et al., 2005). The distribution of NHL subtypes in migrants paralleled that of Hong Kong residents, except for a higher percentage of FL and low-​grade B-​cell NHL and a lower percentage of gastric and nasal NK lymphomas. Similarly, CLL incidence rates for Chinese in Hong Kong (0.28) and Chinese migrants from Hong Kong to British Columbia (0.40) were similar, and were much lower than the background population rate (1.71) (Mak et al., 2014).

Time Trends Using the classic definition of NHL to evaluate long-​term temporal trends, Figure 40–​4 summarizes the dramatic changes over time in NHL incidence, survival, and mortality. Based on US SEER data for both sexes combined and all racial/​ethnic groups, NHL incidence (per 100,000) increased steadily from 10.2 in 1973, to a peak at 21.4 in 2004 and then remained stable, most recently at 20.0 in 2012. From 1973 to 1995, 5-​year relative survival rates gradually increased from 42% to 52%, then increased rapidly to 70% by 2003, and have since remained stable (most recently at 71% in 2012). These changes have greatly impacted age-​adjusted NHL mortality rates (per 100,000), which were 5.7 in 1973, peaked at 8.9 in 1997, and have been declining since that time, most recently at 5.9 in 2012. Prior to 1973, NHL incidence and mortality rates were increasing over most of the twentieth century, as reviewed in the previous edition of this text (Hartge et al., 2006). While the current data appear to support an end to the “epidemic” of NHL that occurred over the last half of the twentieth century, incidence rates appear to have stabilized at a new higher level. While specific rates differ, the temporal trend patterns shown in Figure 40–​4 were broadly observed for whites, blacks, and other racial/​ethnic groups; males and females; and those under age 65 and 65 years and older (data not shown). These temporal trend incidence patterns were independent of HIV infection, and overall, HIV has been estimated to account for only 5.9% of NHL cases diagnosed during 1992–​2009 in the United States (Shiels et  al., 2013). Finally, these broad temporal trend patterns also occurred in Scandinavia (Sandin et  al., 2006), the United Kingdom (Smith et  al., 2015)  and other European countries (Bosetti et al., 2008), and Australia (van Leeuwen et al., 2014). Most of the increase over the twentieth century and the more recent plateau in incidence does not appear to be explained by changes in diagnosis and classification, data quality in cancer registries, or the occurrence of NHL as a second malignancy (Hartge et al., 2006; Sandin et al., 2006). Improving NHL survival rates (Sant et al., 2014)  and declining mortality rates (Bosetti et  al., 2008)  have also been observed in Europe. The observed decline in NHL mortality rates in Western countries since 1998 represents clear progress on this cancer as defined by the Extramural Committee to Assess Measures of Progress Against Cancer (1990).

 73



773

The Non-Hodgkin Lymphomas

North America

Male 14.6

Australia/New Zealand

Female

3.8 11.5

Northern Europe

2.2

3.2

1.9

10.3

More developed regions

3.5

2

3.2

1.9

10.3

Southern Europe

7.8

Western Asia

5.4 6.9 5.9 6.7

Melanesia Polynesia

2.6

6

South America

3.2

6

World

3.2

5.7

South-Eastern Asia

4

5.2

Southern Africa

4

5.6 4.6 5.9

Eastern Africa Micronesia

3.2

4.9

Caribbean

2.8

4.8

Central and Eastern Europe

2.6 4.4 3.8 4.2 2.4 4.3 2.8 4 2.3 3.7 3 3.3 2.3

Middle Africa Central America Less developed regions Eastern Asia Western Africa South-Central Asia 15

10

7.7 7.1 6.6

5

5.5

5.4 3.7 4.8 3.7 4 2.3 4.3 2.1 4.1 2 3.8 2.5 4.1 2.9 3.5 2.9 2.8 0.7 3.6 1.9 3.6 1.5 3.2 2.7 3.3 1.9 2.8 1.8 2.8 1.4 2.4 2 1.8 1.2

7.6

North Africa

8.1

3.3

4.9

Incidence Mortality

10.1

2.3

3.5

10.9

Western Europe

20

2.6

4.5

14.3

10.2

0

5

10

15

20

Figure 40–​3.  Age-​standardized rates (World Population) per 100,000 for NHL incidence and mortality, all ages, by sex, 2012, Globocan.

There are few high-​quality data on long-​term incidence trends in other parts of the world. In Asia, NHL incidence rates have not plateaued, and have shown increases in Japan during 1993–​2008 (Chihara et  al., 2014)  and in Singapore during 1998–​2012 (Lim et al., 2015).

Descriptive Epidemiology Using the WHO Classification As discussed earlier and elsewhere in this volume (Chapters 38, 39, 41), the classification of hematologic malignancies has rapidly evolved over the last 30 years. In this section, we focus on the most recent US data from 2003–​2012, using all 18 SEER registries and the WHO classification system as implemented using the InterLymph classification (Morton et  al., 2007; Turner et  al., 2010). These results include the lymphoid leukemias, but (except where noted) we have not included HL or MM in calculating rates. Racial/​ethnic groupings (white, black, Asian/​Pacific Islander) are based on SEER definitions. The widespread adoption of the WHO classification system in the 1990s now allows us to observe emerging subtype patterns and early time trends. As shown in Table 40–​1, mature B-​cell neoplasms have the highest age-​adjusted incidence rate per 100,000 at 20.50, followed by mature T-​cell neoplasms at 1.97, precursor B-​cell at 1.33, and finally precursor T-​cell at 0.14.

Of the mature B-​cell neoplasms, the highest incidence rates are observed for DLBCL, CLL/​SLL, and FL. Virtually all of the mature B-​cell neoplasms are most common in whites, with the most pronounced differences for whites versus other racial/​ ethnic groups, based on incidence rate ratios (IRR), observed for hairy-​cell leukemia (HCL) (white:black IRR = 0.3; white:Asian/​Pacific Islanders IRR = 0.3), CLL/​SLL (white:Asian/​Pacific Islanders IRR = 0.2), and splenic marginal-​ zone lymphoma (SMZL) (white: Asian/​ Pacific Islanders IRR = 0.3). Mature B-​cell neoplasms are also more common in males than females (IRR = 1.6), with the exceptions of mediastinal large B-​ cell lymphoma (MLBCL) (IRR = 0.7), intravascular large B-​cell lymphoma (IVBCL) (IRR = 1.0) and extranodal marginal-​zone lymphoma (EMZL) (IRR = 1.0); the greatest excess for males is seen for HCL (IRR = 4.6), BL (IRR = 2.9) and MCL (IRR = 2.9). The age-​incidence curves for mature B-​cell malignancies are steepest for CLL/​SLL and DLBCL and much flatter for BL and HCL (Figure 40–​5a). While still a short time frame and still potentially impacted by diagnostic changes, time trends in incidence rates for 2003–​2012 show that EMZL has had the greatest increase, and this was observed in whites and blacks, but not Asian/​Pacific Islanders. The greatest decreases were observed for CLL/​SLL and FL, and these were largely consistent across race/​ethnicity and sex. The CLL/​SLL decline likely reflects better classification of MCL and MZL (drawing from CLL/​SLL) (van Leeuwen et al., 2014) and the change in diagnostic criteria implemented in 2008 that

74

774 25

100%

20

80%

15

60%

10

40%

5

20%

Percent Survival

Age-Adjusted Rates (per 100,000)

PART IV:  Cancers by Tissue of Origin

Incidence Mortality 5-year relative survival

10 20

05 20

00 20

95 19

90 19

85 19

80

0%

19

19

75

0

Figure 40–​4.  Age-adjusted incidence, mortality and 5-year relative survival, 1975–2012, United States (based on SEER9 registries).

reassigned a substantial portion of CLL/​SLL as MBL (Hallek et al., 2008), with resulting impacts on incidence rates (Call et  al., 2014); reasons for the decline in FL are less clear. Mycosis fungoides/​Sezary syndrome (MF/​SS) and peripheral T-​cell lymphoma, not otherwise specified (PTCL-​NOS) are the most common mature T-​cell neoplasms. Compared to whites, mature T-​cell neoplasms are slightly more common in blacks (IRR = 1.3) and slightly less common in Asian/​Pacific Islanders (IRR = 0.9); they are also more common in males compared to females (IRR = 1.7). However, these patterns obscure the substantial heterogeneity among the many diverse subtypes of mature T-​cell neoplasms. Compared to whites, the greatest racial/​ethnic diversity is observed for subcutaneous panniculitis-​ like T-​cell lymphoma (SPTCL) (IRR  =  3.0) and extranodal natural killer/​T-​cell lymphoma (ENKTL) (IRR = 0.4) in blacks and ENKTL (IRR = 2.4) and cutaneous T-​cell lymphoma (CTCL) (IRR = 0.4) in Asian/​Pacific Islanders. The only mature T-​cell subtypes not elevated in men are SPTCL (IRR  =  0.5) and T-​cell large granular lymphocytic leukemia (T-​LGL) (IRR  =  1.0). The age-​incidence curves for mature T-​cell malignancies are steepest for MS/​FF and PTCL-​NOS (Figure  40–​5b). During 2003–​2012, the overall incidence of mature T-​cell neoplasms was increasing, similarly for males and females but more strongly for blacks; in contrast, Asian/​Pacific Islanders showed a non-​significant decline. The greatest increases of specific subtypes were observed for adult T-​cell leukemia/​lymphoma (ATLL) and angioimmunoblastic T-​cell lymphoma (AITL), and the greatest declines were observed for ALCL; rates were too unstable to compare APCs for mature T-​cell subtypes by race/​ethnicity. Incidence of the B-​cell precursor neoplasm B-​ALL/​LBL is higher in whites relative to blacks (IRR  =  0.5) or Asian/​Pacific Islanders (IRR = 0.7), and rates are slightly higher for males (IRR = 1.2). The highest rates are seen in early childhood (ages 1–​4 years), after which there is a precipitous decline to age 30–​34 years, and then a slow but monotonic increase with age (Figure  40–​5c). Age-​adjusted B-​ALL/​ LBL has shown one of the highest rates of increase in the last decade for both whites and blacks, but not Asian/​Pacific Islanders. The rate of T-​cell precursor neoplasm T-​ALL/​LBL is equal in blacks and

whites (IRR = 1.0), but is lower in Asian/​Pacific Islanders (IRR = 0.7 compared to whites). Rates are higher in males (IRR = 2.1), and rates are highest from ages 5–​19 years, and then slowly decrease with age (Figure 40–​5c). Time trends from 2003–​2012 could not be accurately estimated due to the rarity of this neoplasm. Broadly, the WHO-​defined NHL subtypes patterns for incidence, age-​incidence curves, sex ratios, and to some degree time trends have also been observed for cases diagnosed since 2000 in Australia (van Leeuwen et al., 2014), the United Kingdom (Smith et al., 2015), and France (Dandoit et al., 2015), and more broadly across Europe (Sant et al., 2010). Five-​year relative survival rates are highly variable across NHL subtypes, with rates 90% or higher observed for EMZL, HCL, and primary cutaneous anaplastic large cell lymphoma (PCALCL), and rates < 20% for enteropathy-​associated T-​cell lymphoma EATL and T-​cell prolymphocytic leukemia (T-​PLL) (Table 40–​1). Disparities in 5-​year relative survival of greater than 10% for blacks compared to whites are noted for only a few (rarer) B-​cell lymphomas (i.e., WM/​LPL, BL, and HCL) but were more common for T-​cell lymphomas (i.e., T-​ALL/​LBL, AITL, ALCL, CTCL, PCALCL, and ATLL). Disparities for Asian/​Pacific Islanders compared to whites were only observed for PTCL-​NOS, CTCL, and ATLL. Racial disparities in the early adoption of immunochemotherapy for DLBCL have been documented in the United States (Flowers et al., 2012) and may explain some of the disparities for other subtypes. There were only minor differences in relative survival by sex (data not shown). Very similar estimates and patterns for 5-​year relative survival in the United Kingdom (Smith et al., 2015) or net survival in France (Dandoit et al., 2015) for WHO subtypes from similar time periods to these data have been reported, including the lack of major sex differences. Other studies have reported additional variation in relative survival within ethnic subgroups, such as various Asian ethnic groups in the United States (Carreon et  al., 2008)  and black Africans versus black Caribbeans in the United Kingdom (Shirley et al., 2013). Five-​year survival rates for all of the common WHO-​defined subtypes except PTCL and HCL have been improving in the United States since 1998 (Al-​Hamadani et al., 2015).

– 15 14 – 20 19 – 25 24 – 30 29 – 35 34 – 40 39 –4 45 4 – 50 49 – 55 54 – 60 59 – 65 64 – 70 69 – 75 74 – 80 79 –8 4 85 +

10

1 1– 4 5– 9


10  years before diagnosis; and these associations were observed in HIV-​infected (Breen et  al., 2011; Hussain et  al., 2013)  and healthy Western (De Roos et  al., 2012; Hosnijeh et  al., 2016; Purdue et  al., 2009a, 2011, 2015)  and Asian (Bassig et  al., 2015b) populations. Other immune biomarkers linked in at least three studies to NHL risk include soluble CD23 (a surrogate of B-​cell activation) (Breen et al., 2011; De Roos et al., 2012; Hussain et al., 2013; Kaaks et al., 2015; Purdue et al., 2015); B-​cell attracting chemokine (BCA-​1, which promotes chemotaxis to secondary lymphoid tissues) (De Roos et  al., 2012; Hussain et  al., 2013; Purdue et  al., 2013); and TNF (a proinflammatory cytokine) (Edlefsen et al., 2014; Gu et al., 2010; Purdue et al., 2013). Associations are less clear for specific NHL subtypes, in

 79



The Non-Hodgkin Lymphomas

part due to small numbers and lack of detailed subtype data in many of these studies. Reverse causality, particularly related to precursor conditions, has not been fully assessed.

Medical Conditions and Medications Autoimmunity

In autoimmune disease, an individual’s immune system mistakenly targets and destroys healthy tissues and organs. There are more than 80 conditions, mediated by B-​cell and/​or T-​cell responses (Ballotti et al., 2006; Porakishvili et al., 2001; Sweet et al., 2013). NHL risk is increased in several autoimmune diseases, but it is challenging to disentangle disease, treatment, and host effects (Baecklund et al., 2014; Ekstrom Smedby et  al., 2008). Chronic disease effects are likely to include localized inflammation, antigenic stimulation, lymphocyte infiltration and proliferation, immune dysregulation (e.g., BAFF/​BLyS and APRIL cytokines), and target tissue and organ damage such as fibrosis (Baecklund et al., 2014). Many autoimmune diseases, particularly when severe, are treated with immunosuppressive agents that can increase NHL risk via impaired immune surveillance and impaired control of oncogenic viruses (Bouvard et al., 2009; Grulich et al., 2007). Some of these agents may also be cytotoxic (Beaugerie et al., 2009). Finally, host-​related and possibly lifestyle and environmental risk factors may modify the association of autoimmune diseases with NHL. The excess risk of NHL in rheumatoid arthritis (RA), an inflammatory connective tissue disease, is 2-​fold (Askling et al., 2009; Wolfe and Michaud, 2007). A Swedish cohort study reported that markers of RA disease severity, not its treatment, predicted NHL risk, with odds ratios of 70 for those with very high inflammatory activity or disease severity (Baecklund et  al., 2006). However, there is ongoing uncertainty regarding the role of immunosuppressive therapy. Baecklund et  al. (2014) have hypothesized that immune-​modulating therapies such as anti-​TNF for RA may reduce the disease-​related lymphoma risk by reducing cumulative inflammation, but may increase the immunosuppression-​related lymphoma risk, resulting in stable lymphoma incidence rates. Primary Sjögren syndrome is characterized by the progressive destruction of salivary and lacrimal glands and predominantly affects women. InterLymph pooled case-​control analyses identified a marked increased risk of B-​cell NHL in Sjögren syndrome: including DLBCL (OR  =  9), MZL (OR  =  23), and parotid gland MZL (OR  =  1000) (Ekstrom Smedby et al., 2008); LPL/​WM (OR = 14.0) (Vajdic et al., 2014); and FL among women (OR = 3.37) (Linet et al., 2014). In the only population-​based cohort study of primary Sjögren syndrome, the excess risk of NHL relative to the general population was 16-​fold (Theander et al., 2006). A number of clinical and biologic markers of Sjögren disease severity have been identified as risk factors for NHL (Giannouli and Voulgarelis, 2014). Systemic lupus erythematosus (SLE) can affect multiple organs, including the skin, joints, kidneys, and brain. InterLymph pooled case-​ control study data also show an increased risk of B-​cell NHL in SLE, predominantly DLBCL (OR  =  2.74), MZL (OR  =  7.52), and LPL/​ WM (OR = 8.23) (Ekstrom Smedby et al., 2008; Vajdic et al., 2014). A recent meta-​analysis of prospective SLE cohorts reported a meta-​ SIR of 5.7 for NHL overall (Apor et al., 2014). The risk factors for NHL in SLE are unclear; an international case-​cohort study identified increased risk for male gender and older age, but not disease activity, and there was a non-​significantly elevated risk associated with use of cyclophosphamide and the cumulative dose of glucocorticosteroids (Bernatsky et al., 2014). Celiac disease results in damage to the lining of the small intestine, and InterLymph pooled case-​control study data show that it strikingly increases the risk of PTCL (OR = 17.8) (Wang et al., 2014). Consistent with these findings, a meta-​analysis of five observational studies found a 16-​fold increased risk of T-​cell NHL in celiac disease, predominantly enteropathy-​associated T-​cell lymphoma of the small intestine (Tio et al., 2012). Excess NHL risk has also been observed in other autoimmune conditions, including PTCL in psoriasis (OR = 1.97) (Wang et al., 2014);

779

NHL in systemic sclerosis/​scleroderma (meta-​SIR  =  2.26) (Onishi et al., 2013); thyroid lymphoma in Hashimoto’s thyroiditis (SIR ~60) (Holm et al., 1985); and gastrointestinal DLBCL in IBD (OR = 2.70) (Cerhan et  al., 2014b). However, the excess risk of NHL in IBD appears to be associated with thiopurine immunosuppressive therapy (azathioprine or its metabolite) rather than a disease effect (Beaugerie et al., 2009; Jones and Loftus, 2007; Kotlyar et al., 2015). Furthermore, there is an exceptionally high risk (OR ~100) of a rare aggressive lymphoma, HSTL, associated with exposure to use of thiopurines alone or in combination with anti-​TNF, in young men treated for IBD (Deepak et al., 2013; Kotlyar et al., 2011).

Atopic Conditions

Allergy, hay fever, asthma, and eczema are relatively common phenotypic expressions of altered immune function that may influence lymphomagenesis. An InterLymph pooled case-​ control analysis identified a modest reduction in B-​cell NHL risk for those with a history of specific allergies or hay fever and a significant inverse gradient in risk with increasing number of atopic conditions (Vajdic et al., 2009). Significant inverse associations with history of atopy were observed for DLBCL (OR = 0.82), FL (OR = 0.79–​0.85), CLL/​SLL (OR = 0.85), and MCL (OR = 0.63) (Cerhan et al., 2014b; Linet et al., 2014; Slager et  al., 2014; Smedby et  al., 2014). In contrast, asthma without any other atopic condition significantly increased the risk of splenic MZL (OR = 2.28) (Bracci et al., 2014), and eczema without any other atopic condition significantly increased the risk of BL in younger individuals (OR = 2.54) (Mbulaiteye et al., 2014). A history of eczema also significantly increased the risk of T-​cell NHL (OR = 2.62) and NHL originating in the skin (OR = 2.09), but misdiagnosis cannot be excluded in these cases (Vajdic et al., 2009). Strong evidence of an inverse association was recently reported in a case-​control study of like-​sex twin pairs discordant for NHL; hay fever and certain allergies were associated with decreased NHL risk, and there was a significant trend with number of atopic diseases (Wang et al., 2015a). The potential for selection bias, recall bias, and reduced allergic symptoms from pre-​clinical NHL cannot be excluded from case-​control study evidence (Erber et al., 2009a). The US Multiethnic Cohort Study is the largest (N = 939 NHLs) prospective study to examine the association with atopy, and it revealed no significant relationship with risk of NHL or the major NHL subtypes (Erber et  al., 2009a). Other cohorts of people with test-​positive and self-​reported atopic disease have produced predominantly null results, based on small numbers of incident NHLs. The exception is an inverse association between high levels of total IgE and risk of CLL (Ellison-​ Loschmann et al., 2007; Helby et al., 2015; Nieters et al., 2014). On balance, the findings are mixed, and the relationship between atopic disease and NHL development is uncertain.

Type 2 Diabetes

Type 2 diabetes is associated with metabolic dysfunction (including hyperglycemia and hyperinsulinemia) (Nolan et al., 2011) as well as immunologic alterations (including immunosuppression and B-​and T-​cell dysfunction) (Donath and Shoelson, 2011), suggesting a plausible role in NHL etiology. A  meta-​analysis of 13 case-​control and 13 cohort studies (Castillo et  al., 2012)  reported a positive association of type 2 diabetes with risk of NHL (OR = 1.22; 95% CI: 1.07, 1.39), which did not vary by study design or gender, but was only significant in Asian and European studies and not in American studies. Only four studies in the meta-​analysis had data on FL or DLBCL, and there was no association for FL and only an association for DLBCL in European (OR = 1.48; 95% CI: 1.10, 2.01) but not American studies. Of three studies with data on PTCL, an association was only observed in Asian (OR = 2.42; 95% CI: 1.24, 4.72) but not European studies. While there was an association of type 2 diabetes with risk of leukemia (OR = 1.22; 95% CI: 1.03, 1.44), there was no clear association when stratifying on lymphoid versus myeloid leukemia (Castillo et al., 2012). The impact of potential confounding on the overall associations with NHL and leukemia, particularly by body mass index (BMI), physical activity, and smoking, have not been fully evaluated, nor has the impact of diabetes duration, severity, and treatment. Variation by

780

780

PART IV:  Cancers by Tissue of Origin

geographic region could relate to variability found for other risk factors, particularly BMI. NHL subtype associations remain preliminary given the small number of studies and cases.

Hormonal and Reproductive

Sex differences in NHL incidence and the excess risk of autoimmune disease (an NHL risk factor) in women, along with the biologic impacts of estrogens on immune function (Bouman et al., 2005), have led to an interest in the role of hormonal and reproductive characteristics in NHL etiology. In InterLymph pooled analyses of case-​control studies, there were no associations with age at menarche, age at menopause, ever pregnant, age at first birth, number of live births, or oral contraceptive (OC) use with NHL risk (Kane et al., 2012), while there was an inverse association with menopausal hormone therapy use (OR = 0.79; 95% CI: 0.69, 0.90) (Kane et al., 2013). In the InterLymph analysis across 11 subtypes (Morton et al., 2014b), the only statistically significant findings were for OC use before 1970 (OR = 0.78; 95% CI: 0.62, 1.00) and hormone therapy (HT) use started at age 50 years or older (OR = 0.68; 95% CI: 0.52, 0.88) with risk of DLBCL (Cerhan et  al., 2014b). Cohort studies have reported both positive (Cerhan et  al., 2002; Teras et  al., 2013b) and null (Lu et  al., 2011a; Morton et al., 2009; Norgaard et al., 2006), but not inverse, associations of HT use with NHL risk overall; specific risks with DLBCL, FL, and CLL/​ SLL have been mixed. In a post hoc analysis of the Women’s Health Initiative, incidence of NHL was essentially identical in the treatment (conjugated equine estrogen alone or with medroxyprogesterone acetate) and placebo groups (Kato et  al., 2016). Overall, there is little evidence to support a strong role for menstrual or reproductive factors in the etiology of NHL.

Other Medications

The potential role of drugs used in transplantation, to treat autoimmune diseases, and to modulate hormone levels in the etiology of NHL suggests potential roles for other medications as NHL risk factors, particularly those with immunomodulatory effects (Alexander et  al., 2007). The literature has been mixed, and has generally been based on only one or a few studies, which often have a small sample size, particularly for evaluating NHL subtypes. Here, we focus on more recent data regarding aspirin/​non-​steroidal anti-​inflammatory drugs (NSAIDs), statins, antibiotics, and anti-​depressants. Studies in this area are impacted by concerns of reverse causation (early disease impacting the use of certain medications) and inability to disentangle the effect of the medication versus the underlying condition that prompted the medication use (confounding by indication). Based on three meta-​analyses, risk of NHL was not associated with either regular use of NSAIDs (Bernatsky et  al., 2007; Bosetti et  al., 2006) or aspirin (Algra and Rothwell, 2012). More recently, two large cohort studies (Teras et al., 2013a; Walter et al., 2011) also reported no association of aspirin or other NSAIDs with NHL overall, although one of the studies reported a positive association with FL but not other subtypes (Teras et al., 2013a). A study nested in a primary care database found elevated risk of NHL with long-​term use of COX-​2 inhibitors (OR = 1.70; 95% CI: 1.21, 2.40) (Vinogradova et al., 2011). Acetaminophen use has been linked to NHL risk in some (Baker et al., 2005; Becker et al., 2009; Walter et al., 2011) but not all (Holly et al., 1999; Kato et al., 2002) studies. Stronger associations for current over historical NSAID use in cohort studies provide evidence for the impact of reverse causation (Teras et al., 2013a). Besides lowering lipid levels, statins also have anti-​inflammatory effects (Jain and Ridker, 2005)  and have been inversely associated with NHL in both case-​control (Fortuny et al., 2006; Holly et al., 1999; Zhang et al., 2004) and cohort (Beiderbeck et al., 2003; Jacobs et al., 2011)  studies, although a few studies have reported no association (Coogan et  al., 2007; Friedman et  al., 2008). Similar to aspirin and NSAID use, single studies of statin use and NHL risk have had low power to address NHL subtypes. Antibiotic use has shown both a positive (Chang et al., 2005; Kato et al., 2003) and null association (Anderson et al., 2008) with NHL risk in case-​control studies based on self-​reported use. To overcome

the high potential for both recall and selection bias in the latter studies, population-​based pharmacy databases have been used. A study from the Netherlands reported no significant association with ever use (Beiderbeck et al., 2003), while a more recent study from Denmark reported an elevated risk for NHL overall (RR = 1.13; 95% CI: 1.08, 1.19), which was specific to CLL/​SLL (RR = 1.32; 95% CI: 1.20, 1.45), MCL (RR = 1.40; 95% CI: 1.04, 1.88) and ALCL (RR = 1.83; 95% CI: 1.00, 3.36), but not DLBCL, FL, BL, LPL/​ WM, or MALT (Rasmussen et al., 2012). In the latter study, risk was most strongly associated with shorter time since last prescription, raising the potential of reverse causation. While longer-​term use (> 5 years) and number of prescriptions were associated with CLL/​SLL, the potential for reverse causality remains since the CLL precursor MBL is also associated with increased risk of infections (Moreira et al., 2013). Tricyclic antidepressants, but not other types of antidepressants, were associated with increased risk of NHL (RR = 1.53; 95% CI: 1.06, 2.21) in a population-​based cohort study in Denmark (exposure from pharmacy records), and risk was greater with more prescriptions and longer-​term use (Dalton et al., 2008). However, this association was not observed in a population-​ based case-​ control study in Canada (exposure based on self-​report) (Bahl et  al., 2004)  or a case-​control study embedded in an integrated health delivery system in the United States (exposure based on pharmacy records) (Lowry et  al., 2013), although the latter study found elevated risk for CLL/​SLL (OR = 1.5; 95% CI: 1.1, 2.0).

Prior Cancer and Cancer Therapies

Prior cancers may increase risk of NHL through shared risk factor exposures, shared inherited susceptibility, impact of therapy (including direct effects and secondary effects, such as immunosuppression), or through increased detection or other biases. Case-​ control and cohort studies have shown mixed results for history of prior cancer as an NHL risk factor, and have been limited by small sample size, recall and other biases, along with lack of data on specific types of prior cancer and treatment (Alexander et  al., 2007). In a comprehensive analysis of the SEER program based on primary cancers diagnosed 1973–​2000 (Curtis et al., 2006), an excess of NHL or CLL was not commonly observed after most site-​specific cancers, with only statistically significantly elevated risks of NHL after primary cancer of the nasopharynx (O/​E = 2.58), thymus (O/​E = 4.81), Kaposi sarcoma (O/​ E = 75.99), HL (O/​E = 5.81), cutaneous melanoma (O/​E = 1.25), and CML (O/​E = 1.91), while there were lower risks of NHL (O/​E = 0.88) and CLL (O/​E = 0.66) after primary cancer of the female breast. The excess risk of NHL after nasopharynx carcinoma and Kaposi sarcoma may be due to shared infections (EBV and HIV, respectively); after HL it may be due to a common lymphoid etiology, EBV infection, immune dysfunction, or misclassification; and after cutaneous melanoma, thymus cancer and CML may be due to immunosuppression. An explanation for the lower risk of NHL after breast cancer was not immediately evident. In a systematic review of NHL after cancer therapy (Krishnan and Morgan, 2007), risk of secondary NHL increased more than 5 years after the completion of chemotherapy or radiotherapy (RRs 1.5–​20), with latent periods ranging from 10 to 17 years and longer term risk persisting for more than 3 decades. While the majority of data were from HL survivors, excess risk was also clearly observed after testicular cancer. NHLs after treatment were more often clinically aggressive, but NHL subtype data have been limited. Most of the excess risk is thought to be mechanistically linked to cytotoxic drugs and perhaps general therapy-​induced immunosuppression, but not radiotherapy per se, although this has been difficult to completely disentangle due to heterogeneous and limited data (Travis, 2006). There is some limited evidence that radiotherapy alone after solid cancer is associated with small excess NHL in the SEER program during 1981–​2007 (RR = 1.13; 95% CI: 1.06, 2.17) (Kim et al., 2013). Data on NHL subtype risks are limited; CLL (unlike ALL and myeloid leukemias) has not been linked to either prior radiotherapy or chemotherapy (Travis, 2006). The cumulative incidence of NHL after

 781



The Non-Hodgkin Lymphomas

HL decreased by more than 50% between the periods 1965–​1976 and 1989–​2000 in the Netherlands, which was attributed to declining use of alkylating agent-​based chemotherapy for HL (Schaapveld et al., 2015).

Blood Transfusions

The use of allogeneic blood transfusions increased dramatically after World War II until the 1970s, when concerns about transmitting infectious agents, including hepatitis B and C and HIV led to a decline in use (Surgenor and Schnitzer, 1985). Transmission of infectious agents linked to NHL, along with the immunologic impacts of blood transfusions (Buddeberg et  al., 2008), suggests that this exposure could be a risk factor for NHL. In a meta-​analysis of nine case-​control and five cohort studies (Castillo et al., 2010), blood transfusions were associated with an increased risk of NHL (RR = 1.2; 95% CI: 1.07, 1.35). Results were similar for transfusions given before or after 1992 (a surrogate for increased viral screening and leukodepletion after that year). Heterogeneity was observed by study design, with an association observed only for cohort studies. In a subset of the studies with NHL subtype data, transfusion history was only associated with CLL/​SLL (RR  =  1.66), and not DLBCL or FL. Transfusion history was evaluated in the InterLymph NHL Subtypes Project, which included more recent case-​control studies, most not included in the meta-​analysis (Morton et al., 2014a). In contrast to prior results, blood transfusions before 1990 were inversely associated with risk of NHL overall (OR  =  0.76; 95% CI:  0.67, 0.87) but with evidence for heterogeneity (P  =  0.013) (Morton et  al., 2014b); subtypes showing statistically significant inverse associations after adjustment for potential confounding by subtype-​ specific factors included DLBCL (OR  =  0.69) (Cerhan et  al., 2014b), FL (OR  =  0.78) (Linet et  al., 2014), and CLL/​SLL (OR  =  0.69) (Slager et  al., 2014). Using linked population-​based transfusion and cancer registries in Sweden and Denmark, risk of NHL was elevated 1–​5 months after transfusion (SIR = 7.19; 95% CI: 6.73, 7.68) (supporting a detection bias) and then decreased over time such that risk fell to levels similar to that of the background population (Hjalgrim et al., 2007). There was also no evidence that blood transfusions from pre-​cancerous donors were associated with risk of lymphoma (Edgren et al., 2007), or that transfusion recipients’ risk of CLL was affected by post-​donation CLL in the donor (who presumably had MBL at the time of donation) (Hjalgrim et al., 2015). Overall, the role of blood transfusion in the etiology of NHL or CLL remains mixed.

Vaccinations

Vaccines may have a role in the genesis of NHLs by reducing the risk of infection and modifying immune function. However, in all studies of the association with vaccination history, confounding by inherent differences between those receiving and those not receiving vaccinations is possible. Case-​control studies have the added issue of questionable reliability of self-​reported early life vaccination history and recall bias. Indeed, case-​control study findings have been highly inconsistent in the associated vaccine type and the direction of effect (Lankes et al., 2009). Results from the only cohort study, using Danish vaccination records and data linkage, show a significantly reduced risk of lymphoma in young adults associated with Bacille Calmette-​Guerin (BCG), but no association with smallpox vaccination (Villumsen et al., 2009).

Foreign Body Implants

Two population-​based cohort studies have shown elevated risk of ALCL in women with breast implants (RRs = 10–​20), although this is a very rare occurrence (de Jong et al., 2008; Wang et al., 2015b). An evaluation of the cumulative evidence using a structured expert consultation process found that evidence for a positive association of breast implants and ALCL was very strong, although a causal association had not been fully established (Kim et  al., 2011). In a large meta-​analysis, there was no association of NHL risk after total joint arthroplasty (Onega et al., 2006).

781

Occupational and Environment Factors Occupations

A number of occupations have been associated with increased risk of NHL in one or more studies. The most consistent results are for farmers or agricultural workers, dry cleaners, meat-​processing workers, printers, and teachers (Boffetta and de Vocht, 2007); however, results are far from consistent. The strongest evidence for an association with a specific occupational group is for farmers. The causal agents under consideration include pesticides and viral exposures. A  meta-​analysis suggested a modest increased risk (RR = 1.15), with higher risks observed for livestock farmers (RR = 1.31) (Boffetta and de Vocht, 2007). However, an analysis of occupations in the EPIC cohort study found no association with farming occupations (Neasham et  al., 2011). Results from the InterLymph NHL Subtypes Project found only a suggestive elevation in risk overall for any type of farmer (OR = 1.08; 95% CI: 0.98, 1.18), with no significant heterogeneity by NHL subtype (Morton et  al., 2014b). However, an overall increased risk was observed for employment as a general farm worker (OR = 1.28), with no evidence of heterogeneity among subtypes and field crop/​vegetables farmer (OR  =  1.32), for which the risks differed among subtypes, with the highest risks observed for MF/​SS (OR = 2.80). In contrast, a significantly decreased risk was observed for animal farmers (OR = 0.77), with no evidence of heterogeneity among subtypes. Neither the meta-​ analysis of Boffetta and deVocht nor the InterLymph NHL Subtypes pooled analysis found associations with meat workers. However, an increased risk of NHL was observed in the EPIC cohort (HR = 1.30). While a significant association with work as a printer was found in the meta-​analysis (RR = 1.86), this was not found in the InterLymph NHL Subtypes Project (OR = 0.99), with no heterogeneity among subtypes. An association between work as a teacher and NHL was observed in the meta-​ analysis (RR  =  1.47), but with significant heterogeneity between studies, and likely publication bias. In contrast, the InterLymph NHL Subtypes Project found work as a teacher to be inversely associated with NHL (OR = 0.86), with some heterogeneity observed between subtypes. The analysis of the EPIC cohort additionally identified car repair workers as having an increased risk of NHL (HR = 1.51). Of the 33 occupations examined in the InterLymph NHL Subtypes Project, there was a significant association with one or more subtypes for hairdressers (OR = 1.34) and painters (OR = 1.22). Heterogeneity within subtypes was not observed for hairdressers, but was observed for painters, with the increased risks observed for MF/​SS and BL and no effect for the other subtypes.

Pesticides

The association between NHL and pesticides has long been hypothesized, and has been reviewed extensively (Alavanja and Bonner, 2012; Bassig et al., 2012; Merhi et al., 2007). Associations have been reported between NHL and several classes of pesticides, specifically organochlorines, organophosphate and cabamate insecticides, and the organophosphate and phenoxyacetic acid herbicides. However, the results of case-​control and cohort studies examining the risks for individual pesticides and pesticide classes have been inconsistent, and evaluation of risks for NHL subtypes is limited. We focus our update since the last edition of this chapter on the extensive publications from the Agricultural Health Study (AHS), a population-​based prospective cohort study conducted in the United States, a systematic review and meta-​analysis of 44 papers on the association of multiple pesticides with NHL risk (Schinasi and Leon, 2014) and new or updated IARC monograph evaluations on several pesticides (Guyton et  al., 2015; Loomis et al., 2015). In a meta-​analysis of five studies, organochlorine insecticides were associated with an increased risk of NHL (RR  =  1.3) (Schinasi and Leon, 2014). Increased risks for organochlorine insecticides DDT (RR  =  1.3, based on 7 studies) and lindane (RR  =  1.6, based on 4

782

782

PART IV:  Cancers by Tissue of Origin

studies) were also observed, with evidence of increased risk for longer-​ term exposure to lindane. A  recent publication from the AHS also showed an increased risk for NHL in pesticide applicators with long-​ term exposure to lindane (RR = 2.5) and DDT (RR = 1.7) (Alavanja et al., 2014). Due to the long half-​lives of organochlorine insecticides, several case-​control and cohort studies have examined the association between levels of organochlorine insecticides in blood and adipose tissue and risk of NHL. These results have tended to support a weak association for DDT/​DDE, as well as an association with oxychlordane, a metabolite of chlordane, although the results have not been entirely consistent (Bertrand et al., 2010b; Brauner et al., 2012; Cantor et al., 2003; Cocco et al., 2008; De Roos et al., 2005b; Engel et al., 2007; Laden et al., 2010; Quintana et al., 2004; Spinelli et al., 2007). In the recent IARC review, lindane was classified as a Group 1 carcinogen for lymphoma based on sufficient human and animal evidence; and DDT was classified as a probable carcinogen (Group 2a) based on sufficient evidence in animals, but limited evidence in humans (Loomis et al., 2015). The organophosphate herbicide glyphosate was associated with NHL (RR  =  1.5), with a stronger association when restricted to B-​ cell lymphomas (RR = 2.0), and some evidence of greater risk with higher exposure (Schinasi and Leon, 2014). However, the effect was seen only in case-​control studies, and there was no association in the AHS cohort (De Roos et  al., 2005a). A  recent IARC evaluation of organophosphate pesticides concluded that there was limited evidence of the carcinogenicity of glyphosate in humans, although there was sufficient evidence from animal studies and strong mechanistic evidence (Guyton et al., 2015). As a class, organophosphate insecticides (RR  =  1.6), as well as the organophosphate insecticides diazinon (RR = 1.6) and malathion (RR  =  1.8), were all found to be associated with NHL in the meta-​ analysis. The associations were each based on results from a limited number of case-​control studies. In the AHS cohort there was no association with NHL for either malathion or diazinon, or for any other organophosphate insecticide (Alavanja et al., 2014). The recent IARC review on organophosphate pesticides classified diazinon and malathion as probable carcinogens despite limited human evidence, and in the case of diazinon, limited animal evidence. A class of pesticides that have been extensively studied with respect to NHL risk is phenoxyacetic acid herbicides, in particular 2,4-​D (2,4-​ dichlorophenoxyacetic acid). The meta-​analysis of 12 studies showed an increased risk for 2,4-​D (RR = 1.4), with the largest effect found for DLBCL (RR  =  2.0), based on only two studies and with little evidence of a dose–​response association (Schinasi and Leon, 2014). The IARC Working Group recently evaluated 2,4-​D, and based on a meta-​analysis of 11 case-​control studies that found no association with NHL, along with negative results for cohorts of 2,4-​D manufacturing workers (Burns et al., 2011; Kogevinas and Boffetta, 1995), concluded that there was inadequate evidence of cancer risk in humans for 2,4-​ D (Loomis et al., 2015). They also concluded that there was limited evidence of carcinogenicity in animals, leading to an overall classification of “possibly carcinogenic to humans.” It should be pointed out that there was disagreement in the Working Group, with a “substantial minority” of members concluding that there was limited evidence in humans and sufficient evidence in animals of cancer risk. The meta-​analysis also found significant associations with carbamate herbicides (RR  =  1.4) and insecticides (RR  =  1.7), with similar risks found for the two most used carbamate insecticides, carbaryl (RR = 1.7) and carbofuran (RR = 1.6). Analyses of the AHS cohort (results not included in the meta-​analyses) showed no association with either carbaryl or carbofuran. There was an indication of a possible higher risk with greater exposure, but confidence intervals lacked precision (Bonner et al., 2005; Mahajan et al., 2007).

Industrial Solvents

Results from several epidemiological studies have suggested that exposure to some solvents may increase risk of NHL overall or for specific NHL subtypes, although the specific solvents and subtypes affected have not been consistent (Cocco et al., 2010; Fritschi et al., 2005; Kato et al., 2005; Miligi et al., 2006; Orsi et al., 2010; Purdue et al., 2009b;

Tranah et al., 2009; Wang et al., 2009a; Wong et al., 2010). In addition, studies that have examined these associations have varied with respect to the quality of the exposure assessment, with methods ranging from the use of urinary biomarkers of exposure and job exposure matrices (JEMs) to evaluations based solely on self-​reported job titles (Bassig et al., 2013). Also, most studies have not considered specific NHL subtypes, and have had a relatively small number of exposed cases. Two of the most widely studied solvents for NHL risk have been benzene and trichloroethylene (TCE), and recent systematic reviews and meta-​ analyses have been conducted for both exposures. Three reviews were published on the association of benzene with NHL from 2005 to 2007; one review concluded there was no association (Wong and Fu, 2005) and two concluded that there was evidence of a positive association (Mehlman, 2006; Smith et  al., 2007). The first meta-​analysis of benzene reviewed 18 studies (8 case-​control, 10 cohort) and showed no overall effect (OR = 0.96; 95% CI: 0.86, 1.06) (Lamm et  al., 2005). The second meta-​analysis, published 3  years later, included 22 studies (16 case-​control and 6 cohort) (Steinmaus et al., 2008). Studies that reported risks only by job type or industry were excluded, and a controversial adjustment for an observed healthy worker effect bias in cohort studies was performed. The analysis showed an overall association between benzene and NHL (RR = 1.22; 95% CI: 1.02, 1.47), and in 13 studies that provided results for highly exposed workers, a stronger association was observed (RR  =  1.49; 95% CI: 1.12, 1.97). Two additional meta-​ analyses were published in 2010, which included many of the same studies plus two additional case-​control studies published since the prior meta-​ analyses (Alexander and Wagner, 2010; Kane and Newton, 2010). Both analyses concluded little support for an association between benzene exposure (RR = 1.02; 95% CI: 0.94, 1.12; and RR = 1.11; 95% CI: 0.94, 1.30). In addition to systematically reviewing the literature, a meta-​analysis published in 2011 also assessed risk based on study quality (Vlaanderen et al., 2011). The summary relative risk for NHL showed no association (RR = 1.00; 95% CI: 0.89, 1.13), although there was a suggestive positive association when restricting to higher study quality, but this did not reach statistical significance. TCE exposure has been linked to NHL, and in a meta-​analysis of 17 studies (8 case-​control and 9 cohort) (Scott and Jinot, 2011), there was an overall significant association with TCE (RR = 1.23; 95% CI: 1.07, 1.42), which was stronger for cohort studies (RR = 1.33; 95% CI: 1.13, 1.58). When analysis was restricted to workers in the highest TCE exposure group, an even greater effect was observed (RR = 1.43; 95% CI: 1.30, 1.82). A second meta-​analysis of 19 TCE studies (9 case-​control and 10 cohort), which included several new results (Karami et  al., 2013), found an association of occupational TCE exposure with NHL risk (RR = 1.32; 95% CI: 1.14, 1.54), with stronger effects seen for cohort studies (RR = 1.52; 95% CI: 1.29, 1.79). Including studies that utilized chlorinated solvent exposure rather than TCE specific exposure attenuated the risk estimates. In 2012, the IARC upgraded the classification of TCE to a known human carcinogen (Group 1) (Guha et al., 2012), largely based on findings related to kidney cancer. For both TCE and benzene, new studies with more accurate validated exposure metrics are needed, along with the detailed assessment of specific NHL subtypes.

Environmental Contaminants

Levels of environmental contaminants in the diet, air, or water may also be related to NHL. In particular, diet can be a source of exposure to pesticides, polychlorinated biphenyls (PCBs), and other organic compounds. PCBs have been extensively studied in relation to risk of NHL. PCBs have long half-​lives similar to those of organochlorine insecticides, and thus their levels in blood and adipose tissue reflect longer-​term exposure. The strongest associations between measured levels of organochlorines and NHL risk have been with specific congeners of PCBs, although the specific congeners identified have not been consistent. A meta-​analysis of plasma levels of PCBs with NHL risk (Freeman and Kohles, 2012) found that several PCB congeners (PCB congeners 118, 138, 153, 156,

 783



The Non-Hodgkin Lymphomas

170, 180, and 187) were significantly associated with NHL, with a weighted overall OR of 1.43 (95% CI: 1.31, 1.55). In a systematic review, the overall association between PCBs and NHL, including all study designs and not just studies based on plasma levels, concluded that the weight of evidence supports a causal role of PCBs in NHL (Kramer et al., 2012). While IARC classified PCBs as carcinogenic to humans (Group 1), evidence for NHL was considered limited (Lauby-​Secretan et al., 2013). Drinking water can be a source of exposure to many organic compounds. Several recent studies have examined dietary nitrate and its association with NHL. In the NCI-​SEER study no increased risk of NHL with nitrate levels in water or through diet was observed (Ward et al., 2006), however, an increased risk of NHL related to dietary exposure to nitrite was identified (OR = 3.1; 95% CI: 1.7, 5.5). In an analysis of a population-​based case-​control study from Nebraska, dietary nitrate and nitrite were examined for associations with NHL and subtypes. No significant associations were identified, but a non-​significant excess risk of NHL was observed in women with high dietary levels of nitrites. At this time, an association between nitrate exposure in diet and drinking water and NHL cannot be concluded. A recent novel geographic-​based study of benzene exposure based on place of residence showed a clear association of distance between the place of residence and a benzene release site with the risk of NHL, suggesting that environmental exposure to benzene may be related to NHL risk (Bulka et al., 2013).

Radiation

The risk of NHL related to radiation is weak at best, with small increased risk related to high doses observed in atomic bomb survivors and those exposed to therapeutic irradiation (Kim et  al., 2013), but no association with occupational exposure. An analysis of Chernobyl accident recovery workers showed an increased risk in the highest exposed workers, similar to those estimated from A-​ bomb survivors (Kesminiene et al., 2008). A Canadian case-​control study found a 2-​fold increased risk in men who self-​reported any exposure to radium, but with small numbers of exposed subjects (Karunanayake et  al., 2008). A  case-​control study from Australia using a job-​exposure matrix found no increased risk (Karipidis et  al., 2009). At this point, it is not known whether risks exist in specific subtypes other than CLL and ALL, since most subtypes have not been separately evaluated, mostly because many of the studies were initiated long before the WHO classification system was published, as well as the small number of exposed individuals in the studies.

Lifestyle and Personal Factors Tobacco

Tobacco has been extensively studied with respect to NHL, and the results continue to suggest that smokers have about the same risk of NHL as do non-​smokers, but the association differs across NHL subtypes. A recent meta-​analysis of 41 case–​control studies (20,143 cases) and 9 cohort studies (5748 cases) found that ever smoking was associated with a slight increased risk for NHL (OR = 1.05; 95% CI: 1.00, 1.09), with the largest effect for T-​cell tumors (Sergentanis et al., 2013). Recent publications from a cohort of 330,000 workers in the Swedish construction industry (Fernberg et  al., 2006) and 480,000 participants in the EPIC cohort (Nieters et al., 2008) showed no association with smoking overall or for individual subtypes. However, analysis of the 133,000 women in the California Teachers Study showed an increased risk for NHL with pack-​years smoked (OR = 1.32 for 20+ pack years), but no specific associations were identified for any NHL subtypes (Lu et al., 2011b). Results from the InterLymph NHL Subtypes Project suggested heterogeneity between subtypes, with significantly increased risks associated with duration of smoking identified for PTCL (OR = 1.75 per increasing duration category) (Morton et al., 2014b).

783

Alcohol

Overall, there is a growing consensus of a slight reduced risk of NHL related to alcohol consumption, although whether this is a causal association remains unclear. A  meta-​ analysis of 21 case-​ control (including a pooled analysis of 7 studies; Morton et al., 2005) and 8 cohort studies, with a total of 18,759 NHL cases, found a significant reduced risk for drinkers versus non-​drinkers (RR  =  0.85; 95% CI:  0.79, 0.91) (Tramacere et  al., 2012). Estimates were similar between cohort and case-​control studies; however, no dose–​ response association was observed. Results from the InterLymph NHL Subtypes Project found that the inverse association of alcohol consumption with NHL risk overall (OR  =  0.87) and showed weak evidence of heterogeneity (p  =  0.062), with slightly stronger associations for DLBCL, BL, PTCL, and MZL than other subtypes (Morton et  al., 2014b). There were also no strong associations for specific types of alcohol. Overall, while there is a fairly consistent albeit weak inverse association of alcohol use with NHL risk, residual confounding (particularly by diet) and other biases have not been excluded. Also, there is not an established mechanism to explain the association, although the impact of moderate alcohol use on immune function has been suggested (Romeo et al., 2007), as well as repression of mTOR signaling with subsequent decreased lymphoma growth (Hagner et al., 2009).

Diet

The impact of dietary factors on classic carcinogenic pathways (e.g., reactive oxygen species, DNA repair, apoptosis) and exposure to chemical carcinogens through diet (e.g., organochlorines), as well as impacts on immune function and inflammation, have led to an expanding interest in the role of diet in the etiology of NHL, and multiple associations, largely inconsistent, have been reported (Alexander et  al., 2007; Bassig et  al., 2012; Cross and Lim, 2006; Skibola, 2007). Major methodological issues limiting conclusions include small sample sizes, particularly for NHL subtypes; measurement error; potential biases in dietary assessment, particularly from case-​control studies; few data from cohort studies; and limited ability to adjust for potential NHL confounding factors (Cross and Lim, 2006). Here, we focus on the most mature data and some areas with novel leads.

Vegetables, Fruits, and Antioxidants. A recent meta-​ analysis (Chen et al., 2013) of nine case-​control studies and five cohort studies (8718 NHL cases) found that high consumption of vegetables was associated with a decreased risk of NHL compared to individuals with low consumption (RR = 0.81; 95% CI: 0.71, 0.92). However, the association with vegetable consumption was weaker in cohort (RR = 0.90; 95% CI: 0.81,1.00) relative to case-​control (RR = 0.75; 95% CI: 0.60, 0.94) studies, and the inverse association was stronger for DLBCL and FL and was not observed for CLL/​SLL. There was no association of fruit consumption with NHL risk overall or for DLBCL, FL, and CLL/​SLL subtypes. Vegetables are a rich source of antioxidants and phytochemicals, which have antiproliferative and other chemopreventive properties, as well as impacts on immune function; they are also a rich source of folate, which is important in DNA synthesis, repair, and methylation. Several cohort studies have specifically examined risk of NHL in relation to antioxidants. Overall, there is evidence for a protective effect of antioxidant intake and NHL risk, although the specific nutrients involved are not clear. In an analysis of 415 NHL cases from the Iowa Women’s cohort, several antioxidant intakes inferred from diet and supplements were protective for NHL. They included vitamin C, α-​carotene, proanthocyanidins and dietary manganese, with RRs ranging from 0.68 to 0.78 for the highest versus lowest quartile (Thompson et al., 2010). In an analysis of 536 women who developed incident B-​ cell NHL from the California Teachers Cohort, no significant associations were observed, but a weak inverse association was observed for individuals with the highest quartile of an antioxidant index measuring hydroxyl radical absorbance capacity compared to the lowest quartile (RR = 0.68; 95% CI: 0.42, 1.10) (Chang et al., 2011a). A case-​control study using an updated and more extensive measure of hydroxyl

784

784

PART IV:  Cancers by Tissue of Origin

radical absorbance capacity also found a similar inverse association with NHL risk (RR = 0.61; 95% CI: 0.44, 0.84), with little evidence for heterogeneity across common NHL subtypes (Holtan et al., 2012). In an analysis of 1104 cases of NHL from the Women’s Health Initiative, an inverse association with vitamin A intake for NHL overall (highest vs. lowest quartile HR = 0.83; 95% CI: 0.69, 0.99), and an inverse association with vitamin C with risk of DLBCL (HR = 0.69; 95% CI:  0.49, 0.98) was observed (Kabat et  al., 2012). In analyses of 271 NHL cases and 538 matched controls from the Multiethnic Cohort, serum levels of antioxidants (Tertile 3 vs. Tertile 1)  were inversely associated with NHL risk, including carotenoids (OR = 0.66; 95% CI: 0.46, 0.96), lycopene (OR = 0.54; 95% CI: 0.38, 0.78), and α-​cryptoxanthin (OR = 0.53; 95% CI: 0.36, 0.78) (Ollberding et al., 2012). A non-​linear but significant inverse association with serum α-​ tocopherol was also observed (Morimoto et al., 2013). Fewer studies have examined the association between nutrients involved in the one-​ carbon metabolism and NHL risk. In an analysis of 195 NHL cases from the Alpha-​Tocopherol Beta-​Carotene Cancer Prevention Study cohort (Lim et  al., 2006), dietary vitamin B12 was inversely associated with NHL (HR = 0.61; 95% CI: 0.37, 1.00), particularly DLBCL; no other one-​carbon nutrient was associated with NHL, although the power of the study was low. There is some evidence for gene and one-​ carbon nutrient interactions (Lim et al., 2007), although these gene–​ diet interactions await confirmation.

Fat and Protein.  Several studies have examined dietary fat and

protein in NHL risk because macronutrients are believed to interact with and stimulate or suppress the immune system through a number of potential mechanisms (Guthrie and Carroll, 1999). The most consistent findings have been for a positive association of NHL risk with trans fatty acid (Charbonneau et al., 2013; Laake et al., 2013; Zhang et al., 1999), with more mixed but generally supportive data for a positive association with saturated fat and an inverse association with Ω-​3 fatty acid (Chang et al., 2006; Charbonneau et al., 2013; Fritschi et al., 2004; Polesel et al., 2006; Zhang et al., 1999) intakes; the former two types of fats are known to have pro-​inflammatory properties (Bendsen et al., 2011), while Ω-​3 fatty acids are anti-​inflammatory (Wall et al., 2010). Findings for animal protein, meat, and fish are mixed (Bassig et al., 2012; Cross and Lim, 2006), and the largest study of over 3600 incident cases of NHL from the NIH-​AARP Diet and Health Study found no associations of meat or fish intake with risk of NHL overall or for common NHL subtypes (Daniel et al., 2012); similar null findings were found in a pooled analysis of two cohort studies with over 1000 CLL cases (Tsai et al., 2010). There are few data on dietary patterns with NHL risk, but a case-​control study from Nebraska found a dietary pattern high in “meat, fat, and sweets” associated with elevated NHL risk overall (OR = 3.6 for highest quartile; 95% CI: 1.9, 6.8) and for common subtypes (Ollberding et al., 2014), while the Multiethnic Cohort reported an increased risk of FL was observed for men in the highest tertile of consumption of the “fat and meat” dietary pattern compared to men in the lowest tertile based on a 61 cases (OR = 5.16; 95% CI: 1.33, 20.00); no association was observed for NHL overall, other NHL subtypes, or for women (Erber et al., 2009b).

Milk, Dairy, and Vitamin D.  Data on the association of milk

and dairy intake are mixed (Bassig et al., 2012; Cross and Lim, 2006). In the Multiethnic Cohort (Erber et al., 2010), Nurses’ Health Study (Bertrand et  al., 2011), and the California Teachers Study cohort (Chang et al., 2011b), there were no overall associations with vitamin D intake and NHL risk. In a nested case-​control study (1353 cases and 1778 controls) from 10 cohorts in the NCI Cohort Consortium Vitamin D Pooling Project, there was no association of circulating 25-​ hydroxyvitamin D with risk of NHL overall or DLBCL, FL, or CLL/​ SLL (Purdue et al., 2010). In a nested case-​control study within the EPIC cohort (1127 cases and 1127 controls), there was no association of dietary or serum 25-​hydroxyvitamin D with NHL risk overall or with DLBCL or FL, but an inverse association was observed with CLL (RR = 0.40; 95% CI: 0.18, 0.90) (Luczynska et al., 2013). On the other hand, vitamin D insufficiency appears to predict poorer outcomes in DLBCL, CLL, FL, and T-​cell NHL (Drake et al., 2010; Kelly et al.,

2015; Shanafelt et  al., 2011), supporting a potential role in disease progression.

Summary.  While there is mixed but suggestive evidence for an association of diet with NHL risk, more and better studies are needed. Clarification of the role of diet  also requires assessment of other energy balance–​related factors, such as body mass and physical activity, as well as cooking/​processing of foods, dietary patterns, and diet–​gene interactions. Dietary factors may also vary by molecularly defined subtypes, such as the t(14;18) translocation (Chiu et al., 2008). The World Cancer Research Fund/​American Institute for Research 2007 report, Food, Nutrition, Physical Activity, and the Prevention of Cancer, noted the need for more comprehensive and systematic evaluation of the epidemiologic data, particularly for subtypes, and more work on mechanistic underpinnings (World Cancer Research Fund and American Institute for Cancer Research, 2007).

Anthropometrics

Due to the relationship between obesity and both inflammation and immunologic changes, the association between NHL and anthropometric measures such as BMI have been examined in multiple studies. An InterLymph pooled analysis of 18 case-​control studies of 10,453 NHL cases and 16,507 controls found no association overall with increasing BMI and no significant association with obesity per se (Willett et al., 2008). An increased risk of DLBCL (OR = 1.8) with severe obesity (BMI > 40 kg/​m2) was the only association observed. Results from the InterLymph NHL Subtypes Project based on an expanded group of studies also found no overall association with increasing usual BMI, but increased risk for DLBCL (OR = 1.32 per increase in BMI category) (Morton et al., 2014b). A meta-​analysis of 16 cohort studies (17,291 cases) reported a positive association of increasing BMI with NHL risk overall (OR = 1.07 per 5 kg/​m2 increase in BMI) and a significantly increased risk in obese individuals (BMI > 30 kg/​m2) (OR = 1.20) (Larsson and Wolk, 2011); in analysis by subtypes, only DLBCL showed any association with risk (RR  =  1.13 per 5  kg/​m2 increase in BMI). The InterLymph NHL Subtypes Project also found a strong positive association with young adult BMI with NHL risk overall (OR = 1.95 per increase in BMI category), and there was no evidence for heterogeneity across subtypes. For DLBCL, usual adult BMI attenuated after adjusting for young adult BMI and other DLBCL-​specific risk factors (Cerhan et al., 2014b). This association was also observed in an analysis of two large cohorts that found higher young adult BMI was associated with increased risk of all NHL (RR = 1.19 per 5 kg/​m2 in BMI) as well as with DLBCL and FL (Bertrand et al., 2013). In the Million Women Study, higher birthweight, relative size at age 10 years and at age 20 years, and higher recent BMI were all associated with risk of lymphoid malignancy, but the independent contribution of body size at birth and during childhood was small after accounting for adult BMI (Yang et al., 2016).

Physical Activity

There is some evidence that physical activity might reduce the risk of NHL, but the evidence is not consistent. Two recent meta-​analyses of physical activity and NHL showed small non-​significant decreased risks of NHL (OR = 0.91 and 0.92) when the highest level of physical activity was compared the lowest level (Jochem et al., 2014; Vermaete et al., 2013). The effect was slightly stronger in women, although still non-​significant. A pooled analysis of 12 cohort studies that included 6953 NHL cases, the pooled HR for high (90th percentile) versus low (10th percentile) level of leisure-​time physical activity was weakly protective for NHL risk (HR  =  0.90; 95% CI:  0.83, 1.00), which was further attenuated after adjustment for BMI (HR  =  0.94; 95% CI: 0.85, 1.04) (Moore et al., 2016). A recent analysis of a Canadian case-​control study found a decreased risk of NHL in women with high lifetime vigorous physical activity; participants in the highest quartile of activity had a 30% reduction in NHL risk compared to the lowest (Boyle et al., 2015), with the strongest association found in B-​cell lymphomas excluding DLBCL and FL. Recreational sitting time was not associated with the common NHL subtypes in a large cohort study

 785



The Non-Hodgkin Lymphomas

(Teras et al., 2012). Further research is needed to better understand the association between physical activity and NHL.

Hair Dyes

A meta-​analysis of 14 studies (2 cohort and 12 case-​control) found a small increased risk of NHL with hair dye use (OR = 1.23; 95% CI 1.07–1.42) (Takkouche et  al., 2005). The effect was not stronger for use of permanent dye, but there was no analysis of color or years of use. An InterLymph pooled analysis of 4461 NHL cases and 5799 controls (2 studies included in the preceding meta-​analysis, and 2 additional studies) showed an increased risk among women using hair dye before 1980 (OR = 1.3; 95% CI: 1.1, 1.4). This association was restricted to FL and CLL/​SLL subtypes. An association between FL and dark hair dye use started after 1980 was also observed (OR = 1.5; 95% CI:  1.1, 2.0) (Zhang et  al., 2008). It is not clear if these associations reflect risk with long-​term usage or a change in formulation that occurred in the early 1980s. Results from the InterLymph NHL Subtypes Project confirmed the association for FL and CLL/​SLL, and observed an increased risk for long-​term hair dye users for mediastinal DLBCL (OR = 5.0; 95% CI: 1.6, 15.2), although based on a small number of exposed cases (Cerhan et al., 2014b). A clinic-​based case-​control study of 390 NHL cases and 422 controls conducted in Thailand (not included in the InterLymph NHL Subtypes Project) also found an increased risk of NHL related to hair dye use prior to 1980, and for permanent hair dye use in women. Increased risks were observed for both DLBCL and FL subtypes (Sangrajrang et al., 2011).

Sun Exposure

Ultraviolet radiation (UVR) from sunlight is a potential risk factor for NHL because of its established immunosuppressive effects; the geographical and temporal similarities in the incidence of NHL, non-​ melanoma skin cancer, and melanoma; and the co-​occurrence of these neoplasms. However, while the evidence to date is mixed, it favors an inverse association between high levels of personal UVR exposure and NHL risk, with little heterogeneity between subtypes (Morton et al., 2014b). InterLymph pooled case-​control study data show significant, modest reductions in risk for the highest levels of recalled recreational sun exposure for NHL overall (OR = 0.76; 95% CI: 0.63, 0.91; P-​trend 0.01) (Kricker et al., 2008), FL (OR = 0.74; 95% CI: 0.65, 0.86) (Linet et  al., 2014)  and DLBCL (OR  =  0.78; 95% CI:  0.69, 0.89) (Cerhan et  al., 2014b). A  significant association was not observed for either total sun exposure or non-​recreational sun exposure, although ORs for the highest levels of nearly all composite measures were less than 1.0 (Kricker et  al., 2008). The only cohort study to assess self-​reported time outdoors found no association with NHL risk, based on 137 cases (Freedman et  al., 2010). Other measures of personal sun exposure, such as number of sunburns and number of sun-​bathing vacations, have also not predicted NHL risk in prospective cohorts (Chang et al., 2011b; Veierod et al., 2010). An inverse association between residential ambient UVR exposure, as approximated by latitude (van Leeuwen et  al., 2013)  or satellite-​ based measures (Cahoon et al., 2015), and the incidence of NHL overall and of some B-​and T-​cell subtypes has been consistently observed. Nevertheless, cohort study findings for ambient UVR levels and NHL risk are mixed. In the California Teachers Study (women only), high residential UVR levels were associated with reduced risk of NHL overall (RR for highest vs. lowest quartile of minimum UVR = 0.58; 95% CI:  0.42, 0.80), DLBCL (RR  =  0.36; 95% CI:  0.17, 0.78) and CLL/​SLL (RR = 0.46; 95% CI: 0.21, 1.01), but not FL (RR = 0.86; 95% CI: 0.46, 1.62) (Chang et al., 2011b). In contrast, high residential UVR increased the risk of NHL overall, but no individual subtypes, in women enrolled in the US Nurses’ Health Study (RR  =  1.21 for highest vs. lowest quartile at age 15; 95% CI:  1.00, 1.47) (Bertrand et al., 2011). The underlying biological mechanism for an inverse association with UVR exposure is unknown, and findings from observational studies measuring dietary vitamin D intake or circulating 25-​hydroxyvitamin D levels (reviewed earlier) do not support vitamin D as a major mechanism. Immunologic effects of UV-​radiation have been suggested as an alternative mechanism (Norval et al., 2008).

785

Family History and Genetic Factors Familial Aggregation and Family History

While there has been a long history of case reports of familial clustering of lymphomas and leukemias, it has only been relatively recently that these malignancies were considered to have an important inherited genetic component outside of very rare hereditary cancer syndromes (Segel and Lichtman, 2004). In a study of 274 twin pairs, there was a 23-​fold higher risk of NHL in a monozygotic twin and a 14-​fold higher risk in a dizygotic twin of patients with NHL compared to expected background rates, suggesting components of both shared genetics and environment (Mack et al., 1995). In a study of 44,788 twins from Scandinavia (Lichtenstein et al., 2000), there were 274 twins discordant for NHL and none was concordant, so heritability could not be estimated. In contrast for leukemia (N = 305 twin pairs), there was an excess of concordant monozygotic twins compared to dizygotic twins, and heritability was estimated to be 21% (95% CI: 0, 54%); these results have largely been attributed to CLL, as acute lymphoblastic and myeloid leukemias have much weaker familial clustering (Albright et al., 2012). Based on data from the Utah Population Database and the Utah Cancer Registry (Goldgar et al., 1994), the risk of NHL was increased 1.7-​fold in first-​degree relatives of a proband with NHL (familial RR = 1.68; 95% CI: 1.04, 2.48) and the risk of lymphocytic leukemia was > 5-​fold in first-​degree relatives of a proband with lymphocytic leukemia (familial RR  =  5.69; 95% CI:  2.58, 10.0). Using updated data from Utah and a different analytic approach that estimates the Genealogical Index of Familiality (Albright et  al., 2012), excess overall and distant relatedness was observed for NHL and CLL, with the finding for distant relatedness providing additional support for shared genetics in familial clustering. In population-​based data from Sweden and Denmark (Goldin et  al., 2005), first-​degree relatives of cases with NHL had a 1.7-​fold higher risk of developing NHL (95% CI: 1.4, 2.2), while NHL risk was weaker and not statistically significant for first-​degree relatives with HL (RR = 1.4; 95% CI: 1.0, 2.0) or CLL (RR = 1.3; 95% CI: 0.9, 1.9). A pooled analysis of 169,830 first-​degree relatives of 45,406 NHL patients who were diagnosed from 1955 to 2010 in five European countries found risk of NHL was increased 1.6-​fold in siblings (95% CI: 1.3, 1.9) and 1.4-​fold in parent-​ offspring pairs (95% CI: 1.3, 1.5) over the general population (Fallah et al., 2016). Further, familial risk did not significantly change by age at diagnosis of NHL in relatives. In a cohort study of over 120,000 female California teachers (Lu et al., 2009), a history of lymphoma in a first-​degree relative was associated with a 1.7-​fold higher risk of B-​cell NHL (RR = 1.74; 95% CI: 1.16, 2.60). In the InterLymph NHL Subtypes Project, a pooled analysis of 17,471 NHL cases and 23,096 controls from 20 case-​control studies, Morton and colleagues reported a 1.8-​fold increased risk of NHL (95% CI: 1.5, 2.1) for individuals with a first-​degree (blood) relative with NHL; there was also elevated NHL risk for individuals who reported a first-​degree relative with HL (OR = 1.7; 95% CI: 1.2, 2.3) or leukemia (OR = 1.5; 95% CI: 1.3, 1.8), suggesting susceptibility across lymphomas. Furthermore, in the InterLymph NHL Subtypes pooling project, family history of NHL was associated with risk for DLBCL (Cerhan et al., 2014b), FL (Linet et al., 2014), CLL (Slager et al., 2014), MZL (Bracci et al., 2014), MCL (Smedby et al., 2014), LPL/​WM (Vajdic et al., 2014), and PTCL (Wang et al., 2014), and these associations remained unchanged after adjusting for extensive subtype-​specific risk factors, further supporting a role for shared genetics. In a simultaneous analysis of all major NHL subtypes (Morton et al., 2014b), there was no statistically significant heterogeneity across risk of most common NHL subtypes for a family history of NHL (PHomogeneity = 0.52) or HL (PHomogeneity = 0.47). In contrast, there was strong evidence for heterogeneity for a family history of leukemia (PHomogeneity = 3.9 × 10–​5), with family history of leukemia most strongly associated with risk of CLL, LPL/​WM, MCL, and PTCL. Familial aggregation by specific lymphoma subtypes has been more difficult to estimate due to changes in classification and the need for very large studies; recent studies from population-​based registry

786

786

PART IV:  Cancers by Tissue of Origin

studies in Scandanavia have reported some evidence for clustering of risk by NHL subtype. Specifically, first-​degree relatives of the following subtypes have an increased risk of the same subtypes: CLL, 8.5-​ fold (Goldin et al., 2009a); DLBCL, 9.8-​fold (Goldin et al., 2009b); FL, 4-​fold (Goldin et al., 2009b); MCL, 9.0-​fold (Fallah et al., 2016); LPL/​WM, 20-​fold (Kristinsson et  al., 2008); mature T-​cell, 8.2-​fold (Fallah et  al., 2016); and CTCL, 63-​fold (Fallah et  al., 2016). Risk estimates for discordant subtypes in family members have been more varied, but have been associated with a weaker (than for concordant subtypes) but in general elevated risk relative to the background population (Fallah et al., 2016; Goldin et al., 2009b). As summarized earlier and reviewed elsewhere (Cerhan and Slager, 2015), multiple lines of research across different study designs show that a family history of NHL is associated with an increased risk of NHL, familial risk is elevated for multiple lymphoma subtypes, and familial risk does not seem to be confounded by non-​genetic risk factors, although there are likely unidentified risk factors, and clustering of known (and unknown) risk factors within families is difficult to exclude. This suggests at least some shared genetic etiology across the lymphoma subtypes. However, because a family history of a specific lymphoma subtype also appears to be more strongly associated with risk for that specific lymphoma, some genetic factors are likely to be unique to a subtype.

Genetic Risk Factors Linkage Studies.  The only linkage studies that have been con-

ducted in NHL are for CLL (Sellick et al., 2007) and WM (McMaster et al., 2006), and they have not definitively identified genes with large effects. For CLL, significant linkage was identified at 2q21.2, which contains the chemokine receptor (CXCR4) gene and for which rare coding mutations have been identified (Crowther-​Swanepoel et al., 2009).

Genetic Association Studies Candidate Gene. Studies of candidate genes have been pre-

viously reviewed (Cerhan, 2011; Skibola et  al., 2007; Slager et  al., 2013). The choice of a candidate gene has been mainly driven by a priori biologic knowledge of NHL and diseases associated with lymphoma (e.g., infectious or autoimmune), or results identified in other cancers. Candidate gene studies have included pathways related to immune function, cell cycle/​ proliferation, apoptosis, DNA repair, and carcinogen metabolism. Most findings from candidate genes and pathways have failed to replicate, probably due to small sample size (low power), uncontrolled multiple testing (leading to false positive associations), and/​or unrealistic expectations in our ability to choose variants and genes. The most robust findings have been for an LTA-​ TNF haplotype with DLBCL risk (P = 2.93 × 10−​8) (Rothman et al., 2006; Skibola et al., 2010); an SNP (rs3789068) in the proapoptotic BCL2L11 gene with B-​cell NHL risk (OR = 1.21; P = 2.21 × 10−​11) (Nieters et al., 2012); an SNP (rs2266690) in CCNH with CLL risk (OR  =  0.63, P  =  1.62 × 10−​8) (Enjuanes et  al., 2008); and an SNP (rs3132453) in PRRC2A in the HLA class III region with B-​cell NHL risk (OR = 0.68, P = 1.07 × 10−​9) (Nieters et al., 2012).

Genome-​Wide Association Studies.  In contrast to candidate

gene/​pathway studies, GWAS have definitively identified multiple susceptibility loci for lymphoma (Cerhan and Slager, 2015). This success is in part due to the requirement for a stringent level of evidence (e.g., P < 5 × 10−​8) and replication across multiple independent studies. Most GWAS have also been subtype-​specific, with only one GWAS based on all lymphomas (including HL, multiple myeloma, and PTCL) in both the discovery and validation stages (Vijai et al., 2013), and the locus identified in that study (11q12.1 near LPXN) has not replicated in subsequent GWAS. The first GWAS in a lymphoid malignancy was conducted for CLL (Di Bernardo et al., 2008) and to date, GWAS analyses have identified 36 SNPs from 31 loci for CLL (Berndt et al., 2013, 2016; Crowther-​ Swanepoel et al., 2010; Sava et al., 2014; Slager et al., 2011, 2012;

Speedy et al., 2014), which accounts for approximately 16.5% of familial risk of CLL (Berndt et al., 2016). Many of the established SNPs are near or in genes plausibly linked to CLL, including genes involved in apoptosis (including FAS, PMAIP1, BAK1, BCL2, BCL2L11, BMF, CASP8/​CASP10, SERPINB6), telomere function (POT1, TERT, TERC), transcription factors important in B-​ cell function (IRF8, LEF1, PRKD3, SP140, EOMES, LPP), and B-​cell receptor activation (IRF3, HLA-​DQA1). Notably, there has been little evidence of interaction among these SNPs, compatible with independent effects. None of the SNPs has individually shown a strong relationship with age at diagnosis, although cases diagnosed at a younger age tended to carry a greater number of risk alleles (Speedy et al., 2014), supporting the hypothesis that early onset CLL is enriched for genetic susceptibility. In an East Asian population, GWAS-​discovered SNPs for CLL near IRF4 (rs872071), SP140 (rs13397985), and ACOXL (rs17483466) were associated with CLL risk (nominal P < 0.05), with a suggestive association with GRAMD1B (rs735665) (Lan et al., 2010). The minor allele frequencies of these SNPs were much lower than in populations of European descent, supporting the hypothesis that the lower prevalence of CLL-​susceptibility SNPs might explain part of the lower incidence of CLL in East Asian populations. For FL, three early GWAS based on small discovery sets (< 400 cases) identified loci at 6p21.33 (Skibola et  al., 2009)  and 6p21.32 (Conde et al., 2010; Smedby et al., 2011) in the major histocompatibility complex (MHC) associated with FL. In a meta-​analysis of those studies plus a new GWAS of over 2100 cases, the HLA region showed overwhelming association with FL, with over 8000 SNPs achieving genome-​wide significance and a top SNP from this region (rs12195582) reaching P = 5.35 × 10−​100 after additional validation (Skibola et al., 2014). HLA alleles and amino acids (AA) were imputed and the top signal mapped to four linked DRβ1 multi-​allelic AA at positions at 11, 13, 28, and 30, suggesting an important role for DRβ1 peptide presentation in FL. Additional independent signals were also identified in HLA class II (rs17203612) and class I (rs3130437, near HLA-​C); after accounting for all of these signals, no other previously identified SNPs from the MHC achieved genome-​wide significance. Outside of the HLA region, five loci have been identified, including 11q23.3 (near CXCR5), 11q24.3 (near ETS1), 3q28 (in LPP), 18q21.33 (near BCL2), and 8q24 (near PVT1) (Skibola et al., 2014). These genes are linked to B-​cell biology, making them plausible in the etiology of FL. In a GWAS conducted in an East Asian population, a locus at 3q27 (near BCL6 and LPP) for DLBCL was identified (Tan et  al., 2013), although this could not be replicated in independent studies of East Asian (Bassig et  al., 2015a) or European ancestry (Cerhan et  al., 2014a). In a large GWAS of European ancestry (Cerhan et al., 2014a), novel loci identified included 6p25.3 (EXOC2), 6p21.33 (HLA-​B), 2p23.3 (NCOA1), and 8q24.21 (near PVT1 and MYC); the strongest finding after imputing HLA alleles and amino acids (AA) was with HLA-​B*08:01, although this could not be statistically distinguished from the HLA-​B SNP rs2523607 due to high LD. The latter study also estimated that common SNPs, including but not limited to the GWAS-​discovered loci, explained approximately 16% of the variance in DLBCL risk overall. Three of the five GWAS-​discovered SNPs for DLBCL in Europeans were significantly associated with DLBCL in an East Asian population (Bassig et al., 2015a), including EXOC2 (OR = 2.04; P = 3.9 × 10−​10), PVT1 (OR = 1.34; P = 2.1 × 10−​6), and HLA-​B (OR = 3.05; P = 0.009). Overall, MAFs were similar or only modestly lower in the East Asian population for all SNPs except for one of the 8q24 SNPs, which was much rarer. The only GWAS of MZL identified two distinct loci at 6p21.32 (intragenic to BTNL2, in HLA class II) and 6p21.33 (HLA-​B, in HLA class  I); these two loci were in low LD and were statistically independent of each other (Vijai et al., 2015). There was no strong heterogeneity in these results when stratified on MALT versus non-​MALT (splenic MZL and nodal MZL) subtypes, although this was based on a modest sample size. These loci are also associated with autoimmune diseases and immune response, suggesting shared biologic underpinnings with MZL.

 78



The Non-Hodgkin Lymphomas

In summary, GWAS have successfully identified 53 SNPs from 47 genetic loci, mainly associated with specific NHL subtypes, with only two regions—​the HLA region and 8q24—​associated with multiple lymphoma subtypes. The established loci are common (minor allele frequency > 5%) and have small effect sizes (ORs of 0.6 to 2.0), supporting a polygenic model for susceptibility. The GWAS-​identified SNPs that have been identified are largely of unknown function.

Gene–​Environment Studies

Given the lack of validated candidate genes, there have been even fewer robust studies of gene–​environment interaction. The largest studies to date suggest an interaction of B-​ cell-​ mediated autoimmune conditions and the TNF SNP rs1800629 with NHL risk (Wang et  al., 2015c) and BMI and the TNF SNP rs1800629 with DLCBL risk (Kane et al., 2015). There was no association of current cigarette smoking and genetic variation in N-​acetyltransferase enzymes and FL risk (Gibson et al., 2013), but an interaction between organochlorine exposure and the AHR IVS1 + 4640G/​A SNP was observed for overall NHL risk (Ng et al., 2010).

OPPORTUNITIES FOR PREVENTION Although there is convincing evidence for certain infectious agents, lindane, factors related to immune function, and genetic factors in the etiology of NHL, these factors neither explain the majority of cases nor do they easily explain the rapid increase then leveling off of NHL incidence rates. Of these risk factors, only a few are modifiable—​lindane can be avoided, some infections can be prevented through immunization or reduced exposure, and early detection and treatment of infected people can reduce or eliminate risk. Treatment of people with autoimmune disease to reduce chronicity and extent of inflammation is another preventive strategy for these relatively rare but higher risk patients. Minimizing duration and dose of immunosuppression where possible is a more general prevention approach. New strategies to identify and intervene on precursor conditions (e.g., MBL) or intermediate markers (e.g., immune markers) may be another approach to NHL prevention. At this point, other potentially modifiable environmental and lifestyle factors have insufficient evidence to recommend interventions specifically for NHL. Finally, there is no validated genetic risk score to identify high-​risk people in families or the general population.

FUTURE RESEARCH There has been much momentum in our understanding of NHL etiology. This has been greatly facilitated by an increasing number of cohort studies and consortiums such as InterLymph, which allows pooling of multiple studies to achieve large sample sizes needed to evaluate NHL subtype-​specific associations and associations with rarer risk factors. While there has been considerable progress in identifying risk factors, the inability to explain a majority of NHL or much of the epidemic during the last half of the twentieth century suggests that there is not likely to be a single or small number of factors that globally explain all NHL, but rather this may be a case of multifactorial causation related to both environmental and genetic factors (and their interaction) that operate across subtypes in some instances, and are specific to one or a few subtypes in other instances. A high priority agenda is to continue to fully explore etiologic heterogeneity of risk factors that are both common and unique to NHL subtypes. Our understanding of subtypes will also evolve as we better understand underlying lymphoma biology (e.g., cell of origin, MYC status, etc.), and these new findings will need to be incorporated into epidemiologic studies, which will require tissue for phenotyping. Understanding tumor (somatic) genomic changes in NHL, as well as underlying molecular pathways in NHL and its subtypes that are common with other cancers (e.g., ALK, p53, IRF, apoptosis pathways), may also provide clues to new risk factors. Similarly, as we increase our understanding of precursor states to NHL, this will open up new approaches to studying etiology and perhaps prevention strategies.

787

We are still early in our understanding of genetic susceptibility to NHL. There are likely to be additional common variants to be discovered from GWAS, and the role of rare and low-​frequency variants is largely unknown. Genetic variation in the HLA system appears to be particularly promising, and there is a need to address other genetic mechanisms such as copy number variation and epigenetics. Understanding the new genetic loci from a functional perspective is another important area of work, and will need to be accomplished through collaboration with basic scientists and bioinformaticians. Translation of these findings to the clinic and the population is another important research agenda item. There also remains much work in identifying non-​genetic risk factors. Identification of new infectious agents should be pursued. The role of the microbiome has yet to be assessed. New drugs and biologic agents, particularly those that alter immunologic function, will need to be assessed for any role in NHL etiology. Improved exposure assessment for occupational, environmental, dietary, and other personal exposures would help better define the role of these factors in NHL etiology and perhaps identify novel prevention approaches. Gene–​ environment interactions remain promising for understanding NHL etiology, given the established role for both types of factors in this cancer. There is also a growing opportunity to understand genetic and epidemiologic factors in NHL outcome and survivorship, including tertiary prevention. Finally, a majority of studies have been conducted in white populations of European ancestry; future work in other racial/​ ethnic groups, particularly of contrasting NHL incidence or exposure, should be pursued. References Al-​Hamadani M, Habermann TM, Cerhan JR, et al. 2015. Non-​Hodgkin lymphoma subtype distribution, geodemographic patterns, and survival in the US: A longitudinal analysis of the National Cancer Data Base from 1998 to 2011. Am J Hematol, 90(9), 790–​795. PMID: 26096944. Al-​Herz W, Bousfiha A, Casanova JL, et al. 2014. Primary immunodeficiency diseases:  an update on the classification from the international union of immunological societies expert committee for primary immunodeficiency. Front Immunol, 5, 162. PMCID: PMC4001072. Alavanja MC, and Bonner MR. 2012. Occupational pesticide exposures and cancer risk: a review. J Toxicol Environ Health B Crit Rev, 15(4), 238–​ 263. PMID: 22571220 Alavanja MC, Hofmann JN, Lynch CF, et al. 2014. Non-​hodgkin lymphoma risk and insecticide, fungicide and fumigant use in the agricultural health study. PLoS One, 9(10), e109332. PMCID: PMC4206281. Albright F, Teerlink C, Werner TL, and Cannon-​ Albright LA. 2012. Significant evidence for a heritable contribution to cancer predisposition:  a review of cancer familiality by site. BMC Cancer, 12, 138. PMCID: PMC3350420. Alexander DD, Mink PJ, Adami HO, et al. 2007. The non-​Hodgkin lymphomas:  a review of the epidemiologic literature. Int J Cancer, 120(S12), 1–​39. PMID: 17405121. Alexander DD, and Wagner ME. 2010. Benzene exposure and non-​Hodgkin lymphoma:  a meta-​analysis of epidemiologic studies. J Occup Environ Med, 52(2), 169–​189. PMID: 20134348. Algra AM, and Rothwell PM. 2012. Effects of regular aspirin on long-​term cancer incidence and metastasis:  a systematic comparison of evidence from observational studies versus randomised trials. Lancet Oncol, 13(5), 518–​527. PMID: 22440112. Altieri A, Castro F, Bermejo JL, and Hemminki K. 2006. Number of siblings and the risk of lymphoma, leukemia, and myeloma by histopathology. Cancer Epidemiol Biomarkers Prev, 15(7), 1281–​1286. PMID: 16835324. Amin J, Dore GJ, O’Connell DL, et al. 2006. Cancer incidence in people with hepatitis B or C infection:  a large community-​based linkage study. J Hepatol, 45(2), 197–​203. PMID: 16684579. Andersen ES, Omland LH, Jepsen P, et al. 2015. Risk of all-​type cancer, hepatocellular carcinoma, non-​Hodgkin lymphoma and pancreatic cancer in patients infected with hepatitis B virus. J Viral Hepat, 22(10), 828–​834. PMID: 25650146 Anderson LA, Gridley G, Engels EA, et  al. 2008. Antibiotic use and risk of non-​Hodgkin’s lymphoma:  a population-​based case-​control study. Br J Cancer, 98(1), 161–​164. PMCID: PMC2359683. Apor E, O’Brien J, Stephen M, and Castillo JJ. 2014. Systemic lupus erythematosus is associated with increased incidence of hematologic malignancies: a meta-​analysis of prospective cohort studies. Leuk Res, 38(9), 1067–​1071. PMID: 25052307.

78

788

PART IV:  Cancers by Tissue of Origin

Askling J, Baecklund E, Granath F, et  al. 2009. Anti-​tumour necrosis factor therapy in rheumatoid arthritis and risk of malignant lymphomas: relative risks and time trends in the Swedish Biologics Register. Ann Rheum Dis, 68(5), 648–​653. PMID: 18467516. Au WY, Gascoyne RD, Klasa RD, et  al. 2005. Incidence and spectrum of non-​Hodgkin lymphoma in Chinese migrants to British Columbia. Br J Haematol, 128(6), 792–​796. PMID: 15755282. Baecklund E, Iliadou A, Askling J, et al. 2006. Association of chronic inflammation, not its treatment, with increased lymphoma risk in rheumatoid arthritis. Arthritis Rheum, 54(3), 692–​701. PMID: 16508929. Baecklund E, Smedby KE, Sutton LA, Askling J, and Rosenquist R. 2014. Lymphoma development in patients with autoimmune and inflammatory disorders-​-​what are the driving forces? Semin Cancer Biol, 24, 61–​70. PMID: 24333759. Bahl S, Cotterchio M, Kreiger N, and Klar N. 2004. Antidepressant medication use and non-​Hodgkin’s lymphoma risk: no association. Am J Epidemiol, 160(6), 566–​575. PMID: 15353417. Baker JA, Weiss JR, Czuczman MS, et al. 2005. Regular use of aspirin or acetaminophen and risk of non-​Hodgkin lymphoma. Cancer Causes Control, 16(3), 301–​308. PMID: 15947882. Ballotti S, Chiarelli F, and de Martino M. 2006. Autoimmunity: basic mechanisms and implications in endocrine diseases. Part II. Horm Res, 66(3), 142–​152. PMID: 16807509. Bassig BA, Cerhan JR, Au W-​Y, et al. 2015a. Genetic susceptibility to diffuse large B-​cell lymphoma in a pooled study of three Eastern Asian populations. Eur J Haematol, 95(5), 442–​448. PMID: 25611436. Bassig BA, Lan Q, Rothman N, Zhang Y, and Zheng T. 2012. Current understanding of lifestyle and environmental factors and risk of non-​hodgkin lymphoma:  an epidemiological update. J Cancer Epidemiol, 2012, 978930. PMCID: PMC3447374. Bassig BA, Shu XO, Koh WP, et al. 2015b. Soluble levels of CD27 and CD30 are associated with risk of non-​Hodgkin lymphoma in three Chinese prospective cohorts. Int J Cancer, 137(11), 2688–​2695. PMCID: PMC4898881. Bassig BA, Zhang L, Tang X, et  al. 2013. Occupational exposure to trichloroethylene and serum concentrations of IL-​6, IL-​10, and TNF-​alpha. Environ Mol Mutagen, 54(6), 450–​454. PMCID: PMC4360987. Beaugerie L, Brousse N, Bouvier AM, et al. 2009. Lymphoproliferative disorders in patients receiving thiopurines for inflammatory bowel disease: a prospective observational cohort study. Lancet, 374(9701), 1617–​1625. PMID: 19837455. Becker N, Falster MO, Vajdic CM, et al. 2012. Self-​reported history of infections and the risk of non-​Hodgkin lymphoma:  an InterLymph pooled analysis. Int J Cancer, 131(10), 2342–​2348. PMCID: PMC3406230. Becker N, Fortuny J, Alvaro T, et al. 2009. Medical history and risk of lymphoma: results of a European case-​control study (EPILYMPH). J Cancer Res Clin Oncol, 135(8), 1099–​1107. PMID: 19205736. Beiderbeck AB, Holly EA, Sturkenboom MC, et al. 2003. Prescription medications associated with a decreased risk of non-​Hodgkin’s lymphoma. Am J Epidemiol, 157(6), 510–​516. PMID 12631540. Bendsen NT, Stender S, Szecsi PB, et al. 2011. Effect of industrially produced trans fat on markers of systemic inflammation: evidence from a randomized trial in women. J Lipid Res, 52(10), 1821–​1828. PMCID: PMC3173005. Bernatsky S, Lee JL, and Rahme E. 2007. Non-​ Hodgkin’s lymphoma-​ -​ meta-​ analyses of the effects of corticosteroids and non-​ steroidal anti-​ inflammatories. Rheumatology (Oxford), 46(4), 690–​694. PMID: 17148471. Bernatsky S, Ramsey-​Goldman R, Joseph L, et  al. 2014. Lymphoma risk in systemic lupus: effects of disease activity versus treatment. Ann Rheum Dis, 73(1), 138–​142. PMCID: PMC3855611. Berndt SI, Camp NJ, Skibola CF, et al. 2016. Meta-​analysis of genome-​wide association studies discovers multiple loci for chronic lymphocytic leukemia. Nat Commun, 7, 10933. PMCID: PMC4786871. Berndt SI, Skibola CF, Joseph V, et al. 2013. Genome-​wide association study identifies multiple risk loci for chronic lymphocytic leukemia. Nat Genet, 45(8), 868–​876. PMCID: PMC3729927. Bertrand KA, Birmann BM, Chang ET, et  al. 2010a. A  prospective study of Epstein-​Barr virus antibodies and risk of non-​Hodgkin lymphoma. Blood, 116(18), 3547–​3553. PMCID: PMC2981477. Bertrand KA, Chang ET, Abel GA, et al. 2011. Sunlight exposure, vitamin D, and risk of non-​Hodgkin lymphoma in the Nurses’ Health Study. Cancer Causes Control, 22(12), 1731–​1741. PMCID: PMC3240999. Bertrand KA, Giovannucci E, Zhang SM, et  al. 2013. A  prospective analysis of body size during childhood, adolescence, and adulthood and risk of non-​Hodgkin lymphoma. Cancer Prev Res (Phila), 6(8), 864–​873. PMCID: PMC3761937. Bertrand KA, Spiegelman D, Aster JC, et al. 2010b. Plasma organochlorine levels and risk of non-​Hodgkin lymphoma in a cohort of men. Epidemiology, 21(2), 172–​180. PMCID: PMC2957873.

Bhutani M, Polizzotto MN, Uldrick TS, and Yarchoan R. 2015. Kaposi sarcoma-​associated herpesvirus-​associated malignancies:  epidemiology, pathogenesis, and advances in treatment. Semin Oncol, 42(2), 223–​246. PMID: 25843728. Birmann BM, Breen EC, Stuver S, et  al. 2009. Population differences in immune marker profiles associated with human T-​lymphotropic virus type I  infection in Japan and Jamaica. Int J Cancer, 124(3), 614–​621. PMCID: PMC2701897. Boffetta P, and de Vocht F. 2007. Occupation and the risk of non-​Hodgkin lymphoma. Cancer Epidemiol Biomarkers Prev, 16(3), 369–​ 372. PMID: 17372232. Bonner MR, Lee WJ, Sandler DP, et al. 2005. Occupational exposure to carbofuran and the incidence of cancer in the Agricultural Health Study. Environ Health Perspect, 113(3), 285–​289. PMCID: PMC1253753. Booth C, Gilmour KC, Veys P, et al. 2011. X-​linked lymphoproliferative disease due to SAP/​SH2D1A deficiency: a multicenter study on the manifestations, management and outcome of the disease. Blood, 117(1), 53–​62. PMCID: PMC3374620. Bornkamm GW. 2009. Epstein-​Barr virus and its role in the pathogenesis of Burkitt’s lymphoma: an unresolved issue. Semin Cancer Biol, 19(6), 351–​ 365. PMID: 19619654. Bosetti C, Gallus S, and La Vecchia C. 2006. Aspirin and cancer risk:  an updated quantitative review to 2005. Cancer Causes Control, 17(7), 871–​ 888. PMID: 16841255. Bosetti C, Levi F, Ferlay J, et  al. 2008. Incidence and mortality from non-​ Hodgkin lymphoma in Europe:  the end of an epidemic? Int J Cancer, 123(8), 1917–​1923. PMID:18688859. Bouman A, Heineman MJ, and Faas MM. 2005. Sex hormones and the immune response in humans. Hum Reprod Update, 11(4), 411–​ 423. PMID: 15817524. Bouvard V, Baan R, Straif K, et  al. 2009. A review of human carcinogens:  Part B:  biological agents. Lancet Oncol, 10(4), 321–​ 322. PMID: 19350698. Boyle T, Gallagher RP, Gascoyne RD, et al. 2015. Lifetime physical activity and the risk of non-​Hodgkin lymphoma. Cancer Epidemiol Biomarkers Prev, 24(5), 873–​877. PMID: 25934832. Bracci PM, Benavente Y, Turner JJ, et al. 2014. Medical history, lifestyle, family history, and occupational risk factors for marginal zone lymphoma: the InterLymph Non-​Hodgkin Lymphoma Subtypes Project. J Natl Cancer Inst Monogr, 2014(48), 52–​65. PMCID: PMC4207869. Brauner EV, Raaschou-​Nielsen O, Gaudreau E, et  al. 2012. Predictors of adipose tissue concentrations of organochlorine pesticides in a general Danish population. J Expo Sci Environ Epidemiol, 22(1), 52–​59. PMID: 22044925. Breen EC, Hussain SK, Magpantay L, et  al. 2011. B-​cell stimulatory cytokines and markers of immune activation are elevated several years prior to the diagnosis of systemic AIDS-​associated non-​Hodgkin B-​cell lymphoma. Cancer Epidemiol Biomarkers Prev, 20(7), 1303–​ 1314. PMCID: PMC3132317. Buddeberg F, Schimmer BB, and Spahn DR. 2008. Transfusion-​transmissible infections and transfusion-​related immunomodulation. Best Pract Res Clin Anaesthesiol, 22(3), 503–​517. PMID: 18831300. Bulka C, Nastoupil LJ, McClellan W, et al. 2013. Residence proximity to benzene release sites is associated with increased incidence of non-​Hodgkin lymphoma. Cancer, 119(18), 3309–​3317. PMCID: PMC4151314. Burns C, Bodner K, Swaen G, et  al. 2011. Cancer incidence of 2,4-​D production workers. Int J Environ Res Public Health, 8(9), 3579–​3590. PMCID: PMC3194105. Cahoon EK, Pfeiffer RM, Wheeler DC, et  al. 2015. Relationship between ambient ultraviolet radiation and non-​Hodgkin lymphoma subtypes:  a U.S.  population-​based study of racial and ethnic groups. Int J Cancer, 136(5), E432–​E441. PMCID: PMC4268147. Call TG, Norman AD, Hanson CA, et  al. 2014. Incidence of chronic lymphocytic leukemia and high-​ count monoclonal B-​ cell lymphocytosis using the 2008 guidelines. Cancer, 120(13), 2000–​ 2005. PMCID: PMC4124730. Cantor KP, Strickland PT, Brock JW, et  al. 2003. Risk of non-​Hodgkin’s lymphoma and prediagnostic serum organochlorines:  beta-​ hexachlorocyclohexane, chlordane/​heptachlor-​related compounds, dieldrin, and hexachlorobenzene. Environ Health Perspect, 111(2), 179–​183. PMCID: PMC1241347. Carbone A, Cesarman E, Spina M, Gloghini A, and Schulz TF. 2009. HIV-​ associated lymphomas and gamma-​herpesviruses. Blood, 113(6), 1213–​ 1224. PMID: 18955561. Carbone A, Vaccher E, Gloghini A, et  al. 2014. Diagnosis and management of lymphomas and other cancers in HIV-​infected patients. Nat Rev Clin Oncol, 11(4), 223–​238. PMID: 24614140

 789



The Non-Hodgkin Lymphomas

Carreon JD, Morton LM, Devesa SS, et  al. 2008. Incidence of lymphoid neoplasms by subtype among six Asian ethnic groups in the United States, 1996–​2004. Cancer Causes Control, 19(10), 1171–​ 1181. PMCID: PMC2581633 Casabonne D, Benavente Y, Robles C, et al. 2015. Aberrant Epstein-​Barr virus antibody patterns and chronic lymphocytic leukemia in a Spanish multicentric case-​control study. Infect Agent Cancer, 10, 5. PMCID: PMC4429596. Castillo JJ, Dalia S, and Pascual SK. 2010. Association between red blood cell transfusions and development of non-​Hodgkin lymphoma: a meta-​analysis of observational studies. Blood, 116(16), 2897–​2907. PMID: 20625008 Castillo JJ, Mull N, Reagan JL, Nemr S, and Mitri J. 2012. Increased incidence of non-​Hodgkin lymphoma, leukemia, and myeloma in patients with diabetes mellitus type 2:  a meta-​analysis of observational studies. Blood, 119(21), 4845–​4850. PMCID: PMC3367891. Castillo JJ, Reagan JL, Bishop KD, and Apor E. 2014. Viral lymphomagenesis:  from pathophysiology to the rationale for novel therapies. Br J Haematol, 165(3), 300–​315. PMID: 24533596. Cerhan JR. 2011. Host genetics in follicular lymphoma. Best Pract Res Clin Haematol, 24(2), 121–​134. PMCID: PMC3112359. Cerhan JR, Berndt SI, Vijai J, et  al. 2014a. Genome-​wide association study identifies multiple susceptibility loci for diffuse large B cell lymphoma. Nat Genet, 46(11), 1233–​1238. PMCID: PMC4213349. Cerhan JR, Kricker A, Paltiel O, et al. 2014b. Medical history, lifestyle, family history, and occupational risk factors for diffuse large B-​cell lymphoma:  the InterLymph Non-​Hodgkin Lymphoma Subtypes Project. J Natl Cancer Inst Monogr, 2014(48), 15–​25. PMCID: PMC4155465. Cerhan JR, and Slager SL. 2015. Familial predisposition and genetic risk factors for lymphoma. Blood, 126, 2265–​2273. PMCID: PMC26405224. Cerhan JR, Vachon CM, Habermann TM, et  al. 2002. Hormone replacement therapy and risk of non-​Hodgkin lymphoma and chronic lymphocytic leukemia. Cancer Epidemiol Biomarkers Prev, 11, 1466–​ 1471. PMID: 12433728. Chang CM, Stapleton JT, Klinzman D, et  al. 2014. GBV-​C infection and risk of NHL among U.S.  adults. Cancer Res, 74(19), 5553–​ 5560. PMCID: PMC4184918. Chang ET, Balter KM, Torrang A, et  al. 2006. Nutrient intake and risk of non-​Hodgkin’s lymphoma. Am J Epidemiol, 164(12), 1222–​ 1232. PMID: 17005624. Chang ET, Canchola AJ, Clarke CA, et al. 2011a. Dietary phytocompounds and risk of lymphoid malignancies in the California Teachers Study cohort. Cancer Causes Control, 22(2), 237–​249. PMCID: PMC3074494. Chang ET, Canchola AJ, Cockburn M, et al. 2011b. Adulthood residential ultraviolet radiation, sun sensitivity, dietary vitamin D, and risk of lymphoid malignancies in the California Teachers Study. Blood, 118(6), 1591–​ 1599. PMCID: PMC3156047. Chang ET, Smedby KE, Hjalgrim H, et  al. 2005. Medication use and risk of non-​ Hodgkin’s lymphoma. Am J Epidemiol, 162(10), 965–​ 974. PMID: 16192343. Charbonneau B, O’Connor HM, Wang AH, et  al. 2013. Trans fatty acid intake is associated with increased risk of Ω3 fatty acid intake with reduced risk of non-​ Hodgkin lymphoma. J Nutr, 143, 672–​ 681. PMCID: PMC3738236. Chen GC, Lv DB, Pang Z, and Liu QF. 2013. Fruits and vegetables consumption and risk of non-​Hodgkin’s lymphoma: a meta-​analysis of observational studies. Int J Cancer, 133(1), 190–​200. PMID: 23238796. Chen MH, Hsiao LT, Chiou TJ, et al. 2008. High prevalence of occult hepatitis B virus infection in patients with B cell non-​Hodgkin’s lymphoma. Ann Hematol, 87(6), 475–​480. PMID: 18327583. Chihara D, Ito H, Matsuda T, et al. 2014. Differences in incidence and trends of haematological malignancies in Japan and the United States. Br J Haematol, 164(4), 536–​545. PMCID: PMC3907701. Chiu BC, Dave BJ, Ward MH, et al. 2008. Dietary factors and risk of t(14;18)-​ defined subgroups of non-​Hodgkin lymphoma. Cancer Causes Control, 19(8), 859–​867. PMID: 18386141. Clarke CA, Morton LM, Lynch C, et  al. 2013. Risk of lymphoma subtypes after solid organ transplantation in the United States. Br J Cancer, 109(1), 280–​288. PMCID: PMC3708563. Cocco P, Brennan P, Ibba A, et al. 2008. Plasma polychlorobiphenyl and organochlorine pesticide level and risk of major lymphoma subtypes. Occup Environ Med, 65(2), 132–​140. PMID: 17699548. Cocco P, t’Mannetje A, Fadda D, et al. 2010. Occupational exposure to solvents and risk of lymphoma subtypes: results from the Epilymph case-​control study. Occup Environ Med, 67(5), 341–​347. PMID: 20447988. Conde L, Halperin E, Akers NK, et al. 2010. Genome-​wide association study of follicular lymphoma identifies a risk locus at 6p21.32. Nat Genet, 42(8), 661–​664. PMCID: PMC2913472.

789

Coogan PF, Smith J, and Rosenberg L. 2007. Statin use and risk of colorectal cancer. J Natl Cancer Inst, 99(1), 32–​40. PMID: 17202111. Cook LB, Melamed A, Niederer H, et al. 2014. The role of HTLV-​1 clonality, proviral structure, and genomic integration site in adult T-​cell leukemia/​ lymphoma. Blood, 123(25), 3925–​3931. PMCID: PMC4064332. Cross AJ, and Lim U. 2006. The role of dietary factors in the epidemiology of non-​ Hodgkin’s lymphoma. Leuk Lymphoma, 47(12), 2477–​ 2487. PMID: 17169793. Crowther-​Swanepoel D, Broderick P, Di Bernardo MC, et al. 2010. Common variants at 2q37.3, 8q24.21, 15q21.3 and 16q24.1 influence chronic lymphocytic leukemia risk. Nat Genet, 42(2), 132–​136. PMID: 20062064. Crowther-​Swanepoel D, Qureshi M, Dyer MJ, et  al. 2009. Genetic variation in CXCR4 and risk of chronic lymphocytic leukemia. Blood, 114(23), 4843–​4846. PMID: 19812382. Crump C, Sundquist K, Sieh W, Winkleby MA, and Sundquist J. 2012. Perinatal and family risk factors for non-​Hodgkin lymphoma in early life: a Swedish national cohort study. J Natl Cancer Inst, 104(12), 923–​ 930. PMCID: PMC3732249. Curtis RE, Freedman DM, Ron E, et al. 2006. New Malignancies Among Cancer Survivors: SEER Cancer Registries, 1973–​2000. Bethesda, MD: National Cancer Institute, NIH Publ. No. 05-​5302. Dalia S, Chavez J, Castillo JJ, and Sokol L. 2013. Hepatitis B infection increases the risk of non-​Hodgkin lymphoma: a meta-​analysis of observational studies. Leuk Res, 37(9), 1107–​1115. PMID: 23809055. Dalton SO, Poulsen AH, Norgaard M, et  al. 2008. Tricyclic antidepressants and non-​ Hodgkin lymphoma. Epidemiology, 19(4), 546–​ 549. PMID: 18552589. Dandoit M, Mounier M, Guy J, et  al. 2015. The heterogeneity of changes in incidence and survival among lymphoid malignancies in a 30-​year French population-​based registry. Leuk Lymphoma, 56(4), 1050–​1057. PMID: 25166007. Daniel CR, Sinha R, Park Y, et al. 2012. Meat intake is not associated with risk of non-​Hodgkin lymphoma in a large prospective cohort of U.S. men and women. J Nutr, 142(6), 1074–​1080. PMCID: PMC3349978. Dave SS, Wright G, Tan B, et al. 2004. Prediction of survival in follicular lymphoma based on molecular features of tumor-​infiltrating immune cells. N Engl J Med, 351(21), 2159–​2169. PMID: 15548776. de Jong D. 2005. Molecular pathogenesis of follicular lymphoma: a cross talk of genetic and immunologic factors. J Clin Oncol, 23(26), 6358–​6363. PMID: 16155020. de Jong D, Vasmel WL, de Boer JP, et  al. 2008. Anaplastic large-​cell lymphoma in women with breast implants. JAMA, 300(17), 2030–​ 2035. PMID: 18984890. De Roos AJ, Blair A, Rusiecki JA, et  al. 2005a. Cancer incidence among glyphosate-​exposed pesticide applicators in the Agricultural Health Study. Environ Health Perspect, 113(1), 49–​54. PMCID: PMC1253709. De Roos AJ, Hartge P, Lubin JH, et al. 2005b. Persistent organochlorine chemicals in plasma and risk of non-​Hodgkin’s lymphoma. Cancer Res, 65(23), 11214–​11226. PMID: 16322272. De Roos AJ, Martinez-​Maza O, Jerome KR, et  al. 2013. Investigation of epstein-​barr virus as a potential cause of B-​cell non-​Hodgkin lymphoma in a prospective cohort. Cancer Epidemiol Biomarkers Prev, 22(10), 1747–​1755. PMCID: PMC4193346. De Roos AJ, Mirick DK, Edlefsen KL, et al. 2012. Markers of B-​cell activation in relation to risk of non-​Hodgkin lymphoma. Cancer Res, 72(18), 4733–​4743. PMCID: PMC3445724. de Sanjose S, Benavente Y, Vajdic CM, et al. 2008. Hepatitis C and non-​Hodgkin lymphoma among 4784 cases and 6269 controls from the International Lymphoma Epidemiology Consortium. Clin Gastroenterol Hepatol, 6(4), 451–​458. PMCID: PMC3962672. de Sanjose S, Bosch R, Schouten T, et al. 2007. Epstein-​Barr virus infection and risk of lymphoma:  immunoblot analysis of antibody responses against EBV-​related proteins in a large series of lymphoma subjects and matched controls. Int J Cancer, 121(8), 1806–​1812. PMID: 17557295. Deepak P, Sifuentes H, Sherid M, et al. 2013. T-​cell non-​Hodgkin’s lymphomas reported to the FDA AERS with tumor necrosis factor-​alpha (TNF-​alpha) inhibitors: results of the REFURBISH study. Am J Gastroenterol, 108(1), 99–​105. PMID: 23032984. Di Bernardo MC, Crowther-​Swanepoel D, Broderick P, et al. 2008. A genome-​ wide association study identifies six susceptibility loci for chronic lymphocytic leukemia. Nat Genet, 40(10), 1204–​1210. PMID: 18758461. Dolcetti R, Gloghini A, Caruso A, and Carbone A. 2016. A  lymphomagenic role for HIV beyond immune suppression? Blood, 127(11), 1403–​1409. PMCID: PMC4826146. Donath MY, and Shoelson SE. 2011. Type 2 diabetes as an inflammatory disease. Nat Rev Immunol, 11(2), 98–​107. PMID: 21233852.

790

790

PART IV:  Cancers by Tissue of Origin

Drake MT, Maurer MJ, Link BK, et al. 2010. Vitamin D insufficiency and prognosis in non-​Hodgkin’s lymphoma. J Clin Oncol, 28(27), 4191–​4198. PMCID: PMC2953973. Du MQ, Bacon CM, and Isaacson PG. 2007. Kaposi sarcoma-​associated herpesvirus/​human herpesvirus 8 and lymphoproliferative disorders. J Clin Pathol, 60(12), 1350–​1357. PMCID: PMC2095558. Edgren G, Hjalgrim H, Reilly M, et al. 2007. Risk of cancer after blood transfusion from donors with subclinical cancer: a retrospective cohort study. Lancet, 369(9574), 1724–​1730. PMID: 17512857. Edlefsen KL, Martinez-​ Maza O, Madeleine MM, et  al. 2014. Cytokines in serum in relation to future non-​Hodgkin lymphoma risk:  evidence for associations by histologic subtype. Int J Cancer, 135(4), 913–​922. PMID: 24488825. Ekstrom Smedby K, Vajdic CM, Falster M, et  al. 2008. Autoimmune disorders and risk of non-​ Hodgkin lymphoma subtypes:  a pooled analysis within the InterLymph Consortium. Blood, 111(8), 4029–​ 4038. PMCID: PMC2288717. Ellison-​Loschmann L, Benavente Y, Douwes J, et  al. 2007. Immunoglobulin E levels and risk of lymphoma in a case-​control study in Spain. Cancer Epidemiol Biomarkers Prev, 16(7), 1492–​1498. PMID: 17627016. Engel SM, Berkowitz GS, Barr DB, et  al. 2007. Prenatal organophosphate metabolite and organochlorine levels and performance on the Brazelton Neonatal Behavioral Assessment Scale in a multiethnic pregnancy cohort. Am J Epidemiol, 165(12), 1397–​1404. PMID: 17406008. Engels EA, Cho ER, and Jee SH. 2010a. Hepatitis B virus infection and risk of non-​Hodgkin lymphoma in South Korea: a cohort study. Lancet Oncol, 11(9), 827–​834. PMCID: PMC2933963. Engels EA, Pfeiffer RM, Fraumeni JF, Jr., et al. 2011. Spectrum of cancer risk among US solid organ transplant recipients. JAMA, 306(17), 1891–​1901. PMCID: PMC3310893. Engels EA, Pfeiffer RM, Goedert JJ, et al. 2006. Trends in cancer risk among people with AIDS in the United States 1980–​2002. AIDS, 20(12), 1645–​ 1654. PMID: 16868446. Engels EA, Pfeiffer RM, Landgren O, and Moore RD. 2010b. Immunologic and virologic predictors of AIDS-​related non-​hodgkin lymphoma in the highly active antiretroviral therapy era. J Acquir Immune Defic Syndr, 54(1), 78–​84. PMCID: PMC3078556. Engels EA, Rollison DE, Hartge P, et al. 2005. Antibodies to JC and BK viruses among persons with non-​ Hodgkin lymphoma. Int J Cancer, 117(6), 1013–​1019. PMID; 15986438. Enjuanes A, Benavente Y, Bosch F, et al. 2008. Genetic variants in apoptosis and immunoregulation-​related genes are associated with risk of chronic lymphocytic leukemia. Cancer Res, 68(24), 10178–​10186. PMID: 19074885. Enose-​Akahata Y, Abrams A, Johnson KR, Maloney EM, and Jacobson S. 2012. Quantitative differences in HTLV-​I antibody responses:  classification and relative risk assessment for asymptomatic carriers and ATL and HAM/​ TSP patients from Jamaica. Blood, 119(12), 2829–​ 2836. PMCID: PMC3327461. Erber E, Lim U, Maskarinec G, and Kolonel LN. 2009a. Common immune-​ related risk factors and incident non-​Hodgkin lymphoma: the multiethnic cohort. Int J Cancer, 125(6), 1440–​1445. PMCID: PMC2734966. Erber E, Maskarinec G, Gill JK, Park SY, and Kolonel LN. 2009b. Dietary patterns and the risk of non-​Hodgkin lymphoma: the Multiethnic Cohort. Leuk Lymphoma, 50(8), 1269–​1275. PMCID: PMC2903438. Erber E, Maskarinec G, Lim U, and Kolonel LN. 2010. Dietary vitamin D and risk of non-​Hodgkin lymphoma: the multiethnic cohort. Br J Nutr, 103(4), 581–​584. PMCID: PMC2935689. Extramural Committee to Assess Measures of Progress Against Cancer. 1990. Measurement of progress against cancer. J Natl Cancer Inst, 82, 825–​835. PMID: 2332901. Fallah M, Kharazmi E, Pukkala E, et al. 2016. Familial risk of non-​Hodgkin lymphoma by sex, relationship, age at diagnosis and histology: a joint study from five Nordic countries. Leukemia, 30(2), 373–​378. PMID: 26442613. Fazi C, Scarfo L, Pecciarini L, et  al. 2011. General population low-​count CLL-​like MBL persists over time without clinical progression, although carrying the same cytogenetic abnormalities of CLL. Blood, 118(25), 6618–​6625. PMID: 21876118. Ferlay J, Soerjomataram I, Ervik M, et  al. 2013. GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality Worldwide:  IARC CancerBase No. 11 [Internet]. Available from: http://​globocan.iarc.fr. Accessed July 21, 2015. Fernberg P, Edgren G, Adami J, et al. 2011. Time trends in risk and risk determinants of non-​Hodgkin lymphoma in solid organ transplant recipients. Am J Transplant, 11(11), 2472–​2482. PMID: 21883909. Fernberg P, Odenbro A, Bellocco R, et al. 2006. Tobacco use, body mass index and the risk of malignant lymphomas:  a nationwide cohort study in Sweden. Int J Cancer, 118(9), 2298–​2302. PMID: 16331621.

Ferreri AJ, Guidoboni M, Ponzoni M, et  al. 2004. Evidence for an association between Chlamydia psittaci and ocular adnexal lymphomas. J Natl Cancer Inst, 96(8), 586–​594. PMID: 15100336. Flowers CR, Fedewa SA, Chen AY, et al. 2012. Disparities in the early adoption of chemoimmunotherapy for diffuse large B-​cell lymphoma in the United States. Cancer Epidemiol Biomarkers Prev, 21(9), 1520–​1530. PMCID: PMC4155492. Fortuny J, de Sanjose S, Becker N, et al. 2006. Statin use and risk of lymphoid neoplasms: results from the European Case-​Control Study EPILYMPH. Cancer Epidemiol Biomarkers Prev, 15(5), 921–​925. PMID: 16702371. Freedman DM, Kimlin MG, Hoffbeck RW, Alexander BH, and Linet MS. 2010. Multiple indicators of ambient and personal ultraviolet radiation exposure and risk of non-​Hodgkin lymphoma (United States). J Photochem Photobiol B, 101(3), 321–​325. PMCID: PMC2963689. Freeman MD, and Kohles SS. 2012. Plasma levels of polychlorinated biphenyls, non-​Hodgkin lymphoma, and causation. J Environ Public Health, 2012, 258981. PMCID: PMC3346688. Friedman GD, Flick ED, Udaltsova N, et al. 2008. Screening statins for possible carcinogenic risk: up to 9 years of follow-​up of 361,859 recipients. Pharmacoepidemiol Drug Saf, 17(1), 27–​36. PMID: 17944002. Fritschi L, Ambrosini GL, Kliewer EV, and Johnson KC. 2004. Dietary fish intake and risk of leukaemia, multiple myeloma, and non-​ Hodgkin lymphoma. Cancer Epidemiol Biomarkers Prev, 13(4), 532–​ 537. PMID: 15066916. Fritschi L, Benke G, Hughes AM, et  al. 2005. Risk of non-​Hodgkin lymphoma associated with occupational exposure to solvents, metals, organic dusts and PCBs (Australia). Cancer Causes Control, 16(5), 599–​607. PMID: 15986116. Fritz A, Percy C, Jack A, et al. 2000. International Classification of Diseases for Oncology (3rd ed.). Geneva: World Health Organization. Fwu CW, Chien YC, Nelson KE, et al. 2010. Mortality after chronic hepatitis B virus infection: a linkage study involving 2 million parous women from Taiwan. J Infect Dis, 201(7), 1016–​1023. PMID: 20187745. George LC, Rowe M, and Fox CP. 2012. Epstein-​barr virus and the pathogenesis of T and NK lymphoma: a mystery unsolved. Curr Hematol Malig Rep, 7(4), 276–​284. PMID: 22983913. Ghobrial IM, Habermann TM, Macon WR, et al. 2005. Differences between early and late posttransplant lymphoproliferative disorders in solid organ transplant patients:  are they two different diseases? Transplantation, 79(2), 244–​247. PMID: 15665775. Giannouli S, and Voulgarelis M. 2014. Predicting progression to lymphoma in Sjogren’s syndrome patients. Expert Rev Clin Immunol, 10(4), 501–​512. PMID: 24451006 Gibson TM, Engels EA, Clarke CA, et al. 2014a. Risk of diffuse large B-​cell lymphoma after solid organ transplantation in the United States. Am J Hematol, 89(7), 714–​720. PMCID: PMC4069221. Gibson TM, Morton LM, Shiels MS, Clarke CA, and Engels EA. 2014b. Risk of non-​ Hodgkin lymphoma subtypes in HIV-​ infected people during the HAART era:  a population-​based study. AIDS, 28(15), 2313–​2318. PMCID: PMC4260326. Gibson TM, Smedby KE, Skibola CF, et  al. 2013. Smoking, variation in N-​ acetyltransferase 1 (NAT1) and 2 (NAT2), and risk of non-​Hodgkin lymphoma:  a pooled analysis within the InterLymph consortium. Cancer Causes Control, 24(1), 125–​134. PMCID: PMC3529854. Giordano TP, Henderson L, Landgren O, et al. 2007. Risk of non-​Hodgkin lymphoma and lymphoproliferative precursor diseases in US veterans with hepatitis C virus. JAMA, 297(18), 2010–​2017. PMID: 17488966. Glas AM, Kersten MJ, Delahaye LJ, et al. 2005. Gene expression profiling in follicular lymphoma to assess clinical aggressiveness and to guide the choice of treatment. Blood, 105(1), 301–​307. PMID: 15345589. Goldgar DE, Easton DF, Cannon-​ Albright LA, and Skolnick MH. 1994. Systematic population-​based assessment of cancer risk in first-​degree relatives of cancer probands. J Natl Cancer Inst, 86(21), 1600–​1608. PMID: 7932824. Goldin LR, Bjorkholm M, Kristinsson SY, Turesson I, and Landgren O. 2009a. Elevated risk of chronic lymphocytic leukemia and other indolent non-​Hodgkin’s lymphomas among relatives of patients with chronic lymphocytic leukemia. Haematologica, 94(5), 647–​ 653. PMCID: PMC2675676. Goldin LR, Bjorkholm M, Kristinsson SY, Turesson I, and Landgren O. 2009b. Highly increased familial risks for specific lymphoma subtypes. Br J Haematol, 146(1), 91–​94. PMCID: PMC2702464. Goldin LR, Landgren O, McMaster ML, et  al. 2005. Familial aggregation and heterogeneity of non-​ Hodgkin lymphoma in population-​ based samples. Cancer Epidemiol Biomarkers Prev, 14(10), 2402–​ 2406. PMID: 16214923.

 791



The Non-Hodgkin Lymphomas

Grulich AE, Vajdic CM, Falster MO, et al. 2010. Birth order and risk of non-​ hodgkin lymphoma:  true association or bias? Am J Epidemiol, 172(6), 621–​630. PMCID: PMC2950815. Grulich AE, van Leeuwen MT, Falster MO, and Vajdic CM. 2007. Incidence of cancers in people with HIV/​ AIDS compared with immunosuppressed transplant recipients: a meta-​analysis. Lancet, 370(9581), 59–​67. PMID: 17617273. Grulich AE, Wan X, Law MG, et al. 2000. B-​cell stimulation and prolonged immune deficiency are risk factors for non-​Hodgkin’s lymphoma in people with AIDS. AIDS, 14(2), 133–​140. PMID: 10708283 Gu Y, Shore RE, Arslan AA, et al. 2010. Circulating cytokines and risk of B-​cell non-​Hodgkin lymphoma:  a prospective study. Cancer Causes Control, 21(8), 1323–​1333. PMCID: PMC3111139. Guha N, Loomis D, Grosse Y et  al.; International Agency for Research on Cancer Monograph Working Group (2012). Carcinogenicity of trichloroethylene, tetra-​chloroethylene, some other chlorinated solvents, and their metabolites. Lancet Oncol, 13, 1192–​1193. doi:  10.1016/​S1470-​ 2045(12)70485-​0. PMID: 23323277. Guthrie N, and Carroll KK. 1999. Specific versus non-​specific effects of dietary fat on carcinogenesis. Prog Lipid Res, 38(3), 261–​271. PMID: 10664796. Guyton KZ, Loomis D, Grosse Y, et  al. 2015. Carcinogenicity of tetrachlorvinphos, parathion, malathion, diazinon, and glyphosate. Lancet Oncol, 16(5), 490–​491. PMID: 25801782. Hagner PR, Mazan-​Mamczarz K, Dai B, et al. 2009. Alcohol consumption and decreased risk of non-​Hodgkin lymphoma:  role of mTOR dysfunction. Blood, 113(22), 5526–​5535. PMID: 19293424. Hallek M, Cheson BD, Catovsky D, et  al. 2008. Guidelines for the diagnosis and treatment of chronic lymphocytic leukemia:  a report from the International Workshop on Chronic Lymphocytic Leukemia updating the National Cancer Institute-​Working Group  1996 guidelines. Blood, 111(12), 5446–​5456. PMCID: PMC2972576. Hartge P, Bracci PM, Wang SS, Devesa SS, and Holly EA. 2006. Non-​Hodgkin’s lymphoma. In:  Schottenfeld D, and Fraumeni JF, Jr. (Eds.), Cancer Epidemiology and Prevention (3rd ed., pp. 898–​918). New York: Oxford University Press. Helby J, Bojesen SE, Nielsen SF, and Nordestgaard BG. 2015. IgE and risk of cancer in 37 747 individuals from the general population. Ann Oncol, 26(8), 1784–​1790. PMID: 25969367. Hjalgrim H, Edgren G, Rostgaard K et  al. 2007. Cancer incidence in blood transfusion recipients. J Natl Cancer Inst, 99(24), 1864–​ 1874. PMID: 18073377. Hjalgrim H, Rostgaard K, Vasan SK, et  al. 2015. No evidence of transmission of chronic lymphocytic leukemia through blood transfusion. Blood, 126(17), 2059–​2061. PMID: 26302757. Hodson A, Laydon DJ, Bain BJ, Fields PA, and Taylor GP. 2013. Pre-​morbid human T-​lymphotropic virus type I proviral load, rather than percentage of abnormal lymphocytes, is associated with an increased risk of aggressive adult T-​cell leukemia/​lymphoma. Haematologica, 98(3), 385–​388. PMCID: PMC3659941. Holly EA, Lele C, Bracci PM, and McGrath MS. 1999. Case-​control study of non-​Hodgkin’s lymphoma among women and heterosexual men in the San Francisco Bay Area, California. Am J Epidemiol, 150(4), 375–​389. PMID: 10453814. Holm LE, Blomgren H, and Lowhagen T. 1985. Cancer risks in patients with chronic lymphocytic thyroiditis. N Engl J Med, 312(10), 601–​ 604. PMID: 3838363. Holtan SG, O’Connor HM, Fredericksen ZS, et  al. 2012. Food-​frequency questionnaire-​ based estimates of total antioxidant capacity and risk of non-​ Hodgkin lymphoma. Int J Cancer, 131(5), 1158–​ 1168. PMCID: PMC3306533. Horner MJ, Ries LAG, Krapcho M, et al. 2009. SEER Cancer Statistics Review, 1975-​2006, National Cancer Institute. Bethesda, MD, http://​seer.cancer. gov/​csr/​1975_​2006/​, based on November 2008 SEER data submission, posted to the SEER website. Hosnijeh FS, Portengen L, Spath F, et al. 2016. Soluble B-​cell activation marker of sCD27 and sCD30 and future risk of B-​cell lymphomas: a nested case-​ control study and meta-​analyses. Int J Cancer, 138(10), 2357–​2367. PMID: 26684261. Huang Y, de Leval L, and Gaulard P. 2013. Molecular underpinning of extranodal NK/​T-​cell lymphoma. Best Pract Res Clin Haematol, 26(1), 57–​74. PMID: 23768641. Hussain SK, Hessol NA, Levine AM, et al. 2013. Serum biomarkers of immune activation and subsequent risk of non-​Hodgkin B-​cell lymphoma among HIV-​infected women. Cancer Epidemiol Biomarkers Prev, 22(11), 2084–​ 2093. PMCID: PMC3833437. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. 2012. Pharmaceuticals. Volume 100 A.  A  review of human

791

carcinogens. IARC Monogr Eval Carcinog Risks Hum 100(Pt A), 1–​401. PMCID: PMC4781347. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. 2014. Trichloroethylene, tetrachloroethylene, and some other chlorinated agents. IARC Monogr Eval Carcinog Risks Hum, 106, 1–​512. PMCID: PMC4781308. Imai K, Morio T, Zhu Y, et al. 2004. Clinical course of patients with WASP gene mutations. Blood, 103(2), 456–​464. PMID: 12969986. Ishitsuka K, and Tamura K. 2014. Human T-​cell leukaemia virus type I  and adult T-​cell leukaemia-​lymphoma. Lancet Oncol, 15(11), e517–​e526. PMID: 25281470. Iwanaga M, Watanabe T, Utsunomiya A, et al. 2010. Human T-​cell leukemia virus type I  (HTLV-​1) proviral load and disease progression in asymptomatic HTLV-​1 carriers: a nationwide prospective study in Japan. Blood, 116(8), 1211–​1219. PMID: 20448111. Iwanaga M, Watanabe T, and Yamaguchi K. 2012. Adult T-​ cell leukemia:  a review of epidemiological evidence. Front Microbiol, 3, 322. PMCID: PMC3437524. Jacobs EJ, Newton CC, Thun MJ, and Gapstur SM. 2011. Long-​term use of cholesterol-​lowering drugs and cancer incidence in a large United States cohort. Cancer Res, 71(5), 1763–​1771. PMID: 21343395. Jaffe ES, Harris NHL, Stein H, et al. 2008. Introduction and overview of the classification of lymphoid neoplasms. In: Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H, Thiele J, and Vardiman JW (Eds.), WHO Classification of Tumours of the Haematopoietic and Lymphoid Tissues. Lyon: IARC. Jain MK, and Ridker PM. 2005. Anti-​inflammatory effects of statins: clinical evidence and basic mechanisms. Nat Rev Drug Discov, 4(12), 977–​987. PMID: 16341063. Jochem C, Leitzmann MF, Keimling M, Schmid D, and Behrens G. 2014. Physical activity in relation to risk of hematologic cancers: a systematic review and meta-​analysis. Cancer Epidemiol Biomarkers Prev, 23(5), 833–​846. PMID: 24633143. Jones JL, and Loftus EV, Jr. 2007. Lymphoma risk in inflammatory bowel disease: is it the disease or its treatment? Inflamm Bowel Dis, 13(10), 1299–​ 1307. PMID: 17600819. Kaaks R, Sookthai D, Luczynska A, et al. 2015. Lag times between lymphoproliferative disorder and clinical diagnosis of chronic lymphocytic leukemia:  a prospective analysis using plasma soluble CD23. Cancer Epidemiol Biomarkers Prev, 24(3), 538–​545. PMID: 25542829. Kabat GC, Kim MY, Wactawski-​Wende J, et  al. 2012. Intake of antioxidant nutrients and risk of non-​Hodgkin’s Lymphoma in the Women’s Health Initiative. Nutr Cancer, 64(2), 245–​254. PMID: 22211937. Kahl BS, and Yang DT. 2016. Follicular lymphoma: evolving therapeutic strategies. Blood, 127(17), 2055–​2063. PMID: 26989204. Kane E, Skibola CF, Bracci PM, et al. 2015. Non-​Hodgkin lymphoma, body mass index, and cytokine polymorphisms:  a pooled analysis from the InterLymph Consortium. Cancer Epidemiol Biomarkers Prev, 24(7), 1061–​1070. PMCID: PMC4490950. Kane EV, Bernstein L, Bracci PM, et al. 2013. Postmenopausal hormone therapy and non-​Hodgkin lymphoma: a pooled analysis of InterLymph case-​ control studies. Ann Oncol, 24(2), 433–​441. PMCID: PMC3551484. Kane EV, and Newton R. 2010. Benzene and the risk of non-​Hodgkin lymphoma: a review and meta-​analysis of the literature. Cancer Epidemiol, 34(1), 7–​12. PMID: 20075019 Kane EV, Roman E, Becker N, et  al. 2012. Menstrual and reproductive factors, and hormonal contraception use:  associations with non-​Hodgkin lymphoma in a pooled analysis of InterLymph case-​control studies. Ann Oncol, 23(9), 2362–​2374. PMCID: PMC3425371. Karami S, Bassig B, Stewart PA, et al. 2013. Occupational trichloroethylene exposure and risk of lymphatic and haematopoietic cancers:  a meta-​ analysis. Occup Environ Med, 70(8), 591–​599. PMID: 23723297. Karipidis KK, Benke G, Sim MR, et  al. 2009. Non-​Hodgkin lymphoma and occupational radiation exposure assessed using local data. Occup Med (Lond), 59(6), 437–​439. PMID: 19578076. Karube K, Scarfo L, Campo E, and Ghia P. 2014. Monoclonal B cell lymphocytosis and “in situ” lymphoma. Semin Cancer Biol, 24, 3–​ 14. PMID: 23999128. Karunanayake CP, McDuffie HH, Dosman JA, Spinelli JJ, and Pahwa P. 2008. Occupational exposures and non-​Hodgkin’s lymphoma: Canadian case-​ control study. Environ Health, 7, 44. PMCID: PMC2531101. Kato I, Chlebowski RT, Hou L, et al. 2016. Menopausal estrogen therapy and non-​Hodgkin’s lymphoma: a post-​hoc analysis of women’s health initiative randomized clinical trial. Int J Cancer, 138(3), 604–​611. PMID: 26365326. Kato I, Koenig KL, Baptiste MS, et  al. 2003. History of antibiotic use and risk of non-​Hodgkin’s lymphoma (NHL). Int J Cancer, 107(1), 99–​105. PMID: 12925963.

792

792

PART IV:  Cancers by Tissue of Origin

Kato I, Koenig KL, Shore RE, et  al. 2002. Use of anti-​inflammatory and non-​ narcotic analgesic drugs and risk of non-​ Hodgkin’s lymphoma (NHL) (United States). Cancer Causes Control, 13(10), 965–​ 974. PMID: 12588093. Kato I, Koenig KL, Watanabe-​Meserve H, et al. 2005. Personal and occupational exposure to organic solvents and risk of non-​Hodgkin’s lymphoma (NHL) in women (United States). Cancer Causes Control, 16(10), 1215–​ 1224. PMID: 16215872. Kelly JL, Salles G, Goldman B, et al. 2015. Low serum vitamin D levels are associated with inferior survival in follicular lymphoma:  a prospective evaluation in SWOG and LYSA studies. J Clin Oncol, 33(13), 1482–​ 1490. PMCID: PMC4404425. Kempen JH, Daniel E, Dunn JP, et al. 2009. Overal and cancer related mortality among patients with ocular inflammation treated with immunosuppresive drugs:  retrospective cohort study. BMJ, 339(b2480). PMCID: PMC2714688. Kesminiene A, Evrard AS, Ivanov VK, et  al. 2008. Risk of hematological malignancies among Chernobyl liquidators. Radiat Res, 170(6), 721–​735. PMCID: PMC2904977. Kiesewetter B, and Raderer M. 2013. Antibiotic therapy in nongastrointestinal MALT lymphoma: a review of the literature. Blood, 122(8), 1350–​1357. PMID: 23770778. Kim B, Roth C, Young VL, et al. 2011. Anaplastic large cell lymphoma and breast implants:  results from a structured expert consultation process. Plast Reconstr Surg, 128(3), 629–​639. PMID: 21502904. Kim CJ, Freedman DM, Curtis RE, Berrington de Gonzalez A, and Morton LM. 2013. Risk of non-​Hodgkin lymphoma after radiotherapy for solid cancers. Leuk Lymphoma, 54(8), 1691–​1697. PMCID: PMC3862256. Kinch A, Baecklund E, Backlin C, et  al. 2014. A  population-​based study of 135 lymphomas after solid organ transplantation: the role of Epstein-​Barr virus, hepatitis C and diffuse large B-​cell lymphoma subtype in clinical presentation and survival. Acta Oncol, 53(5), 669–​679. PMID: 24164103. Klein U, and Dalla-​Favera R. 2008. Germinal centres: role in B-​cell physiology and malignancy. Nat Rev Immunol, 8(1), 22–​33. PMID: 18097447. Kogevinas M, and Boffetta P. 1995. Occupational exposure to carcinogens and cancer occurrence in Europe. Med Lav, 86(3), 236–​242. PMID 7565284. Kotlyar DS, Lewis JD, Beaugerie L, et al. 2015. Risk of lymphoma in patients with inflammatory bowel disease treated with azathioprine and 6-​ mercaptopurine:  a meta-​analysis. Clin Gastroenterol Hepatol, 13(5), 847–​858; e844; quiz e848–​e850. PMID: 24879926. Kotlyar DS, Osterman MT, Diamond RH, et al. 2011. A systematic review of factors that contribute to hepatosplenic T-​cell lymphoma in patients with inflammatory bowel disease. Clin Gastroenterol Hepatol, 9(1), 36–​41 e31. PMID: 20888436 Krajden M, Yu A, Braybrook H, et al. 2010. GBV-​C/​hepatitis G virus infection and non-​Hodgkin lymphoma: a case control study. Int J Cancer, 126(12), 2885–​2892. PMID: 19904755 Kramer S, Hikel SM, Adams K, Hinds D, and Moon K. 2012. Current status of the epidemiologic evidence linking polychlorinated biphenyls and non-​Hodgkin lymphoma, and the role of immune dysregulation. Environ Health Perspect, 120(8), 1067–​1075. PMCID: PMC3440083. Kricker A, Armstrong BK, Hughes AM, et al. 2008. Personal sun exposure and risk of non Hodgkin lymphoma: a pooled analysis from the Interlymph Consortium. Int J Cancer, 122(1), 144–​154. PMID: 17708556. Krishnan B, and Morgan GJ. 2007. Non-​Hodgkin lymphoma secondary to cancer chemotherapy. Cancer Epidemiol Biomarkers Prev, 16(3), 377–​380. PMID: 17372233. Kristinsson SY, Bjorkholm M, Goldin LR, et  al. 2008. Risk of lymphoproliferative disorders among first-​degree relatives of lymphoplasmacytic lymphoma/​Waldenstrom macroglobulinemia patients: a population-​based study in Sweden. Blood, 112(8), 3052–​3056. PMCID: PMC2569164. Kuppers R. 2005. Mechanisms of B-​cell lymphoma pathogenesis. Nat Rev Cancer, 5(4), 251–​262. PMID: 15803153. Laake I, Carlsen MH, Pedersen JI, et al. 2013. Intake of trans fatty acids from partially hydrogenated vegetable and fish oils and ruminant fat in relation to cancer risk. Int J Cancer, 132(6), 1389–​1403. PMID: 22821174. Laden F, Bertrand KA, Altshul L, et  al. 2010. Plasma organochlorine levels and risk of non-​ Hodgkin lymphoma in the Nurses’ Health Study. Cancer Epidemiol Biomarkers Prev, 19(5), 1381–​ 1384. PMCID: PMC2866053. Lamm SH, Engel A, and Byrd DM. 2005. Non-​Hodgkin lymphoma and benzene exposure: a systematic literature review. Chem Biol Interact, 153–​ 154, 231–​237. PMID: 15885679. Lan Q, Au WY, Chanock S et al. 2010. Genetic susceptibility for chronic lymphocytic leukemia among Chinese in Hong Kong. Eur J Haematol, 85(6), 492–​495. PMCID: PMC2980583.

Landgren O, Albitar M, Ma W, et  al. 2009. B-​cell clones as early markers for chronic lymphocytic leukemia. N Engl J Med, 360(7), 659–​667. PMID: 19213679. Lankes HA, Fought AJ, Evens AM, Weisenburger DD, and Chiu BC. 2009. Vaccination history and risk of non-​Hodgkin lymphoma:  a population-​ based, case-​ control study. Cancer Causes Control, 20(5), 517–​ 523. PMCID: PMC3446202. Larsson SC, and Wolk A. 2011. Body mass index and risk of non-​Hodgkin’s and Hodgkin’s lymphoma: a meta-​analysis of prospective studies. Eur J Cancer, 47(16), 2422–​2430. PMID: 21733676. Lauby-​Secretan B, Loomis D, Grosse Y, El Ghissassi F, Bouvard V, Benbrahim-​ Tallaa L, Guha N, Baan R, Mattock H, Straif K; WHO International Agency for Research on Cancer. 2013. Carcinogenicity of polychlorinated biphenyls and polybrominated biphenyls. Lancet Oncol, 14(4), 287–​288. doi:  10.1016/​S1470-​2045(13)70104-​9. No abstract available. PMID: 23499544. Lauletta G, Russi S, Conteduca V, et  al. 2013 Impact of cryoglobulinemic syndrome on the outcome of chronic hepatitis C virus infection:  a 15-​ year prospective study. Medicine (Baltimore), 92(5), 245–​ 256. PMCID: PMC4553977. Lecuit M, Abachin E, Martin A et al. 2004. Immunoproliferative small intestinal disease associated with Campylobacter jejuni. N Engl J Med, 350(3), 239–​248. PMID: 14724303. Lenz G, and Staudt LM. 2010. Aggressive lymphomas. N Engl J Med, 362(15), 1417–​1429. PMID: 20393178. Lichtenstein P, Holm NV, Verkasalo PK, et al. 2000. Environmental and heritable factors in the causation of cancer-​-​analyses of cohorts of twins from Sweden, Denmark, and Finland. N Engl J Med, 343(2), 78–​85. PMID: 10891514. Lim RB, Loy EY, Lim GH, et al. 2015. Gender and ethnic differences in incidence and survival of lymphoid neoplasm subtypes in an Asian population:  secular trends of a population-​based cancer registry from 1998 to 2012. Int J Cancer, 137(11), 2674–​2687. PMID: 26061168. Lim U, Wang SS, Hartge P, et al. 2007. Gene-​nutrient interactions among determinants of folate and one-​carbon metabolism on the risk of non-​Hodgkin lymphoma:  NCI-​SEER case-​control study. Blood, 109(7), 3050–​3059. PMCID: PMC1852210. Lim U, Weinstein S, Albanes D, et  al. 2006. Dietary factors of one-​carbon metabolism in relation to non-​Hodgkin lymphoma and multiple myeloma in a cohort of male smokers. Cancer Epidemiol Biomarkers Prev, 15(6), 1109–​1114. PMID: 16775167. Limpens J, Stad R, Vos C, et  al. 1995. Lymphoma-​associated translocation t(14;18) in blood B cells of normal individuals. Blood, 85(9), 2528–​2536. PMID: 7727781. Linet MS, Vajdic CM, Morton LM, et al. 2014. Medical history, lifestyle, family history, and occupational risk factors for follicular lymphoma:  the InterLymph Non-​Hodgkin Lymphoma Subtypes Project. J Natl Cancer Inst Monogr, 2014(48), 26–​40. PMCID: PMC4155461. Loomis D, Guyton K, Grosse Y, et al. 2015. Carcinogenicity of lindane, DDT, and 2,4-​ dichlorophenoxyacetic acid. Lancet Oncol, 16(8), 891–​ 892. PMID: 26111929. Lowry SJ, Chubak J, Press OW, McKnight B, and Weiss NS. 2013. Risk of non-​Hodgkin lymphoma in relation to tricyclic antidepressant use. Ann Epidemiol, 23(6), 349–​354. PMCID: PMC3718633. Lu Y, Sullivan-​Halley J, Cozen W, et al. 2009. Family history of haematopoietic malignancies and non-​Hodgkin’s lymphoma risk in the California Teachers Study. Br J Cancer, 100(3), 524–​526. PMCID: PMC2658543. Lu Y, Wang SS, Reynolds P, et al. 2011b. Cigarette smoking, passive smoking, and non-hodgkin lymphoma risk: evidence from the California teachers Study. Am J Epidemiol, 174(5), 563–573. Lu Y, Wang SS, Sullivan-​Halley J, et al. 2011a. Oral contraceptives, menopausal hormone therapy use and risk of B-​cell non-​Hodgkin lymphoma in the California Teachers Study. Int J Cancer, 129(4), 974–​982. PMCID: PMC3258672. Luczynska A, Kaaks R, Rohrmann S, et al. 2013. Plasma 25-​hydroxyvitamin D concentration and lymphoma risk: results of the European Prospective Investigation into Cancer and Nutrition. Am J Clin Nutr, 98(3), 827–​838. PMID: 23885049. Mack TM, Cozen W, Shibata DK, et al. 1995. Concordance for Hodgkin’s disease in identical twins suggesting genetic susceptibility to the young-​adult form of the disease. N Engl J Med, 332(7), 413–​418. PMID: 7824015. Magrath I. 2012. Epidemiology:  clues to the pathogenesis of Burkitt lymphoma. Br J Haematol, 156(6), 744–​756. PMID: 22260300. Mahajan R, Blair A, Coble J, et al. 2007. Carbaryl exposure and incident cancer in the Agricultural Health Study. Int J Cancer, 121(8), 1799–​1805. PMID: 17534892.

 793



The Non-Hodgkin Lymphomas

Mak V, Ip D, Mang O, et al. 2014. Preservation of lower incidence of chronic lymphocytic leukemia in Chinese residents in British Columbia:  a 26-​ year survey from 1983 to 2008. Leuk Lymphoma, 55(4), 824–​ 827. PMID: 23909397. Malecka A, Tierens A, Ostlie I, et al. 2015. Primary diffuse large B-​cell lymphoma associated with clonally-​ related monoclonal B lymphocytosis indicates a common precursor cell. Haematologica, 100(10), e415–​e418. PMCID: PMC260011788. Mamessier E, Broussais-​Guillaumot F, Chetaille B, et  al. 2014. Nature and importance of follicular lymphoma precursors. Haematologica, 99(5), 802–​810. PMCID: PMC4008113. Marcucci F, and Mele A. 2011. Hepatitis viruses and non-​ Hodgkin lymphoma:  epidemiology, mechanisms of tumorigenesis, and therapeutic opportunities. Blood, 117(6), 1792–​1798. PMID: 20959600. Marti GE, Carter P, Abbasi F, et al. 2003. B-​cell monoclonal lymphocytosis and B-​cell abnormalities in the setting of familial B-​cell chronic lymphocytic leukemia. Cytometry B Clin Cytom, 52(1), 1–​12. PMID: 12599176. Marti GE, Rawstron AC, Ghia P, et  al. 2005. Diagnostic criteria for monoclonal B-​ cell lymphocytosis. Br J Haematol, 130(3), 325–​ 332. PMID: 16042682. Mbulaiteye SM, Morton LM, Sampson JN, et  al. 2014. Medical history, lifestyle, family history, and occupational risk factors for sporadic Burkitt lymphoma/​leukemia:  the Interlymph Non-​Hodgkin Lymphoma Subtypes Project. J Natl Cancer Inst Monogr, 2014(48), 106–​ 114. PMCID: PMC4155458. McMaster ML, Goldin LR, Bai Y, et al. 2006. Genomewide linkage screen for Waldenstrom macroglobulinemia susceptibility loci in high-​risk families. Am J Hum Genet, 79(4), 695–​701. PMCID: PMC1592553. Mehlman MA. 2006. Causal relationship between non-​Hodgkin’s lymphoma and exposure to benzene and benzene-​containing solvents. Ann N Y Acad Sci, 1076, 120–​128. PMID: 17119197. Mellemkjaer L, Hammarstrom L, Andersen V, et al. 2002. Cancer risk among patients with IgA deficiency or common variable immunodeficiency and their relatives: a combined Danish and Swedish study. Clin Exp Immunol, 130(3), 495–​500. PMCID: PMC1906562. Mensah FK, Willett EV, Simpson J, Smith AG, and Roman E. 2007. Birth order and sibship size: evaluation of the role of selection bias in a case-​control study of non-​Hodgkin’s lymphoma. Am J Epidemiol, 166(6), 717–​723. PMID: 17591593. Merhi M, Raynal H, Cahuzac E, et  al. 2007. Occupational exposure to pesticides and risk of hematopoietic cancers:  meta-​analysis of case-​ control studies. Cancer Causes Control, 18(10), 1209–​1226. PMID: 17874193. Mesnard B, De Vroey B, Maunoury V, and Lecuit M. 2012. Immunoproliferative small intestinal disease associated with Campylobacter jejuni. Dig Liver Dis, 44(9), 799–​800. PMID: 22552252. Mesri EA, Feitelson MA, and Munger K. 2014. Human viral oncogenesis:  a cancer hallmarks analysis. Cell Host Microbe, 15(3), 266–​282. PMCID: PMC3992243. Michot JM, Canioni D, Driss H, et al. 2015. Antiviral therapy is associated with a better survival in patients with hepatitis C virus and B-​cell non-​Hodgkin lymphomas, ANRS HC-​13 lympho-​C study. Am J Hematol, 90(3), 197–​ 203. PMID: 25417909. Miligi L, Costantini AS, Benvenuti A, et  al. 2006. Occupational exposure to solvents and the risk of lymphomas. Epidemiology, 17(5), 552–​ 561. PMID: 16878041. Moore SC, Lee IM, Weiderpass E, et al. 2016. Association of leisure-​time physical activity with risk of 26 types of cancer in 1.44 million adults. JAMA Intern Med, 176(6), 816–​825. PMID: 27183032. Moreira J, Rabe KG, Cerhan JR, et al. 2013. Infectious complications among individuals with clinical monoclonal B-​ cell lymphocytosis (MBL):  a cohort study of newly diagnosed cases compared to controls. Leukemia, 27(1), 136–​141. PMID: 22781591. Morimoto Y, Ollberding NJ, Cooney RV, et  al. 2013. Prediagnostic serum tocopherol levels and the risk of non-​hodgkin lymphoma: the multiethnic cohort. Cancer Epidemiol Biomarkers Prev, 22(11), 2075–​ 2083. PMCID: PMC3819232. Morton LM, Sampson JN, Cerhan JR, et  al. 2014a. Rationale and design of the International Lymphoma Epidemiology Consortium (InterLymph) non-​Hodgkin Lymphoma Subtypes Project. J Natl Cancer Inst Monogr, 2014(48), 1–​14. PMCID: PMC4155460. Morton LM, Slager SL, Cerhan JR, et  al. 2014b. Etiologic heterogeneity among non-​Hodgkin lymphoma subtypes: the InterLymph Non-​Hodgkin Lymphoma Subtypes Project. J Natl Cancer Inst Monogr, 2014(48), 130–​ 144. PMCID: PMC4155467. Morton LM, Turner JJ, Cerhan JR, et al. 2007. Proposed classification of lymphoid neoplasms for epidemiologic research from the Pathology Working

793

Group of the International Lymphoma Epidemiology Consortium (InterLymph). Blood, 110(2), 695–​708. PMCID: PMC1924473. Morton LM, Wang SS, Richesson DA, et al. 2009. Reproductive factors, exogenous hormone use and risk of lymphoid neoplasms among women in the National Institutes of Health-​AARP Diet and Health Study Cohort. Int J Cancer, 124(11), 2737–​2743. PMCID: PMC2701156. Morton LM, Zheng T, Holford TR, et al. 2005. Alcohol consumption and risk of non-​Hodgkin lymphoma: a pooled analysis. Lancet Oncol, 6(7), 469–​ 476. PMID: 15992695. Mullighan CG. 2013. Genome sequencing of lymphoid malignancies. Blood, 122(24), 3899–​3907. PMID: 24041576. Na R, Laaksonen MA, Grulich AE, et al. 2015 Longitudinal dose and type of immunosuppression in a national cohort of Australian liver, heart, and lung transplant recipients, 1984–​2006. Clin Transplant, 39, 978–​990. PMID: 26278925. Nagel D, Vincendeau M, Eitelhuber AC, and Krappmann D. 2014. Mechanisms and consequences of constitutive NF-​kappaB activation in B-​cell lymphoid malignancies. Oncogene, 33(50), 5655–​5665. PMID: 24469030. National Cancer Institute. Surveillance Research Program, National Cancer Institute SEER* Stat Software (www.seer.cancer.gov/​seerstat) version 8.2.1., 2015. Neasham D, Sifi A, Nielsen KR, et  al. 2011. Occupation and risk of lymphoma:  a multicentre prospective cohort study (EPIC). Occup Environ Med, 68(1), 77–​81. PMID: 20884795. Ng CH, Janoo-​Gilani R, Sipahimalani P, et  al. 2010. Interaction between organochlorines and the AHR gene, and risk of non-​Hodgkin lymphoma. Cancer Causes Control, 21(1), 11–​22. doi: 10.1007/​s10552-​009-​9429-​5. Epub 2009 Sep 26. Nieters A, Conde L, Slager SL, et al. 2012. PRRC2A and BCL2L11 gene variants influence risk of non-​Hodgkin lymphoma:  results from the InterLymph consortium. Blood, 120(23), 4645–​4648. PMCID: PMC3512239. Nieters A, Luczynska A, Becker S, et al. 2014. Prediagnostic immunoglobulin E levels and risk of chronic lymphocytic leukemia, other lymphomas and multiple myeloma-​results of the European Prospective Investigation into Cancer and Nutrition. Carcinogenesis, 35(12), 2716–​ 2722. PMCID: PMC4247516. Nieters A, Rohrmann S, Becker N, et al. 2008. Smoking and lymphoma risk in the European prospective investigation into cancer and nutrition. Am J Epidemiol, 167(9), 1081–​1089. PMID: 18321867. Nogai H, Dorken B, and Lenz G. 2011. Pathogenesis of non-​Hodgkin’s lymphoma. J Clin Oncol, 29(14), 1803–​1811. PMID: 21483013. Nolan CJ, Damm P, and Prentki M. 2011. Type 2 diabetes across generations:  from pathophysiology to prevention and management. Lancet, 378(9786), 169–​181. PMID: 21705072. Nolen BM, Breen EC, Bream JH, et al. 2014. Circulating mediators of inflammation and immune activation in AIDS-​related non-​Hodgkin lymphoma. PloS One, 9(6), e99144. PMCID: PMC4055650. Norgaard M, Poulsen AH, Pedersen L, et  al. 2006. Use of postmenopausal hormone replacement therapy and risk of non-​Hodgkin’s lymphoma:  a Danish population-​based cohort study. Br J Cancer, 94(9), 1339–​1341. PMCID: PMC2361418. Norval M, McLoone P, Lesiak A, and Narbutt J. 2008. The effect of chronic ultraviolet radiation on the human immune system. Photochem Photobiol, 84(1), 19–​28. PMID: 18173697. Nyboe Andersen N, Pasternak B, Basit S, et  al. 2014. Association between tumor necrosis factor-​alpha antagonists and risk of cancer in patients with inflammatory bowel disease. JAMA, 311(23), 2406–​ 2413. PMID: 24938563. Ollberding NJ, Aschebrook-​Kilfoy B, Caces DB, et al. 2014. Dietary patterns and the risk of non-​Hodgkin lymphoma. Public Health Nutr, 17(7), 1531–​1537. PMID: 23659580. Ollberding NJ, Maskarinec G, Conroy SM, et  al. 2012. Prediagnostic circulating carotenoid levels and the risk of non-​Hodgkin lymphoma:  the Multiethnic Cohort. Blood, 119(24), 5817–​5823. PMCID: PMC3382940. Onega T, Baron J, and MacKenzie T. 2006. Cancer after total joint arthroplasty: a meta-​analysis. Cancer Epidemiol Biomarkers Prev, 15(8), 1532–​ 1537. PMID: 16896045. Onishi A, Sugiyama D, Kumagai S, and Morinobu A. 2013. Cancer incidence in systemic sclerosis: meta-​analysis of population-​based cohort studies. Arthritis Rheum, 65(7), 1913–​1921. PMID: 23576072. Orsi L, Monnereau A, Dananche B, et  al. 2010. Occupational exposure to organic solvents and lymphoid neoplasms in men: results of a French case-​ control study. Occup Environ Med, 67(10), 664–​672. PMID: 20837648. Peveling-​Oberhag J, Arcaini L, Hansmann ML, and Zeuzem S. 2013. Hepatitis C-​associated B-​cell non-​Hodgkin lymphomas:  epidemiology, molecular signature and clinical management. J Hepatol, 59(1), 169–​ 177. PMID: 23542089.

794

794

PART IV:  Cancers by Tissue of Origin

Pinzone MR, Berretta M, Cacopardo B, and Nunnari G. 2015. Epstein-​barr virus-​and Kaposi sarcoma-​associated herpesvirus-​related malignancies in the setting of human immunodeficiency virus infection. Semin Oncol, 42(2), 258–​271. PMID: 25843730. Polesel J, Clifford GM, Rickenbach M, et al. 2008. Non-​Hodgkin lymphoma incidence in the Swiss HIV Cohort Study before and after highly active antiretroviral therapy. AIDS, 22(2), 301–​306. PMID: 18097233. Polesel J, Talamini R, Montella M, et al. 2006. Linoleic acid, vitamin D and other nutrient intakes in the risk of non-​Hodgkin lymphoma: an Italian case-​control study. Ann Oncol, 17(4), 713–​718. PMID: 16556850. Porakishvili N, Mageed R, Jamin C, et al. 2001. Recent progress in the understanding of B-​cell functions in autoimmunity. Scand J Immunol, 54(1–​2), 30–​38. PMID: 11439145. Purdue MP, Freedman DM, Gapstur SM, et  al. 2010. Circulating 25-​ hydroxyvitamin D and risk of non-​ hodgkin lymphoma:  Cohort Consortium Vitamin D Pooling Project of Rarer Cancers. Am J Epidemiol, 172(1), 58–​69. PMCID: PMC2892540. Purdue MP, Hofmann JN, Kemp TJ, et  al. 2013. A  prospective study of 67 serum immune and inflammation markers and risk of non-​Hodgkin lymphoma. Blood, 122(6), 951–​957. PMCID: PMC3739038. Purdue MP, Lan Q, Bagni R, et al. 2011. Prediagnostic serum levels of cytokines and other immune markers and risk of non-​hodgkin lymphoma. Cancer Res, 71(14), 4898–​4907. PMCID: PMC3138883. Purdue MP, Lan Q, Kemp TJ, et al. 2015. Elevated serum sCD23 and sCD30 up to two decades prior to diagnosis associated with increased risk of non-​ Hodgkin lymphoma. Leukemia, 29(6), 1429–​1431. PMID: 25567136. Purdue MP, Lan Q, Martinez-​Maza O, et  al. 2009a. A  prospective study of serum soluble CD30 concentration and risk of non-​Hodgkin lymphoma. Blood, 114(13), 2730–​2732. PMCID: PMC2756127. Purdue MP, Severson RK, Colt JS, et  al. 2009b. Degreasing and risk of non-​Hodgkin lymphoma. Occup Environ Med, 66(8), 557–​ 560. PMCID: PMC2995288. Quinlan SC, Pfeiffer RM, Morton LM, and Engels EA. 2011. Risk factors for early-​onset and late-​onset post-​transplant lymphoproliferative disorder in kidney recipients in the United States. Am J Hematol, 86(2), 206–​209. PMCID: PMC3311225. Quintana PJ, Delfino RJ, Korrick S, et  al. 2004. Adipose tissue levels of organochlorine pesticides and polychlorinated biphenyls and risk of non-​Hodgkin’s lymphoma. Environ Health Perspect, 112(8), 854–​861. PMCID: PMC1242012. Rasmussen ML, Hjalgrim H, Molgaard-​Nielsen D, Wohlfahrt J, and Melbye M. 2012. Antibiotic use and risk of non-​Hodgkin lymphomas. Int J Cancer, 131(7), E1158–​E1165. PMID: 22552810. Rastogi T, Devesa S, Mangtani P, et al. 2008. Cancer incidence rates among South Asians in four geographic regions: India, Singapore, UK and US. Int J Epidemiol, 37(1), 147–​160. PMID: 18094016 Rawstron AC, Bennett FL, O’Connor SJ, et al. 2008. Monoclonal B-​cell lymphocytosis and chronic lymphocytic leukemia. N Engl J Med, 359(6), 575–​583. PMID: 18687638. Rawstron AC, Yuille MR, Fuller J, et al. 2002. Inherited predisposition to CLL is detectable as subclinical monoclonal B-​lymphocyte expansion. Blood, 100(7), 2289–​2290. PMID: 12239136. Resnick ES, Moshier EL, Godbold JH, and Cunningham-​Rundles C. 2012. Morbidity and mortality in common variable immune deficiency over 4 decades. Blood, 119(7), 1650–​1657. PMCID: PMC3286343. Rickinson AB. 2014. Co-​ infections, inflammation and oncogenesis:  future directions for EBV research. Semin Cancer Biol, 26, 99–​ 115. PMID: 24751797. Robbiani DF, Deroubaix S, Feldhahn N, et  al. 2015. Plasmodium infection promotes genomic instability and AID-​dependent B cell lymphoma. Cell, 162(4), 727–​737. PMCID: PMC4538708. Robbins HA, Shiels MS, Pfeiffer RM, and Engels EA. 2014. Epidemiologic contributions to recent cancer trends among HIV-​infected people in the United States. AIDS, 28(6), 881–​890. PMID: 24300545. Rollinson S, Levene AP, Mensah FK, et al. 2003. Gastric marginal zone lymphoma is associated with polymorphisms in genes involved in inflammatory response and antioxidative capacity. Blood, 102(3), 1007–​1011. PMID: 12676777. Rollison DE, Engels EA, Halsey NA, et al. 2006. Prediagnostic circulating antibodies to JC and BK human polyomaviruses and risk of non-​Hodgkin lymphoma. Cancer Epidemiol Biomarkers Prev, 15(3), 543–​ 550. PMID: 16537714. Romeo J, Warnberg J, Nova E, et  al. 2007. Moderate alcohol consumption and the immune system:  a review. Br J Nutr, 98 Suppl 1, S111–​S115. PMID: 17922947. Rossi D, Sala L, Minisini R, et  al. 2009a. Occult hepatitis B virus infection of peripheral blood mononuclear cells among treatment-​naive patients

with chronic lymphocytic leukemia. Leuk Lymphoma, 50(4), 604–​611. PMID: 19373659. Rossi D, Sozzi E, Puma A, et al. 2009b. The prognosis of clinical monoclonal B cell lymphocytosis differs from prognosis of Rai 0 chronic lymphocytic leukaemia and is recapitulated by biological risk factors. Br J Haematol, 146(1), 64–​75. PMID: 19438485. Rothman N, Skibola CF, Wang SS, et al. 2006. Genetic variation in TNF and IL10 and risk of non-​Hodgkin lymphoma: a report from the InterLymph Consortium. Lancet Oncol, 7(1), 27–​38. PMID: 16389181. Roulland S, Faroudi M, Mamessier E, et al. 2011. Early steps of follicular lymphoma pathogenesis. Adv Immunol, 111, 1–​46. PMID: 21970951. Roulland S, Kelly RS, Morgado E, et al. 2014. t(14;18) translocation: a predictive blood biomarker for follicular lymphoma. J Clin Oncol, 32(13), 1347–​1355. PMID: 24687831. Sandin S, Hjalgrim H, Glimelius B, et al. 2006. Incidence of non-​Hodgkin’s lymphoma in Sweden, Denmark, and Finland from 1960 through 2003: an epidemic that was. Cancer Epidemiol Biomarkers Prev, 15(7), 1295–​1300. PMID: 16835326. Sangrajrang S, Renard H, Kuhaprema T, et  al. 2011. Personal use of hair dyes: increased risk of non-​Hodgkin’s lymphoma in Thailand. Asian Pac J Cancer Prev, 12(9), 2393–​2396. PMID: 22296389. Sant M, Allemani C, Tereanu C, et al. 2010. Incidence of hematologic malignancies in Europe by morphologic subtype: results of the HAEMACARE project. Blood, 116(19), 3724–​3734. PMID: 20664057. Sant M, Minicozzi P, Mounier M, et  al. 2014. Survival for haematological malignancies in Europe between 1997 and 2008 by region and age: results of EUROCARE-​5, a population-​based study. Lancet Oncol, 15(9), 931–​ 942. PMID: 25030467 Sava GP, Speedy HE, Di Bernardo MC, et  al. 2014. Common variation at 12q24.13 (OAS3) influences chronic lymphocytic leukemia risk. Leukemia, 29(3), 748–​751. PMCID: PMC4360210. Schaapveld M, Aleman BM, van Eggermond AM, et al. 2015. Second cancer risk up to 40  years after treatment for Hodgkin’s lymphoma. N Engl J Med, 373(26), 2499–​2511. PMID: 26699166. Schinasi L, and Leon ME. 2014. Non-​Hodgkin lymphoma and occupational exposure to agricultural pesticide chemical groups and active ingredients:  a systematic review and meta-​analysis. Int J Environ Res Public Health, 11(4), 4449–​4527. PMCID: PMC4025008. Scott CS, and Jinot J. 2011. Trichloroethylene and cancer: systematic and quantitative review of epidemiologic evidence for identifying hazards. Int J Environ Res Public Health, 8(11), 4238–​4272. PMCID: PMC3228569. Scott DW, and Gascoyne RD. 2014. The tumour microenvironment in B cell lymphomas. Nat Rev Cancer, 14(8), 517–​534. PMID: 25008267. Segel GB, and Lichtman MA. 2004. Familial (inherited) leukemia, lymphoma, and myeloma:  an overview. Blood Cells Mol Dis, 32(1), 246–​ 261. PMID: 14757442. Sellick GS, Goldin LR, Wild RW, et al. 2007. A high-​density SNP genome-​wide linkage search of 206 families identifies susceptibility loci for chronic lymphocytic leukemia. Blood, 110(9), 3326–​3333. PMCID: PMC2200912. Sergentanis TN, Kanavidis P, Michelakos T, and Petridou ET. 2013. Cigarette smoking and risk of lymphoma in adults: a comprehensive meta-​analysis on Hodgkin and non-​Hodgkin disease. Eur J Cancer Prev, 22(2), 131–​ 150. PMID: 22759975. Shaffer AL, Rosenwald A, and Staudt LM. 2002. Lymphoid malignancies: the dark side of B-​cell differentiation. Nat Rev Immunol, 2(12), 920–​932. PMID: 12461565. Shah KV. 2007. SV40 and human cancer: a review of recent data. Int J Cancer, 120(2), 215–​223. PMID: 17131333. Shanafelt TD, Drake MT, Maurer MJ, et al. 2011. Vitamin D insufficiency and prognosis in chronic lymphocytic leukemia. Blood, 117(5), 1492–​1498. PMCID: PMC3056589. Shanafelt TD, Kay NE, Jenkins G, et al. 2009. B-​cell count and survival: differentiating chronic lymphocytic leukemia from monoclonal B-​ cell lymphocytosis based on clinical outcome. Blood, 113(18), 4188–​4196. PMCID: PMC2676080. Shankland KR, Armitage JO, and Hancock BW. 2012. Non-​Hodgkin lymphoma. Lancet, 380(9844), 848–​857. PMID: 22835603. Shiels MS, Engels EA, Linet MS, et al. 2013. The epidemic of non-​Hodgkin lymphoma in the United States:  disentangling the effect of HIV, 1992–​2009. Cancer Epidemiol Biomarkers Prev, 22(6), 1069–​ 1078. PMCID: PMC3698875. Shirley MH, Sayeed S, Barnes I, Finlayson A, and Ali R. 2013. Incidence of haematological malignancies by ethnic group in England, 2001–​7. Br J Haematol, 163(4), 465–​477. PMID: 24033296. Silverberg MJ, Chao C, Leyden WA, et  al. 2011. HIV infection, immunodeficiency, viral replication, and the risk of cancer. Cancer Epidemiol Biomarkers Prev, 20(12), 2551–​2559. PMCID: PMC3237725.

 795



The Non-Hodgkin Lymphomas

Skibola CF. 2007. Obesity, diet and risk of non-​Hodgkin lymphoma. Cancer Epidemiol Biomarkers Prev, 16(3), 392–​395. PMCID: PMC2819159. Skibola CF, Berndt SI, Vijai J, et  al. 2014. Genome-​wide association study identifies five susceptibility loci for follicular lymphyoma outside the HLA region. Am J Hum Genet, 95(4), 462–​471. PMCID: PMC4185120. Skibola CF, Bracci PM, Halperin E, et al. 2009. Genetic variants at 6p21.33 are associated with susceptibility to follicular lymphoma. Nat Genet, 41(8), 873–​875. PMCID: PMC2823809. Skibola CF, Bracci PM, Nieters A, et al. 2010. Tumor necrosis factor (TNF) and lymphotoxin-​alpha (LTA) polymorphisms and risk of non-​Hodgkin lymphoma in the InterLymph Consortium. Am J Epidemiol, 171(3), 267–​276. PMCID: PMC2842204. Skibola CF, Curry JD, and Nieters A. 2007. Genetic susceptibility to lymphoma. Haematologica, 92(7), 960–​969. PMCID: PMC2819165. Slager SL, Benavente Y, Blair A, et al. 2014. Medical history, lifestyle, family history, and occupational risk factors for chronic lymphocytic leukemia/​ small lymphocytic lymphoma:  the InterLymph Non-​ Hodgkin Lymphoma Subtypes Project. J Natl Cancer Inst Monogr, 2014(48), 41–​ 51. PMCID: PMC4155456. Slager SL, Caporaso NE, de Sanjose S, and Goldin LR. 2013. Genetic susceptibility to chronic lymphocytic leukemia. Semin Hematol, 50(4), 296–​302. PMCID: PMC3834539. Slager SL, Rabe KG, Achenbach SJ, et  al. 2011. Genome-​wide association study identifies a novel susceptibility locus at 6p21.3 among familial CLL. Blood, 117(6), 1911–​1916. PMCID: PMC3056639. Slager SL, Skibola CF, Di Bernardo MC, et  al. 2012. Common variation at 6p21.31 (BAK1) influences the risk of chronic lymphocytic leukemia. Blood, 120(4), 843–​846. PMCID: PMC3412347. Smedby KE, Foo JN, Skibola CF, et al. 2011. GWAS of follicular lymphoma reveals allelic heterogeneity at 6p21.32 and suggests shared genetic susceptibility with diffuse large B-​cell lymphoma. PLoS Genet, 7(4), e1001378. PMCID: PMC3080853. Smedby KE, Sampson JN, Turner JJ, et al. 2014. Medical history, lifestyle, family history, and occupational risk factors for mantle cell lymphoma: the InterLymph Non-​Hodgkin Lymphoma Subtypes Project. J Natl Cancer Inst Monogr, 2014(48), 76–​86. PMCID: PMC4155462. Smith A, Crouch S, Lax S et  al. 2015. Lymphoma incidence, survival and prevalence 2004–​2014: sub-​type analyses from the UK’s Haematological Malignancy Research Network. Br J Cancer, 112(9), 1575–​ 1584. PMCID: PMC4453686. Smith MT, Jones RM, and Smith AH. 2007. Benzene exposure and risk of non-​ Hodgkin lymphoma. Cancer Epidemiol Biomarkers Prev, 16(3), 385–​ 391. PMID: 17337645. Speedy HE, Di Bernardo MC, Sava GP, et al. 2014. A genome-​wide association study identifies multiple susceptibility loci for chronic lymphocytic leukemia. Nature Genet, 46(1), 56–​60. PMID: 24292274. Spinelli JJ, Ng CH, Weber JP, et  al. 2007. Organochlorines and risk of non-​ Hodgkin lymphoma. Int J Cancer, 121(12), 2767–​2775. PMID: 17722095. Starzl TE, Nalesnik MA, Porter KA, et al. 1984. Reversibility of lymphomas and lymphoproliferative lesions developing under cyclosporin-​ steroid therapy. Lancet, 1(8377), 583–​587. PMCID: PMC2987704. Steidl C, Lee T, Shah SP, et al. 2010. Tumor-​associated macrophages and survival in classic Hodgkin’s lymphoma. N Engl J Med, 362(10), 875–​885. PMCID: PMC2897174. Steinmaus C, Smith AH, Jones RM, and Smith MT. 2008. Meta-​analysis of benzene exposure and non-​ Hodgkin lymphoma:  biases could mask an important association. Occup Environ Med, 65(6), 371–​378. PMCID: PMC4353490. Strati P, and Shanafelt TD. 2015. Monoclonal B-​cell lymphocytosis and early-​ stage chronic lymphocytic leukemia: diagnosis, natural history, and risk stratification. Blood, 126(4), 454–​462. PMCID: PMC4624440. Suarez F, Mahlaoui N, Canioni D, et  al. 2015. Incidence, presentation, and prognosis of malignancies in ataxia-​telangiectasia:  a report from the French national registry of primary immune deficiencies. J Clin Oncol, 33(2), 202–​208. PMID: 25488969. Sundquist K, Sundquist J, and Ji J. 2014. Risk of hepatocellular carcinoma and cancers at other sites among patients diagnosed with chronic hepatitis B virus infection in Sweden. J Med Virol, 86(1), 18–​22. PMID: 24038002. Surgenor DM, and Schnitzer SS. 1985. The Nation’s Blood Resource: A Summary Report (NIH publication no. 85-​2028). Bethesda, MD: National Institutes of Health. Sweet RA, Cullen JL, and Shlomchik MJ. 2013. Rheumatoid factor B cell memory leads to rapid, switched antibody-​ forming cell responses. J Immunol, 190(5), 1974–​1981. PMCID: PMC3578004. Swerdlow SH, Berger F, Pileri SA, et  al. 2008a. Lymphoplasmacytic lymphoma. In: Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H, Thiele J, and Vardima JW (Eds.), WHO Classification of Tumours

795

of Haematopoietic and Lymphoid Tissues (4th ed., pp. 194–​ 195). Lyon: IARC. Swerdlow SH, Campo E, Harris NL, et  al. 2008b. WHO Classification of Tumours of the Haematopoietic and Lymphoid Tissues. Lyon: IARC Press. Swerdlow SH, Campo E, Pileri SA, et  al. 2016. The 2016 revision of the World Health Organization classification of lymphoid neoplasms. Blood, 127(20), 2375–​2390. PMCID: PMC4874220. Takkouche B, Etminan M, and Montes-​Martinez A. 2005. Personal use of hair dyes and risk of cancer:  a meta-​analysis. JAMA, 293(20), 2516–​2525. PMID: 15914752. Tan DE, Foo JN, Bei JX, et al. 2013. Genome-​wide association study of B cell non-​Hodgkin lymphoma identifies 3q27 as a susceptibility locus in the Chinese population. Nat Genet, 45(7), 804–​807. PMID: 23749188. Teras LR, Gapstur SM, Diver WR, Birmann BM, and Patel AV. 2012. Recreational physical activity, leisure sitting time and risk of non-​Hodgkin lymphoid neoplasms in the American Cancer Society Cancer Prevention Study II Cohort. Int J Cancer, 131(8), 1912–​1920. PMID: 22275172. Teras LR, Gapstur SM, Patel AV, et  al. 2013a. Aspirin and other nonsteroidal anti-​inflammatory drugs and risk of non-​hodgkin lymphoma. Cancer Epidemiol Biomarkers Prev, 22(3), 422–​428. PMID: 23329726. Teras LR, Patel AV, Hildebrand JS, and Gapstur SM. 2013b. Postmenopausal unopposed estrogen and estrogen plus progestin use and risk of non-​ Hodgkin lymphoma in the American Cancer Society Cancer Prevention Study-​II Cohort. Leuk Lymphoma, 54(4), 720–​725. PMID: 22916741. Teras LR, Rollison DE, Pawlita M, et  al. 2015. Epstein-​Barr virus and risk of non-​ Hodgkin lymphoma in the cancer prevention study-​ II and a meta-​ analysis of serologic studies. Int J Cancer, 136(1), 108–​ 116. PMID: 24831943. Theander E, Henriksson G, Ljungberg O, et  al. 2006. Lymphoma and other malignancies in primary Sjogren’s syndrome:  a cohort study on cancer incidence and lymphoma predictors. Ann Rheum Dis, 65(6), 796–​803. PMCID: PMC1798187. Thieblemont C, Bertoni F, Copie-​Bergman C, Ferreri AJ, and Ponzoni M. 2014. Chronic inflammation and extra-​nodal marginal-​zone lymphomas of MALT-​type. Semin Cancer Biol, 24, 33–​42. PMID: 24333758. Thompson CA, Habermann TM, Wang AH, et  al. 2010. Antioxidant intake from fruits, vegetables and other sources and risk of non-​Hodgkin’s lymphoma: the Iowa Women’s Health Study. Int J Cancer, 126(4), 992–​1003. PMCID: PMC2798902. Tio M, Cox MR, and Eslick GD. 2012. Meta-​analysis: coeliac disease and the risk of all-​cause mortality, any malignancy and lymphoid malignancy. Aliment Pharmacol Ther, 35(5), 540–​551. PMID: 22239821 Torres HA, Nevah MI, Barnett BJ, et al. 2012. Hepatitis C virus genotype distribution varies by underlying disease status among patients in the same geographic region: a retrospective multicenter study. J Clin Virol, 54(3), 218–​222. PMID: 22459004. Tramacere I, Pelucchi C, Bonifazi M, et al. 2012. Alcohol drinking and non-​ Hodgkin lymphoma risk: a systematic review and a meta-​analysis. Ann Oncol, 23(11), 2791–​2798. PMID: 22357444. Tranah GJ, Holly EA, and Bracci PM. 2009. Solvent exposure and non-​ Hodgkin lymphoma:  no risk in a population-​based study in the San Francisco Bay Area. Cancer Epidemiol Biomarkers Prev, 18(11), 3130–​ 3132. PMCID: PMC2782419. Travis LB. 2006. The epidemiology of second primary cancers. Cancer Epidemiol Biomarkers Prev, 15(11), 2020–​2026. PMID: 17057028. Tsai HT, Cross AJ, Graubard BI, et al. 2010. Dietary factors and risk of chronic lymphocytic leukemia and small lymphocytic lymphoma:  a pooled analysis of two prospective studies. Cancer Epidemiol Biomarkers Prev, 19(10), 2680–​2684. PMCID: PMC3501724. Tsukasaki K, and Tobinai K. 2014. Human T-​cell lymphotropic virus type I-​ associated adult T-​cell leukemia-​lymphoma:  new directions in clinical research. Clin Cancer Res, 20(20), 5217–​5225. PMID: 25320371. Turner JJ, Morton LM, Linet MS, et al. 2010. InterLymph hierarchical classification of lymphoid neoplasms for epidemiologic research based on the WHO classification (2008): update and future directions. Blood, 116(20), e90–​e98. PMCID: PMC2993636. Vajdic CM, Falster MO, de Sanjose S, et al. 2009. Atopic disease and risk of non-​Hodgkin lymphoma:  an InterLymph pooled analysis. Cancer Res, 69(16), 6482–​6489. PMCID: PMC2758272. Vajdic CM, Landgren O, McMaster ML, et  al. 2014. Medical history, lifestyle, family history, and occupational risk factors for lymphoplasmacytic lymphoma/​ Waldenstrom’s macroglobulinemia:  the InterLymph Non-​Hodgkin Lymphoma Subtypes Project. J Natl Cancer Inst Monogr, 2014(48), 87–​97. PMCID: PMC4155457. Vajdic CM, Mao L, van Leeuwen MT, et  al. 2010. Are antibody deficiency disorders associated with a narrower range of cancers than other forms of immunodeficiency? Blood, 116(8), 1228–​1234. PMID: 20466855.

796

796

PART IV:  Cancers by Tissue of Origin

van Leeuwen MT, Grulich AE, Webster AC, et al. 2009a. Immunosuppression and other risk factors for early and late non-​Hodgkin lymphoma after kidney transplantation. Blood, 114(3), 630–​637. PMID: 19443660. van Leeuwen MT, Turner JJ, Falster MO, et al. 2013. Latitude gradients for lymphoid neoplasm subtypes in Australia support an association with ultraviolet radiation exposure. Int J Cancer, 133(4), 944–​951. PMID: 23382012 van Leeuwen MT, Turner JJ, Joske DJ, et al. 2014. Lymphoid neoplasm incidence by WHO subtype in Australia 1982–​2006. Int J Cancer, 135(9), 2146–​2156. PMID: 24639369. van Leeuwen MT, Vajdic CM, Middleton MG, et  al. 2009b. Continuing declines in some but not all HIV-​associated cancers in Australia after widespread use of antiretroviral therapy. AIDS, 23(16), 2183–​ 2190. PMCID: PMC2873230. van Leeuwen MT, Webster AC, McCredie MR, et al. 2010. Effect of reduced immunosuppression after kidney transplant failure on risk of cancer:  population based retrospective cohort study. BMJ, 340, c570. PMCID: PMC2820609. Veierod MB, Smedby KE, Lund E, Adami HO, and Weiderpass E. 2010. Pigmentary characteristics, UV radiation exposure, and risk of non-​ Hodgkin lymphoma: a prospective study among Scandinavian women. Cancer Epidemiol Biomarkers Prev, 19(6), 1569–​1576. PMID:20501763. Vendrame E, Hussain SK, Breen EC, et al. 2014. Serum levels of cytokines and biomarkers for inflammation and immune activation, and HIV-​associated non-​Hodgkin B-​cell lymphoma risk. Cancer Epidemiol Biomarkers Prev, 23(2), 343–​349. PMCID: PMC3948172. Vermaete NV, Wolter P, Verhoef GE, et al. 2013. Physical activity and risk of lymphoma: a meta-​analysis. Cancer Epidemiol Biomarkers Prev, 22(7), 1173–​1184. PMID: 23677574. Vijai J, Kirchhoff T, Schrader KA, et al. 2013. Susceptibility Loci associated with specific and shared subtypes of lymphoid malignancies. PLoS Genet, 9(1), e1003220. PMCID: PMC3547842. Vijai J, Wang Z, Berndt SI, et al. 2015. A genome-​wide association study of marginal zone lymphoma shows association to the HLA region. Nat Commun, 6, 5751. PMCID: PMC4287989. Villumsen M, Sorup S, Jess T, et al. 2009. Risk of lymphoma and leukaemia after bacille Calmette-​Guerin and smallpox vaccination: a Danish case-​ cohort study. Vaccine, 27(49), 6950–​6958. PMID: 19747577. Vinogradova Y, Coupland C, and Hippisley-​ Cox J. 2011. Exposure to cyclooxygenase-​2 inhibitors and risk of cancer: nested case-​control studies. Br J Cancer, 105(3), 452–​459. PMCID: PMC3172909. Vlaanderen J, Lan Q, Kromhout H, Rothman N, and Vermeulen R. 2011. Occupational benzene exposure and the risk of lymphoma subtypes:  a meta-​analysis of cohort studies incorporating three study quality dimensions. Environ Health Perspect, 119(2), 159–​167. PMCID: PMC3040601. Vockerodt M, Yap LF, Shannon-​ Lowe C, et  al. 2015. The Epstein-​ Barr virus and the pathogenesis of lymphoma. J Pathol, 235(2), 312–​322. PMID: 25294567 Wall R, Ross RP, Fitzgerald GF, and Stanton C. 2010. Fatty acids from fish: the anti-​inflammatory potential of long-​chain omega-​3 fatty acids. Nutr Rev, 68(5), 280–​289. PMID: 20500789. Walter RB, Milano F, Brasky TM, and White E. 2011. Long-​term use of acetaminophen, aspirin, and other nonsteroidal anti-​ inflammatory drugs and risk of hematologic malignancies:  results from the prospective Vitamins and Lifestyle (VITAL) study. J Clin Oncol, 29(17), 2424–​2431. PMCID: PMC3107756. Wang J, Mack TM, Hamilton AS, et  al. 2015a. Common immune-​related exposures/​conditions and risk of non-​Hodgkin lymphoma: a case-​control study of disease-​discordant twin pairs. Am J Epidemiol, 182(5), 417–​425. PMCID: PMC4552265. Wang R, Zhang Y, Lan Q, et al. 2009a. Occupational exposure to solvents and risk of non-​Hodgkin lymphoma in Connecticut women. Am J Epidemiol, 169(2), 176–​185. PMCID: PMC2727253. Wang SS, Carreon JD, Hanchard B, Chanock S, and Hisada M. 2009b. Common genetic variants and risk for non-​Hodgkin lymphoma and adult T-​cell lymphoma/​leukemia in Jamaica. Int J Cancer, 125(6), 1479–​1482. PMID: 19533685.

Wang SS, Deapen D, Voutsinas J, et al. 2015b. Breast implants and anaplastic large cell lymphomas among females in the California Teachers Study cohort. Br J Haematol, 174(3), 480–​483. PMCID: PMC4829485. Wang SS, Flowers CR, Kadin ME, et al. 2014. Medical history, lifestyle, family history, and occupational risk factors for peripheral T-​cell lymphomas: the InterLymph Non-​Hodgkin Lymphoma Subtypes Project. J Natl Cancer Inst Monogr, 2014(48), 66–​75. PMCID: PMC4155466. Wang SS, Vajdic CM, Linet MS, et  al. 2015c. Associations of non-​ Hodgkin Lymphoma (NHL) risk with autoimmune conditions according to putative NHL loci. Am J Epidemiol, 181(6), 406–​421. PMCID: PMC4402340. Ward MH, Cerhan JR, Colt JS, and Hartge P. 2006. Risk of non-​Hodgkin lymphoma and nitrate and nitrite from drinking water and diet. Epidemiology, 17(4), 375–​382. PMID: 16699473. Watson IR, Takahashi K, Futreal PA, and Chin L. 2013. Emerging patterns of somatic mutations in cancer. Nat Rev Genet, 14(10), 703–​ 718. PMCID: PMC4014352. Weisenburger DD. 1992. Pathological classification of non-​Hodgkin’s lymphoma for epidemiological studies. Cancer Res, 52(19 Suppl), 5456s–​ 5462s; discussion 5462s–​5464s. PMID: 1394154. Willett EV, Morton LM, Hartge P, et  al. 2008. Non-​Hodgkin lymphoma and obesity: a pooled analysis from the InterLymph Consortium. Int J Cancer, 122(9), 2062–​2070. PMCID: PMC3928289. Williams CJ, Peyrin-​Biroulet L, and Ford AC. 2014. Systematic review with meta-​ analysis:  malignancies with anti-​ tumour necrosis factor-​ alpha therapy in inflammatory bowel disease. Aliment Pharmacol Ther, 39(5), 447–​458. PMID: 24444171. Wolfe F, and Michaud K. 2007. The effect of methotrexate and anti-​tumor necrosis factor therapy on the risk of lymphoma in rheumatoid arthritis in 19,562 patients during 89,710 person-​years of observation. Arthritis Rheum, 56(5), 1433–​1439. PMID: 17469100. Wong O, and Fu H. 2005. Exposure to benzene and non-​Hodgkin lymphoma, an epidemiologic overview and an ongoing case-​control study in Shanghai. Chem Biol Interact, 153–​154, 33–​41. PMID: 15935798. Wong O, Harris F, Armstrong TW, and Hua F. 2010. A hospital-​based case-​ control study of non-​Hodgkin lymphoid neoplasms in Shanghai: analysis of environmental and occupational risk factors by subtypes of the WHO classification. Chem Biol Interact, 184(1–​2), 129–​146. PMID: 19900422. World Cancer Research Fund and American Institute for Cancer Research. 2007. Food, Nutrition, Physical Activity, and the Prevention of Cancer: A Global Perspective. Washington, DC: AICR. Xochelli A, Kalpadakis C, Gardiner A, et al. 2014. Clonal B-​cell lymphocytosis exhibiting immunophenotypic features consistent with a marginal-​ zone origin: is this a distinct entity? Blood, 123(8), 1199–​1206. PMID: 24300853. Yang TO, Cairns BJ, Kroll ME, et  al. 2016. Body size in early life and risk of lymphoid malignancies and histological subtypes in adulthood. Int J Cancer, 139(1), 42–​49. PMCID: PMC4855635. Yates WB, Vajdic CM, Na R, McCluskey PJ, and Wakefield D. 2015. Malignancy risk in patients with inflammatory eye disease treated with systemic immunosuppressive therapy:  a tertiary referral cohort study. Ophthalmology, 122(2), 265–​273. PMID: 25312044. Zhang S, Hunter DJ, Rosner BA, et al. 1999. Dietary fat and protein in relation to risk of non-​Hodgkin’s lymphoma among women. J Natl Cancer Inst, 91(20), 1751–​1758. PMID: 10528026. Zhang Y, Holford TR, Leaderer B, et  al. 2004. Prior medical conditions and medication use and risk of non-​ Hodgkin lymphoma in Connecticut United States women. Cancer Causes Control, 15(4), 419–​428. PMID: 15141141. Zhang Y, Sanjose SD, Bracci PM, et  al. 2008. Personal use of hair dye and the risk of certain subtypes of non-​Hodgkin lymphoma. Am J Epidemiol, 167(11), 1321–​1331. PMCID: PMC4025953. Zucca E, Bertoni F, Vannata B, and Cavalli F. 2014. Emerging role of infectious etiologies in the pathogenesis of marginal zone B-​cell lymphomas. Clin Cancer Res, 20(20), 5207–​5216. PMID: 25320370.

 79

41

Multiple Myeloma MARK P. PURDUE, JONATHAN N. HOFMANN, ELIZABETH E. BROWN, AND CELINE M. VACHON

OVERVIEW Multiple myeloma (MM) is the most common malignancy arising from plasma cells, fully differentiated B lymphocytes that produce the immunoglobulin (Ig) heavy-​and light-​chain molecules comprising antibodies. MM is characterized by an overproduction of clonal plasma cells in the bone marrow and, in most cases, monoclonal secretion of IgG, IgA, or light-​chain Ig. Symptoms of end organ damage (hypercalcemia [C]‌, renal failure [R], anemia [A], or bone lesions [B]), herein referred to as CRAB features, were traditionally a necessary criterion for diagnosing MM; however, improvements in treatment and diagnostic techniques have led to updated diagnostic criteria, enabling intervention among patients before the onset of organ damage. Multiple myeloma is an important cause of lymphoid malignancy (LM) mortality in Western populations. In the United States in 2015, MM was estimated to account for approximately one in every five newly diagnosed LMs, and one in every three LM-​related deaths. Survival rates have increased substantially over the past two decades, likely attributable to improvements in treatment. Despite these improvements, myeloma remains one of the more lethal LMs, with a 5-​year relative survival rate of 47%. With no established curative therapy for MM, prevention represents a potentially important strategy for reducing deaths from this malignancy, although few modifiable risk factors have been identified. Obesity has been consistently associated with increased MM risk, while several other medical, occupational, and lifestyle factors are also suspected to affect risk. Non-​modifiable risk factors include male sex, African American race, family history of MM, and genetic susceptibility loci. This chapter summarizes our understanding to date of MM pathology, survival, natural history, incidence and mortality patterns, genetics, and etiology. Other, rarer plasma cell malignancies, including plasma cell leukemia and Waldenström macroglobulinemia, are not reviewed in this chapter.

PATHOLOGY Multiple myeloma is a malignancy of post-​germinal center B cells. It is characterized by cellular resistance to apoptosis, leading to enhanced survival and accumulation of clonally expanded, cytogenetically heterogeneous, antibody-​ producing plasma cells in the bone marrow and extramedullary sites (Mahindra et  al., 2010). Although the origin of the myeloma progenitor cell remains controversial (i.e., single clone versus genetically diverse clonal subsets with varying proliferative capacities) (Chapman et al., 2011; Egan et al., 2012), malignant transformation is a multistage process that is highly dependent on the cellular and non-​cellular compartments of the bone marrow microenvironment (Basak et  al., 2009; Hideshima et  al., 2007). Plasma cell activation, survival, and pleiotropic signaling cascades mediate homing and adhesion of clonal plasma cells to stromal cells and the extracellular matrix in the bone marrow microenvironment (Hideshima et al., 2004; Uchiyama et al., 1993). The physical interaction of tumor cells with the bone marrow stromal cells results in an upregulation of cell cycle regulatory and anti-​apoptotic proteins, as well as a paracrine cascade of cytokines, chemokines, growth factors, and inter-​cellular

adhesion molecules, which optimize the bone marrow milieu for continued support of tumor cell growth, survival, and resistance to therapy (Hideshima et al., 2001; Nefedova et al., 2004), as well as osteoclastogenesis (Roodman, 2010) and angiogenesis (Jakob et al., 2006; Ribatti et  al., 2006). Ultimately, this process of myelomagenesis promotes phenotypic changes, including the formation of lytic bone lesions, skeletal destruction and other end organ damage, which are used to characterize symptomatic disease.

DIAGNOSIS AND CLASSIFICATION Multiple myeloma is preceded by two asymptomatic premalignant plasma cell proliferative disorders known as monoclonal gammopathy of undetermined significance (MGUS) and smoldering multiple myeloma (SMM) (Landgren et  al., 2009b; Landgren and Waxman, 2010). At present, MGUS, SMM, and MM are classified on the basis of predominantly three clinicopathologic features including serum monoclonal protein, percent clonal bone marrow plasma cells, and myeloma-​ defining events including end organ damage (Rajkumar et al., 2014) (Table 41–1). The diagnosis of MGUS requires a serum M-​protein < 3 g/​dL, clonal bone marrow plasma cells < 10%, and the absence of myeloma-​defining events such as end organ damage attributed to the plasma cell proliferative disorder (i.e., CRAB features). The defining clinicopathologic features of SMM include a threshold level of serum and/​or urine M-​protein ≥ 3 g/​dL or ≥ 500mg per 24 hours, respectively, and/​or a proportion of clonal plasma cells present in the bone marrow of between 10% and 59% (Rajkumar et al., 2014) (Table 41–1). In contrast, an MM diagnosis requires any one or more myeloma-​defining events, including evidence of end organ damage attributed to the underlying plasma cell proliferative disorder (Rajkumar et  al., 2014) or a biopsy-​proven bone or extramedullary plasmacytoma (Dimopoulos et al., 1999; Hill et al., 2014; Paiva et al., 2014; Warsame et  al., 2012). However, in the absence of end organ damage, clonal bone marrow plasma cells ≥ 60%, serum FLC ratio ≥ 100, or more than one focal bone lesion detected using magnetic resonance imaging (MRI) can constitute the MM diagnosis (Hillengass et al., 2010; Kastritis et al., 2013, 2014; Larsen et al., 2013; Rajkumar et al., 2011) (Table 41–1).

Molecular Classification The myeloma plasma cell genome is characterized by numerous chromosomal abnormalities, including gains and losses of whole chromosomes, chromosomal translocations, and distinct chromosomal deletions and amplifications. Myeloma tumors are broadly divided into two groups based on the pattern of whole-​chromosome gains or losses (Chng et al., 2005; Cremer et al., 2005; Gutierrez et al., 2004; Smadja et al., 1998; Wuilleme et al., 2005). The majority of myeloma tumors are hyperdiploid (trisomies of odd-​numbered chromosomes, giving rise to 48 to 74 chromosomes), whereas the remaining tumors are non-​hyperdiploid (Fonseca et al., 2004). Overall, non-​hyperdiploid myeloma is associated with a worse prognosis than hyperdiploid myeloma (Avet-​Loiseau et  al., 2007; Zhan et  al., 2006). Non-​ hyperdiploid tumors are characterized by reciprocal chromosomal

797

798

798

PART IV:  Cancers by Tissue of Origin

Table 41–1.  Diagnostic Criteria for Monoclonal Gammopathy of Undetermined Significance, Light-​Chain Monoclonal Gammopathy of Undetermined Significance, Smoldering Multiple Myeloma, and Multiple Myeloma

epidemiologic studies should incorporate data on tumor genetic characteristics to explore the possible existence of etiologic heterogeneity across MM subtypes.

Plasma Cell Dyscrasia

The World Health Organization Classification of Lymphoid Malignancies

Case Definition

Monoclonal Gammopathy of Undetermined Significance

• Serum monoclonal protein, < 30 g/​L • Clonal bone marrow plasma cells, < 10% • Absence of end-​organ damage, other myeloma defining event or amyloidosis attributed to the underlying plasma cell proliferative disorder End-​organ damage (CRAB features) includes the following: • Hypercalcemia: serum calcium > 0.25 mmol/​L (> 1mg/​dL) higher than the upper limit of normal or > 2.75 mmol/​L (> 11 mg/​dL) • Renal insufficiency: creatinine clearance < 40 mL per minute or serum creatinine > 177 μmol/​L (> 2 mg/​dL) • Anemia: hemoglobin of > 20 g/​L below the lower limit of normal or < 100 g/​L • Bone lesions: ≥ 1 osteolytic lesion identified using skeletal radiography, CT or PET-​CT

Light-​Chain Monoclonal Gammopathy of Undetermined Significance

• Abnormal serum free light chain ratio (outside of the reference range of 0.26–​1.65, or 0.37–​ 3.10 in individuals with renal failure) and increased serum concentration of the involved light-​chain (free κ > 19.4 mg/​L or free λ > 26.3 mg/​L) • No serum monoclonal protein expression or underlying plasma cell proliferative disorder

Smoldering Multiple Myeloma

• Serum monoclonal protein, ≥ 30 g/​L or urinary monoclonal protein ≥ 500 mg per 24 hours and/​or clonal bone marrow plasma cells 10%–​59% • Absence of end-​organ damage or other myeloma defining event attributed to the underlying plasma cell proliferative disorder

Multiple Myeloma

• Clonal bone marrow plasma cells, ≥ 10% or biopsy-​proven bony or extramedullary plasmacytoma and any one or more of the following myeloma defining events: Myeloma-​defining events: • Evidence of end-​organ damage (CRAB features) attributed to the underlying proliferative plasma cell disorder • Any one or more of the following clinicopathologic features including: • Clonal bone marrow plasma cells ≥ 60% • Serum free light chain ratio ≥ 100 • > 1 focal bone lesions identified by MRI

Adapted from The Lancet Oncology, Vol. 15, Rajkumar et al, “International Myeloma Working Group updated criteria for the diagnosis of multiple myeloma”, e538–​548, 2014, with permission from Elsevier.

translocations of the IgH locus at 14q32.3 (Bergsagel et al., 1996) with other chromosomal regions, most commonly 11q13 (with transcriptional activation of CCND1), 4p16 (CCND3), and 16p23 (MAF) (Fonseca et  al., 2003). Losses or gains of specific chromosomal regions are also used to classify disease and prognosis, including chromosome 13 monosomy (Avet-​Loiseau et al., 1999; Fonseca et al., 2001), deletions of chromosome 17p (Chng et al., 2007; Debes-​Marun et al., 2003) and 1p (Debes-​Marun et al., 2003), and amplification of chromosome 1q (Carrasco et al., 2006; Hanamura et al., 2006). It is not known whether MM molecular subgroups differ in etiology. One study observed differences by sex in patterns of MM genetic characteristics, with IgH translocations more common in women and hyperdiploidy more common in men (Boyd et al., 2011). Future

In the World Health Organization (WHO) classification of tumors of hematopoietic and lymphoid tissues, published in 2001 and updated in 2008, MGUS and plasma cell myelomas (including MM, SMM, and indolent myeloma) are classified within the category of mature B-​cell lymphoid neoplasms (International Agency for Research on Cancer [IARC], 2008). The most notable change to the updated 2008 classification was the recommendation to combine SMM and indolent myeloma into a single category, termed “asymptomatic plasma cell myeloma,” in recognition of the fact that these two asymptomatic conditions fall along a continuum (Campo et al., 2011). On the basis of this cell-​lineage-​based classification, some epidemiologic studies now include MM as a subtype of non-​Hodgkin lymphoma (NHL). However, given the many differences between MM and other lymphomas in descriptive epidemiologic patterns and etiology (Morton et al., 2014), subtypes should be analyzed separately where possible. The current edition of the WHO classification does not further subclassify plasma cell myeloma on the basis of cytogenetic abnormalities, although the WHO working groups recommend that cytogenetic analyses be performed where possible. The inclusion of molecularly defined MM subtypes may be revisited for future editions as the evidence in this area continues to evolve.

STAGES OF MULTIPLE MYELOMA DEVELOPMENT Monoclonal Gammopathy of Undetermined Significance (MGUS) Multiple myeloma evolves from a precursor state in almost all cases (Dispenzieri et al., 2010; Landgren et al., 2009b; Weiss et al., 2009), including MGUS and SMM. Clinical subtypes of MGUS with different progression to intermediate phenotypes and ultimately malignancy have been defined on the basis of the type of immunoglobulin involved: non-​IgM (IgG or IgA) MGUS, IgM MGUS, and light-​chain MGUS (LC-​MGUS). LC-​MGUS is defined as having an abnormal free light chain ratio with increased levels of the involved light chain (κ or λ), but no evidence of immunoglobulin heavy chain expression on immunofixation (Dispenzieri et  al., 2010). The rate of progression and type of eventual malignancy varies by MGUS subtype; IgG MGUS and IgA (non-​IgM) MGUS progress to MM at a low rate of 1% per year (Kyle et  al., 2002). IgM MGUS progresses to the advanced premalignant state of Waldenström macroglobulinemia at a rate of 1.5% per year, and infrequently progresses to IgM MM (Kyle et al., 2003; Schuster et al., 2010). LC-​MGUS, accounting for 20% of MGUS, progresses to a subtype of MM called light-​chain MM (LC-​ MM) at a rate of 0.3% per year (Dispenzieri et al., 2010). MGUS is often found incidentally during the evaluation of a variety of symptoms and disorders. Reported estimates of MGUS prevalence vary widely, ranging from 0.05% to 6.1% (Wadhera and Rajkumar, 2010). This wide variation reflects differences in distributions of age, sex, and race, sensitivity of laboratory technique, and diagnostic criteria. In a large population-​ based study of 21,463 predominantly white residents of Olmsted County, Minnesota, the prevalence of MGUS was estimated at 3.2% among people aged 50 or older (Kyle et al., 2006). In a later report, the prevalence of LC-​MGUS subtype in this same population was found to be 0.8%, increasing the overall MGUS prevalence to 4.2% (Dispenzieri et al., 2010). A recent study of 12,482 participants in the US National Health and Nutritional Examination Survey reported an overall prevalence of 2.4% among persons aged 50+, with a higher prevalence in the North and Midwest regions of the United States compared to the South and West (3.1% vs. 2.1%, respectively) (Landgren et al., 2014).

 79



799

Multiple Myeloma

MGUS prevalence is consistently higher among men than women, and is highest among black populations, moderate among predominantly white populations, and lowest among Asian populations (Kyle et al., 2006; Landgren et al., 2014; Wadhera and Rajkumar, 2010; Wu et al., 2013). The patterns in MGUS prevalence and MM incidence across racial groups likely reflect differences in lifestyle, environmental, and/​ or genetic factors (Landgren et al., 2014; Wu et al., 2013). Few studies have attempted to calculate MGUS incidence; one recent study in Olmsted County estimated the annual incidence at 50 years of age to be 120 and 60 cases per 100,000 persons among men and women, respectively (Therneau et al., 2012). Over the last few years, advances in the understanding of the biology of MGUS, together with large MGUS cohorts, have resulted in models to estimate the risk of progression to MM according to MGUS subtype. In the Mayo Clinic model, an abnormal serum free light chain ratio (i.e., the ratio of free immunoglobulin κ to λ light chains in the serum), non-​IgG MGUS isotype, and a high serum M protein level (≥ 1.5 g/​dL) are three major risk factors for the progression of MGUS to MM. Patients with an abnormal serum FLC ratio, non-​IgG MGUS, and a high serum M protein level (≥ 1.5 g/​dL) have a cumulative risk of progression of 58% at 20  years (high-​risk MGUS), compared to 5% when none of the risk factors was present (low-​risk MGUS), 21% where one is present, and 37% for two (Rajkumar et al., 2005). A Spanish model (PETHEMA) uses multiparameter flow cytometry to identify the proportion of aberrant plasma cells (aPC) in the bone marrow. The features of aPCs include decreased CD38 expression, expression of CD56, and the absence of CD19 and/​or CD45. Risk factors for progression in this model include ≥ 95% aPCs/​total bone marrow plasma cells and DNA aneuploidy (hypo-​or hyerpdiploidy). The presence of 0, 1, or 2 factors correspond with 5-​year cumulative risks of progression from MGUS to MM of 2%, 10%, and 46%, respectively (Perez-​Persona et al., 2007).

Smoldering Multiple Myeloma (SMM) SMM is an intermediate clinical stage between MGUS and MM but is relatively uncommon and generally is associated with a shorter time for progression to MM. The rates of progression from SMM to MM are 10% per year for the first 5 years, 3% per year over the next 5 years, and 1% per year after a decade (Kyle et al., 2007), resulting

in a cumulative risk of progression of 73% at 15  years. However, there is heterogeneity in SMM progression; 50% of SMM are likely to progress to MM within 5 years and likely define a subgroup with early MM without clinical symptoms, while one-​third of SMM will not progress even after 10 years of diagnosis (Kyle et al., 2007) and have a premalignant state similar to MGUS (Rajkumar et al., 2015). The MM definition has recently been revised to include those SMM with the highest likelihood of progression to MM (greater than 80% risk of progression in 2 years) so as not to delay treatment until end organ damage is present (Rajkumar et al., 2015). A randomized clinical trial has shown that treatment of SMM can delay progression to active disease and increase survival (Mateos et al., 2013). There are several other promising biomarkers that with validation will help to further stratify risk of progression for SMM to MM in the future. These include high circulating plasma cells, increases in serum M-​protein within 6 months, t(4;14) chromosome translocation, 1q amplification, 17p deletion, high bone marrow plasma cell proliferative rate, abnormal plasma cell immunophenotype, immunoparesis, and unexplained decrease in creatinine clearance, accompanied by a rise in urinary monoclonal protein or serum free light chain concentrations (Rajkumar et al., 2014). Both MGUS and SMM provide a model to study the conversion of premalignancy to MM malignancy (Rajkumar, 2009). In particular, MGUS is easily detected by blood tests and can be monitored, and therefore offers a noninvasive precursor disorder for use in larger epidemiologic studies.

INCIDENCE AND MORTALITY Demographic Patterns Multiple myeloma makes up less than 1% of all new cancer diagnoses globally, with incidence rates generally higher in more-​versus less-​ developed countries (Figure 41–1) (Popat et  al., 2013). In a review of international cancer incidence patterns between 2003 to 2007, the US black population had the highest incidence of MM in the world, with intermediate rates of MM generally observed in North America, Europe, and Oceania, and lower rates experienced in Latin American and Asian populations (IARC, 2014a). Differential case ascertainment due to varying access to diagnostic tests is likely an important factor

North America

USA, SEER 18: Black USA, SEER 18: Hispanic White USA, SEER 18: non-Hispanic White Canada USA, SEER 18: Asian/Pacific Islander

Asia

Israel, Jews China, Beijing India, Mumbai Japan, Osaka

South America

Europe

Oceania

Brazil, Sao Paulo USA, Puerto Rico Colombia, Cali Costa Rica France, Bas-Rhin Italy, Turin UK, England Switzerland, Vaud Sweden Male Female

Australia, New South Wales New Zealand

0

2

4

6

8

Age-Standardized Incidence (per 100,000)

10

Figure  41–1. Age-​ adjusted (world standard) incidence of multiple myeloma in selected geographic locations, 2003–​2007. Source:  Forman D, Bray F, Brewster DH, Gombe Mbalawa C, Kohler B, Piñeros M, Steliarova-​Foucher E, Swaminathan R and Ferlay J, editors. Cancer incidence in five continents, Vol. X (electronic version). Lyon:  International Agency for Research on Cancer. Available from:  http://​ci5.iarc. fr, accessed April  2015.) Source:  US Surveillance, Epidemiology and End Results Program (SEER). SEER-​18 Research Data.

80

800

PART IV:  Cancers by Tissue of Origin 100

70 60 50 40 30 20

16 Rate per 100,000 person-years

80

White Males White Females Black Males Black Females

14 12 10 8 6 4

10

2

0

0 2000

2002

35

–3 40 9 –4 45 4 –4 50 9 –5 55 4 –5 60 9 –6 65 4 –6 70 9 –7 75 4 –7 80 9 –8 4 85 +

Rate per 100,000 person-years

90

18

Age

Figure 41–2.  Age-​specific multiple myeloma incidence rates in the United States by sex and race (white and black only), 2008–​2012. Source:  US Surveillance, Epidemiology and End Results Program (SEER). SEER-​18 Research Data.

underlying the international variation in incidence, although differences across populations in genetic susceptibility and/​or the prevalence of causal risk factors may also play a role. In the United States, approximately 26,850 cases of MM were diagnosed in 2015, with 11,240 deaths attributable to this malignancy (American Cancer Society [ACS], 2015). Figure 41–2 presents age-​ specific incidence rates of MM between 2008 and 2012 for 18 cancer registries within the US Surveillance, Epidemiology, and End Results (SEER) Program, stratified by sex and race (white and black only) (SEER, 2015a). Incidence increases markedly with age and is higher for men than women, and for blacks than whites. Among non-​black populations, incidence is slightly higher for Hispanic versus non-​Hispanic whites, and lower for Asians (Figure 41–1). In an analysis of US Asians within the California Cancer Registry, MM incidence rates did not notably differ by birthplace (foreign vs. US), neighborhood enclave status, or socioeconomic status (SES) among men or women (Clarke et al., 2011). The reasons for the excess incidence rates of MM and MGUS among blacks compared to other racial groups have not yet been identified. Racial differences in obesity, a risk factor for MM, is one possibility, given the 50% higher prevalence among non-​Hispanic blacks compared to non-​Hispanic whites (Centers for Disease and Prevention, 2009). However, a study of MGUS prevalence within the population-​based Southern Community Cohort Study (SCCS) cohort found the increased risk of MGUS among blacks versus whites to be independent of obesity (Landgren et  al., 2010). Genetic differences between racial populations are also suspected to play a role. One possible contributor is the hyperphosphorylated paratarg-​7 protein carrier state, an autosomal dominantly inherited risk factor for MGUS and MM that is more prevalent among both African American cases and controls compared to Europeans (Zwick et  al., 2014), although the etiologic relevance of this finding is still unclear. Additional studies of genetic and other risk factors across racial groups are needed to better understand the factors underlying this racial disparity.

Trends In this edition of Cancer Epidemiology and Prevention, we summarize the secular trends in MM incidence and mortality within the SEER Program for the periods 2000 through 2012 and 1990 through 2012, respectively (SEER, 2015a); a summary of earlier trends can be

2004

2006 Year

2008

2010

2012

Black Males

Black Females

White Males

White Females

Asian or Pacific Islander Males

Asian or Pacific Islander Females

Figure  41–3. Age-​adjusted (2000 US standard) multiple myeloma incidence trends in the United States by sex and race (white, black, Asian/​ Pacific Islander), 2000–​2012. Source: US Surveillance, Epidemiology and End Results Program (SEER). SEER-​18 Research Data.

found in the previous edition (De Roos et al., 2006). Within the SEER Program, the age-​standardized incidence of MM increased by 7.8% between 2000 and 2012 (annual percent change [APC] 0.8%; 95% confidence interval [CI] = 0.5%, 1.2%). As shown in Figure 41–3, incidence trends were generally similar across race-​and sex-​defined population subgroups, with APCs ranging from 0.3% to 1.2% for black and white men and women and male US Asians; for female Asians, a non-​ significant APC of –​1.1% was observed (95% CI = –​2.8%, 0.7%). In other countries, MM incidence increased between the 1990s and late 2000s in both the United Kingdom and Taiwan, particularly among older age groups, while rates in Malmo, Sweden, remained stable across this time period (Renshaw et al., 2010; Turesson et al., 2010b; Tzeng et al., 2013). The reported increases in some populations may be a reflection of improvements in case ascertainment, particularly in the elderly. Multiple myeloma is an important cause of lymphoid malignancy (LM) mortality in Western populations, accounting for one in every three LM-​related deaths (ACS, 2015). MM mortality rates in the United States declined by 10.4% between 1990 and 2012 (APC  –​ 0.9%; 95% CI = –​1.0%, –​0.7%) (SEER, 2015b). Statistically significant declines in mortality were observed among black and white men and women, with APCs ranging from –​0.7% to –​1.3%, while mortality did not significantly change among Asians (Figure 41–4). It is likely that this decrease at least partly reflects improvements in survival resulting from recent therapeutic advances.

SURVIVAL The case-​fatality rate of MM is high, with a 5-​year relative survival rate of 47% in the United States for the years 2004–​2010 (Siegel et  al., 2015), although response rates, progression-​free survival, and overall survival have increased significantly in past decades (Brenner et al., 2008; Kristinsson et al., 2007; Kumar et al., 2008; Turesson et al., 2010a). While it is possible that lead-​time bias from increased detection of early-​stage MM due to improved diagnostic tests and evolving classification criteria may play a role, the observed increase in survival is likely to also reflect real improvements

 801



Multiple Myeloma 12

Rate per 100,000 person-years

10

8

6

4

2

0

1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 Year Black Males White Females

Black Females Asian or Pacific Islander Males

White Males Asian or Pacific Islander Females

Figure 41–4.  Age-​adjusted (2000 US standard) multiple myeloma mortality trends in the United States by sex and race (white, black, Asian/​Pacific Islander), 1990–​2012. Source: US National Center for Health Statistics.

in targeted therapeutic modalities. In the time period of 1950–​1959, before the introduction of alkylating agents, the median survival of myeloma patients was less than 1 year (Osgood, 1960). In subsequent 10-​year periods through 2005, the median overall survival improved from 24.3 to 56.3 months (Turesson et al., 2010a) following the introduction of interferon, high-​dose treatment with autologous stem cell transplantation (ASCT), and, more recently, therapeutic agents with novel mechanisms of action used both alone and in combination as a first-​line treatment or in patients with relapsed or refractory disease. It is unclear whether MM survival for blacks differs from that of other racial groups. Before the advent of ASCT, retrospective findings from a randomized clinical trial conducted as part of the Southwestern Oncology Group demonstrated no significant difference in survival among blacks and whites, even after adjusting for stage of disease (Modiano et  al., 1996). Similarly, two independent evaluations of patients referred to treatment centers for ASCT found comparable survival by race (Hari et  al., 2010; Verma et  al., 2008). In contrast, data from population-​based SEER studies demonstrate significantly higher disease-​specific and relative survival rates in blacks compared to whites (Waxman et al., 2010). However, in the time periods examined (1973–​2007) using 5-​year relative survival rates before 1994 and after 1999 (selected based on the years that ASCT and thalidomide were introduced), significant improvements in survival were noted in whites (but not in blacks), particularly among patients diagnosed before 70 years of age (Pulte et al., 2012; Waxman et al., 2010). More recently, findings from the Eastern Cooperative Oncology Group show that among patients treated with immunomodulatory drugs, improved overall survival was also observed for blacks versus whites (Rajkumar et  al., 2010). Differences in survival by race may reflect disparities associated with access to health care or differences in underlying tumor biology, such as chromosomal abnormalities (Greenberg et al., 2015) or response to treatment, which are known prognostic factors.

RISK STRATIFICATION AND PROGNOSTIC FACTORS Multiple myeloma prognosis in newly diagnosed patients is associated with chromosomal abnormalities present in clonal myeloma plasma cells, stage of disease, age, race, comorbidities, performance status, and response to treatment (Russell and Rajkumar, 2011). Chromosomal

801

abnormalities are important predictors of prognosis; high-​risk disease, defined by the presence of t(14;16), t(14;20), del(1p), or del(17p) (Avet-​Loiseau et al., 2007; Fonseca et al., 2003; Moreau et al., 2002; Zhan et  al., 2006), is associated with a median overall survival of 24.5 months, whereas standard-​risk disease is defined by the absence of these markers and is associated with a median overall survival of 50.5 months (Table 41–2) (Kumar et al., 2012; Mikhael et al., 2013). Independent of the tumor genome, the Durie-​Salmon (DS) staging system (Durie and Salmon, 1975)  and International Staging System (ISS) (Greipp et al., 2005) have been widely used to classify myeloma prognosis on the basis of several laboratory parameters and radiologic evidence of bone lesions. However, the use of chromosomal abnormalities independent of staging has been insufficient to adequately predict prognosis and to inform clinical decision-​making (Rajkumar, 2014). Therefore, the ISS was recently modified to include serum lactate dehydrogenase (LDH), a marker of high tumor cell proliferation/​ mass, and chromosomal abnormalities in order to improve the positive predictive value of myeloma prognostic factors (Palumbo et al., 2015). Improvements in risk stratification schemas continue to evolve. Future studies that consider the combined effects of chromosomal abnormalities and disease stage together with host factors are warranted.

HOST FACTORS Familial Aggregation Evidence from epidemiologic, family, and genome-​ wide association studies (GWAS) suggests a genetic component underlying MM and MGUS etiology. Several Swedish epidemiological studies have estimated the familial risk of MM to be approximately 2.5 (Altieri et al., 2006; Hemminki et al., 2004; Landgren et al., 2006a). In a large pedigree study of 39 families with multiple cases of MM or related disorders, 17 had MM-​affected members in multiple generations, suggesting an autosomal dominant mode of transmission (Lynch et  al., 2005). A decrease in the age at MM diagnosis in successive generations has been observed in some family studies (Altieri et al., 2006; Deshpande et al., 1998; Grosbois et al., 1999; Lynch et al., 2005) but not others (Daugherty et al., 2005; Landgren et al., 2006a). The data from family studies are consistent with the numerous epidemiologic studies that have investigated family history of MM and show a 2-​to 4-​fold increased risk of MM among those with an affected first-​degree relative (Alexander et al., 2007; Greenberg et al., 2009), resulting in positive family history as one of the few accepted risk factors for MM. Although studied to a lesser extent, family history also may be a risk factor among other races, including US black (Brown et al., 1999; VanValkenburg et al., 2015) and Chinese (Wang et al., 2012) populations. Black families with MM and MGUS in multiple generations have been described (Jain et al., 2009; Lynch et al., 2008). Two population-​based case-​control studies observed a higher risk associated with familial MM for blacks (OR  =  17.3 and 20.9) than whites (OR = 1.5 and 2.04), although the differences were based on small numbers and were not statistically significant (Brown et al., 1999; VanValkenburg et al., 2015). Whether inherited factors contribute to the racial disparity in MM risk remains largely unknown. Studies to date suggest MGUS and MM share a common underlying genetic predisposition. Three large family studies investigating the aggregation of MM and MGUS have observed a 2–​3-​fold increased risk of MGUS in first-​degree relatives of MM or MGUS probands (Kristinsson et al., 2009; Landgren et al., 2009b; Vachon et al., 2009). Findings from these studies are inconsistent as to whether familial risk differs by MGUS isotype. There have also been reports of aggregation of MM and MGUS with other B-​cell malignancies as well as solid tumors (Alexander et  al., 2007; Frank et  al., 2015; Greenberg et  al., 2012). In a large family-​based study in Sweden, first-​degree relatives of MM patients had increased risk of acute lymphoblastic leukemia (relative risk [RR]  =  2.1), while first-​ degree relatives of MGUS patients were found to have an increased risk of lymphoplasmacytic lymphoma/​ Waldenström macroglobulinemia (LPL/​WM) (RR  =  4.0) and CLL

802

Table 41–2.  Multiple Myeloma Prognostic Factors Prognostic Factor Chromosomal abnormality   Standard risk               Intermediate risk   High risk*                         Disease stage   Durie-​Salmon Staging System     I                                         II           III                                       A   B          

International Staging System   I       II    

 

 

                   

Revised-​International Staging System   R-​I               R-​II   R-​III            

III

Host factors                      

   

   

Tumor factors        

 

 

Definition Trisomies (hyperdiploid) t(11;14)(q13;q32), CCND1 t(6;14)(6p21;q32), CCND3 t(4;14)(4p16.3;q32), FGFR3, MMSET del(17p), TP53 t(14;16)(16p23;q32), MAF t(14;20)(16p23;q32), MAFB Chromosome 1: deletion of 1p or amplification of 1q High-​risk signature in gene expression profiling   All of the following: Hemoglobin, >10 g/​dL Serum calcium, normal or 100 kU/​L vs. < 20 kU/​L, OR 0.40; 95% CI  =  0.21–​0.79); this study also assessed levels of specific IgE against inhalant allergens, which were not associated with MM risk. While in the Danish study the association with total IgE was observed after > 10 years of follow-​up, in the EPIC study it was most apparent among those diagnosed within 5 years of blood collection. As such, these findings may be indicative of moderate immunodeficiency among those who are subsequently diagnosed with MM, possibly as a result of the developing tumor. As reviewed in the previous edition of this chapter and elsewhere (Alexander et al., 2007), there is little evidence to suggest that atopic diseases other than allergies are associated with MM risk; recent reports have found no association with either asthma (Brown et  al., 2008; Landgren et  al., 2006c) or hay fever (Landgren et  al., 2006c; Soderberg et al., 2004).

Obesity, Energy Balance, and Metabolic Syndrome

There is considerable evidence that excess body weight is associated with an increased risk of MM. A meta-​analysis of cohort studies observed consistent evidence of a modest but statistically significant positive association between body mass index (BMI) at study entry and future MM risk (per 5 kg/​m2 increase in BMI, summary RR 1.12; 95% CI = 1.08–​1.16) (Wallin and Larsson, 2011), and a similar association with MM mortality was observed in a more recent pooled analysis of prospective studies (HR 1.09; 95% CI = 1.03–​1.16) (Teras et al., 2014). In the latter investigation and in other cohort studies of older adults, high BMI in early adulthood was associated with an increased risk of MM later in life (Hofmann et al., 2013; Teras et al., 2014; Troy et al., 2010). Interestingly, the prevalence of MGUS has been reported to be significantly elevated among obese versus normal-​weight women, further supporting a role for obesity early in MM development (Landgren et al., 2010). Higher waist circumference was also associated with MM incidence in one cohort study (Blair et al., 2005) and with MM mortality in a pooled cohort analysis (Teras et al., 2014), although waist circumference was not associated with MM risk in several other cohorts (Britton et al., 2008; Hofmann et al., 2013; Lu et al., 2010; MacInnis et  al., 2005). Waist-​to-​hip ratio has generally not been associated with risk of MM (Blair et al., 2005; Britton et al., 2008; Hofmann et al., 2013; Lu et al., 2010; MacInnis et al., 2005; Teras et al., 2014). Although the underlying biological mechanisms linking obesity and MM remain unclear, recent evidence suggests a potential role of adiponectin, a metabolic hormone with anti-​inflammatory properties that is typically underexpressed in obese individuals (Dalamaga and Christodoulatos, 2015; Hofmann et al., 2012, 2016). Few other obesity-​or immune-​related biomarkers have been associated with MM risk prospectively, although there is some evidence that prediagnosis circulating levels of fasting insulin-​like growth factor binding protein-​1 and soluble interleukin-​6 receptor are elevated in the years immediately preceding MM diagnosis (Birmann et al., 2012). Another prospective investigation that included a small number of MM cases (n = 39) found no association with future MM risk for plasma levels of soluble (s)CD27 or sCD30, markers of B-​cell activation that have been implicated in the development of other lymphoid malignancies (Hosnijeh et al., 2016). The association between physical activity and MM has been evaluated in several recent cohort studies (Birmann et  al., 2007; Hofmann et al., 2013; Teras et al., 2012), and overall there is limited evidence of an association. In a recent meta-​analysis of physical activity and hematologic cancers (Jochem et al., 2014), no association with MM risk was observed. While there may truly be no relation between physical activity and MM risk, it is also possible that misclassification in the assessment of physical activity might have

805

obscured associations that could be present. With respect to other measures of physical activity, longer leisure time spent sitting (> 6 hours vs. < 3 hours per day) was associated with an increased risk of MM among women, but not among men, in the CPS-​II cohort (Patel et al., 2015; Teras et al., 2012). A recent meta-​ analysis of 10 case-​ control and cohort studies reported a borderline statistically significant increased risk of MM among individuals with a history of type-​2 diabetes mellitus (Castillo et al., 2012). However, this association was not observed consistently across studies, and findings may have been confounded by obesity or other factors.

Female Reproductive Factors

Differences by sex in the incidence of MM has driven interest in whether female reproductive factors (e.g., age at menarche/​menopause, age at first birth, and number of births) or exogenous hormone use might confer protection against MM, although historically the evidence of associations with these factors has been limited (Alexander et  al., 2007). Findings for parity have been inconsistent in recent case-​control and cohort studies, with one case-​control study reporting an increased MM risk in relation to higher parity (Wang et al., 2013), another case-​control study reporting a reduced risk (Costas et al., 2012), and a cohort study finding no evidence of an association (Morton et al., 2009). Whereas two relatively small case-​control studies have reported a reduced risk of MM among postmenopausal hormone replacement therapy users (Altieri et al., 2004; Landgren et al., 2006c), no association was observed in two recent cohort studies (Morton et  al., 2009; Teras et  al., 2013b). A  more recent pooled investigation of case-​control studies in the International Multiple Myeloma Consortium (IMMC) found no association with either reproductive history or hormone therapy use (Costas et  al., 2015). Taken together, these findings suggest that female reproductive history and exogenous hormone use are unlikely to play an important role in the etiology of MM.

ENVIRONMENTAL FACTORS Radiation The epidemiologic evidence evaluating the association between ionizing radiation exposure and MM risk is inconsistent (IARC, 2012b). Early findings from the Life Span Study (LSS) of Atomic Bomb survivors were suggestive of increased MM (Ichimaru et al., 1982; Shimizu et al., 1990). However, subsequent analyses with longer follow-​up and updated diagnostic criteria have shown no association (Hsu et al., 2013; Preston et al., 1994). A re-​review of LSS MM mortality data identified a small number of early high-​dose cases where the assignment of MM as the cause of death was not confirmed by subsequent in-​depth hematological review (Hsu et al., 2013). When these suspect cases were excluded from analysis, no association with MM mortality was observed. As summarized in the previous edition of this book, case-​control and cohort investigations of nuclear industry workers, who experience lower but more protracted doses of ionizing radiation, have generally suggested an increased risk of MM with radiation exposure (De Roos et  al., 2006). More recent cohort findings within the nuclear power industry further support an association (Cardis et al., 2007; Leuraud et al., 2015; Muirhead et al., 2009; Schubauer-​Berigan et al., 2015). No clear evidence of association with MM has been observed in cohort investigations of other radiation-​exposed occupations, such as uranium miners exposed to radon and radiology technologists (Darby et  al., 1995; Rericha et al., 2006; Schubauer-​Berigan et al., 2009; Vacquier et al., 2008), although an excess number of MM deaths was reported in a cohort of female radium dial workers (Stebbings et al., 1984). Studies of diagnostic and therapeutic irradiation have generated inconsistent evidence of an association with MM. An increased relative risk of MM mortality (RR 1.62; 95% CI = 1.07–​2.46) was observed in a cohort study of 15,577 ankylosing spondylitis patients, of whom over 90% had received X-​ray treatment (Weiss et al., 1994). A subsequent linkage-​based analysis of over 4 million male US veterans also

806

806

PART IV:  Cancers by Tissue of Origin

observed significant associations with a prior diagnosis of ankylosing spondylitis for both MM (RR 2.29; 95% CI = 1.55–​3.40) and MGUS (RR 2.02; 95% CI = 1.14–​3.56) (Brown et al., 2008). However, it is unclear from these studies whether the increased risk is a consequence of irradiation versus potential promyelomagenic effects arising from the condition itself (O’Neill et al., 1997). A statistically significant excess of MM mortality was observed in a cohort study of 2067 women given X-​ray therapy for metropathia haemorrhagica (SMR 2.59; 95% CI = 1.19–​4.92) (Darby et al., 1994), although findings from a subsequent investigation of women who underwent irradiation for benign gynecologic conditions do not support an association (Sakata et al., 2012). Similarly, a SEER analysis of pelvic radiotherapy and second-​primary cancer risk did not observe an association for MM (Wright et al., 2010). Few studies have investigated the association between MM and exposure to sunlight and ultraviolet radiation. No consistent pattern of association with sun exposure or tanning bed use was observed in two case-​control studies (Boffetta et al., 2008; Grandin et al., 2008). Higher residential solar UV levels were associated with reduced MM risk in one US cohort (Chang et al., 2011), but not another (Lin et al., 2012). In summary, the published epidemiologic evidence to date does not clearly support an association between radiation exposure and MM risk, although the findings for nuclear industry workers warrant continued investigation.

Occupational Exposures Occupations

An excess of MM has been observed in certain occupations, in particular among individuals working in various agricultural settings (Lope et al., 2008; Perrotta et al., 2008, 2012, 2013), as discussed in greater detail in the following. Some but not all studies have also found an elevated MM risk among cleaning workers (Perrotta et al., 2012, 2013), machinists and metal processers (Ghosh et al., 2011; Gold et al., 2010; Perrotta et al., 2013), carpenters (Ghosh et al., 2011), food and beverage service workers (Gold et al., 2010), and bakers (Lope et al., 2008).

Agricultural Work

A summary of 21 case-​control studies published between 1970 and 2007 noted fairly consistent evidence of an increased risk of MM among farmers (Perrotta et al., 2008). More recently, an investigation in a multicenter case-​control study in six European countries observed a statistically significant increased risk of MM among farmers (OR 1.8; 95% CI = 1.1–​3.0) (Perrotta et al., 2012). A larger pooled analysis including five case-​control studies found that risk of MM was somewhat elevated among gardeners and nursery workers (OR 1.5; 95% CI = 0.9–​2.3) but not among farmers (Perrotta et al., 2013). In a historical cohort that included almost 3 million workers in Sweden (Lope et  al., 2008), an increased risk of MM was observed among men employed in the agricultural sector, and risk was somewhat elevated among women with agricultural occupations. Furthermore, an elevated prevalence of MGUS has been observed among US farmers in the Agricultural Health Study (AHS), a large prospective cohort in Iowa and North Carolina (Landgren et  al., 2009a), and in a French agricultural cohort (Lecluse et al., 2015). Limited data are available for risk of MM or MGUS in relation to specific agricultural exposures (e.g., individual pesticides, livestock or other animals, organic dusts and allergens, diesel exhaust, and solvents). A meta-​analysis by Perrotta et al. (2008) found that MM risk was increased among those ever occupationally exposed to pesticides (summary OR 1.47; 95% CI = 1.11–​1.94), and another meta-​analysis by Merhi et al. (2007) that focused on reports published more recently (between 1990 and 2005)  found a borderline statistically significant association with occupational pesticide exposure (summary OR 1.16; 95% CI = 0.99–​1.36). Findings from a population-​based case-​control study among men residing in Canada suggest an elevated risk of MM among those exposed to carbamate insecticides and fungicides (Kachuri et al., 2013; Pahwa et al., 2012). In the AHS, an increased risk of MM was observed among farmers exposed to permethrin, a

widely used pyrethroid insecticide (high lifetime-​ days of use vs. unexposed, RR 3.1; 95% CI = 1.5–​6.2; Ptrend = 0.002) (Alavanja et al., 2014). A  recent report from the US Air Force Health Study noted an elevated prevalence of MGUS among Vietnam War veterans who were exposed to Agent Orange (a defoliant contaminated with the carcinogen 2,3,7,8-​tetrachlorodibenzo-​p-​dioxin) while conducting aerial herbicide spray missions (Landgren et al., 2015). With respect to agricultural exposures other than pesticides, raising sheep has also been associated with an increased risk of MM in the AHS (Beane Freeman et al., 2012) and elsewhere (Baris et al., 2004; Perrotta et al., 2008), and exposure to other farm animals may also be associated with MM (Perrotta et al., 2008; Svec et al., 2005). Growing up on a farm was associated with a non-​statistically significant elevated risk of MM among farmers and their spouses in the AHS (Hofmann et al., 2015).

Benzene

Benzene is a flammable, volatile chemical widely used in many industries to produce chemicals, plastics, resins, and other materials, among other applications. It is also naturally found in crude oil, and is a component of gasoline engine exhaust and cigarette smoke. Benzene is classified by the International Agency for Research on Cancer (IARC) as a known human carcinogen based on evidence that exposure causes acute myeloid leukemia (AML) (IARC, 2012a). The IARC review also noted evidence of a positive association with other hematopoietic malignancies, including MM. A  2011 meta-​analysis of cohort studies investigating occupational benzene exposure observed a weak non-​significant association with MM overall (RR 1.12; 95% CI = 0.98–​1.27), although stronger summary associations were observed in analyses restricted to studies that had observed an association with AML (RR 1.56; 95% CI = 1.11–​2.21) and with better exposure assessment quality (RR 1.48; 95% CI = 0.96–​2.27) (Vlaanderen et al., 2011). A recent case-​cohort study of Norwegian offshore oil industry workers provided further support for an association, demonstrating a dose–​repose relationship with cumulative exposure (RR 3.25; 95% CI = 1.00–​10.00 for highest tertile vs. unexposed; Ptrend = 0.024) (Stenehjem et al., 2015). The evidence from case-​ control studies regarding benzene exposure and MM is largely null, although most studies employed relatively crude exposure assessment methods. Two case-​control studies that collected detailed benzene-​related occupational data and employed expert-​based exposure assessment found weak, non-​significant associations with MM at high exposure (Cocco et al., 2010; Costantini et al., 2008).

Chlorinated Solvents

Trichloroethylene (TCE) and other chlorinated solvents have been employed for many decades for metal degreasing, dry cleaning, and other industrial applications, although concerns over the health effects and environmental impact have led to a decline in their use. TCE has been classified by IARC as a known human carcinogen causing kidney cancer, although associations with liver cancer and non-​Hodgkin lymphoma have also been observed (IARC, 2014b). The overall epidemiologic evidence for MM does not support an association with TCE exposure, with a recent meta-​analysis involving findings from 11 studies (9 cohort, 2 case-​control) having observed a summary RR of 1.05 (95% CI = 0.88–​1.27) (Karami et al., 2013). Three case-​control studies of MM assessed exposure to TCE and other chlorinated solvents through an expert industrial hygienist review of detailed workplace exposure data (Cocco et  al., 2010; Gold et  al., 2011; Seidler et  al., 2007). One study observed exposure–​response relationships for TCE, perchloroethylene, and 1,1,1-​trichloroethane (Gold et  al., 2011). The other two studies, however, observed no association with TCE or perchloroethylene (Cocco et  al., 2010; Seidler et  al., 2007). A  pooled analysis of three Nordic cohorts of TCE-​exposed workers also observed no association with MM (Hansen et al., 2013).

Hair Dye Use As described in the previous edition of this book (De Roos et  al., 2006)  and reviewed more recently elsewhere (Baris et  al., 2013), there is little evidence of an association between hair dye use and MM risk. In a meta-​analysis of studies published prior to January 2005

 807



Multiple Myeloma

(Takkouche et al., 2005), no overall association between hair dye use and MM was observed. More recent reports on the relation between hair dye use and MM in case-​control (de Sanjose et al., 2006; Koutros et al., 2009) and cohort (Mendelsohn et al., 2009) studies have been similarly null.

Medication Use There is limited evidence to suggest that the use of certain over-​the-​ counter or prescription medications may influence the risk of MM development. In an investigation of aspirin use and MM risk in the Health Professionals Follow-​up Study and the Nurses’ Health Study, Birmann et  al. (2013) reported a reduced future risk of MM among participants in the highest categories of cumulative average use (≥ 5 adult strength pills/​week, HR 0.61; 95% CI = 0.39–​0.94) and duration of regular use (≥ 11 years, HR 0.63; 95% CI = 0.41–​0.95) of aspirin compared with non-​users. In contrast, an investigation in the CPS-​ II cohort by Teras et  al. (2013a) observed an increased risk of MM among participants in the highest category of current aspirin use (60 + pills/​month regardless of product strength) compared to those with no reported non-​steroidal anti-​inflammatory drug (NSAID) use (HR 1.97; 95% CI  =  1.06–​3.67). In the Vitamins and Lifestyle (VITAL) cohort, Walter et al. (2011) observed an increased risk of plasma cell disorders (predominantly MM) in relation to use of acetaminophen in the 10  years prior to baseline, but little evidence of associations with long-​term use of aspirin, non-​aspirin NSAIDs, or ibuprofen. With respect to the evidence from case-​control studies, no associations were observed with regular aspirin use (Landgren et al., 2006c; Moysich et  al., 2007)  or with ever use (Nuyujukian et  al., 2014)  or regular use (Landgren et al., 2006c) of NSAIDs. Regular use of acetaminophen was evaluated in one of these studies (Moysich et  al., 2007)  and was associated with increased MM risk; this association was strongest among those with greater frequency and duration of acetaminophen use. In a case-​control study conducted among women in Connecticut (Landgren et al., 2006c), use of antilipid statins was associated with a reduced risk of MM, and MM risk was increased in relation to use of prednisone, insulin, and gout medication. The association with statin use was not confirmed in a subsequent meta-​analysis of findings from three studies of MM (Yi et al., 2014), and another recent report from a case-​control study in Los Angeles County (Nuyujukian et al., 2014)  found no association between history of steroid use and MM risk, although data on prednisone use specifically were not available. Use of erythromycin, which has been implicated previously for MM (Selby et  al., 1989), was associated with MM risk among men but not among women in the Los Angeles County case-​control study (Nuyujukian et  al., 2014). A  prospective registry-​ based study in Denmark (Rasmussen et al., 2012) observed an increased risk of MM in relation to history of antibiotic use, in particular for use of macrolides (such as erythromycin). However, this association was restricted to the two years prior to MM diagnosis, suggesting a potential role of prodromal immune-​related effects of MGUS or undetected MM (Rasmussen et al., 2012).

807

MM was not associated with consumption of red meat, processed meat, offals, poultry, eggs, dairy products, or vegetables in an analysis within the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort, although increased total fruit consumption was associated with reduced MM risk (OR 0.60; 95% CI = 0.37–​0.98 for fourth vs. first quartile; Ptrend = 0.01) (Rohrmann et al., 2007, 2011). In a meta-​analysis of five case-​control studies, frequent fish consumption was associated with a reduced risk of MM (summary OR 0.65; 95% CI = 0.46–​0.91 for highest intake category vs. lowest) (Wang et  al., 2015). A subsequent pooled analysis of two cohorts observed participants who ate fish but not meat to have a weak reduced relative risk of MM compared to meat-​eating participants, although the confidence intervals were wide and included the null (RR 0.80; 95% CI = 0.35–​1.81) (Key et  al., 2014). In other cohort investigations, a borderline statistically significant association with dietary acrylamide intake was observed for men (RR 1.14; 95% CI = 1.01–​1.27 per 10 ug acrylamide/​day) but not women (RR 0.92; 95% CI = 0.77–​1.11) in the Netherlands Cohort Study on Diet and Cancer (Bongers et al., 2012), while investigations of aspartame and non-​sugar sweeteners in four cohorts have shown no consistent evidence of an association with MM (Lim et al., 2006; McCullough et al., 2014; Schernhammer et al., 2012).

Smoking and Alcohol Smoking has been conclusively shown not to be a risk factor for MM (Andreotti et al., 2015; Kroll et al., 2012; Psaltopoulou et al., 2013). The overall epidemiologic evidence for alcohol consumption is suggestive of a weak protective effect. A meta-​analysis of 26 cohort and case-​control studies reported a summary relative risk of 0.88 (95% CI = 0.79–​0.99) for ever consumption of alcohol (Psaltopoulou et al., 2015). In a pooled analysis of six case-​control studies, the inverse association was stronger when comparing current to never drinkers (OR 0.57; 95% CI = 0.68–​0.95) and did not materially differ by type of alcoholic beverage, although no exposure–​response relationships with increasing frequency or duration of alcohol consumption were observed (Andreotti et  al., 2013). Arguably the strongest epidemiologic evidence regarding alcohol consumption and MM risk comes from an investigation within the UK Million Women Study, including over 1500 incident cases (Kroll et al., 2012). In this analysis, a weak, statistically significant reduction in relative risk was observed per 10 g increase in alcohol consumption per day, both overall (RR 0.86; 95% CI = 0.77–​0.96) and after excluding the first 3 years of follow-​up (RR 0.83; 95% CI = 0.74–​0.94). A similar inverse association with alcohol consumption has been observed for other B-​cell malignancies (Kroll et al., 2012; Morton et al., 2005). It is unclear how alcohol consumption would protect against MM; some suggested mechanisms associated with light to moderate consumption that have been proposed include the inhibition in expression of proinflammatory cytokines and chemokines and stimulation of DNA repair (Andreotti et  al., 2013). The possibility of confounding from other exposures as an explanation for this association cannot be ruled out.

OPPORTUNITIES FOR PREVENTION Diet Early investigations of dietary factors and MM were mainly conducted within case-​control studies, the findings from which have been inconsistent (De Roos et al., 2006). Over the past decade, dietary analyses involving MM from a small number of cohort studies have been published. In a recent pooled analysis of two cohorts, vegetarians were found to have a 77% lower risk of MM compared to meat-​eating participants (RR 0.23; 95% CI = 0.09–​0.60) (Key et al., 2014). However, the epidemiologic evidence regarding vegetable and meat consumption and MM risk from other cohort studies is equivocal. Within the NIH–​AARP Diet and Health Study cohort, MM was weakly associated with consumption of processed meat (RR 1.30; 95% CI = 0.98–​ 1.71 for fifth vs. first quartile; Ptrend = 0.01), but not with intake of red meat, dairy food, or total calcium (Cross et al., 2007; Park et al., 2009).

With no established curative therapy for MM, prevention represents a potentially important strategy for reducing deaths from this malignancy, although few modifiable risk factors have been identified. Given the large body of evidence relating excess weight to increased risk, initiatives to reduce obesity, in principle, represent one opportunity for prevention of MM, along with a host of other obesity-​related cancers and chronic diseases. However, the lifestyle, economic, and societal barriers to reducing obesity are manifest; no country has achieved a significant reduction in obesity prevalence over the past three decades (Ng et al., 2014). There is tremendous interest in the potential to intervene among patients with MGUS and SMM most likely to progress to active disease (Korde et  al., 2011; Rajkumar, 2009). At present there is inadequate evidence to support the adoption of such interventions,

80

808

PART IV:  Cancers by Tissue of Origin

although promising findings were reported from a randomized trial of SMM patients, in which treatment with lenalidomide and dexamethasone was associated with significantly longer progression-​free and overall survival (Mateos et al., 2013). Several other clinical trials are ongoing.

FUTURE RESEARCH There are several promising research directions with the potential to provide important advances in our understanding of the etiology and pathogenesis of MM. With the recognition that nearly all MM is preceded by MGUS, there is a need for more studies seeking to identify factors associated with discrete stages of myelomagenesis, such as MGUS development or progression from MGUS and SMM to MM; findings from such research may help to advance our understanding of MM pathogenesis, improve clinical risk stratification models, and identify targets for early intervention. Continued research in germline and somatic genomics, utilizing new technological and bioinformatic resources and larger numbers of population-​and family-​based studies, is likely to identify new genes associated with MM susceptibility, response to treatment, and prognosis. Additional research into molecular subtypes of MM, extended to investigations of epidemiologic and genetic risk factors, could potentially uncover etiologically distinct patterns across disease subgroups. Molecular epidemiologic investigations utilizing banked biospecimens within cohorts may also provide important insight into biologic mechanisms underlying risk factors for MM. Finally, more classical, molecular, and genetic epidemiologic studies including black participants are needed in order to understand the basis for the high incidence of MM in this racial group. References ACS. 2015. Cancer facts & ­figures 2015. Atlanta: American Cancer Society. Alavanja MC, Hofmann JN, Lynch CF, et al. 2014. Non-​hodgkin lymphoma risk and insecticide, fungicide and fumigant use in the agricultural health study. PLoS One, 9(10), e109332. PMCID: PMC4206281. Alexander DD, Mink PJ, Adami HO, et al. 2007. Multiple myeloma: a review of the epidemiologic literature. Int J Cancer, 120(Suppl 12), 40–​61. PMID: 17405120. Altieri A, Chen B, Bermejo JL, Castro F, and Hemminki K. 2006. Familial risks and temporal incidence trends of multiple myeloma. Eur J Cancer, 42(11), 1661–​1670. PMID: 16753294. Altieri A, Gallus S, Franceschi S, et al. 2004. Hormone replacement therapy and risk of lymphomas and myelomas. Eur J Cancer Prev, 13(4), 349–​ 351. PMID: 15554564. Anderson LA, Gadalla S, Morton LM, et al. 2009. Population-​based study of autoimmune conditions and the risk of specific lymphoid malignancies. Int J Cancer, 125(2), 398–​405. PMCID: PMC2692814. Andreotti G, Birmann B, De Roos AJ, et al. 2013. A pooled analysis of alcohol consumption and risk of multiple myeloma in the international multiple myeloma consortium. Cancer Epidemiol Biomarkers Prev, 22(9), 1620–​ 1627. PMCID: PMC3769449. Andreotti G, Birmann BM, Cozen W, et al. 2015. A pooled analysis of cigarette smoking and risk of multiple myeloma from the international multiple myeloma consortium. Cancer Epidemiol Biomarkers Prev, 24(3), 631–​ 634. PMCID: PMC4355157. Avet-​Loiseau H, Attal M, Moreau P, et al. 2007. Genetic abnormalities and survival in multiple myeloma: the experience of the Intergroupe Francophone du Myelome. Blood, 109(8), 3489–​3495. PMID: 17209057. Avet-​Loiseau H, Li JY, Morineau N, et al. 1999. Monosomy 13 is associated with the transition of monoclonal gammopathy of undetermined significance to multiple myeloma. Intergroupe Francophone du Myelome. Blood, 94(8), 2583–​2589. PMID: 10515861. Baris D, Brown LM, Andreotti G, and Devesa SS. 2013. Epidemiology of multiple myeloma. In: Wiernik PH, Goldman JM, Dutcher JP, and Kyle RA (Eds.), Neoplastic diseases of the blood, 5th ed. (pp. 547–​563). New York: Springer. Baris D, Silverman DT, Brown LM, et al. 2004. Occupation, pesticide exposure and risk of multiple myeloma. Scand J Work Environ Health, 30(3), 215–​ 222. PMID: 15250650.

Basak GW, Srivastava AS, Malhotra R, and Carrier E. 2009. Multiple myeloma bone marrow niche. Curr Pharm Biotechnol, 10(3), 345–​346. PMID: 19355944. Beane Freeman LE, Deroos AJ, Koutros S, et al. 2012. Poultry and livestock exposure and cancer risk among farmers in the agricultural health study. Cancer Causes Control, 23(5), 663–​670. PMCID: PMC3337970. Bergsagel PL, Chesi M, Nardini E, et al. 1996. Promiscuous translocations into immunoglobulin heavy chain switch regions in multiple myeloma. Proc Natl Acad Sci U S A, 93(24), 13931–​13936. PMCID: PMC19472. Biggar RJ, Kirby KA, Atkinson J, et  al. 2004. Cancer risk in elderly persons with HIV/​AIDS. J Acquir Immune Defic Syndr, 36(3), 861–​868. PMID: 15213571. Birmann BM, Giovannucci E, Rosner B, Anderson KC, and Colditz GA. 2007. Body mass index, physical activity, and risk of multiple myeloma. Cancer Epidemiol Biomarkers Prev, 16(7), 1474–​ 1478. PMCID: PMC3860750. Birmann BM, Giovannucci EL, Rosner BA, and Colditz GA. 2013. Regular aspirin use and risk of multiple myeloma: a prospective analysis in the Health Professionals Follow-​up Study and Nurses’ Health Study. Cancer Prev Res, 7(1), 33–​41. Birmann BM, Neuhouser ML, Rosner B, et al. 2012. Prediagnosis biomarkers of insulin-​like growth factor-​1, insulin, and interleukin-​6 dysregulation and multiple myeloma risk in the Multiple Myeloma Cohort Consortium. Blood, 120(25), 4929–​4937. PMCID: PMC3525019. Blair CK, Cerhan JR, Folsom AR, and Ross JA. 2005. Anthropometric characteristics and risk of multiple myeloma. Epidemiology, 16(5), 691–​694. PMID: 16135948. Boffetta P, van der Hel O, Kricker A, et al. 2008. Exposure to ultraviolet radiation and risk of malignant lymphoma and multiple myeloma:  a multicentre European case-​control study. Int J Epidemiol, 37(5), 1080–​1094. PMID: 18511490. Bongers ML, Hogervorst JG, Schouten LJ, et al. 2012. Dietary acrylamide intake and the risk of lymphatic malignancies: the Netherlands Cohort Study on diet and cancer. PLoS One, 7(6), e38016. PMCID: PMC3377662. Boyd KD, Ross FM, Chiecchio L, et al. 2011. Gender disparities in the tumor genetics and clinical outcome of multiple myeloma. Cancer Epidemiol Biomarkers Prev, 20(8), 1703–​1707. PMCID: PMC4545514. Brenner H, Gondos A, and Pulte D. 2008. Recent major improvement in long-​ term survival of younger patients with multiple myeloma. Blood, 111(5), 2521–​2526. PMID: 17901246. Britton JA, Khan AE, Rohrmann S, et  al. 2008. Anthropometric characteristics and non-​Hodgkin’s lymphoma and multiple myeloma risk in the European Prospective Investigation into Cancer and Nutrition (EPIC). Haematologica, 93(11), 1666–​1677. PMID: 18835833. Broderick P, Chubb D, Johnson DC, et al. 2012. Common variation at 3p22.1 and 7p15.3 influences multiple myeloma risk. Nat Genet, 44(1), 58–​61. PMID: 22120009. Brown LM, Gridley G, Check D, and Landgren O. 2008. Risk of multiple myeloma and monoclonal gammopathy of undetermined significance among white and black male United States veterans with prior autoimmune, infectious, inflammatory, and allergic disorders. Blood, 111(7), 3388–​3394. PMCID: PMC2275008. Brown LM, Linet MS, Greenberg RS, et al. 1999. Multiple myeloma and family history of cancer among blacks and whites in the U.S. Cancer, 85(11), 2385–​2390. PMID: 10357409. Camp NJ, Werner TJ, and Cannon-​Albright LA. 2008. Familial myeloma. New Engl J Med, 359(16), 1734–​1735. PMID: 18923179. Campa D, Martino A, Varkonyi J, et al. 2015. Risk of multiple myeloma is associated with polymorphisms within telomerase genes and telomere length. Int J Cancer, 136(5), E351–​358. PMID: 25066524. Campo E, Swerdlow SH, Harris NL, et  al. 2011. The 2008 WHO classification of lymphoid neoplasms and beyond: evolving concepts and practical applications. Blood, 117(19), 5019–​5032. PMCID: PMC3109529. Cardis E, Vrijheid M, Blettner M, et al. 2007. The 15-​Country Collaborative Study of Cancer Risk among Radiation Workers in the Nuclear Industry: estimates of radiation-​related cancer risks. Radiat Res, 167(4), 396–​416. PMID: 17388693. Carrasco DR, Tonon G, Huang Y, et  al. 2006. High-​resolution genomic profiles define distinct clinico-​pathogenetic subgroups of multiple myeloma patients. Cancer Cell, 9(4), 313–​325. PMID: 16616336. Castillo JJ, Mull N, Reagan JL, Nemr S, and Mitri J. 2012. Increased incidence of non-​Hodgkin lymphoma, leukemia, and myeloma in patients with diabetes mellitus type 2:  a meta-​analysis of observational studies. Blood, 119(21), 4845–​4850. PMCID: PMC3367891. Centers for Disease Control and Prevention (CDC). 2009. Differences in prevalence of obesity among black, white, and Hispanic adults:  United

 809



Multiple Myeloma

States, 2006–​2008. MMWR Morb Mortal Wkly Rep, 58(27), 740–​744. PMID: 19609247. Chang ET, Canchola AJ, Cockburn M, et al. 2011. Adulthood residential ultraviolet radiation, sun sensitivity, dietary vitamin D, and risk of lymphoid malignancies in the California Teachers Study. Blood, 118(6), 1591–​ 1599. PMCID: PMC3156047. Chapman MA, Lawrence MS, Keats JJ, et al. 2011. Initial genome sequencing and analysis of multiple myeloma. Nature, 471(7339), 467–​472. PMCID: PMC3560292. Chng WJ, Price-​Troska T, Gonzalez-​Paz N, et al. 2007. Clinical significance of TP53 mutation in myeloma. Leukemia, 21(3), 582–​584. PMID: 17215851. Chng WJ, Van Wier SA, Ahmann GJ, et al. 2005. A validated FISH trisomy index demonstrates the hyperdiploid and nonhyperdiploid dichotomy in MGUS. Blood, 106(6), 2156–​2161. PMCID: PMC1895145. Chubb D, Weinhold N, Broderick P, et al. 2013. Common variation at 3q26.2, 6p21.33, 17p11.2 and 22q13.1 influences multiple myeloma risk. Nat Genet, 45(10), 1221–​1225. PMID: 23955597. Clarke CA, Glaser SL, Gomez SL, et al. 2011. Lymphoid malignancies in U.S. Asians: incidence rate differences by birthplace and acculturation. Cancer Epidemiol Biomarkers Prev, 20(6), 1064–​1077. PMCID: PMC3111874. Cocco P, t’Mannetje A, Fadda D, et al. 2010. Occupational exposure to solvents and risk of lymphoma subtypes: results from the Epilymph case-​control study. Occup Environ Med, 67(5), 341–​347. PMID: 20447988. Costantini AS, Benvenuti A, Vineis P, et al. 2008. Risk of leukemia and multiple myeloma associated with exposure to benzene and other organic solvents: evidence from the Italian Multicenter Case-​control study. Am J Ind Med, 51(11), 803–​811. PMID: 18651579. Costas L, Casabonne D, Benavente Y, et  al. 2012. Reproductive factors and lymphoid neoplasms in Europe: findings from the EpiLymph case-​control study. Cancer Causes Control, 23(1), 195–​206. PMID: 22116538. Costas L, Lambert BH, Birmann BM, et  al. 2015. A pooled analysis of reproductive factors, exogenous hormone use, and risk of multiple myeloma among women in the International Multiple Myeloma Consortium. Cancer Epidemiol Biomarkers Prev, 25(1), 217–​ 221. PMID: 26464426. Cremer FW, Bila J, Buck I, et  al. 2005. Delineation of distinct subgroups of multiple myeloma and a model for clonal evolution based on interphase cytogenetics. Genes Chromosomes Cancer, 44(2), 194–​ 203. PMID: 16001433. Cross AJ, Leitzmann MF, Gail MH, et al. 2007. A prospective study of red and processed meat intake in relation to cancer risk. PLoS Med, 4(12), e325. PMCID: PMC2121107. Dalamaga M, and Christodoulatos GS. 2015. Adiponectin as a biomarker linking obesity and adiposopathy to hematologic malignancies. Horm Mol Biol Clin Investig, 23(1), 5–​20. PMID: 26057219. Darby SC, Reeves G, Key T, Doll R, and Stovall M. 1994. Mortality in a cohort of women given X-​ray therapy for metropathia haemorrhagica. Int J Cancer, 56(6), 793–​801. PMID: 8119768. Darby SC, Whitley E, Howe GR, et al. 1995. Radon and cancers other than lung cancer in underground miners:  a collaborative analysis of 11 studies. J Natl Cancer Inst, 87(5), 378–​384. PMID: 7853419. Daugherty SE, Pfeiffer RM, Mellemkjaer L, Hemminki K, and Goldin LR. 2005. No evidence for anticipation in lymphoproliferative tumors in population-​based samples. Cancer Epidemiol Biomarkers Prev, 14(5), 1245–​1250. PMID: 15894680. De Roos AJ, Baris D, Weiss NS, and Herrington L. 2006. Multiple myeloma. In:  Schottenfeld D, and Fraumeni JF (Eds.), Cancer epidemiology and prevention (3rd ed., pp. 919–​945). New York: Oxford University Press. de Sanjose S, Benavente Y, Nieters A, et al. 2006. Association between personal use of hair dyes and lymphoid neoplasms in Europe. Am J Epidemiol, 164(1), 47–​55. PMID: 16731576. Debes-​Marun CS, Dewald GW, Bryant S, et  al. 2003. Chromosome abnormalities clustering and its implications for pathogenesis and prognosis in myeloma. Leukemia, 17(2), 427–​436. PMID: 12592343. Deshpande HA, Hu XP, Marino P, Jan NA, and Wiernik PH. 1998. Anticipation in familial plasma cell dyscrasias. Br J Haematol, 103(3), 696–​703. PMID: 9858219. Dimopoulos MA, Kiamouris C, and Moulopoulos LA. 1999. Solitary plasmacytoma of bone and extramedullary plasmacytoma. Hematol Oncol Clin North Am, 13(6), 1249–​1257. PMID: 10626148. Dispenzieri A, Katzmann JA, Kyle RA, et al. 2010. Prevalence and risk of progression of light-​chain monoclonal gammopathy of undetermined significance: a retrospective population-​based cohort study. Lancet, 375(9727), 1721–​1728. PMCID: PMC2904571. Durie BG, and Salmon SE. 1975. A clinical staging system for multiple myeloma:  correlation of measured myeloma cell mass with presenting

809

clinical features, response to treatment, and survival. Cancer, 36(3), 842–​ 854. PMID: 1182674. Egan JB, Shi CX, Tembe W, et al. 2012. Whole-​genome sequencing of multiple myeloma from diagnosis to plasma cell leukemia reveals genomic initiating events, evolution, and clonal tides. Blood, 120(5), 1060–​1066. PMCID: PMC3412329. Engels EA, Clarke CA, Pfeiffer RM, et  al. 2013. Plasma cell neoplasms in US solid organ transplant recipients. Am J Transplant, 13(6), 1523–​1532. PMCID: PMC3676887. Engels EA, Pfeiffer RM, Fraumeni JF, Jr., et al. 2011. Spectrum of cancer risk among US solid organ transplant recipients. JAMA, 306(17), 1891–​1901. PMCID: PMC3310893. Fonseca R, Barlogie B, Bataille R, et al. 2004. Genetics and cytogenetics of multiple myeloma:  a workshop report. Cancer Res, 64(4), 1546–​1558. PMID: 14989251. Fonseca R, Debes-​Marun CS, Picken EB, et al. 2003. The recurrent IgH translocations are highly associated with nonhyperdiploid variant multiple myeloma. Blood, 102(7), 2562–​2567. PMID: 12805059. Fonseca R, Oken MM, Harrington D, et  al. 2001. Deletions of chromosome 13 in multiple myeloma identified by interphase FISH usually denote large deletions of the q arm or monosomy. Leukemia, 15(6), 981–​986. PMID: 11417487. Fordyce EJ, Wang Z, Kahn AR, et  al. 2000. Risk of cancer among women with AIDS in New  York City. AIDS Public Policy J, 15(3–​4), 95–​104. PMID: 12189715. Frank C, Fallah M, Chen T, et al. 2016. Search for familial clustering of multiple myeloma with any cancer. Leukemia, 30(3), 627–​632. PMID: 26449663. Ghosh S, McLaughlin JR, Spinelli JJ, et  al. 2011. Multiple myeloma and occupational exposures: a population-​based case-​control study. J Occup Environ Med, 53(6), 641–​646. PMID: 21654434. Goedert JJ, Cote TR, Virgo P, et al. 1998. Spectrum of AIDS-​associated malignant disorders. Lancet, 351(9119), 1833–​1839. PMID: 9652666. Gold LS, Milliken K, Stewart P, et  al. 2010. Occupation and multiple myeloma:  an occupation and industry analysis. Am J Ind Med, 53(8), 768–​779. PMCID: PMC3085317. Gold LS, Stewart PA, Milliken K, et al. 2011. The relationship between multiple myeloma and occupational exposure to six chlorinated solvents. Occup Environ Med, 68(6), 391–​399. PMCID: PMC3094509. Grandin L, Orsi L, Troussard X, et  al. 2008. UV radiation exposure, skin type and lymphoid malignancies: results of a French case-​control study. Cancer Causes Control, 19(3), 305–​315. PMCID: PMC2323431. Grass S, Preuss KD, Ahlgrimm M, et  al. 2009. Association of a dominantly inherited hyperphosphorylated paraprotein target with sporadic and familial multiple myeloma and monoclonal gammopathy of undetermined significance: a case-​control study. Lancet Oncol, 10(10), 950–​956. PMID: 19767238. Greenberg AJ, Cousin M, Kumar S, et al. 2013a. Differences in the distribution of cytogenetic subtypes between multiple myeloma patients with and without a family history of monoclonal gammopathy and multiple myeloma. Eur J Haematol, 91(3), 193–​195. PMCID: PMC3762589. Greenberg AJ, Lee AM, Serie DJ, et  al. 2013b. Single-​nucleotide polymorphism rs1052501 associated with monoclonal gammopathy of undetermined significance and multiple myeloma. Leukemia, 27(2), 515–​516. PMCID: PMC3707297. Greenberg AJ, Philip S, Paner A, et al. 2015. Racial differences in primary cytogenetic abnormalities in multiple myeloma: a multi-​center study. Blood Cancer J, 5, e271. PMID: 25555162. Greenberg AJ, Rajkumar SV, and Vachon CM. 2012. Familial monoclonal gammopathy of undetermined significance and multiple myeloma:  epidemiology, risk factors, and biological characteristics. Blood, 119(23), 5359–​5366. PMCID: PMC3369675. Greenberg PL, Sun Z, Miller KB, et  al. 2009. Treatment of myelodysplastic syndrome patients with erythropoietin with or without granulocyte colony-​ stimulating factor:  results of a prospective randomized phase 3 trial by the Eastern Cooperative Oncology Group (E1996). Blood, 114(12), 2393–​2400. PMCID: PMC2746469. Greipp PR, San Miguel J, Durie BG, et al. 2005. International staging system for multiple myeloma. J Clin Oncol, 23(15), 3412–​3420. PMID: 15809451. Grosbois B, Jego P, Attal M, et  al. 1999. Familial multiple myeloma:  report of fifteen families. Br J Haematol, 105(3), 768–​770. PMID: 10354144. Grulich AE, Li Y, McDonald A, et al. 2002. Rates of non-​AIDS-​defining cancers in people with HIV infection before and after AIDS diagnosis. AIDS, 16(8), 1155–​1161. PMID: 12004274. Grulich AE, van Leeuwen MT, Falster MO, and Vajdic CM. 2007. Incidence of cancers in people with HIV/​ AIDS compared with immunosuppressed transplant recipients: a meta-​analysis. Lancet, 370(9581), 59–​67. PMID: 17617273.

810

810

PART IV:  Cancers by Tissue of Origin

Grulich AE, Wan X, Law MG, Coates M, and Kaldor JM. 1999. Risk of cancer in people with AIDS. AIDS, 13(7), 839–​843. PMID: 10357384. Gutierrez NC, Garcia JL, Hernandez JM, et  al. 2004. Prognostic and biologic significance of chromosomal imbalances assessed by comparative genomic hybridization in multiple myeloma. Blood, 104(9), 2661–​2666. PMID: 15238415. Hanamura I, Stewart JP, Huang Y, et al. 2006. Frequent gain of chromosome band 1q21 in plasma-​cell dyscrasias detected by fluorescence in situ hybridization: incidence increases from MGUS to relapsed myeloma and is related to prognosis and disease progression following tandem stem-​ cell transplantation. Blood, 108(5), 1724–​1732. PMCID: PMC1895503. Hansen J, Sallmen M, Selden AI, et al. 2013. Risk of cancer among workers exposed to trichloroethylene:  analysis of three Nordic cohort studies. J Natl Cancer Inst, 105(12), 869–​877. PMID: 23723420. Hari PN, Majhail NS, Zhang MJ, et al. 2010. Race and outcomes of autologous hematopoietic cell transplantation for multiple myeloma. Biol Blood Marrow Transplant, 16(3), 395–​402. PMCID: PMC2833344. Helby J, Bojesen SE, Nielsen SF, and Nordestgaard BG. 2015. IgE and risk of cancer in 37 747 individuals from the general population. Ann Oncol, 26(8), 1784–​1790. PMID: 25969367. Hemminki K, Li X, and Czene K. 2004. Familial risk of cancer:  data for clinical counseling and cancer genetics. Int J Cancer, 108(1), 109–​114. PMID: 14618624. Hideshima T, Bergsagel PL, Kuehl WM, and Anderson KC. 2004. Advances in biology of multiple myeloma: clinical applications. Blood, 104(3), 607–​ 618. PMID: 15090448. Hideshima T, Chauhan D, Podar K, et al. 2001. Novel therapies targeting the myeloma cell and its bone marrow microenvironment. Semin Oncol, 28(6), 607–​612. PMID: 11740818. Hideshima T, Mitsiades C, Tonon G, Richardson PG, and Anderson KC. 2007. Understanding multiple myeloma pathogenesis in the bone marrow to identify new therapeutic targets. Nat Rev Cancer, 7(8), 585–​598. PMID: 17646864. Hill QA, Rawstron AC, de Tute RM, and Owen RG. 2014. Outcome prediction in plasmacytoma of bone:  a risk model utilizing bone marrow flow cytometry and light-​ chain analysis. Blood, 124(8), 1296–​ 1299. PMID: 24939658. Hillengass J, Fechtner K, Weber MA, et  al. 2010. Prognostic significance of focal lesions in whole-​ body magnetic resonance imaging in patients with asymptomatic multiple myeloma. J Clin Oncol, 28(9), 1606–​1610. PMID: 20177023. Hofmann JN, Birmann BM, Teras LR, et al. 2016. Low levels of circulating adiponectin are associated with multiple myeloma risk in overweight and obese individuals. Cancer Res, 76(7), 1935–​1941. PMID: 26921332. Hofmann JN, Hoppin JA, Lynch CF, et al. 2015. Farm characteristics, allergy symptoms, and risk of non-​hodgkin lymphoid neoplasms in the agricultural health study. Cancer Epidemiol Biomarkers Prev, 24(3), 587–​594. PMCID: PMC4355385. Hofmann JN, Liao LM, Pollak MN, et al. 2012. A prospective study of circulating adipokine levels and risk of multiple myeloma. Blood, 120(22), 4418–​4420. PMID: 23007407. Hofmann JN, Moore SC, Lim U, et  al. 2013. Body mass index and physical activity at different ages and risk of multiple myeloma in the NIH-​AARP diet and health study. Am J Epidemiol, 177(8), 776–​786. PMCID: PMC3668425. Hosnijeh FS, Portengen L, Spath F, et al. 2016. Soluble B-​cell activation marker of sCD27 and sCD30 and future risk of B-​cell lymphomas: a nested case-​ control study and meta-​analyses. Int J Cancer, 138(10), 2357–​2367. PMID: 26684261. Hsu WL, Preston DL, Soda M, et  al. 2013. The incidence of leukemia, lymphoma and multiple myeloma among atomic bomb survivors: 1950–​2001. Radiat Res, 179(3), 361–​382. PMCID: PMC3875218. IARC. 2008. WHO classification of tumours of haematopoietic and lymphoid tissues, 4th ed. Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H, Thiele J, Vardiman JW (Eds). Lyon, France: International Agency for Research on Cancer. IARC. 2012a. Benzene: chemical agents and related occupations. A review of human carcinogens, Vol. 100 F. Lyon, France: International Agency for Research on Cancer. IARC. 2012b. Radiation. A review of human carcinogens, Vol. 100 D. Lyon, France: International Agency for Research on Cancer. IARC. 2014a. Cancer incidence in five continents, Vol. X. IARC Scientific Publication No. 164. Forman D, Bray F, Brewster DH, Gombe Mbalawa C, Kohler B, Pineros M, Steliarova-​Foucher E, Swaminathan R, and Ferlay J (Eds.). Lyon, France: International Agency for Research on Cancer. IARC. 2014b. Trichloroethylene, tetrachloroethylene, and some other chlorinated agents, Vol. 106. Lyon, France: International Agency for Research on Cancer.

Ichimaru M, Ishimaru T, Mikami M, and Matsunaga M. 1982. Multiple myeloma among atomic bomb survivors in Hiroshima and Nagasaki, 1950–​76:  relationship to radiation dose absorbed by marrow. J Natl Cancer Inst, 69(2), 323–​328. PMID: 6955540. Jain M, Ascensao J, and Schechter GP. 2009. Familial myeloma and monoclonal gammopathy: a report of eight African American families. Am J Hematol, 84(1), 34–​38. PMID: 19037864. Jakob C, Sterz J, Zavrski I, et al. 2006. Angiogenesis in multiple myeloma. Eur J Cancer, 42(11), 1581–​1590. PMID: 16797965. Jiang Y, Villeneuve PJ, Wielgosz A, et  al. 2010. The incidence of cancer in a population-​based cohort of Canadian heart transplant recipients. Am J Transplant, 10(3), 637–​645. PMID: 20121725. Jochem C, Leitzmann MF, Keimling M, Schmid D, and Behrens G. 2014. Physical activity in relation to risk of hematologic cancers: a systematic review and meta-​analysis. Cancer Epidemiol Biomarkers Prev, 23(5), 833–​846. PMID: 24633143. Jourdan M, Caraux A, De Vos J, et al. 2009. An in vitro model of differentiation of memory B cells into plasmablasts and plasma cells including detailed phenotypic and molecular characterization. Blood, 114(25), 5173–​5181. PMCID: PMC2834398. Kachuri L, Demers PA, Blair A, et al. 2013. Multiple pesticide exposures and the risk of multiple myeloma in Canadian men. Int J Cancer, 133(8), 1846–​1858. PMID: 23564249. Karami S, Bassig B, Stewart PA, et al. 2013. Occupational trichloroethylene exposure and risk of lymphatic and haematopoietic cancers:  a meta-​ analysis. Occup Environ Med, 70(8), 591–​599. PMID: 23723297. Kastritis E, Moulopoulos LA, Terpos E, Koutoulidis V, and Dimopoulos MA. 2014. The prognostic importance of the presence of more than one focal lesion in spine MRI of patients with asymptomatic (smoldering) multiple myeloma. Leukemia, 28(12), 2402–​2403. PMID: 25079172. Kastritis E, Terpos E, Moulopoulos L, et al. 2013. Extensive bone marrow infiltration and abnormal free light chain ratio identifies patients with asymptomatic myeloma at high risk for progression to symptomatic disease. Leukemia, 27(4), 947–​953. PMID: 23183429. Key TJ, Appleby PN, Crowe FL, et  al. 2014. Cancer in British vegetarians: updated analyses of 4998 incident cancers in a cohort of 32,491 meat eaters, 8612 fish eaters, 18,298 vegetarians, and 2246 vegans. Am J Clin Nutr, 100 Suppl 1, 378S–​385S. PMCID: PMC4144109. Kirchhof MG, Chau LA, Lemke CD, et  al. 2008. Modulation of T cell activation by stomatin-​ like protein 2. J Immunol, 181(3), 1927–​ 1936. PMCID: PMC2913160. Korde N, Kristinsson SY, and Landgren O. 2011. Monoclonal gammopathy of undetermined significance (MGUS) and smoldering multiple myeloma (SMM):  novel biological insights and development of early treatment strategies. Blood, 117(21), 5573–​5581. PMCID: PMC3316455. Koutros S, Baris D, Bell E, et  al. 2009. Use of hair colouring products and risk of multiple myeloma among US women. Occup Environ Med, 66(1), 68–​70. PMCID: PMC2756225. Kristinsson SY, Bjorkholm M, Goldin LR, et al. 2009. Patterns of hematologic malignancies and solid tumors among 37,838 first-​degree relatives of 13,896 patients with multiple myeloma in Sweden. Int J Cancer, 125(9), 2147–​2150. PMCID: PMC2737604. Kristinsson SY, Landgren O, Dickman PW, Derolf AR, and Bjorkholm M. 2007. Patterns of survival in multiple myeloma: a population-​based study of patients diagnosed in Sweden from 1973 to 2003. J Clin Oncol, 25(15), 1993–​1999. PMID: 17420512. Kroll ME, Murphy F, Pirie K, et al. 2012. Alcohol drinking, tobacco smoking and subtypes of haematological malignancy in the UK Million Women Study. Br J Cancer, 107(5), 879–​887. PMCID: PMC3425977. Krynitz B, Edgren G, Lindelof B et  al. 2013. Risk of skin cancer and other malignancies in kidney, liver, heart and lung transplant recipients 1970 to 2008: a Swedish population-​based study. Int J Cancer, 132(6), 1429–​ 1438. PMID: 22886725. Kumar S, Fonseca R, Ketterling RP, et  al. 2012. Trisomies in multiple myeloma:  impact on survival in patients with high-​risk cytogenetics. Blood, 119(9), 2100–​2105. PMCID: PMC3311247. Kumar SK, Rajkumar SV, Dispenzieri A, et al. 2008. Improved survival in multiple myeloma and the impact of novel therapies. Blood, 111(5), 2516–​ 2520. PMCID: PMC2254544. Kyle RA, Remstein ED, Therneau TM, et al. 2007. Clinical course and prognosis of smoldering (asymptomatic) multiple myeloma. New Engl J Med, 356(25), 2582–​2590. PMID: 17582068. Kyle RA, Therneau TM, Rajkumar SV, et  al. 2006. Prevalence of monoclonal gammopathy of undetermined significance. N Engl J Med, 354(13), 1362–​1369. PMID: 16571879. Kyle RA, Therneau TM, Rajkumar SV, et al. 2002. A long-​term study of prognosis in monoclonal gammopathy of undetermined significance. N Engl J Med, 346(8), 564–​569. PMID: 11856795.

 81



Multiple Myeloma

Kyle RA, Therneau TM, Rajkumar SV, et  al. 2003. Long-​term follow-​up of IgM monoclonal gammopathy of undetermined significance. Blood, 102(10), 3759–​3764. PMID: 12881316. Landgren O, Graubard BI, Katzmann JA, et al. 2014. Racial disparities in the prevalence of monoclonal gammopathies:  a population-​based study of 12,482 persons from the National Health and Nutritional Examination Survey. Leukemia, 28(7), 1537–​1542. PMCID: PMC4090286. Landgren O, Kyle RA, Hoppin JA, et al. 2009a. Pesticide exposure and risk of monoclonal gammopathy of undetermined significance in the Agricultural Health Study. Blood, 113(25), 6386–​6391. PMCID: PMC2710931. Landgren O, Kyle RA, Pfeiffer RM, et  al. 2009b. Monoclonal gammopathy of undetermined significance (MGUS) consistently precedes multiple myeloma:  a prospective study. Blood, 113(22), 5412–​ 5417. PMCID: PMC2689042. Landgren O, Linet MS, McMaster ML, et  al. 2006a. Familial characteristics of autoimmune and hematologic disorders in 8,406 multiple myeloma patients:  a population-​based case-​control study. Int J Cancer, 118(12), 3095–​3098. PMID: 16395700. Landgren O, Rajkumar SV, Pfeiffer RM, et  al. 2010. Obesity is associated with an increased risk of monoclonal gammopathy of undetermined significance among black and white women. Blood, 116(7), 1056–​1059. PMCID: PMC2938127. Landgren O, Rapkin JS, Mellemkjaer L, et al. 2006b. Respiratory tract infections in the pathway to multiple myeloma: a population-​based study in Scandinavia. Haematologica, 91(12), 1697–​1700. PMID: 17145609. Landgren O, Shim YK, Michalek J, et al. 2015. Agent Orange exposure and monoclonal gammopathy of undetermined significance:  An Operation Ranch Hand Veteran Cohort Study. JAMA Oncol, 1(8), 1061–​ 1068. PMID: 26335650. Landgren O, and Waxman AJ. 2010. Multiple myeloma precursor disease. JAMA, 304(21), 2397–​2404. PMID: 21119086. Landgren O, Zhang Y, Zahm SH, et al. 2006c. Risk of multiple myeloma following medication use and medical conditions:  a case-​control study in Connecticut women. Cancer Epidemiol Biomarkers Prev, 15(12), 2342–​ 2347. PMID: 17132770. Larsen JT, Kumar SK, Dispenzieri A, et al. 2013. Serum free light chain ratio as a biomarker for high-​risk smoldering multiple myeloma. Leukemia, 27(4), 941–​946. PMCID: PMC3629951. Lecluse Y, Comby E, Mariotte D, et al. 2015. Prevalence of monoclonal gammopathy of undetermined significance (MGUS) among farmers involved in open field farming and/​or cattle breading in France. Leuk Lymphoma, 1–​4. PMID: 26689498. Leuraud K, Richardson DB, Cardis E, et  al. 2015. Ionising radiation and risk of death from leukaemia and lymphoma in radiation-​monitored workers (INWORKS):  an international cohort study. Lancet Haematology, 2, 6. Lim U, Subar AF, Mouw T, et al. 2006. Consumption of aspartame-​containing beverages and incidence of hematopoietic and brain malignancies. Cancer Epidemiol Biomarkers Prev, 15(9), 1654–​1659. PMID: 16985027. Lin SW, Wheeler DC, Park Y, et al. 2012. Prospective study of ultraviolet radiation exposure and risk of cancer in the United States. Int J Cancer, 131(6), E1015–​1023. PMCID: PMC3402606. Lindqvist EK, Goldin LR, Landgren O, et  al. 2011. Personal and family history of immune-​related conditions increase the risk of plasma cell disorders:  a population-​based study. Blood, 118(24), 6284–​6291. PMCID: PMC3236117. Liu M, Hsu J, Chan C, Li Z, and Zhou Q. 2012. The ubiquitin ligase Siah1 controls ELL2 stability and formation of super elongation complexes to modulate gene transcription. Mol Cell, 46(3), 325–​334. PMCID: PMC3360964. Lope V, Perez-​Gomez B, Aragones N, et al. 2008. Occupation, exposure to chemicals, sensitizing agents, and risk of multiple myeloma in Sweden. Cancer Epidemiol Biomarkers Prev, 17(11), 3123–​3127. PMID: 18990753. Lu Y, Sullivan-​Halley J, Henderson KD, et  al. 2010. Anthropometric characteristics and multiple myeloma risk. Epidemiology, 21(2), 272–​273. PMID: 20160567. Luo Z, Lin C, and Shilatifard A. 2012. The super elongation complex (SEC) family in transcriptional control. Nature Rev Mol Cell Biol, 13(9), 543–​ 547. PMID: 22895430. Lynch HT, Ferrara K, Barlogie B, et al. 2008. Familial myeloma. N Engl J Med, 359(2), 152–​157. PMCID: PMC2775509. Lynch HT, Watson P, Tarantolo S, et al. 2005. Phenotypic heterogeneity in multiple myeloma families. J Clin Oncol, 23(4), 685–​693. PMID: 15681516. MacInnis RJ, English DR, Hopper JL, and Giles GG. 2005. Body size and composition and the risk of lymphohematopoietic malignancies. J Natl Cancer Inst, 97(15), 1154–​1157. PMID: 16077074. Mahale P, Yanik EL, and Engels EA. 2016. Herpes zoster and risk of cancer in the elderly U.S. population. Cancer Epidemiol Biomarkers Prev, 25(1), 28–​35. PMID: 26578536.

811

Mahindra A, Hideshima T, and Anderson KC. 2010. Multiple myeloma: biology of the disease. Blood Rev, 24 (Suppl 1), S5–​11. PMID: 21126636. Martino A, Campa D, Jurczyszyn A, et al. 2014. Genetic variants and multiple myeloma risk:  IMMEnSE validation of the best reported associations-​-​ an extensive replication of the associations from the candidate gene era. Cancer Epidemiol Biomarkers Prev, 23(4), 670–​674. PMID: 24521996. Martino A, Sainz J, Buda G, et al. 2012. Genetics and molecular epidemiology of multiple myeloma:  the rationale for the IMMEnSE consortium (review). Int J Oncol, 40(3), 625–​638. PMID: 22159523. Mateos MV, Hernandez MT, Giraldo P, et al. 2013. Lenalidomide plus dexamethasone for high-​risk smoldering multiple myeloma. N Engl J Med, 369(5), 438–​447. PMID: 23902483. McCullough ML, Teras LR, Shah R, et  al. 2014. Artificially and sugar-​ sweetened carbonated beverage consumption is not associated with risk of lymphoid neoplasms in older men and women. J Nutr, 144(12), 2041–​ 2049. PMID: 25342696. McShane CM, Murray LJ, Engels EA, Landgren O, and Anderson LA. 2014a. Common community-​acquired infections and subsequent risk of multiple myeloma:  a population-​based study. Int J Cancer, 134(7), 1734–​1740. PMCID: PMC3947153. McShane CM, Murray LJ, Landgren O, et al. 2014b. Prior autoimmune disease and risk of monoclonal gammopathy of undetermined significance and multiple myeloma:  a systematic review. Cancer Epidemiol Biomarkers Prev, 23(2), 332–​342. PMID: 24451437. Mendelsohn JB, Li QZ, Ji BT, et al. 2009. Personal use of hair dye and cancer risk in a prospective cohort of Chinese women. Cancer Sci, 100(6), 1088–​ 1091. PMCID: PMC2768318. Merhi M, Raynal H, Cahuzac E, et al. 2007. Occupational exposure to pesticides and risk of hematopoietic cancers:  meta-​analysis of case-​control studies. Cancer Causes Control, 18(10), 1209–​1226. PMID: 17874193. Mikhael JR, Dingli D, Roy V, et  al. 2013. Management of newly diagnosed symptomatic multiple myeloma:  updated Mayo Stratification of Myeloma and Risk-​Adapted Therapy (mSMART) consensus guidelines 2013. Mayo Clin Proc, 88(4), 360–​376. PMID: 23541011. Mitchell JS, Johnson DC, Litchfield K, et al. 2015. Implementation of genome-​ wide complex trait analysis to quantify the heritability in multiple myeloma. Sci Rep, 5, 12473. PMCID: PMC4513545. Modiano MR, Villar-​Werstler P, Crowley J, and Salmon SE. 1996. Evaluation of race as a prognostic factor in multiple myeloma:  an ancillary of Southwest Oncology Group Study 8229. J Clin Oncol, 14(3), 974–​977. PMID: 8622048. Moreau P, Facon T, Leleu X, et  al. 2002. Recurrent 14q32 translocations determine the prognosis of multiple myeloma, especially in patients receiving intensive chemotherapy. Blood, 100(5), 1579–​ 1583. PMID: 12176874. Morgan GJ, Johnson DC, Weinhold N, et al. 2014. Inherited genetic susceptibility to multiple myeloma. Leukemia, 28(3), 518–​524. PMID: 24247655. Morton LM, Slager SL, Cerhan JR, et  al. 2014. Etiologic heterogeneity among non-​Hodgkin lymphoma subtypes: the InterLymph Non-​Hodgkin Lymphoma Subtypes Project. J Natl Cancer Inst Monogr, 2014(48), 130–​ 144. PMCID: PMC4155467. Morton LM, Wang SS, Richesson DA, et al. 2009. Reproductive factors, exogenous hormone use and risk of lymphoid neoplasms among women in the National Institutes of Health-​AARP Diet and Health Study Cohort. Int J Cancer, 124(11), 2737–​2743. PMCID: PMC2701156. Morton LM, Zheng T, Holford TR, et  al. 2005. Alcohol consumption and risk of non-​Hodgkin lymphoma: a pooled analysis. Lancet Oncol, 6(7), 469–​476. PMID: 15992695. Moysich KB, Bonner MR, Beehler GP, et al. 2007. Regular analgesic use and risk of multiple myeloma. Leuk Res, 31(4), 547–​551. PMID: 16962170. Muirhead CR, O’Hagan JA, Haylock RG, et  al. 2009. Mortality and cancer incidence following occupational radiation exposure: third analysis of the National Registry for Radiation Workers. Br J Cancer, 100(1), 206–​212. PMCID: PMC2634664. Murphy G, Dawsey SM, Engels EA, et al. 2015. Cancer risk after pernicious anemia in the US elderly population. Clin Gastroenterol Hepatol, 13(13), 2282–​2289. PMID: 26079040. Nefedova Y, Cheng P, Alsina M, Dalton WS, and Gabrilovich DI. 2004. Involvement of Notch-​1 signaling in bone marrow stroma-​mediated de novo drug resistance of myeloma and other malignant lymphoid cell lines. Blood, 103(9), 3503–​3510. PMID: 14670925. Ng M, Fleming T, Robinson M, et  al. 2014. Global, regional, and national prevalence of overweight and obesity in children and adults during 1980–​ 2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet, 384(9945), 766–​781. PMID: 24880830. Nieters A, Luczynska A, Becker S, et al. 2014. Prediagnostic immunoglobulin E levels and risk of chronic lymphocytic leukemia, other lymphomas and multiple myeloma-​results of the European Prospective Investigation

812

812

PART IV:  Cancers by Tissue of Origin

into Cancer and Nutrition. Carcinogenesis, 35(12), 2716–​ 2722. PMCID: PMC4247516. Nuyujukian DS, Voutsinas J, Bernstein L, and Wang SS. 2014. Medication use and multiple myeloma risk in Los Angeles County. Cancer Causes Control, 25(9), 1233–​1237. PMCID: PMC4156901. O’Neill TW, Harrison BJ, Yin AL, and Holt PJ. 1997. Ankylosing spondylitis associated with IgA lambda chain myeloma. Br J Rheumatol, 36(3), 401–​ 402. PMID: 9133979. Osgood EE. 1960. The survival time of patients with plasmocytic myeloma. Cancer Chemother Rep, 9, 1–​10. PMID: 13731417. Pahwa P, Karunanayake CP, Dosman JA, et al. 2012. Multiple myeloma and exposure to pesticides:  a Canadian case-​control study. J Agromedicine, 17(1), 40–​50. PMID: 22191502. Paiva B, Chandia M, Vidriales MB, et al. 2014. Multiparameter flow cytometry for staging of solitary bone plasmacytoma: new criteria for risk of progression to myeloma. Blood, 124(8), 1300–​1303. PMID: 24876564. Palumbo A, Avet-​Loiseau H, Oliva S, et al. 2015. Revised international staging system for multiple myeloma: a report from International Myeloma Working Group. J Clin Oncol, 33(26), 2863–​2869. PMID: 26240224. Park Y, Leitzmann MF, Subar AF, Hollenbeck A, and Schatzkin A. 2009. Dairy food, calcium, and risk of cancer in the NIH-​AARP Diet and Health Study. Arch Intern Med, 169(4), 391–​401. PMCID: PMC2796799. Patel AV, Hildebrand JS, Campbell PT, et al. 2016. Leisure-​time spent sitting and site-​specific cancer incidence in a large US cohort. Cancer Epidemiol Biomarkers Prev, 24(9), 1350–​1359. PMID: 26126627. Perez-​Persona E, Vidriales MB, Mateo G, et al. 2007. New criteria to identify risk of progression in monoclonal gammopathy of uncertain significance and smoldering multiple myeloma based on multiparameter flow cytometry analysis of bone marrow plasma cells. Blood, 110(7), 2586–​2592. PMID: 17576818. Perrotta C, Kleefeld S, Staines A, et  al. 2013. Multiple myeloma and occupation:  a pooled analysis by the International Multiple Myeloma Consortium. Cancer Epidemiol, 37(3), 300–​305. PMID: 23403129. Perrotta C, Staines A, and Cocco P. 2008. Multiple myeloma and farming: a systematic review of 30  years of research. Where next? J Occup Med Toxicol, 3, 27. PMCID: PMC2628921. Perrotta C, Staines A, Codd M, et  al. 2012. Multiple myeloma and lifetime occupation:  results from the EPILYMPH study. J Occup Med Toxicol, 7(1), 25. PMID: 23241100. Popat K, McQueen K, and Feeley TW. 2013. The global burden of cancer. Best Pract Res Clin Anaesthesiol, 27(4), 399–​408. PMID: 24267547. Preston DL, Kusumi S, Tomonaga M, et al. 1994. Cancer incidence in atomic bomb survivors. Part III. Leukemia, lymphoma and multiple myeloma, 1950–​1987. Radiat Res, 137(2 Suppl), S68–​S97. PMID: 8127953. Preuss KD, Pfreundschuh M, Ahlgrimm M, et  al. 2009. A frequent target of paraproteins in the sera of patients with multiple myeloma and MGUS. Int J Cancer, 125(3), 656–​661. PMID: 19405124. Preuss KD, Pfreundschuh M, Fadle N, et  al. 2011. Hyperphosphorylation of autoantigenic targets of paraproteins is due to inactivation of PP2A. Blood, 118(12), 3340–​3346. PMID: 21791414. Psaltopoulou T, Sergentanis TN, Kanellias N, et al. 2013. Tobacco smoking and risk of multiple myeloma: a meta-​analysis of 40 observational studies. Int J Cancer, 132(10), 2413–​2431. PMID: 23047159. Psaltopoulou T, Sergentanis TN, Sergentanis IN, et al. 2015. Alcohol intake, alcoholic beverage type and multiple myeloma risk: a meta-​analysis of 26 observational studies. Leuk Lymphoma, 56(5), 1484–​1501. PMID: 25146435. Pulte D, Redaniel MT, Brenner H, and Jeffreys M. 2012. Changes in survival by ethnicity of patients with cancer between 1992–​ 1996 and 2002–​ 2006:  is the discrepancy decreasing? Ann Oncol, 23(9), 2428–​ 2434. PMID: 22396445. Quinlan SC, Morton LM, Pfeiffer RM, et al. 2010. Increased risk for lymphoid and myeloid neoplasms in elderly solid-​organ transplant recipients. Cancer Epidemiol Biomarkers Prev, 19(5), 1229–​1237. PMCID: PMC2866098. Rajkumar SV. 2009. Prevention of progression in monoclonal gammopathy of undetermined significance. Clin Cancer Res, 15(18), 5606–​ 5608. PMCID: PMC2759099. Rajkumar SV. 2014. Multiple myeloma:  2014 update on diagnosis, risk-​ stratification, and management. Am J Hematol, 89(10), 999–​ 1009. PMID: 25223428. Rajkumar SV, Dimopoulos MA, Palumbo A, et al. 2014. International Myeloma Working Group updated criteria for the diagnosis of multiple myeloma. Lancet Oncol, 15(12), e538–​548. PMID: 25439696. Rajkumar SV, Jacobus S, Callander NS, et  al. 2010. Lenalidomide plus high-​ dose dexamethasone versus lenalidomide plus low-​ dose dexamethasone as initial therapy for newly diagnosed multiple myeloma: an open-​ label randomised controlled trial. Lancet Oncol, 11(1), 29–​ 37. PMCID: PMC3042271.

Rajkumar SV, Kyle RA, Therneau TM, et  al. 2005. Serum free light chain ratio is an independent risk factor for progression in monoclonal gammopathy of undetermined significance. Blood, 106(3), 812–​ 817. PMCID: PMC1895159. Rajkumar SV, Landgren O, and Mateos MV. 2015. Smoldering multiple myeloma. Blood, 125(20), 3069–​3075. PMCID: PMC4432003. Rajkumar SV, Larson D, and Kyle RA. 2011. Diagnosis of smoldering multiple myeloma. N Engl J Med, 365(5), 474–​475. PMCID: PMC3773467. Rasmussen ML, Hjalgrim H, Molgaard-​Nielsen D, Wohlfahrt J, and Melbye M. 2012. Antibiotic use and risk of non-​Hodgkin lymphomas. Int J Cancer, 131(7), E1158–​1165. PMID: 22552810. Renshaw C, Ketley N, Moller H, and Davies EA. 2010. Trends in the incidence and survival of multiple myeloma in South East England 1985–​2004. BMC Cancer, 10, 74. PMCID: PMC2837016. Rericha V, Kulich M, Rericha R, Shore DL, and Sandler DP. 2006. Incidence of leukemia, lymphoma, and multiple myeloma in Czech uranium miners:  a case-​ cohort study. Environ Health Perspect, 114(6), 818–​ 822. PMCID: PMC1480508. Ribatti D, Nico B, and Vacca A. 2006. Importance of the bone marrow microenvironment in inducing the angiogenic response in multiple myeloma. Oncogene, 25(31), 4257–​4266. PMID: 16518413. Rohrmann S, Becker N, Linseisen J, et al. 2007. Fruit and vegetable consumption and lymphoma risk in the European Prospective Investigation into Cancer and Nutrition (EPIC). Cancer Causes Control, 18(5), 537–​549. PMID: 17443415. Rohrmann S, Linseisen J, Jakobsen MU, et al. 2011. Consumption of meat and dairy and lymphoma risk in the European Prospective Investigation into Cancer and Nutrition. Int J Cancer, 128(3), 623–​634. PMID: 20473877. Roodman GD. 2010. Pathogenesis of myeloma bone disease. J Cell Biochem, 109(2), 283–​291. PMID: 20014067. Russell SJ, and Rajkumar SV. 2011. Multiple myeloma and the road to personalised medicine. Lancet Oncol, 12(7), 617–​619. PMID: 21664869. Sakata R, Kleinerman RA, Mabuchi K, et al. 2012. Cancer mortality following radiotherapy for benign gynecologic disorders. Radiat Res, 178(4), 266–​ 279. PMCID: PMC3471655. Schernhammer ES, Bertrand KA, Birmann BM, et al. 2012. Consumption of artificial sweetener-​and sugar-​containing soda and risk of lymphoma and leukemia in men and women. Am J Clin Nutr, 96(6), 1419–​1428. PMCID: PMC3497928. Schubauer-​Berigan MK, Daniels RD, Bertke SJ, Tseng CY, and Richardson DB. 2015. Cancer mortality through 2005 among a pooled cohort of U.S. nuclear workers exposed to external ionizing radiation. Radiat Res, 183(6), 620–​631. PMID: 26010709. Schubauer-​Berigan MK, Daniels RD, and Pinkerton LE. 2009. Radon exposure and mortality among white and American Indian uranium miners: an update of the Colorado Plateau cohort. Am J Epidemiol, 169(6), 718–​730. PMID: 19208723. Schuster SR, Rajkumar SV, Dispenzieri A, et  al. 2010. IgM multiple myeloma:  disease definition, prognosis, and differentiation from Waldenstrom’s macroglobulinemia. Am J Hematol, 85(11), 853–​ 855. PMCID: PMC3736855. SEER. 2015a. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) Research Data (1973–​2012), National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2015, based on the November 2014 submission. SEER. 2015b. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database:  Populations—​ Total U.S. (1969–​2013) Single Ages to 85+, Katrina/​Rita Adjustment—​Linked to County Attributes—​Total U.S., 1969–​2013 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released October 2014. Seidler A, Mohner M, Berger J, et al. 2007. Solvent exposure and malignant lymphoma: a population-​based case-​control study in Germany. J Occup Med Toxicol, 2, 2. PMCID: PMC1851965. Selby JV, Friedman GD, and Fireman BH. 1989. Screening prescription drugs for possible carcinogenicity: eleven to fifteen years of follow-​up. Cancer Res, 49(20), 5736–​5747. PMID: 2571410. Shaffer AL, Emre NC, Lamy L, et  al. 2008. IRF4 addiction in multiple myeloma. Nature, 454(7201), 226–​231. PMCID: PMC2542904. Shaffer AL, Emre NC, Romesser PB, and Staudt LM. 2009. IRF4: Immunity. Malignancy! Therapy? Clin Cancer Res, 15(9), 2954–​ 2961. PMCID: PMC2790720. Shimizu Y, Kato H, and Schull WJ. 1990. Studies of the mortality of A-​bomb survivors. 9.  Mortality, 1950–​ 1985:  Part  2. Cancer mortality based on the recently revised doses (DS86). Radiat Res, 121(2), 120–​141. PMID: 2305030.

 813



Multiple Myeloma

Siegel RL, Miller KD, and Jemal A. 2015. Cancer statistics, 2015. CA Cancer J Clin, 65(1), 5–​29. PMID: 25559415. Smadja NV, Fruchart C, Isnard F, et al. 1998. Chromosomal analysis in multiple myeloma: cytogenetic evidence of two different diseases. Leukemia, 12(6), 960–​969. PMID: 9639426. Soderberg KC, Hagmar L, Schwartzbaum J, and Feychting M. 2004. Allergic conditions and risk of hematological malignancies in adults:  a cohort study. BMC Public Health, 4, 51. PMCID: PMC534807. Stebbings JH, Lucas HF, and Stehney AF. 1984. Mortality from cancers of major sites in female radium dial workers. Am J Ind Med, 5(6), 435–​459. PMID: 6731445. Stenehjem JS, Kjaerheim K, Bratveit M, et al. 2015. Benzene exposure and risk of lymphohaematopoietic cancers in 25 000 offshore oil industry workers. Br J Cancer, 112(9), 1603–​1612. PMCID: PMC4453669. Svec MA, Ward MH, Dosemeci M, Checkoway H, and De Roos AJ. 2005. Risk of lymphatic or haematopoietic cancer mortality with occupational exposure to animals or the public. Occup Environ Med, 62(10), 726–​735. PMCID: PMC1740863. Swaminathan B, Thorleifsson G, Joud M, et al. 2015. Variants in ELL2 influencing immunoglobulin levels associate with multiple myeloma. Nat Commun, 6, 7213. PMCID: PMC4455110. Takkouche B, Etminan M, and Montes-​Martínez A. 2005. Personal use of hair dyes and risk of cancer: a meta-​analysis. JAMA, 293(20), 2516–​2525. Teras LR, Gapstur SM, Diver WR, Birmann BM, and Patel AV. 2012. Recreational physical activity, leisure sitting time and risk of non-​Hodgkin lymphoid neoplasms in the American Cancer Society Cancer Prevention Study II Cohort. Int J Cancer, 131(8), 1912–​1920. PMID: 22275172. Teras LR, Gapstur SM, Patel AV, et  al. 2013a. Aspirin and other nonsteroidal anti-​inflammatory drugs and risk of non-​hodgkin lymphoma. Cancer Epidemiol Biomarkers Prev, 22(3), 422–​428. PMID: 23329726. Teras LR, Kitahara CM, Birmann BM, et  al. 2014. Body size and multiple myeloma mortality:  a pooled analysis of 20 prospective studies. Br J Haematol, 166(5), 667–​676. PMCID: PMC4134758. Teras LR, Patel AV, Hildebrand JS, and Gapstur SM. 2013b. Postmenopausal unopposed estrogen and estrogen plus progestin use and risk of non-​ Hodgkin lymphoma in the American Cancer Society Cancer Prevention Study-​II Cohort. Leuk Lymphoma, 54(4), 720–​725. PMID: 22916741. Therneau TM, Kyle RA, Melton LJ, 3rd et al. 2012. Incidence of monoclonal gammopathy of undetermined significance and estimation of duration before first clinical recognition. Mayo Clin Proc, 87(11), 1071–​1079. PMCID: PMC3541934. Troy JD, Hartge P, Weissfeld JL, et al. 2010. Associations between anthropometry, cigarette smoking, alcohol consumption, and non-​Hodgkin lymphoma in the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial. Am J Epidemiol, 171(12), 1270–​1281. PMCID: PMC2915494. Turesson I, Velez R, Kristinsson SY, and Landgren O. 2010a. Patterns of improved survival in patients with multiple myeloma in the twenty-​ first century:  a population-​based study. J Clin Oncol, 28(5), 830–​834. PMCID: PMC2834396. Turesson I, Velez R, Kristinsson SY, and Landgren O. 2010b. Patterns of multiple myeloma during the past 5 decades:  stable incidence rates for all age groups in the population but rapidly changing age distribution in the clinic. Mayo Clin Proc, 85(3), 225–​230. PMCID: PMC2843108. Tzeng HE, Lin CL, Tsai CH, et  al. 2013. Time trend of multiple myeloma and associated secondary primary malignancies in Asian patients:  a Taiwan population-​based study. PLoS One, 8(7), e68041. PMCID: PMC3699496. Uchiyama H, Barut BA, Mohrbacher AF, Chauhan D, and Anderson KC. 1993. Adhesion of human myeloma-​derived cell lines to bone marrow stromal cells stimulates interleukin-​6 secretion. Blood, 82(12), 3712–​3720. PMID: 8260708. Vachon CM, Kyle RA, Therneau TM, et al. 2009. Increased risk of monoclonal gammopathy in first-​degree relatives of patients with multiple myeloma or monoclonal gammopathy of undetermined significance. Blood, 114(4), 785–​790. PMCID: PMC2716020. Vacquier B, Caer S, Rogel A, et al. 2008. Mortality risk in the French cohort of uranium miners: extended follow-​up 1946–​1999. Occup Environ Med, 65(9), 597–​604. PMID: 18096654. Vajdic CM, McDonald SP, McCredie MR, et  al. 2006. Cancer incidence before and after kidney transplantation. JAMA, 296(23), 2823–​ 2831. PMID: 17179459. Vajdic CM, and van Leeuwen MT. 2009. Cancer incidence and risk factors after solid organ transplantation. Int J Cancer, 125(8), 1747–​ 1754. PMID: 19444916. Vangsted A, Klausen TW, and Vogel U. 2012. Genetic variations in multiple myeloma I: effect on risk of multiple myeloma. Eur J Haematol, 88(1), 8–​30. PMID: 21883480.

813

Van Valkenburg ME, Pruitt GI, Brill IK, et al. 2016. Family history of hematologic malignancies and risk of multiple myeloma: differences by race and clinical features. Cancer Causes Control, 27(1), 81–​91. PMID: 26596855. Verma PS, Howard RS, and Weiss BM. 2008. The impact of race on outcomes of autologous transplantation in patients with multiple myeloma. Am J Hematol, 83(5), 355–​358. PMID: 18186525. Villeneuve PJ, Schaubel DE, Fenton SS, et al. 2007. Cancer incidence among Canadian kidney transplant recipients. Am J Transplant, 7(4), 941–​948. PMID: 17331115. Vlaanderen J, Lan Q, Kromhout H, Rothman N, and Vermeulen R. 2011. Occupational benzene exposure and the risk of lymphoma subtypes:  a meta-​analysis of cohort studies incorporating three study quality dimensions. Environ Health Perspect, 119(2), 159–​167. PMCID: PMC3040601. Wadhera RK, and Rajkumar SV. 2010. Prevalence of monoclonal gammopathy of undetermined significance: a systematic review. Mayo Clin Proc, 85(10), 933–​942. PMCID: PMC2947966. Wallin A, and Larsson SC. 2011. Body mass index and risk of multiple myeloma: a meta-​analysis of prospective studies. Eur J Cancer, 47(11), 1606–​1615. PMID: 21354783. Walter RB, Milano F, Brasky TM, and White E. 2011. Long-​term use of acetaminophen, aspirin, and other nonsteroidal anti-​ inflammatory drugs and risk of hematologic malignancies:  results from the Prospective Vitamins and Lifestyle (VITAL) Study. J Clin Oncol, 29(17), 2424–​2431. PMCID: PMC3107756. Wang Q, Wang Y, Ji Z, et  al. 2012. Risk factors for multiple myeloma:  a hospital-​based case-​control study in Northwest China. Cancer Epidemiol, 36(5), 439–​444. PMID: 22673750. Wang SS, Voutsinas J, Chang ET, et  al. 2013. Anthropometric, behavioral, and female reproductive factors and risk of multiple myeloma:  a pooled analysis. Cancer Causes Control, 24(7), 1279–​1289. PMCID: PMC3684420. Wang YZ, Wu QJ, Zhu J, and Wu L. 2015. Fish consumption and risk of myeloma:  a meta-​analysis of epidemiological studies. Cancer Causes Control, 26(9), 1307–​1314. PMID: 26156047. Warsame R, Gertz MA, Lacy MQ, et al. 2012. Trends and outcomes of modern staging of solitary plasmacytoma of bone. Am J Hematol, 87(7), 647–​651. PMCID: PMC4440242. Waxman AJ, Mink PJ, Devesa SS, et al. 2010. Racial disparities in incidence and outcome in multiple myeloma:  a population-​based study. Blood, 116(25), 5501–​5506. PMCID: PMC3031400. Weinhold N, Johnson DC, Chubb D, et al. 2013. The CCND1 c.870G>A polymorphism is a risk factor for t(11;14)(q13;q32) multiple myeloma. Nat Genet, 45(5), 522–​525. PMID: 23502783. Weinhold N, Johnson DC, Rawstron AC, et  al. 2014. Inherited genetic susceptibility to monoclonal gammopathy of unknown significance. Blood, 123(16), 2513–​2517; quiz 2593. PMID: 24449210. Weinhold N, Meissner T, Johnson DC, et  al. 2015. The 7p15.3 (rs4487645) association for multiple myeloma shows strong allele-​specific regulation of the MYC-​interacting gene CDCA7L in malignant plasma cells. Haematologica, 100(3), e110–​113. PMCID: PMC4349291. Weires M, Bermejo JL, Sundquist J, and Hemminki K. 2011. Clustering of concordant and discordant cancer types in Swedish couples is rare. Eur J Cancer, 47(1), 98–​106. PMID: 20655736. Weiss BM, Abadie J, Verma P, Howard RS, and Kuehl WM. 2009. A monoclonal gammopathy precedes multiple myeloma in most patients. Blood, 113(22), 5418–​5422. PMCID: PMC2689043. Weiss HA, Darby SC, and Doll R. 1994. Cancer mortality following X-​ray treatment for ankylosing spondylitis. Int J Cancer, 59(3), 327–​ 338. PMID: 7927937. Wright JD, St Clair CM, Deutsch I, et al. 2010. Pelvic radiotherapy and the risk of secondary leukemia and multiple myeloma. Cancer, 116(10), 2486–​ 2492. PMID: 20209618. Wu SP, Minter A, Costello R, et  al. 2013. MGUS prevalence in an ethnically Chinese population in Hong Kong. Blood, 121(12), 2363–​2364. PMCID: PMC3606071. Wuilleme S, Robillard N, Lode L, et  al. 2005. Ploidy, as detected by fluorescence in situ hybridization, defines different subgroups in multiple myeloma. Leukemia, 19(2), 275–​278. PMID: 15538401. Yi X, Jia W, Jin Y, and Zhen S. 2014. Statin use is associated with reduced risk of haematological malignancies:  evidence from a meta-​analysis. PLoS One, 9(1), e87019. Zhan F, Huang Y, Colla S, et al. 2006. The molecular classification of multiple myeloma. Blood, 108(6), 2020–​2028. PMCID: PMC1895543. Zwick C, Held G, Auth M, et  al. 2014. Over one-​third of African-​American MGUS and multiple myeloma patients are carriers of hyperphosphorylated paratarg-​ 7, an autosomal dominantly inherited risk factor for MGUS/​MM. Int J Cancer, 135(4), 934–​938. PMID: 24443359.

814

 815

42

Bone Cancers LISA MIRABELLO, ROCHELLE E. CURTIS, AND SHARON A. SAVAGE

OVERVIEW Cancers arising from bone or cartilage account for about 0.2% of malignant neoplasms. They are histologically heterogeneous with multiple rare subtypes. As a group, there are few environmental risk factors for bone cancers, with the exception of the strong association between therapeutic radiation and increased risk of osteosarcoma. The genetic etiology is also better understood in osteosarcoma, although there have been limited studies in other types of bone cancers. The Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer Institute began collecting cancer incidence and survival data in 1973 (Howlader et al., 2015). SEER currently provides data on 18 cancer registries, representing 28% of the US population. This database identified more than 40 different subtypes of bone sarcomas diagnosed in the years 2001–​2012. Given the limited number of ecologic, epidemiologic, and genetic studies of the very rare bone cancer subtypes, this chapter is focused on the three major histologic types: osteosarcoma, Ewing sarcoma, and chondrosarcoma.

DEMOGRAPHICS AND DESCRIPTIVE EPIDEMIOLOGY Incidence During the calendar years 2001–​2012, there were 9418 cases of bone cancer reported to the 18 SEER cancer registries (Table 42–1; total age-​ adjusted incidence rate [IR] = 9.42 cases/​1,000,000). Osteosarcoma was the most common malignant tumor of the bone with 3047 cases (IR = 3.05), followed by chondrosarcoma with 2379 cases (IR = 2.37), Ewing family of tumors with 1230 cases (denoted Ewing sarcoma, IR = 1.23), and chordoma with 943 cases (IR = 0.94). Other bone tumor subtypes were uncommon, with most cases of osteosarcoma and chondrosarcoma reported as “not otherwise Specified,” or “NOS.” Male rates predominated over females for practically all bone cancer subtypes (Table 42–1; overall incidence rate ratio [IRR] = 1.37), although the male-​to-​female ratio varied from 1.27:1 for osteosarcoma and 1.30:1 for chondrosarcoma to 1.61:1 for Ewing sarcoma.

Age-​Specific Incidence Patterns

Osteosarcoma is notable in having a bimodal age-​distribution of incidence with peaks in adolescence and late in life, and it is the most common bone cancer encountered in children and adolescents (Anfinsen et al., 2011; Hung et al., 2014; Mirabello et al., 2009a, 2009b; Savage and Mirabello, 2011). Male rates were nearly equivalent to those of females for ages < 15  years, although rates were slightly higher for females than for males at ages 10–​14  years (Figure 42–1). IRs rose to a peak among females at ages 10–​14 years (IR = 8.90), with a later higher peak at 15–​19 years among men (IR = 10.7), correlating with the pubertal growth spurt, which occurs at a younger age in females than in males (Mirabello et al., 2009b; Savage and Mirabello, 2011). After age 20  years, osteosarcoma rates for both males and females declined and remained relatively constant for ages up to 60  years. Incidence rates climbed at older ages to reach a second peak in the seventh and eighth decades of life (males, ages 80–​84, IR = 5.35; females ages 75–​79, IR = 2.69). Mirabello et al. (2009b) reported that the most common anatomic site for early onset osteosarcoma was in the lower long bones (74.5%), whereas 11% occurred in the upper long bones.

The long bones remained the most common site for osteosarcoma occurring at older ages but at a reduced percentage (43% for ages 25–​ 59 and 27% for age ≥ 60 years). Among osteosarcomas diagnosed in SEER at ages ≥ 60 years, 161 of 428 cases (37.6%) were second or later primaries, confirming previous reports that second or later primary osteosarcomas accounted for a large percentage of this subtype at older ages (Mirabello et al., 2009b). Chondrosarcoma is infrequent in childhood, and rates rise with advancing age (Anfinsen et  al., 2011; Hung et  al., 2014). In SEER, male IRs predominated over female IRs at younger age of diagnoses (< 25 years) and at older ages (≥ 55 years), with no clear trend by gender for ages 25–​54 years. The age distribution of Ewing sarcoma was similar to that of osteosarcoma at younger ages, with a peak IR at ages 15–​19 years for males (IR = 5.43) and at ages 10–​14 for females (IR = 3.00), after which rates declined and few tumors were seen at older ages; male rates were uniformly greater than females for all age groups (Anfinsen et al., 2011; Hung et al., 2014). Chordoma was especially uncommon at ages under 25 years with rates progressively increasing to reach a peak at 70–​79 years of age, with males predominating in most age groups (data not shown; Hung et al., 2014; McMaster et al., 2001; Smoll et al., 2013). Anfinsen et al. (2011) evaluated age-​period-​cohort incidence trends in nine SEER registries for the period 1976–​2005. Incidence rates were stable over the 30-​year period for Ewing sarcoma, for males with osteosarcoma or chondrosarcoma, and for females with osteosarcoma or chondrosarcoma among those with tumor diagnosed during 1976–​ 1995. A significant increase in chondrosarcoma rates among females was seen in the latest period evaluated, 1996–​2005. Among patients with osteosarcoma as a first primary at ages over 60 years, a decline in rates was noted over that time period (1976–​2005).

Racial and Ethnic Differences

Table 42–2 describes the differences in age-​adjusted incidence rates for the major subtypes of bone cancer according to race and ethnicity. For all histologic types combined, Hispanic whites, blacks, and Asian/​ Pacific Islanders (APIs) were found to have lower IR than non-​Hispanic whites, with the ratio for Hispanic whites compared to non-​Hispanic whites reaching statistical significance (IRR = 0.89). However, rates by race/​ethnicity differed by bone subtype, with osteosarcomas occurring more frequently among Hispanic whites (IRR = 1.12) and blacks (IRR = 1.26) as compared to non-​Hispanic whites. SEER data showed that chondrosarcomas were less common among blacks, Hispanic whites, and APIs as compared with non-​Hispanic whites, with incidence rate ratios consistently below 0.75. Ewing sarcoma was notable for its rare occurrence among blacks (IR = 0.32) with low rates also seen for APIs (IR  =  0.72). The highest Ewing sarcoma rates were observed for non-​Hispanic whites (IR = 1.68), with somewhat lower rates among Hispanic whites (IR = 1.05).

International Incidence Patterns

Valery et  al. (2015) recently evaluated international variation in bone cancer incidence by morphologic subtype. Overall, the authors reported limited differences in the worldwide incidence patterns for osteosarcomas and chondrosarcomas, although differences in rates were greater for Ewing sarcoma, especially in areas where numbers were small. Overall age-​adjusted incidence rates (standardized to the Segi world standard population and converted here to rates

815

816

816

PART IV:  Cancers by Tissue of Origin

Table 42–1.  Age-​Adjusted Incidence Rates and Incidence Rate Ratios of Bone Cancer, Overall and by Gender, SEER-​18, 2001–​2012 Total Bone Classification and ICD-​O-​3 Code Morphology Total Bone Cancers ICCC VIII Classified Total Bone Cancers VIII(a) Total Osteosarcoma (9180–​9194)   Osteosarcoma, NOS (9180)   Chondroblastic osteosarcoma (9181)   Fibroblastic osteosarcoma (9182)   Telangiectatic osteosarcoma (9183)   Osteosarcoma in Paget disease (9184)   Small cell osteosarcoma (9185)   Central osteosarcoma (9186)   Other osteosarcoma (9187, 9192–​9194) VIII(b) Total Chondrosarcoma (9220–​9243)   Chondrosarcoma, NOS (9220)   Juxtacortical chondrosarcoma (9221)   Chondroblastoma, malignant (9230)   Mesenchymal chondrosarcoma (9240)   Clear cell chondrosarcoma (9242)   Dedifferentiated chondrosarcoma (9243) VIII(c) Total Ewing Sarcoma, Related Bone Sarcomas (9260, 9363–​9365)   Ewing sarcoma, Askin tumor of bone (9260, 9365)   PNET (9363–9364) VIII(d) Total Other Specified Bone Tumors (8810–​8830, 9250, 9261–​9342, 9370–​9372)   Chordoma (9370–​9372)   Fibrosarcoma (8810–​8812)   Fibrous histiocytoma (8830)   Giant cell tumor of bone (9250)   Adamantinoma of long bones (9261)   Odontogenic (9270–​9275, 9280–​9282, 9290, 9300–​9302, 9310–​9312, 9320–​9322, 9330, 9340–​9342) VIII(e) Total Unspecified bone   Unspecified (8000-8005)   Sarcoma (8800-8805) Other Bone Cancers (Site=bone, not classified as ICCC VIII=bone)† Hemangiosarcoma and other blood vessel bone tumors (9120,9130,9133)   Myxoid chondrosarcoma (9231)   Squamous cell carcinoma (8070-8074)   Giant cell sarcoma (8802)   All other bone tumors, not classified as ICCC VIII=bone

Male

Female

No.

IR

No.

IR

No.

IR

M:F IRR

9418 8825 3047 2045 441 154 96 27 30 75 179 2379 2035 32 24 41 48 199 1230

9.42 8.82 3.05 2.05 0.44 0.16 0.10 0.03 0.03 0.08 0.18 2.37 2.02 0.03 0.02 0.04 0.05 0.20 1.23

5273 4926 1696 1145 250 74 62 18 17 43 87 1277 1071 25 20 18 38 105 770

10.96 10.19 3.42 2.32 0.50 0.15 0.12 0.04 0.03 0.09 0.17 2.69 2.26 0.05 0.04 0.04 0.08 0.23 1.51

4145 3899 1351 900 191 80 34 9 13 32 92 1102 964 7 4 23 10 94 460

8.02 7.56 2.70 1.80 0.38 0.16 0.07 ~ ~ 0.06 0.18 2.09 1.83 ~ ~ 0.04 ~ 0.17 0.94

1.37* 1.35* 1.27* 1.29* 1.30* 0.92 1.80* ~ ~ 1.35 0.91 1.30* 1.23* ~ ~ 0.79 ~ 1.35* 1.61*

1163 67 1546

1.16 0.07 1.55

727 43 877

1.42 0.09 1.86

436 24 669

0.89 0.05 1.27

1.60* 1.75* 1.46*

943 64 127 184 44 184

0.94 0.06 0.13 0.18 0.04 0.19

540 33 69 89 23 123

1.16 0.07 0.15 0.18 0.04 0.26

403 31 58 95 21 61

0.76 0.06 0.11 0.19 0.04 0.12

1.53* 1.07 1.42 0.98 1.05 2.19*

623 326 297 593

0.63 0.33 0.30 0.60

306 139 167 347

0.71 0.34 0.37 0.77

317 187 130 246

0.57 0.32 0.24 0.46

1.11* 1.04 1.53* 1.67*

115

0.12

71

0.16

44

0.08

1.96*

164 137 40 137

0.17 0.14 0.04 0.14

100 75 25 76

0.22 0.17 0.06 0.16

64 62 15 61

0.12 0.11 0.03 0.11

1.74* 1.57* 2.13* 1.36

Abbreviations: SEER-18, 18 cancer registry areas of the Surveillance, Epidemiology and End Results Program; ICD-O-3, International Classification of Diseasae for Oncology, third edition; ICCC, International Classification of Childhood Cancer, Third Edition; No., number of cases; IR, incidence rate; IRs are per 1,000,000 and age-adjusted to the 2000 US Standard Population (19 age groups - Census P25-1130) standard; Confidence intervals (Tiwari mod) are 95% for rates. M:F IRR, Male to Female IR ratio; ~, IRs not specified for G; OR 6.7, 95% CI 1.06–41.6, P = 0.04) and rs1042522 (Ex4+119C>G, P72R; OR 7.5, 95% CI 1.2–46.3, P = 0.03) IGF2R Two linked IGF2R SNPs, rs998075 (Ex16+88G>A) and rs998074 (IVS16+15C>T), associated with increased risk (haplotype OR 2.04, 95% CI 1.29–3.24, P = 0.006) MTHFR MTHFR C677T not associated with osteosarcoma MDM2 Rs2279744 (SNP309T>G) was associated with high grade osteosarcoma in females (OR 1.96, 95% CI 1.04–3.67, P = 0.04) TP53 Rs1042522 (Ex4+119C>G, P72R) was not associated with risk of osteosarcoma, but was associated with survival (hazard ratio 2.90, 95% CI 1.28–6.66, P = 0.02) TGFBR1 TGFBR1*6A was associated with increased susceptibility (OR 4.59, 95% CI 2.33–7.97, P = 0.0018) 8q24 region Strongest association with increased risk at rs896324 (OR 1.75, 95% CI 1.13–2.69, P = 0.01) GSTM1, No polymorphisms were signicantly associated GSTM3, with risk; SNPs were significantly associated GSTT1 with clinical outcome 39 Telomere 7 linked SNPs in TERF1 were associated with biology osteosarcoma risk after correction for multiple genes tests (OR 0.48, 0.53, 95% CI 0.33–0.76, Gene P = 0.0009) 161 DNA 6 SNPs within or near FANCM (OR 1.95–2.00, repair P = 0.003), MDM2 (OR 0.62, P = 0.003, or genes MPG (OR 4.80, P = 0.004) were significantly associated with osteosarcoma after correction for multiple tests 62 Growth/​ 6 SNPs within or near FGF2 (OR 2.12, hormone P = 0.002), FGFR3 (OR 1.51, P = 0.007), metabolism GH1 (OR 1.61 and 0.51, P = 0.001), GNRH2 genes (OR 1.60, P = 0.002), or IGF1 (OR 0.53, P = 0.002) were significantly associated with OS after correction for multiple tests 28 Bone No significant associations with osteosarcoma formation genes 4 Ribosomal No significant associations with osteosarcoma protein genes CTLA-​4 rs231775 (+49G>A) was associated with an increased risk of OS (OR 1.41, 95% CI 1.07– 1.87, P = 0.015) CTLA-​4 rs231775 (+49G>A) was associated with increased risk of OS (OR 1.32, 95% CI 1.03– 1.69, P = 0.029) CD86 rs1129055 (+1057G>A) was associated with an increased susceptibility to OS (OR 1.43, 95% CI 1.08–1.88, P = 0.011) (continued)

824

Table 42–7. Continued First Author, Year

N Cases/ ​N Controls

Country

Adjustments/​Matching Factors† N SNPs

Ancestry

Hu, 2011

168/​168

China

Chinese

None/​Age, gender, and population

1

Lu et al., 2011

110/​226

North China

Chinese

Liu, 2012

326/​433

China

Han Chinese

Age, gender, and family history/​Age and population None/​Age, gender, and population

Naumov, 2012

26/​96‡

Russia

Russian

None/​Similar age, gender and population

Musselman, 2012

290/​290

Variable

Variable

None/​None*

Savage, 2013

941/​3291 International, European multi-​ institutional study

Mirabello, 2015

227/​708§ International, European, Study center¶/​all multi-​ African, and cases, similar age, institutional Brazilian gender and ethnicity study

Gender, country, eigenvectors/​ Ethnicity

Gene(s)

Main Association Finding

rs334354 (Int7G24A) was associated with an increased risk of OS (homozygous OR 2.89, 95% CI 1.46, 4.92, P trend = 0.001) 2 GSTM1, Null GSTT1 genotype was significantly GSTT1 associated with an increased risk of osteosarcoma (OR 2.07, 95% CI 1.19, 3.04) 3 LOX −22G>C and 473G>A polymorphisms were significantly increased in osteosarcoma cases (OR 3.88, 95% CI 1.94, 7.78, P = 4.18x10-​ 5 ,and OR 1.38, 95% CI 1.07, 1.78, P = 0.013, respectively) 6 FGFR3, rs6599400 in FGFR3 and rs1690916 in MDM2 FGF2, were significantly associated with risk of bone MDM2, sarcomas, including osteosarcoma (OR 2.15, IGF1, GH1, 95% CI 1.06, 4.24, and OR 0.39, 95% CI 0.19, GNRH2 0.78, respectively) 798 42 estrogen 3 SNPs within or near AR (RR 8.26, P < 0.0001), metabolism IGF2R (RR 0.0001, P < 0.0001), or IGFALS and insulin-​ (RR 0.22, P < 0.0001) were significantly like growth associated with osteosarcoma after correction factor/​ for multipe tests growth hormone genes 698,968 Genome-​wide 2 loci reached genome-​wide significance: 1 SNP association in the GRM4 gene (rs1906953, OR 1.57, 95% study CI 1.35–1.83, P = 8.1 × 10–​9) and rs7591996 and rs10208273 in a gene desert at 2p25.2 (OR 1.39, 95% CI 1.23–1.54, P = 1.0 × 10–​8, and OR 1.35, 95% CI 1.21–1.52, P = 2.9 × 10–​7, respectively) 447,040 Genome-​wide 1 locus reached genome-​wide significance in association all populations: rs7034162, and intronic SNP study in the nuclear factor I/​B gene (NFIB) on chromosome 9p24.1 (OR 2.43, 95% CI 1.83–3.24, P = 1.2 × 10–​9) TGFBR1

Abbreviations: N: number of individuals or SNPs; SNP: single nucleotide polymorphism; OR: odds ratio; CI: confidence interval; RR: relative risk † Adjustments in the statistical model for association testing/​case-​control matching factors or characteristics are given. * Case-​parent study, population stratification is inherently corrected for. ‡ Examined 68 bone tumor cases, of which 26 were osteosarcoma cases. § Comparisons were of cases with and without metastasis at diagnosis, numbers for these groups are shown here. ¶ Association results were evaluated separate for each ethnic group and combined with meta-​analyses.

Table 42–8.  Association Studies Evaluating Common SNPs and Ewing Sarcoma First Author, Year

N Cases / ​N Controls

Country

Ancestry

Ruza, 2003 53/​143

Spain

Caucasian

Postel-​ Vinay, 2011

France

European

Silva, 2012 24/​200§

Southern Brazil

Brazilian

DuBois, 2012

135/​200/​285

US

European

Thurow, 2013

24/​91

Southern Brazil

Brazilian

401/​684‡

Adjustments/​Matching Factors† N SNPs

Gene(s)

None/​Similar age, gender, and geographic region Population substructure

3 2 1 286,966

VDR ESR1 COL1A1 Genome-​wide association study

Geographically matched cases and controls

3

EWSR1

21

EWSR1

1

TP53

1

MDM2

Geographically matched cases and controls

Main Association Finding No associations with Ewing sarcoma No associations with Ewing sarcoma No association with Ewing sarcoma rs9430161 (P = 1.4 × 10e−20; OR = 2.2) located 25 kb upstream of TARDBP, and rs224278 (P = 4.0 × 10e−17; OR = 1.7) located 5 kb upstream of EGR2 and, to a lesser extent, rs4924410 at 15q15 (P = 6.6 × 10e−9; OR = 1.5) rs4820804 homozygous TT was associated with Ewing sarcoma, P = 0.02 EWSR1 SNPs not associated with Ewing sarcoma. High degree of variation between white and African-​American groups No association between TP53 Arg72Pro and Ewing sarcoma. MDM2 T309G promoter polymorphism associated with Ewing sarcoma

Abbreviations: OR: odds ratio; kb: kilobase; VDR: vitamin D receptor; ESR1: estrogen receptor 1; COL1A1: collagen type 1 alpha 1; TARDBP: TAR DNA-​binding protein; EGR2: early growth response 2. ‡ 401 French individuals with Ewing, 684 French controls in primary analyses; 3668 European ancestry controls added to address population substructure; 64 and 282 Ewing sarcoma cases from two different studies were genotyped at 24 loci in the replication study. § Also included 54 family members. * 135 white Ewing sarcoma patients, 200 white controls with Wilms tumor, and 285 cancer-​free African-​Americans.

 825



Bone Cancers

at these loci were associated with expression levels of TARDBP, ADO, and EGR2, suggesting a role in the etiology of Ewing sarcoma. A follow-​up of the Ewing sarcoma GWAS found that EGR2 knockdown inhibited cell proliferation and growth (Grunewald et al., 2015). Targeted deep sequencing of the EGR2 locus in Ewing sarcoma cases and controls identified 291 risk-​associated SNPs, and further work showed that one SNP, rs79965208, changed the epigenetic characteristics of an active regulatory element at a region where EWSR1-​FLI1 are bound. This suggested cooperation between the somatic EWSR1-​ FLI1 fusion protein and germline transcriptional regulation by EGR2.

Chondrosarcoma Chondrosarcoma is a cartilage-​ forming neoplasm composed of malignant cells from cartilage without osteoid formation (Leddy and Holmes, 2014). In contrast to other bone cancers, chondrosarcoma occurs in middle-​aged to elderly adults (Qasem and DeYoung, 2014). Hereditary forms of multiple exostoses (HME) are autosomal dominant inherited disorders of cartilage formation (Pansuriya et al., 2010; Pedrini et al., 2011). Patients with HME typically develop osteochondromas, but are also at increased risk of chondrosarcoma due to malignant transformation (Table 42–6). The enchondromatoses, such as Ollier disease and Mafucci syndrome, are a spectrum of disorders in which patients have multiple benign cartsupilage tumors of the bone and skeletal deformities. These tumors may be at risk of transformation to chondrosarcoma or osteosarcoma. Other than a very small study of a SNP in the MTHFR gene (Ozger et al., 2008), studies of rare or common germline genetic variants and chondrosarcoma risk have not yet been conducted.

FUTURE DIRECTIONS The heterogeneity and relative rarity of bone cancers create significant complexity in the design and interpretation of etiologic studies. Other than therapeutic radiation, the contribution of environmental exposures to these tumors is not understood. Host factors, including puberty, height, and birth weight, contribute to osteosarcoma and Ewing sarcoma in adolescents and young adults but the biological mechanisms of these associations are not clear. A great deal has been learned about both rare and common germline genetic variants in osteosarcoma and Ewing sarcoma, but chondrosarcoma and other rare bone tumors are less well studied. Large, longitudinal cohort studies of bone cancers are required to better understand their etiology. References Anfinsen KP, Devesa SS, Bray F, et  al. 2011. Age-​period-​cohort analysis of primary bone cancer incidence rates in the United States (1976-​2005). Cancer Epidemiol Biomarkers Prev, 20(8), 1770–​1777. PMID: 21724855. Arora R, Kontopantelis E, Alston R, et al. 2011. Relationship between height at diagnosis and bone tumours in young people: a meta-​analysis. Cancer Causes Control, 22(5), 681–​688. Barnette P, Scholl R, Blandford M, et al. 2004. High-​throughput detection of glutathione S-​transferase polymorphic alleles in a pediatric cancer population. Cancer Epidemiol Biomarkers Prev, 13(2), 304–​313. Bassin EB, Wypij D, Davis RB, and Mittleman MA. 2006. Age-​specific fluoride exposure in drinking water and osteosarcoma (United States). Cancer Causes Control, 17 (4), 421–​428. PMID: 16596294. Beck A. 1922. Zur Frage des Roritgensarkoms, zugleich ein Beitrag zur Pathogenese des sarkoms. Munchen Med Wchnschr, 69, 623–​624. Berrington de Gonzalez A, Curtis RE, Kry SF, et al. 2011. Proportion of second cancers attributable to radiotherapy treatment in adults:  a cohort study in the US SEER cancer registries. Lancet Oncol, 12(4), 353–​360. PMCID: PMC3086738. Berrington de Gonzalez A, Gilbert E, Curtis R, et al. 2013. Second solid cancers after radiation therapy:  a systematic review of the epidemiologic studies of the radiation dose-​response relationship. Int J Radiat Oncol Biol Phys, 86(2), 224–​233. PMCID: PMC3816386. Berrington de Gonzalez A, Kutsenko A, and Rajaraman P. 2012. Sarcoma risk after radiation exposure. Clin Sarcoma Res, 2(1), 18. PMCID: PMC3507855. Bolland MJ, and Cundy T. 2013. Paget’s disease of bone: clinical review and update. J Clin Pathol, 66(11), 924–​927. PMID: 24043712.

825

Boria I, Garelli E, Gazda H, et  al. 2010. The ribosomal basis of Diamond-​ Blackfan anemia:  mutation and database update. Hum Mutat, 31, 1269–​1279. Broström L, Adamson U, Filipsson R, and Hall K. 1979. Longitudinal growth and dental development in osteosarcoma patients. Acta Orthop Scand, 51, 755–​759. Buckley J, Pendergrass T, Buckley C, et  al. 1998. Epidemiology of osteosarcoma and Ewing’s sarcoma in children: a study of 305 cases by the Children’s Cancer Group. Cancer, 83, 1440–​1448. Burningham Z, Hashibe M, Spector L, and Schiffman JD. 2012. The epidemiology of sarcoma. Clin Sarcoma Res, 2(1), 14. PMCID: PMC3564705. Calvert G, Randall R, Jones K, et al. 2012. At-​risk populations for osteosarcoma: the syndromes and beyond. Sarcoma, 2012, 152382. Chen S, Yang L, Pu F, et al. 2015. High birth weight increases the risk for bone tumor: a systematic review and meta-​analysis. Int J Environ Res Public Health, 12(9), 11178–​11195. PMID: PMC4586668. Choi EY, Gardner JM, Lucas DR, McHugh JB, and Patel RM. 2014. Ewing sarcoma. Semin Diagn Pathol, 31(1), 39–​47. PMID: 24680181. Chompret A, Abel A, Stoppa-​Lyonnet D, et al. 2001. Sensitivity and predictive value of criteria for p53germline mutation screening. J Med Genetics, 38(1), 43–​47. Cool W, Grimer R, Carter S, Tillman R, and Davies A. 1998. Longitudinal growth following treatment for osteosarcoma. Sarcoma, 2, 115–​119. Cote GM, and Choy E. 2013. Update in treatment and targets in Ewing sarcoma. Hematol Oncol Clin North Am, 27(5), 1007–​1019. PMID: 24093173. Cotterill S, Wright C, Pearce M, and Craft A. 2004. Stature of young people with malignant bone tumors. Pediatr Blood Cancer, 42, 59–​63. Douglass CW, and Joshipura K. 2006. Caution needed in fluoride and osteosarcoma study. Cancer Causes Control, 17(4), 481–​ 482. PMID: 16596300. Duchman KR, Gao Y, and Miller BJ. 2015a. Prognostic factors for survival in patients with Ewing’s sarcoma using the Surveillance, Epidemiology, and End Results (SEER) Program database. Cancer Epidemiol, 39(2), 189–​ 195. PMID: 25595632. Duchman KR, Gao Y, and Miller BJ. 2015b. Prognostic factors for survival in patients with high-​grade osteosarcoma using the Surveillance, Epidemiology, and End Results (SEER) Program database. Cancer Epidemiol, 39(4), 593–​599. PMID: 26002013. Eyre R, Feltbower RG, Mubwandarikwa E, et al. 2009. Incidence and survival of childhood bone cancer in northern England and the West Midlands, 1981–​2002. Br J Cancer, 100(1), 188–​193. PMCID: PMC2634696. Farrar JE, Vlachos A, Atsidaftos E, et al. 2011. Ribosomal protein gene deletions in Diamond-​Blackfan anemia. Blood, 118, 6943–​6951. Fraumeni JF. 1967. Stature and malignant tumors of the bone in childhood and adolescence. Cancer, 20, 967–​973. Gazda HT, Preti M, Sheen MR, et al. 2012. Frameshift mutation in p53 regulator RPL26 is associated with multiple physical abnormalities and a specific pre-​ribosomal RNA processing defect in diamond-​blackfan anemia. Human Mutation, 33, 1037–​1044. Gazzano E, Bergandi L, Riganti C, et al. 2010. Fluoride effects: the two faces of janus. Curr Med Chem, 17(22), 2431–​2441. PMID: 20491635. Gelberg K, Fitzgerald E, Hwang S, and Dubrow R. 1997. Growth and development and other risk factors for osteosarcoma in children and young adults. Int J Epidemiol, 26, 272–​278. Giuffrida AY, Burgueno JE, Koniaris LG, et al. 2009. Chondrosarcoma in the United States (1973 to 2003): an analysis of 2890 cases from the SEER database. J Bone Joint Surg Am, 91(5), 1063–​1072. PMID: 19411454. Glasser D, Duane K, Lane J, Healey J, and Caparros-​Sison B. 1991. The effect of chemotherapy on growth in the skeletally immature individual. Clin Orthop Relat Res, 262, 93–​100. Gomez NC, and Davis IJ. 2015. Linking germline and somatic variation in Ewing sarcoma. Nat Genet, 47(9), 964–​965. PMID: 26313223. Goode A, and Layfield R. 2010. Recent advances in understanding the molecular basis of Paget disease of bone. J Clin Pathol, 63(3), 199–​ 203. PMID: 19858527. Goodman M, McMaster J, Drash A, et al. 1978. Metabolic and endocrine alterations in osteosarcoma patients. Cancer, 42, 603–​610. Grunewald TG, Bernard V, Gilardi-​Hebenstreit P, et al. 2015. Chimeric EWSR1-​ FLI1 regulates the Ewing sarcoma susceptibility gene EGR2 via a GGAA microsatellite. Nat Genet, 47(9), 1073–​1078. PMCID: PMC4591073. Hansen MF, Nellissery MJ, and Bhatia P. 1999. Common mechanisms of osteosarcoma and Paget’s disease. J Bone Miner Res, 14(Suppl 2), 39–​44. PMID: 10510212. Hartley A, Birch J, McKinney P, et al. 1988. The inter-​regional epidemiological study of childhood cancer (IRESCC): case control study of children with bone and soft tissue sarcomas. Br J Cancer, 58, 838–​842. He J, Wang J, Wang D, et  al. 2013. Association analysis between genetic variants of MDM2 gene and osteosarcoma susceptibility in Chinese. Endocrine Journal, 60(11), 1215–​1220.

826

826

PART IV:  Cancers by Tissue of Origin

Henderson TO, Rajaraman P, Stovall M, et  al. 2012. Risk factors associated with secondary sarcomas in childhood cancer survivors: a report from the Childhood Cancer Survivor Study. Int J Radiat Oncol Biol Phys, 84(1), 224–​230. PMCID: PMC3423483. Holly EA, Aston DA, Ahn DK, and Kristiansen JJ. 1992. Ewing’s bone sarcoma, paternal occupational exposure, and other factors. Am J Epidemiol, 135(2), 122–​129. Howlader N, Noone AM, Krapcho M, Garshell J, Miller D, Altekruse SF, Kosary CL, Yu M, Ruhl J, Tatalovich Z, Mariotto A, Lewis DR, Chen HS, Feuer EJ, Cronin KA (Eds). 2015. SEER Cancer Statistics Review, 1975–​2012. National Cancer Institute, Bethesda, MD, November 2014 SEER data submission. Hu Y, Pan Y, Li W, et al. 2010. Association between TGFBR1*6A and osteosarcoma: a Chinese case-​control study. BMC Cancer, 10(1), 169. Hu Y-​S, Pan Y, Li W-​H, et al. 2011. Int7G24A variant of transforming growth factor-​beta receptor 1 is associated with osteosarcoma susceptibility in a Chinese population. Med Oncol, 28(2), 622–​625. Hum L, Kreiger N, and Finkelstein MM. 1998. The relationship between parental occupation and bone cancer risk in offspring. Int J Epidemiol, 27(5), 766–​771. PMID: 9839731. Hung GY, Horng JL, Yen HJ, et al. 2014. Incidence patterns of primary bone cancer in Taiwan (2003–​ 2010): a population-​ based study. Ann Surg Oncol, 21(8), 2490–​2498. PMCID: PMC4082651. Infante-​Rivard C, and Weichenthal S. 2007. Pesticides and childhood cancer: an update of Zahm and Ward’s 1998 review. J Toxicol Environ Health B Crit Rev, 10(1–​2), 81–​99. PMID: 18074305. Jiang C, Chen H, Shao L, and Dong Y. 2014. GRM4 gene polymorphism is associated with susceptibility and prognosis of osteosarcoma in a Chinese Han population. Med Oncol (Northwood, London, England), 31(7), 50. PMID: PMC4079940. Jiang Y, Janku F, Subbiah V, et al. 2013. Germline PTPRD mutations in Ewing sarcoma: biologic and clinical implications. Oncotarget, 4(6), 884–​889. PMCID: PMC3757245. Kansara M, and Thomas D. 2007. Molecular pathogenesis of osteosarcoma. DNA Cell Biol, 26(1), 1–​18. Koshkina N, Kleinerman E, Li G, et al. 2007. Exploratory analysis of Fas gene polymorphisms in pediatric osteosarcoma patients. J Pediatr Hematol Oncol, 29, 815–​821. Labutina EV, Kuznetsova IS, Hunter N, Harrison J, and Koshurnikova NA. 2013. Radiation risk of malignant neoplasms in organs of main deposition for plutonium in the cohort of Mayak workers with regard to histological types. Health Phys, 105(2), 165–​176. PMID: 23799501. Laurin N, Brown JP, Morissette J, and Raymond V. 2002. Recurrent mutation of the gene encoding sequestosome 1 (SQSTM1/​p62) in Paget disease of bone. Am J Hum Genet, 70(6), 1582–​1588. PMCID: PMC379146. Leddy LR, and Holmes RE. 2014. Chondrosarcoma of bone. Cancer Treat Res, 162, 117–​130. PMID: 25070233. Lewis DR, and Ries LAG. 2007. Cancers of the bone and joint. SEER survival monograph: cancer survival among adults: U.S. SEER Program, 1988–​ 2001, patient and tumor characteristics, Ries LAG, Keel GE, Eisner MP, Lin YD, Horner M-​J (Eds.) NIH Pub. No. 07-​6215 ed (pp. 81–​87). Bethesda, MD: National Cancer Institute, SEER Program. Li F, and Fraumeni JF. 1969a. Rhabdomyosarcoma in children: epidemiologic study and identification of a familial cancer syndrome. J Natl Cancer Inst, 43(6), 1365–​1373. Li F, and Fraumeni JF. 1969b. Soft-​tissue sarcomas, breast cancer, and other neoplasms: a familial syndrome? Ann Intern Med, 71(4), 747–​752. Lima G, Gomes E, Nunes R, et al. 2006. Osteosarcoma and acromegaly: a case report and review of the literature. J Endocrinol Invest, 29(11), 1006–​1011. Liu Y, He Z, Feng D, et  al. 2011. Cytotoxic T-​lymphocyte antigen-​4 polymorphisms and susceptibility to osteosarcoma. DNA Cell Biol, 30(12), 1051–​1055. Liu Y, Lv B, He Z, et al. 2012. Lysyl oxidase polymorphisms and susceptibility to osteosarcoma. PLoS One, 7(7), e41610. Lohmann D. 2010. Retinoblastoma. Adv Exp Med Biol, 685, 220–​227. Longhi A, Pasini A, Cicognami A, et al. 2005. Height as a risk factor for osteosarcoma. J Pediatr Hematol Oncol, 27, 314–​318. Lu X, Yang W, Wan Z, Li J, and Bi Z. 2011. Glutathione S-​transferase polymorphisms and bone tumor risk in China. Asian Pac J Cancer Prev, 12(12), 3357–​3360. Malkin D. 2011. Li-​Fraumeni syndrome. Genes Cancer, 2(4), 475–​484. Martland HS, and Humphries RE. 1929. Osteogenic sarcoma in dial painters using luminous paint. Arch Pathol Lab Med, 7(3), 406–​417. PMID: WOS:000202343700004. McMaster ML, Goldstein AM, Bromley CM, Ishibe N, and Parry DM. 2001. Chordoma:  incidence and survival patterns in the United States, 1973–​ 1995. Cancer Causes Control, 12(1), 1–​11. PMID: 11227920.

Miller RW, Boice JD, Jr., and Curtis RE. 2006. Bone cancer. In: Schottenfeld D, and Fraumeni JF, Jr. (Eds.), Cancer epidemiology and prevention (3rd edition, pp. 946–958). New York: Oxford University Press. Mirabello L, Berndt S, Seratti G, et  al. 2010. Genetic variation at chromosome 8q24 in osteosarcoma cases and controls. Carcinogenesis, 31(8), 1400–​1404. Mirabello L, Koster R, Moriarity BS, et al. 2015a. A genome-​wide scan identifies variants in NFIB associated with metastasis in patients with osteosarcoma. Cancer Discovery, 5(9), 920–​931. Mirabello L, Macari ER, Jessop L, et al. 2014. Whole-​exome sequencing and functional studies identify RPS29 as a novel gene mutated in multicase Diamond-​Blackfan anemia families. Blood, 124(1), 24–​32. Mirabello L, Pfeiffer R, Murphy G, et al. 2011a. Height at diagnosis and birth-​ weight as risk factors for osteosarcoma. Cancer Causes Control, 22(6), 899–​908. Mirabello L, Richards E, Duong L, et al. 2011b. Telomere length and variation in telomere biology genes in individuals with osteosarcoma. Int J Mol Epidemiol Genet, 2(1), 19–​29. Mirabello L, Troisi RJ, and Savage SA. 2009a. International osteosarcoma incidence patterns in children and adolescents, middle ages and elderly persons. Int J Cancer, 125(1), 229–​234. PMCID: PMC3048853. Mirabello L, Troisi RJ, and Savage SA. 2009b. Osteosarcoma incidence and survival rates from 1973 to 2004:  data from the Surveillance, Epidemiology, and End Results Program. Cancer, 115(7), 1531–​1543. PMCID: PMC2813207. Mirabello L, Yeager M, Mai PL, et al. 2015b. Germline TP53 variants and susceptibility to osteosarcoma. JNCI, 107(7). Mirabello L, Yu K, Berndt S, et  al. 2011c. A comprehensive candidate gene approach identifies genetic variation associated with osteosarcoma. BMC Cancer, 11, 209. Musselman J, Bergemann T, Ross J, et al. 2012. Case-​parent analysis of variation in pubertal hormone genes and pediatric osteosarcoma: a Children’s Oncology Group (COG) study. Int J Mol Epidemiol Genet, 3(4), 286–​293. Nasterlack M. 2007. Pesticides and childhood cancer:  an update. Int J Hyg Environ Health, 210(5), 645–​657. PMID: 17434797. Naumov V, Generozov E, Solovyov Y, Aliev M, and Kushlinsky N. 2012. Association of FGFR3 and MDM2 gene nucleotide polymorphisms with bone tumors. Bull Exp Biol Med, 153(6), 869–​873. O’Neill KA, Murphy MF, Bunch KJ, et al. 2015. Infant birthweight and risk of childhood cancer: international population-​based case control studies of 40 000 cases. Int J Epidemiol, 44(1), 153–​168. Olivier M, Goldgar D, Sodha N, et  al. 2003 Li-​Fraumeni and related syndromes:  correlation between tumor type, family structure, and TP53 genotype. Cancer Res, 63(20), 6643–​6650. Operskalski E, Preston-​Martin S, Henderson B, and Visscher B. 1987. A case-​ control study of osteosarcoma in young persons. Am J Epidemiol 126, 118–​126. Operskalski EA, Preston-​Martin S, Henderson BE, and Visscher BR. 1987. A case-​control study of osteosarcoma in young persons. Am J Epidemiol 126(1), 118–​126. Ozger H, Kilicoglu O, Yilmaz H, et al. 2008. Methylenetetrahydrofolate reductase C677T gene polymorphism in osteosarcoma and chondrosarcoma patients. Folia Biol (Praha), 54(2), 53–​57. PMID: 18498722. Pansuriya TC, Kroon HM, and Bovee JV. 2010. Enchondromatosis:  insights on the different subtypes. Int J Clin Exp Pathol, 3(6), 557–​ 569. PMCID: PMC2907117. Patio-​Garcia A, Sotillo-​Pieiro E, Modesto C, and Sierrases-​Maga L. 2000. Analysis of the human tumour necrosis factor-​alpha (TNFalpha) gene promoter polymorphisms in children with bone cancer. J Med Genet, 37, 789–​792. Pedrini E, Jennes I, Tremosini M, et al. 2011. Genotype-​phenotype correlation study in 529 patients with multiple hereditary exostoses:  identification of “protective” and “risk” factors. J Bone Joint Surg Am, 93(24), 2294–​ 2302. PMID: 22258776. Pendergrass T, Foulkes M, Robison L, and Nesbit M. 1984. Stature and Ewing’s sarcoma in childhood. Am J Pediatr Hematol Oncol, 6(1), 33–​39. Pollak M. 2012. The insulin and insulin-​like growth factor receptor family in neoplasia: an update. Nat Rev Cancer, 12(3), 159–​169. Postel-​ Vinay S, Veron AS, Tirode F, et  al. 2012. Common variants near TARDBP and EGR2 are associated with susceptibility to Ewing sarcoma. Nat Genet, 44(3), 323–​327. PMID: 22327514. Potratz J, Jurgens H, Craft A, and Dirksen U. 2012. Ewing sarcoma: biology-​ based therapeutic perspectives. Pediatr Hematol Oncol, 29(1), 12–​27. PMID: 22304007. Preston DL, Ron E, Tokuoka S, et  al. 2007. Solid cancer incidence in atomic bomb survivors:  1958–​ 1998. Radiat Res, 168(1), 1–​ 64. PMID: 17722996.

 827



Bone Cancers

Pui C, Dodge R, George S, and Green A. 1987. Height at diagnosis of malignancies. Arch Dis Child, 62, 495–​499. Qasem SA, and DeYoung BR. 2014. Cartilage-​forming tumors. Semin Diagn Pathol, 31(1), 10–​20. PMID: 24680178. Randhawa R, and Cohen P. 2005. The role of the insulin-​like growth factor system in prenatal growth. Molec Gen Metab, 86(1), 84–​90. Rizzo JD, Curtis RE, Socie G, et  al. 2009. Solid cancers after allogeneic hematopoietic cell transplantation. Blood, 113(5), 1175–​ 1183. PMCID: PMC2635083. Ruza E, Sotillo E, Sierrasesúmaga L, Azcona C, and Patiño-​Garcia A. 2003. Analysis of polymorphisms of the vitamin D receptor, estrogen receptor, and collagen Iα1 genes and their relationship with height in children with bone cancer. J Pediatr Hematol Oncol, 25, 780–​786. Rytting M, Pearson P, Raymond A, et al. 2000. Osteosarcoma in preadolescent patients. Clin Orthop Relat Res, 373, 39–​50. Salinas-​Souza C, Petrilli A, and de Toledo S. 2010. Glutathione S-​transferase polymorphisms in osteosarcoma patients. Pharmacogenet Genomics, 20(8), 507–​515. Samartzis D, Nishi N, Hayashi M, et al. 2011. Exposure to ionizing radiation and development of bone sarcoma: new insights based on atomic-​bomb survivors of Hiroshima and Nagasaki. J Bone Joint Surg Am, 93(11), 1008–​1015. PMID: 21984980. Sankaran VG, Ghazvinian R, Do R, et al. 2012. Exome sequencing identifies GATA1 mutations resulting in Diamond-​Blackfan anemia. J Clin Invest, 122(7), 2439–​2443. Savage SA, Burdett L, Troisi R, et  al. 2007a. Germ-​line genetic variation of TP53 in osteosarcoma. Pediatr Blood Cancer, 49, 28–​33. Savage SA, and Mirabello L. 2011. Using epidemiology and genomics to understand osteosarcoma etiology. Sarcoma, 2011, 1–13. PMCID: PMC3061299. Savage SA, Mirabello L, Wang Z, et al. 2013. Genome-​wide association study identifies two susceptibility loci for osteosarcoma. Nat Genet, 45(7), 799–​803. Savage SA, Woodson K, Walk E, et  al. 2007b. Analysis of genes critical for growth regulation identifies Insulin-​like Growth Factor 2 Receptor variations with possible functional significance as risk factors for osteosarcoma. Cancer Epidemiol Biomarkers Prev, 16, 1667–​1674. Schwartz B, Benadjaoud MA, Clero E, et al. 2014. Risk of second bone sarcoma following childhood cancer:  role of radiation therapy treatment. Radiat Environ Biophys, 53(2), 381–​390. PMCID: PMC3996275. Scranton P, DeCicco F, Totten R, and Yunis E. 1975. Prognostic factors in osteosarcoma: a review of 20 years’ experience at the University of Pittsburgh Health Center Hospitals. Cancer, 36, 2179–​2191. Siitonen HA, Sotkasiira J, Biervliet M, et al. 2008. The mutation spectrum in RECQL4 diseases. Eur J Hum Genet, 17(2), 151–​158. Silventoinen K, Sammalisto S, Perola M, et  al. 2003. Heritability of adult body height: a comparative study of twin cohorts in eight countries. Twin Research, 6(5), 399–​408. Smoll NR, Gautschi OP, Radovanovic I, Schaller K, and Weber DC. 2013. Incidence and relative survival of chordomas:  the standardized mortality ratio and the impact of chordomas on a population. Cancer, 119(11), 2029–​2037. PMID: 23504991. Sokolnikov ME, Gilbert ES, Preston DL, et  al. 2008. Lung, liver and bone cancer mortality in Mayak workers. Int J Cancer, 123(4), 905–​911. PMCID: PMC4020282. Spector LG, Puumala SE, Carozza SE, et al. 2009. Cancer risk among children with very low birth weights. Pediatrics, 124(1), 96–​104. Thomas DM, and Ballinger ML. 2015. Etiologic, environmental and inherited risk factors in sarcomas. J Surg Oncol, 111(5), 490–​495. PMID: 25335907. Tinat J, Bougeard G, Baert-​Desurmont S, et al. 2009. 2009 Version of the Chompret Criteria for Li Fraumeni syndrome. J Clin Oncol, 27(26), e108–​e109. Toffoli G, Biason P, Russo A, et al. 2009. Effect of TP53 Arg72Pro and MDM2 SNP309 polymorphisms on the risk of high-​grade osteosarcoma development and survival. Clin Cancer Res, 15, 3550–​3556. Troisi R, Masters M, Joshipura K, et  al. 2006. Perinatal factors, growth and development, and osteosarcoma risk. Br J Cancer, 95, 1603–​1607. Troisi R, Stephansson O, Jacobsen J, et al. 2013. Perinatal characteristics and bone cancer risk in offspring:  a Scandinavian population-​based study. Acta Oncologica, 53(6), 830–​838.

827

United Nations Scientific Committee on the Effects of Atomic Radiation. 2006. UNSCEAR. Effects of ionizing radiation. Volume I: Report to the General Assembly, with scientific annexes. New York: United Nations. Valery PC, Holly EA, Sleigh AC, et  al. 2005. Hernias and Ewing’s sarcoma family of tumours:  a pooled analysis and meta-​analysis. Lancet Oncol, 6(7), 485–​490. Valery PC, Laversanne M, and Bray F. 2015. Bone cancer incidence by morphological subtype: a global assessment. Cancer Causes Control, 26(8), 1127–​1139. PMID: 26054913. Valery PC, McWhirter W, Sleigh A, Williams G, and Bain C. 2002. Farm exposures, parental occupation, and risk of Ewing’s sarcoma in Australia:  a national case-​ control study. Cancer Causes Control, 13(3), 263–​ 270. PMID: 12020108. Valery PC, McWhirter W, Sleigh A, Williams G, and Bain C. 2003. A national case-​control study of Ewing’s sarcoma family of tumours in Australia. Int J Cancer, 105(6), 825–​830. Valery PC, McWhirter W, Sleigh A, Williams G, and Bain C. 2003. A national case-​control study of Ewing’s sarcoma family of tumours in Australia. Int J Cancer, 105(6), 825–​830. PMID: 12767069. Vassilopoulou-​Sellin R, Wallis C, and Samaan N. 1985. Hormonal evaluation in patients with osteosarcoma. J Surg Oncol, 28, 209–​213. Vinson F, Merhi M, Baldi I, Raynal H, and Gamet-​ Payrastre L. 2011. Exposure to pesticides and risk of childhood cancer: a meta-​analysis of recent epidemiological studies. Occup Environ Med, 68(9), 694–​702. PMID: 21606468. Virtanen A, Pukkala E, and Auvinen A. 2006. Incidence of bone and soft tissue sarcoma after radiotherapy: a cohort study of 295,712 Finnish cancer patients. Int J Cancer, 118(4), 1017–​1021. PMID: 16152578. Vlachos A, Rosenberg P, Atsidaftos E, Alter B, and Lipton J. 2012. The incidence of neoplasia in Diamond Blackfan anemia:  a report from the Diamond Blackfan anemia registry. Blood, 119(16), 3815–​3819. Wang W, Song H, Liu J, Song B, and Cao X. 2011a. CD86 + 1057G/​A polymorphism and susceptibility to osteosarcoma. DNA Cell Biol, 30(11), 925–​929. Wang W, Wang J, Song H, et  al. 2011b. Cytotoxic T-​lymphocyte antigen-​4 +49G/​A polymorphism is associated with increased risk of osteosarcoma. Genet Test Mol Biomarkers, 15(7–​8), 503–​506. Wang Z, Wen P, Luo X, et  al. 2013. Association of the vascular endothelial growth factor (VEGF) gene single-​nucleotide polymorphisms with osteosarcoma susceptibility in a Chinese population. Tumor Biology, 35(4), 3605–​3610. Winn DM, Li FP, Robison LL, et al. 1992. A case-​control study of the etiology of Ewing’s sarcoma. Cancer Epidemiol Biomarkers Prev, 1(7), 525–​532. PMID: 1302564. Withrow S, and Khanna C. 2010. Bridging the gap between experimental animals and humans in osteosarcoma. In: Jaffe N, Bruland OS, and Bielack S (Eds.), Pediatric and adolescent osteosarcoma (Vol. 152, pp. 439–​446). New York: Springer. Wong FL, Boice JD, Jr., Abramson DH, et  al. 1997. Cancer incidence after retinoblastoma: radiation dose and sarcoma risk. JAMA, 278(15), 1262–​ 1267. PMID: 9333268. Wong JR, Morton LM, Tucker MA, et  al. 2014. Risk of subsequent malignant neoplasms in long-​term hereditary retinoblastoma survivors after chemotherapy and radiotherapy. J Clin Oncol, 32(29), 3284–​ 3290. PMCID: PMC4178525. Wu LC, Kleinerman RA, Curtis RE, Savage SA, and Berrington de Gonzalez AB. 2012. Patterns of bone sarcomas as a second malignancy in relation to radiotherapy in adulthood and histologic type. Cancer Epidemiol Biomarkers Prev, 21(11), 1993–​1999. PMCID: 3493823. Yang W, He M, Zhao J, and Wang Z. 2014. Association of ITGA3 gene polymorphisms with susceptibility and clinicopathological characteristics of osteosarcoma. Med Oncol, 31(2), 1–​8. Zahm SH, and Ward MH. 1998. Pesticides and childhood cancer. Environ Health Perspect, 106(Suppl 3), 893–​908. PMCID: PMC1533072. Zhi L-​Q, Ma W, Zhang H, Zeng S-​X, and Chen B. 2013. Association of RECQL5 gene polymorphisms and osteosarcoma in a Chinese Han population. Tumor Biology, 35(4), 3255–​3259.

82

 829

43

Soft Tissue Sarcoma MARIANNE BERWICK AND CHARLES WIGGINS

OVERVIEW Soft tissue sarcoma (STS) is a rare tumor, occurring in approximately 1.0 to 4.0 of every 100,000 individuals worldwide. Soft tissue sarcomas can form anywhere in the body, including muscle, tendons, fat, blood vessels, lymph vessels, nerves, and tissues around joints. They are most common in the head, neck, arms, legs, trunk, and abdomen. Prognosis is generally poor, with a relative survival rate of approximately 65% at 5 years, with little difference by race. Approximately 12,310 cases of STS will be diagnosed in the United States in 2016, and approximately 4990 deaths from STS will occur. The etiology of STS is still poorly understood, which makes prevention of this relatively rare cancer difficult. A major complication in studying STS is the great number—​more than 100—​of histologies. Newer investigations are evaluating molecular characteristics and prognostic factors but continue to be hampered by a lack of standardized histology.

TUMOR CLASSIFICATION Soft tissue sarcoma is a cancer of mesenchymal tissue other than bone and cartilage: connective, blood, lymph, adipose, nerves, and muscle tissue. Sarcomas generally develop as deeply located masses and infrequently as superficial lesions (Enzinger and Weiss, 1996). STS occurs throughout the body at any anatomical site, at cutaneous, subcutaneous, and deep soft tissue as well as viscera. Approximately 50% of STS arise on the extremities, 40% on the trunk and retroperitoneum, and 10% on the head or neck. Pathologic features relevant to management and clinical outcome include histological type, grade, size, depth, stage, and extent of resection. Soft tissue sarcoma is “bedeviled” by the rarity of the disease combined with multiple histologic classifications, with some disagreement among those making the classification (Ray-​Coquard et al., 2012). In 2002, most experts decided to change the most common histologic subtype, malignant fibrous histiocytoma (MFH), to undifferentiated pleomorphic sarcoma. In 2013, the World Health Organization (WHO) consensus committee updated the STS histological definitions and those for mitoses (Fletcher et al., 2013). According to the WHO, the histological type of sarcoma does not necessarily provide adequate information for prognosis. Site, however, is a major risk factor for survival (Borden et al., 2003). For example, in extremity and truncal sarcomas there is usually good local control, whereas in retroperitoneal lesions local progression is more common.

Stage and Grade Stage and grade are perhaps better indicators of prognosis. Staging provides information on the extent of the tumor (NCI–​ PDQ, 2015) (Table 43–1).

Histopathology Soft tissue sarcoma is a particularly complex tumor to study. It is rare throughout the world, 1% of all tumors, and the histopathologic classification and cancer registry coding are inconsistent. In addition, histopathologists frequently disagree on histologic subtype (Harris

et al., 1991), diluting the ability of epidemiologists to investigate the etiology. The Surveillance, Epidemiology, and End Results (SEER) registry and International Association for Research on Cancer (IARC) classify STS both by anatomic and histologic site, due to the diverse nature of soft tissue tumors and their distribution throughout the body. This dual classification system generates confusion because certain tumors will be coded by tissue type, whereas others will be coded by anatomic site (Lynge et  al., 1987). For example, “connective tissue neoplasm” excludes soft tissue neoplasms occurring at organ sites. Embryologically, the soft tissue is derived principally from the mesoderm, with some contribution from the neuroectoderm (Enzinger and Weiss, 1996). There are a great many subtypes of soft tissue sarcoma, which may vary according to pathologist. The relatively recent elimination of a major subtype, malignant fibrous histiocytoma (MFH), noted earlier, has made analyses of trends, particularly retrospective analyses, more difficult. Histopathological classification of sarcomas is based on cell type (round, spindle, epithelioid, or pleomorphic) and tumor cell lineage (adipocytic, smooth muscle, vascular, skeletal muscle, for example) (Hameed M, 2014). The major histologic categories of STS are summarized in Table 43–2. The actual coding is beyond the scope of this chapter, but can be found at www.who.int/​classifications/​icd/​adaptations/​oncology/​en.

MOLECULAR GENETIC CHARACTERISTICS OF TUMORS An experienced pathologist bases the correct classification of STS on integration of clinical, radiographic, and pathological findings. Genetic factors are important as well. Currently, the great majority of STS throughout the world is diagnosed or confirmed microscopically (Parkin et  al., 2002). Tissue culture, histochemical stains, and cytogenetic analyses often supplement classification. About 20% of STS show characteristic genetic aberrations (Hameed, 2014). In soft tissue sarcomas there are often recurring genetic events, many in the form of chromosomal translocations that result in specific oncogenic fusion genes (Cheah and Billings, 2012). Identification of such gene signatures forms the basis for the development of targeted therapies. These sarcomas can be divided into two general molecular categories: (1) those with simple karyotypes, which are translocation-​associated sarcomas, and (2) tumors with complex karyotypes, showing genomic instability with multiple copy number aberrations. Specific gene amplifications are common, such as MDM2 and CDK4 in liposarcomas and C-​MYC in some angiosarcomas (Crago and Singer 2011; Guo et al., 2010; Taylor et al., 2011). Identification of specific fusions is beyond the scope of this review (see Hameed 2014 for more specifics); however, these are often used as diagnostic among many sarcoma subtypes. On the other hand, “promiscuity of the genes with the ability to form multiple fusion partners (e.g., EWSR1) remains a diagnostic caveat” (Hameed, 2014). Somatic mutations are additional important molecular targets for treatment. The targeting of gastrointestinal stromal tumors (GIST) tumors with a c-​KIT mutation, which can be treated with imatinib (Gleevac), has led the way to new paradigms and attempts to find other key targetable mutations, such as STS patients with a BRCA1 haplotype successfully treated with trabectedin (Laroche-​ Clary et al., 2015).

829

830

830

PART IV:  Cancers by Tissue of Origin

Table 43–​1.  Stage and Grade in Soft Tissue Sarcoma Staging Stage

Grade

Size

Depth

Comments

Stage I IA IB

1, or cannot be assessed 1, or cannot be assessed

≤ 5 cm > 5 cm

Superficial or deep Superficial or deep

Stage II IIA IIB

2–​3 2

≤ 5 cm > 5 cm

Superficial or deep Superficial or deep

Stage III

3, or any grade if spread to lymph nodes

> 5 cm

Superficial or deep

May have spread to nearby lymph nodes from any grade or size

Stage IV

Any grade (1–​3)

Any size

Superficial or deep

May have spread to nearby lymph nodes from any grade or size but has spread distantly, e.g., to lungs

Grade Grade 1 Grade 2

Score* 2–​3 4–​5

Grade 3

*Grade score is based on addition of the following: Differentiation = 1–​3, “mitotic count” = 1–​3, and “necrosis” = 0–​2

6+

TUMOR PROGRESSION MODELS Although our knowledge of the genome has increased exponentially and technological advances have been impressive, tumor biologists are still limited in their ability to characterize STS genetically by several issues, most importantly, the lack of clear and reproducible histologic definitions. The majority of studies of prognosis in STS have been hampered by potential selection bias, low power, and inadequate statistical methods (Maretty-​Nielsen, 2014). However, there are multiple new papers describing genomic evaluations. The prognosis for patients with adult soft tissue sarcomas depends on several factors, including: age, tumor size, histologic grade, mitotic activity and stage of the tumor (Singer et al., 2011). Factors associated with a poorer prognosis include the following:  age greater than 60,

tumors larger than 5 cm, and high-​grade histology with high mitotic activity (Vraa et al., 1998). Although low-​grade tumors are usually curable by surgery alone, higher-​grade sarcomas (as determined by the mitotic index and by the presence of hemorrhage and necrosis) are associated with higher local-​ treatment failure rates and increased metastatic potential. A  small retrospective study of 57 STS patients found that mitotic count and the amount of viable tumor following neoadjuvant systemic chemotherapy for primary localized, high-​ grade STS correlated positively with poorer survival (Andreaou et  al., 2015). An interesting quality of life study was part of a randomized clinical trial of pazopanib among 369 patients (123 placebo and 246 pazopanib), where Coens and colleagues (2015) found that those receiving pazopanib reported significantly poorer symptom scores, but no

Table 43–​2.  Histopathologic Categories of Soft Tissue Sarcoma Category

Tissue of Origin

Leiomyosarcoma Liposarcoma Fibroblastic Undifferentiated pleomorphic sarcoma Dermatofibrosarcoma protuberans (DFSP) Fibrosarcoma Myxofibrosarcoma Rhabdomyosarcoma Embryonal Alveolar pleomorphic rhabdomyosarcoma Soft tissue Ewing’s sarcoma

Smooth muscle cells Fatty tissue Cells in fibrous tissue

Angiosarcoma

Cells that make up the walls of blood or lymphatic vessels Blood vessel Lymph vessel Soft tissues or internal organs (liver, lungs) Cells that cover the nerves

Haemangiosarcoma Lympagniosarcoma Haemangioendothelioma Malignant peripheral nerve sheath tumors (MPNST) Neurofibrosarcoma Malignant Schwannoma Gastrointestinal stromal tumors (GIST) Kaposi Sarcoma (KS)

Starts in skin

Skeletal muscle

Cells close to joints and tendons

From Schwann cells, the fatty sheath that covers nerves Connective tissue supporting digestive organs Endothelial cells

Anatomic Site

Anywhere, but common in arms or legs Formerly malignant fibrous histiocytoma (MFH) Very rare, adults in their 30s

Anywhere on the body, most common STS in childhood More common in children and young people Younger adults Rare, tends to occur in adults Can occur anywhere in the body, more common in young adults

Most in stomach; treated with imatinib Most in skin, but also internal organs

 831

831

Soft Tissue Sarcoma

PRECANCEROUS OR PRECURSOR LESIONS While there are no known precursor lesions to soft tissue sarcoma, there are many benign forms of STS arising from similar tissues. Benign tumors outnumber sarcomas by 100-​fold (Enzinger and Weiss, 1996). These do not invade locally, resemble normal tissue, and are unlikely to recur after excision. Malignant change of benign sarcoma only occurs rarely (King et al., 1979). Interestingly, leiomyomas and leiomyosarcomas develop from the lining of the digestive tract, but leiomyosarcomas do not arise from the benign leiomyomas. Lipomas and liposarcomas develop from fatty tissue, but cytogenetic studies have shown that liposarcomas do not usually develop from lipomas (Enzinger and Weiss, 1996).

DESCRIPTIVE EPIDEMIOLOGY Age, Sex, Race/​Ethnicity As with most tumors, increasing age is associated with higher incidence; however, STS can occur at any age, and the median age is approximately 58  years (Burningham et  al., 2012). Although older STS patients experienced more venous thromboembolic events in a large group of 3480 STS patients (Shantakumar et  al., 2015), older patients have also benefited more from radiotherapy (Yuen et  al., 2015). Age is also important in that adolescent and young adult survivors of soft tissue sarcoma have significant long-​term mortality due to second malignant neoplasms and non-​cancer causes (Youn et  al., 2014), often due to cancer-​directed therapy for the original primary (88% of deaths were due to the original sarcoma).

3

Incidence Rate

2

1 Mortality Rate 0 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

difference in “global quality of life,” even though there was no overall survival benefit. An additional factor associated with poor prognosis, or progression, was a high neutrophil:lymphocyte ratio. In an Austrian study of 340 STS patients with diverse histologies classified by the preoperative neutrophil:lymphocyte ratio (neutrophil count/​white cell count–​ neutrophil count), a high ratio was significantly associated with poor overall survival (hazard ratio [HR] 1.60; 95% confidence interval [CI] 1.07–​2.42) (Szkandera et al., 2015). Such data suggest a role for host immune function in the pathogenesis of STS. Other factors, such as anatomic site and comorbidity, appear to affect progression or survival. In a large French study of 719 STS patients with a variety of histologies, late local recurrence (i.e., > 5 years) was independently associated with internal trunk location and tumor size > 100 mm (Toulmonde M et al., 2014). Further, a study of 349 patients found that those (12%) who had comorbidities, as defined in the Charlson comorbidity index (Charlson et al., 1987), had poor disease-​free survival (Kang et al., 2015). In a study of 136 advanced STS patients, a favorable BRCA1 haplotype was significantly associated with an improved 6-​month non-​ progression rate; thus this BRCA1 haplotype might be a DNA repair biomarker (Laroche-​ Clary et  al., 2015). Reduced expression of hMSH2, another marker of repair, correlated with poor survival in STS patients (Taubert et al., 2003). Finally, MEK (mitogen-​activated protein kinase kinase enzyme) inhibition has a promising role in many types of STS (Dodd et al., 2013).

4 Age-Adjusted Rate per 100,000



Year of Diagnosis

Figure 43–​1.  Incidence and mortality of soft tissue sarcoma in the United States (SEER).

black females having the highest rate among females and non-​Hispanic white males having the highest rate among males (Table 43–3). Beginning in 1992, the SEER Program began to report more detailed information on race and ethnicity (Table 43–3). Incidence rates for non-​ Hispanic whites, Hispanics, and blacks were generally similar. Rates for American Indians and Asian and Pacific Islanders were slightly lower than in the aforementioned groups. Temporal increases were observed during the period 1992–​2012 in each of the expanded races/​ethnicities, and by sex, though not all increases achieved statistical significance. Again, rates for males were higher for males than females in each of the expanded racial/​ethnic categories.

International

Incidence rates for connective and soft tissue sarcomas are relatively low worldwide, ranging from approximately 0.5 to 4.0 per 100,000 person-​years. Rates are consistently higher for males than females in most countries, but the observed differences by sex are not of great magnitude (Figure 43–2). A comparison of male and female incidence rates shows wide variation among countries, as shown in Figure 43–2. Although international rates are very similar to those in the United States, these vary by area of the world (Table 43–4). Five-​year survival is similar throughout the world at 65%.

Temporal Trends Since 1992, there have been slight increases in the incidence of STS overall (Figure 43–1) and for most subtypes (Table 43–3) (SEER, 2015). While there is no screening for STS, in that most tumors are discovered adventitiously or when very large and impeding function, it is possible that more attention has been paid to STS and thus more diagnoses have been made. Again, based on the small numbers, this is difficult to tease out. However, the annual percent change in incidence of soft tissue sarcomas was 4.74% between 2008 and 2012, as noted in Table 43–3, and differs by broad category of histologic type. During the time period 1992–​2012, incidence rates increased for many histologic subcategories of soft tissue sarcomas, including soft tissue tumors and sarcomas, lipomatous neoplasms (ICDO-​3 histologies 8810–​8839), myomatous neoplasms, and blood vessel tumors. In contrast, rates for synovial-​like neoplasms were relative stable over the study period, while rates of fibromatous neoplasms declined (Table 43–5).

Incidence United States

Collectively, soft tissue sarcomas are relatively rare tumors, with a combined average-​annual incidence rate of 2.9 for females and 4.0 for males per 100,000 person-​years (Howlader et  al., 2016). During the time period 1973–​2012, there was a modest, but statistically significant increase in incidence rates for soft tissue sarcomas among all races (whites, blacks, and others) and both sexes (Figure 43–1). Rates were generally higher for males than females in all racial groups, with

Survival Five-​year relative survival depends heavily on stage at diagnosis: 83% for localized STS, 54% for regional stage STS, and 16% for those with distant spread. GIST tumors have a 76% survival rate (ACS 2015). STS of the skin had an excellent survival rate, approximately 90% and higher, while STS of the mediastinum and heart had survival rates less than 15% (Stiller et al., 2013). Rates obviously differ widely by type of STS. Alamanda and colleagues (2014) evaluated

832

Table 43–​3.  Average Annual Age-​Adjusted Incidence Rates per 100,000 (US 2000 Standard) for Soft Tissue Sarcoma in the United States by Expanded Race/​Ethnicity Categories, Sex, and Time Period of Diagnosis, 1992–​2012 Ancestry Non-​Hispanic White

Sex

1992–​1997

1998–​2002

2003–​2007

2008–​2012

APC*

Total

2.7 (2.6–​2.8)** 3.3 (3.1–​3.4) 2.3 (2.2–​2.5) 3.1 (2.7–​3.5) 3.7 (3.1–​4.5) 2.5 (2.0–​3.0) 3.0 (2.7–​3.4) 3.5 (3.0–​4.0) 2.7 (2.3–​3.1) 2.1 (1.3–​3.2) 2.2 (1.0–​4.0) 2.0 (1.1–​3.5) 2.3 (2.0–​2.6) 2.5 (2.1–​3.0) 2.0 (1.7–​2.4)

3.0 (2.9–​3.1) 3.7 (3.5–​3.9) 2.5 (2.3–​2.6) 3.0 (2.6–​3.5) 3.8 (3.1–​4.6) 2.5 (2.0–​3.0) 3.1 (2.8–​3.4) 3.5 (3.0–​4.0) 2.8 (2.4–​3.2) 2.2 (1.4–​3.2) 2.5 (1.4–​4.2) 1.9 (0.9–​3.3) 2.3 (2.1–​2.6) 3.1 (2.7–​3.7) 1.7 (1.4–​2.0)

3.3 (3.2–​3.4) 4.1 (3.9–​4.2) 2.7 (2.6–​2.9) 3.1 (2.7–​3.5) 3.2 (2.7–​3.9) 2.9 (2.4–​3.4) 3.3 (3.0–​3.6) 3.6 (3.1–​4.1) 3.1 (2.7–​3.5) 2.3 (1.5–​3.3) 2.5 (1.2–​4.5) 2.2 (1.3–​3.6) 2.6 (2.3–​2.9) 2.9 (2.4–​3.3) 2.4 (2.1–​2.8)

3.4 (3.3–​3.6) 4.1 (3.9–​4.3) 2.9 (2.8–​3.1) 3.2 (2.9–​3.6) 3.9 (3.4–​4.5) 2.6 (2.2–​3.1) 3.4 (3.1–​3.7) 3.7 (3.2–​4.2) 3.2 (2.9–​3.6) 2.5 (1.8–​3.4) 3.1 (1.9–​4.8) 2.1 (1.3–​3.2) 2.6 (2.3–​2.8) 2.9 (2.6–​3.4) 2.3 (2.0–​2.6)

1.6*

Male Female Hispanic White

Total Male Female

Black

Total Male Female

American Indian

Total Male Female

Asian and Pacific Islander

Total Male Female

1.5* 1.5* 0.2 0.2 0.3 1.1* 0.7 1.4* 1.3 2.7 0.1 1.0* 0.7 1.3

Residents of nine core areas of the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) Program * Annual percent change ** 95% confidence interval

Connective and Soft Tissue Average annual age-adjusted incidence rates per 100,000 Japan-Osaka China-Beijing Thailand-Bangkok Russia-St. Petersburg China-Shanghai Korea Singapore Costa Rica Irelane Austria Uganda-Kyadondo Philippines-Manila Zimbabwe-Harare Brasil-Sao Paulo Netherlands Canada Australia-Queensland Sweden Germany-Munich Columbia-Cali Denmark USA-SEER Israel 0.0

0.5

1.0

1.5

2.0

Age-adjusted rate per 100,000 Male

Figure 43–​2.  Age-​adjusted incidence rates in selected countries by sex.

Female

2.5

3.0

3.5

 83



833

Soft Tissue Sarcoma

Table 43–​4.  Age-​Standardized Incidence per 100,000 Throughout the World, Standardized to the World Population Rate Area

Male Rate Ranges

Female Rate Ranges

1.8–​2.4 1.9–​3.6

1.1–​2.0 1.6–​2.9

1.2–​1.6 3.1–​3.2 2.0–​2.5

1.0–​1.4 2.1–​2.1 1.6–​1.9

Africa Central and South America Asia Israel Europe Source: Forman et al. (2014).

397 patients with extremity soft tissue sarcoma and found that increasing age, tumor size, and tumor grade increased the hazard of death. Smoking and alcohol abuse, diabetes, hyperlipidemia, hypertension, chronic obstructive pulmonary disease (COPD), and wound complications had no significant effect on sarcoma-​specific survival. On the other hand, comorbidity based on the Charlson Index (Charlson et al., 1987) was associated with poor outcome in an extremity STS series of 349 patients in Korea (Kang et al., 2015), as well as older age, high tumor grade, and large tumor size.

ETIOLOGIC FACTORS Little information is available about other demographic characteristics, such as marital status, as STS has been difficult to study etiologically, as noted previously.

Exposures A major problem with all studies of the etiology of soft tissue sarcoma is the low statistical power to find an association between risk factors and STS, due to the low incidence and misclassification inherent in the histology. Partially due to the rarity of this tumor, different histologic subtypes are often grouped together in studies, resulting in a loss of specificity and possibly masking the behavior of the different subtypes. This problem is further complicated in studies of environmental etiology by the difficulty of making accurate environmental measurements. It is most often not possible, except in some unusual occupational cohorts, to have an accurate history of prior environmental

exposure. Thus, many investigators have relied on self-​reported occupational histories as well as self-​reported exposures, which may suffer from misclassification. In some cases, more accurate estimates of exposure can be made with the assistance of an expert industrial hygienist (Piacitelli et al., 2000).

Phenoxy Herbicides, Dioxin, and Pesticides

A controversial risk factor often investigated in relation to the development of STS is Polychlorinated dibenzo-​p-​dioxin (2,3,7,8-​TCDD), a contaminant of industrial processes. Numerous epidemiologic studies have examined occupational groups and those accidentally exposed to putative risk substances, such as dioxin, to substantiate or refute earlier reports of associations with soft tissue sarcoma (Hardell, 1977). As dioxin is a contaminant of industrial processes, it is difficult to measure indirectly by means of exposures to those processes. Therefore, it is not surprising that there continues to be conflicting evidence of an association between exposure to dioxin and STS. This question is, however, important because there is still a poor understanding of the etiology of soft tissue sarcoma. The polychlorinated dibenzo-​p-​dioxin (2,3,7,8-​TCDD) consists of a group of 75 congeners. Dioxin is stored in the lipids in the body and can be measured in the lipid fraction of serum (Schecter, 1994). In most, but not all areas of the world, dioxin levels (as measured in serum) have declined. Contaminated feed clearly increases the concentration of dioxin in the fat of chickens and pigs (Neuberger et al., 2000). In Spain, blood levels have been steadily increasing and could reflect an increase of exposure from foods or other unidentified sources (Gonzales et al., 2001). The US Environmental Protection Agency has revised upward the estimated risks of cancer, particularly soft tissue sarcoma, due to dioxin in the food chain, even though emissions of such chemicals and blood levels have decreased dramatically in the past 10–​15 years (Aylward and Hays, 2002). A very small number of STS deaths among 1615 workers exposed to dioxins in Michigan resulted in an SMR (standardized mortality ratio) of 4.1 (95% CI = 1.1–​10.5) with increasing deaths with increasing exposure. However, these results were based on only four deaths. Tuomisto and Tuomisto (2012) reported that with few cases and the high potential for misclassification in diagnosis, there is extremely limited evidence for this association.

Vinyl Chloride

Vinyl chloride seems to have very specific effects and has been weakly associated with the development of angiosarcoma (e.g., Charbotel et al., 2014) and is classified as a Group 2B carcinogen by IARC, that

Table 43–​5.  Average Annual Age-​Adjusted Incidence Rates per 100,000 (US 2000 Standard) for Soft Tissue Sarcoma in the United States by Broad Categories of Histologic Type and Time Period of Diagnosis, 1992–​2012 Time Period of Diagnosis ICDO-​3 Category

Description

8800–​8809

Soft tissue tumors and sarcomas, NOS

8810–​8839

Fibromatous neoplasms

8850–​8889

Lipomatous neoplasms

8890–​8929

Myomatous neoplasms

9040–​9049

Synovial-​like neoplasms

9120–​9169

Blood vessel tumors

Others

Other sites/​types

1992–​1997

1998–​2002

2003–​2007

2008–​2012

APC*

0.39 (0.36–​0.42)** 0.86 (0.81–​0.91) 0.42 (0.39–​0.46) 0.43 (0.40–​0.47) 0.13 (0.11–​0.15) 0.10 (0.09–​0.12) 0.41 (0.38–​0.45)

0.46 (0.42–​0.49) 0.80 (0.75–​0.85) 0.47 (0.43–​0.50) 0.49 (0.45–​0.53) 0.15 (0.13–​0.18) 0.15 (0.13–​0.18) 0.47 (0.43–​0.51)

0.63 (0.58–​0.67) 0.70 (0.66–​0.75) 0.47 (0.43–​0.51) 0.55 (0.52–​0.60) 0.16 (0.14–​0.19) 0.13 (0.11–​0.15) 0.59 (0.55–​0.64)

0.78 (0.74–​0.83) 0.59 (0.55–​0.63) 0.59 (0.56–​0.64) 0.54 (0.50–​0.58) 0.15 (0.13–​0.17) 0.16 (0.14–​0.18) 0.55 (0.51–​0.59)

4.74*

Residents of nine core areas of the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) Program * Annual percent change ** 95% confidence interval

–​2.34* 2.17* 1.48* 1.00 2.32* 2.28*

834

PART IV:  Cancers by Tissue of Origin

1

200

0

0

Number of Incident Cases

400

2011

2

2009

600

2007

3

2005

800

2003

4

2001

1000

1999

5

1997

1200

1995

6

1993

Some types of Kaposi’s sarcoma (KS) are known to be associated with HIV (human immunodeficiency virus) infection and AIDS (acquired immunodeficiency syndrome). It can be caused by a sexually transmitted virus called human herpes virus 8 (HHV 8) or Kaposi’s sarcoma herpes virus (KSHV). The incidence of KS mirrors KSHV prevalence and is strongly associated with HIV co-​infection (Bhutani et al., 2015; Gramolelli and Schulz, 2015). In fact, the striking increase in incidence of KS in the early 1990s in the United States diminished following widespread dissemination of the highly active antiretroviral therapy (HAART), as noted in Figure 43–3 (SEER, 2015). KS has also been associated with Jewish, Italian or Mediterranean, Eastern European, and West African ancestry; these tend to be more indolent and occur among older individuals (Schwartz et al., 2008). In the past, other viral infections, such as herpes zoster, chicken pox, and mumps (Franceschi and Serraino, 1992; Serraino et  al.,

The role of immunosuppression as an etiologic factor in soft tissue sarcoma has been long suspected (Vineis and Zahm, 1988). Dioxin and similar environmental chemicals are immunotoxic (Vos et  al., 1997–​1998), so the extent to which dioxin is associated with the

1991

Viruses

Immune Factors: Immunosuppression

1989

Although diet has been little studied, Serraino et al. (1991) found positive associations with dairy product consumption and oil, and negative associations with whole grain bread and pasta, as did Tavani et al. (1997). A Norwegian study (Juzeniene et al., 2015) found little difference between serum vitamin D and daily vitamin D intake between soft tissue sarcoma patients and those with benign soft tissue tumors. However, the World Cancer Research Fund/​American Institute for Cancer Research (2015) suggests that studies investigating possible links between STS and food, nutrition, and physical activity would be extremely difficult due to the diversity and rarity of STS. Due to the small numbers of soft tissue sarcomas and the misclassification inherent in dietary studies, it is unlikely that epidemiologic studies can address the association between diet and STS unless there were an extremely large effect.

1987

Diet

1985

Several lines of evidence suggest that exogenous hormone use might be associated with soft tissue sarcoma. Wagner et al. (2014) evaluated 634 cases of STS and found that oral contraceptive use was inversely associated with STS. Early studies showed that high levels of body fat seem to be associated with increased risk (e.g., Tavani et al., 2000), although not all studies (Serraino et al., 1991) found that.

1983

Exogenous Hormonal Factors

1981

Tobacco use, either smokeless tobacco or smoking, has been inconsistently associated with soft tissue sarcoma (e.g., Serraino et al., 1991).

1979

Tobacco Exposure

A creative approach taken to understand the etiology of STS has been to conduct studies of the relationship between STS and other cancers using information from tumor registries. In these studies, individuals are evaluated for the incidence of cancers both prior to and subsequent to their diagnosis of STS. The underlying hypothesis of such studies is that strong associations might yield clues as to shared environmental or genetic factors, although only a few clues are yet available. Several studies of Li-​Fraumeni syndrome families have been conducted with similar goals. Follow-​up studies of childhood cancers have found a high incidence of soft tissue sarcoma (Ognjanovic et al., 2012). In a population-​based study of 6112 individuals diagnosed with GIST tumors during 2001–​2011, Murphy and colleagues (2015) found a striking increase in other tumors both prior to and after GIST diagnosis. Seventeen percent had additional cancers, including other sarcomas, neuroendocrine-​carcinoid tumors, non-​Hodgkin lymphoma, and colorectal adenocarcinoma. Prior to the diagnosis of GIST, these patients had been diagnosed significantly more often with esophageal cancer, bladder adenocarcinoma, melanoma, and prostate adenocarcinoma. After a GIST diagnosis, significantly increased numbers of ovarian carcinoma, small intestine adenocarcinoma, papillary thyroid cancer, renal cell carcinoma hepatobilliary adenocarcinoma, gastric adenocarcinoma, and pancreatic adenocarcinoma were diagnosed. Such findings may help further clinical screening and treatment strategies. In the relatives of probands who form an unselected cohort of 402 breast cancer patients, Bennet et  al. (2002) found an increased incidence of STS in mothers only. Berking and Brady (1997) reported on a large number of sarcomas in association with melanoma that were notable for a family history of cancer. Hemminki and Li (2001) reviewed soft tissue sarcomas in Swedish registries in parents and by histology. Parental stomach cancer and endocrine gland cancer in parents were associated with offspring fibrosarcoma, and parental breast cancer was associated with leiomyosarcoma.

1977

Although the incidence is low, the role of therapeutic radiation in inducing soft tissue sarcoma is well established. Several population-​ based registries have supplied dose–​response estimates for the role of radiation in the development of sarcoma. Samartzis et  al. (2013) evaluated ionizing radiation in the Japanese atomic bomb survivors and found low levels of ionizing radiation associated with STS incidence and lower 5-​year survival. Berrington de Gonzalez et al. (2012) reported increased STS incidence associated with ionizing radiation, with a linear dose response for radiation in childhood. Using cancer registry data, several investigators have identified women with breast cancer who were treated with radiation and followed for subsequent diagnoses of sarcoma (e.g., Yap et al., 2002). It should be noted that the risk estimates are likely smaller than the reported estimates due to the probable incomplete exposure information collected by the SEER registries. However, there appears to be an approximately 10-​year lag time from radiation exposure to the development of STS tumors.

Cofactors

1975

Ionizing Radiation

1991) were associated with STS; however, to our knowledge, no new studies have addressed this issue.

1973

is, “limited evidence for this site.” Collins et  al. (2014) have evaluated more than 73,000 death certificates of North American workers employed in the manufacture of vinyl chloride resins, and found 13 deaths of angiosarcoma among exposed workers (all 13 occurred at a single plant with high vinyl chloride exposure). The mean latency was 36.5 years (range 24–​56 years). No angiosarcoma deaths occurred among exposed workers after 1974 when exposures were reduced.

Age-Adjusted Incidence Rate per 100,000

834

Year of Diagnosis Cases

Rate

Figure  43–​3. Kaposi sarcoma:  annual number of new cases and age-​ adjusted incidence rates among residents of nine core areas of the SEER Program (1981–​2012).

 835



Soft Tissue Sarcoma

development of STS may suggest a role for immunosuppression by other means. Earlier studies reported an excess of STS in patients receiving immunosuppression for renal transplantation and other conditions (Hoover and Fraumeni, 1973). Since that time, sporadic case reports have continued to document associations between transplantation and STS occurrence (e.g., Hussein et al., 2014; Krol et al., 2015; Wieczorek-​Godlewska et al., 2014; Wu et al., 2014). Kaposi sarcoma in the United States and Africa has been strongly associated with HIV infection (de Martel et  al., 2015), known to induce immunosuppression. In the United States, as AIDS treatment is now more effective, there are fewer cases, but KS is still the major cancer among persons with HIV (Silverberg et  al., 2015). Recently, strong associations have been noted between leiomyosarcoma and AIDS, mostly in children (Mbulaiteye et al., 2003). After non-​Hodgkin lymphoma, the second leading cancer in children is soft tissue sarcoma. In fact, Biggar et al. (2000) reported a risk of more than 1900 for leiomyosarcoma in children with AIDS in Uganda. Within 2 years of an AIDS diagnosis in all age groups (Frisch et al., 2001), the risk for soft tissue sarcoma is 3.6 (2.5–​5.3), which the authors cite as evidence of immunosuppression.

Genetic Susceptibility Familial Aggregation

Tumor suppressor genes are important to the pathogenesis of soft tissue sarcoma—​most prominent among them are TP53 and RB1, the hereditary retinoblastoma gene. Germline p53 mutations in families with multiple tumors are diagnostic of Li-​Fraumeni syndrome (Malkin et al., 1990). In the first place, the Li-​Fraumeni syndrome was recognized in four families with an autosomal dominant pattern of STS, breast cancer, and other tumors among young individuals (Li and Fraumeni, 1969). Families have typically been identified through an increased number of both bone and soft tissue sarcoma cases in conjunction with a specific constellation of other tumors, such as breast, brain, adrenal cortical carcinoma, Wilms’s tumor, and leukemia. Among Li-​Fraumeni families, approximately 20% develop soft tissue sarcoma and 28% breast cancer. In those families with p53 mutations, tumors tend to appear in very young individuals: 85% of the sarcomas (including bone) occurred under the age of 20 (Varley et  al., 1997). Clusters of families occur that appear to be Li-​Fraumeni syndrome but do not have germline mutations and thus suggest other syndromes or an underlying genetic etiology (Bell et al., 1999; Li et al., 1997). The range of cancers has been enlarged to include a number of other tumors associated with the Li-​Fraumeni syndrome less often, such as stomach, ovary, colorectal, lymphoma, melanoma, endometrial, thyroid, pancreas, prostate, and cervix (Nichols et al., 2001). In addition, individuals within such families tend to develop multiple primary tumors (Hisada et al., 1998; Strong et al., 1987), particularly soft tissue sarcoma arising in a radiation field. The identification of the Li-​Fraumeni syndrome has been critically important for understanding the etiology of soft tissue sarcoma and cancer in general. However, germline TP53 mutations do not account for a large proportion of soft tissue sarcomas, having been found in approximately 4% of unselected STS cases (Mitchell et  al., 2013). Both point mutations and deletions and insertions in TP53 have been identified in soft tissue tumors themselves. Studies of families with children with STS have demonstrated the probable genetic basis of such early cancer incidence (e.g., Birch, 1990; Hwang et al., 2003; Strong, 1989). In individuals with hereditary retinoblastoma, there is a significantly increased risk for second cancers, particularly leiomyosarcoma and osteosarcomas (Wong et al., 2014). Germline mutations in the retinoblastoma gene (RB1) on chromosome 13q14 seem to be responsible for hereditary retinoblastoma as well as sarcomas (Hansen et  al., 1985). Retinoblastoma is typically treated with radiation, and there is a dose–​response relationship between radiation dose and sarcoma risk, evident at doses about 3 Gy (Wong, 1997), and increasing to a relative risk of 10.7 at doses of 60 Gy or higher.

835

Genetic Syndromes

Soft tissue sarcomas occur with greater frequency in patients with the following inherited syndromes (Singer et al., 2011): nevoid basal cell carcinoma (Gorlin syndrome:  PTC mutations), Gardner syndrome (APC mutation), Li-​ Fraumeni syndrome (TP53 mutation), tuberous sclerosis (Bourneville disease:  TSC1 or TSC2 mutation), von Recklinghausen disease (neurofibromatosis type 1:  NF1 mutation), and Werner syndrome (adult progeria: WRN mutation). TP53 and RB1 abnormalities often occur together (Stratton et  al., 1990), indicating that coincident inactivation of more than one tumor suppressor gene may be required for tumor development. Other genetic syndromes have been associated with soft tissue sarcoma. Individuals with NF1 (the neurofibromatosis type 1 gene) mutations have a 7%–​ 14% lifetime risk of developing a sarcoma, usually neurosarcoma or fibrosarcoma. Dodd et al. (2013) found that NF1 appears to be mutated in a variety of soft tissue sarcomas. While genetic syndromes in the germline are critical to understand for the etiology of STS, it may be helpful to also evaluate somatic mutations in the tumor. Interestingly, the type of TP53 mutation in a tumor seems to be related to prognosis:  non-​frameshift mutations have significantly poorer outcome (Taubert et  al., 1996). Barretina et al. (2010) had previously shown that TP53 and NF1 were frequently mutated in STS tumors and might present new targets for therapy. These types of molecular alterations my lead to etiologic insights. It is abundantly clear that GIST tumors are remarkably distinct and present a uniform expression profile (Allander et al., 2001). Because a therapy (imatinib) for those GIST tumors with a mutated proto-​oncogene, KIT, has been extremely successful, molecular characterization of STS will probably continue to be high-​priority research. Furthermore, the durable response to therapy for c-​KIT mutations in GIST tumors is a paradigm for the role of molecular pathology to elucidate mechanisms of carcinogenesis, as well as additional therapeutic targets.

Genetic Polymorphisms There is a burgeoning literature examining the role of genetic variation with STS etiology and outcome. Berwick et  al. (2004) found that a single nucleotide polymorphism (SNP) in the Ah receptor was associated with poor prognosis among 120 patients. Miao et al. (2015) found that FPGS polymorphisms were associated with risk for primary retroperitoneal liposarcoma among 138 Chinese patients and 131 controls. Further, Zhang et al. (2015) identified an MDM2 SNP through a systematic computerized search with a 1.42-​fold higher risk (95% CI = 1.08–​1.85) based on a meta-​analysis of five independent studies. Unfortunately, both due to the rarity of the disease and the inconsistent histologic definition, it is difficult to use retrospectively collected data to evaluate SNPs, which require large sample sizes to overcome issues with false discovery.

Endogenous Hormonal Factors Hormonal factors may play a role in the etiology of soft tissue sarcoma. Fioretti et  al. (2000) and Molife et  al. (2001) have evaluated gender-​related risk factors, suggesting that female hormones might be a profitable area to explore further. Wagner et al. (2014) reported that sex hormones and growth hormones, as well as insulin-​like growth factor 1 (IGF-​1), were possibly associated with STS at multiple anatomic sites, based on the reduced risk among those with short stature at puberty in a Swedish population-​based study.

Chronic Repair Processes Froehner and Wirth (2001) have suggested that chronic repair processes may increase risk for STS. A  decreased repair DNA repair capacity, measured by the mutagen sensitivity assay (Hsu et  al., 1991), was associated with the development of STS in a small case control study (Berwick et  al., 2001). New evidence suggests further associations with DNA repair capacity (Laroche-​Clary et  al., 2015),

836

836

PART IV:  Cancers by Tissue of Origin

as STS patients with a BRCA1 haplotype (one or more AAAG alleles) responded well to trabectedin.

OPPORTUNITIES FOR PREVENTION Unfortunately, due to the lack of etiologic understanding, preventive measures cannot yet be undertaken except among the high-​risk groups for STS that include Li-​Fraumeni families, childhood survivors of cancer, and individuals who have received more than 5 Gy radiation. These groups will continue to receive surveillance for the development of new cancers.

FUTURE RESEARCH The future looks optimistic if the incorporation of molecular techniques in epidemiologic studies of STS continues. The molecular characterization of tumors is ongoing in many centers and may help define as well as provide insights in the etiology and outcome and treatment of STS. Once tumor subtypes can be consistently identified, then patients can be appropriately categorized for etiologic research. It is likely that tumor-​specific mutations will be an adjunct to exposure measures, pathologic characterization, and genetic susceptibility factors that will be able to accurately and reproducibly define the pathways to tumor development, thus providing targets for preventive measures as well as therapeutic intervention. References Alamanda VK, Moore DC, Song Y, Schwartz HS, Holt GE. 2014. Obesity does not affect survival outcomes in extremity soft tissue sarcoma. Clin Orthop Relat Res 472:2799–​2806. PMCID: PMC4117870. Allander SV, Nipponen NN, Ringer M, et  al. 2001. Gastrointestinal stromal tumors with KIT mutations exhibit a remarkably homogenous gene expression profile. Cancer Res 61:8624–​8628. PMID: 11751374. American Cancer Society. 2015. Cancer facts and fi­gures  2015. Atlanta, GA:  American Cancer Society. Available online. Last accessed September 2015. Andreaou D, Werner M, Pink D, et al. 2015. Prognostic relevance of the mitotic count and the amount of viable tumour after neoadjuvant chemotherapy for primary, localized, high-​ grade soft tissue sarcoma. Br J Cancer 112:455–​460. PMCID: PMC4453655. Aylward LL, Hays SM. 2002. Temporal trends in human TCDD body burden: decreases over three decades and implications for exposure levels. J Expo Anal Environ Epidemiol 12:3119–​3328. PMID: 12198580. Barretina J, Taylor BS, Banerji S, et al. 2010. Subtypespecific genomic alterations define new targets for soft-​tissue sarcoma therapy. Nature Genetics 42:715–​721. PMCID: PMC2911503. Bell DW, Varley JM, Szydlo TE, et  al. 1999. Heterozygous germ line hCHK2 mutations in Li-​Fraumeni syndrome. Science 286:2528–​2531. PMID: 10617473. Bennett KE, Howell A, Evans DGR, Birch JM. 2002. A follow-​up study of breast and other cancers in families of an unselected series of breast cancer patients. Br J Cancer 86:718–​722. PMID: 11875732. PMC2375308. Berking C, Brady MS. 1997. Cutaneous melanoma in patients with sarcoma. Cancer 79 843–​846. PMID: 9024723. Berrington de Gonzalez A, Kutsenko A, Rajaraman P. 2012. Sarcoma risk after radiation exposure. Clin Sarcoma Res 2:18. PMCID: PMC3507855. Berwick M, Song Y, Jordan R, Brady MS, Orlow I. 2001. Mutagen sensitivity as an indicator of soft tissue sarcoma risk. Environ Mol Mutagen 38:223–​ 226. PMID: 11746758. Berwick M, Matullo G, Song YS, et al. 2004. Association between aryl hydrocarbon receptor genotype and survival in soft tissue sarcoma. J Clin Oncol 22:3997–​4001. PMID: 15459223. Bhutani M, Pollizzotto MN, Uldrick TS, Yarchoan R. 2015. Kaposi Sarcoma-​ associated herpesvirus-​associated malignancies: epidemiology, pathogenesis, and advances in treatment. Semin Oncol 42:223–​246. PMID: 25843728. Biggar BJ, Frisch M, Goedert JJ. 2000. Risk of cancer in children with AIDS. JAMA 284:205–​209. PMID: 10889594. Birch JM, Hartley AL, Blair V, et al. 1990. Cancer in the families of children with soft tissue sarcoma. Cancer 66:2239–​2248. PMID: 2224780. Borden EC, Baker LH, Bell RS, et  al. 2003. Soft tissue sarcoma of adults: state of the translational science. Clin Cancer Res 9:1941–​1956. PMID: 12796356.

Burningham Z, Hashibe M, Spector L, Schiffman JD. 2012. The epidemiology of sarcoma. Clin Sarc Res 2:14. PMCID: PMC3564705. Charbotel B, Fervers B, Droz JP. 2014. Occupational exposures in rare cancers: a critical review of the literature. Crit Rev Oncol Hematol 90:99–​ 134. PMID: 24387944. Charlson MD, Pompei P, Ales KL, MacKenzie CR. 1987. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis, 40:373–​383. PMID: 3088089. Cheah AL, Billings SD. 2012. The role of molecular testing in the diagnosis of cutaneous soft tissue tumors. Semin Cutan Med and Surg 31:221–​233. PMID: 23174492. Coens C, van der Graef WTA, Blay J-​Y, et al. 2015. Health-​related quality-​of-​ life results from PALETTE: a randomized, double-​blind, phase 3 trial of pazopaib versus placebo in patient with soft tissue sarcoma whose disease has progressed during or after prior chemotherapy—​a European organization for research and treatment of cancer soft tissue and bone sarcoma group global network study (EORTC 62072). Cancer 121:2933–​2941. PMID: 26033286. Collins JJ, Jammer B, Sladeczek FM, Bodnar CM, Salomon SS. 2014. Surveillance for angiosarcoma of the liver among vinyl chloride workers. J Occup Environ Med 56:1207–​1209. PMID: 25376416. Crago AM, Singer S. 2011. Clinical and molecular approaches to well-​ differentiated and dedifferentiated liposarcomas. Curr Opin Oncol 23:373–​378. PMCID: PMC3253354. de Martel C, Shiels MS, Franceschi S, et al. 2015. Cancers attributable to infections among adults with HIV in the United States. AIDS 29:2173–​2181. PMCID: PMC4636914. Dodd RD, Mito JK, Eward WC, et al. 2013. NF1 deletion generates multiple subtypes of soft-​tissue sarcoma that respond to MEK inhibition. Mol Cancer Ther 12:1906–​1917. PMCID: PMC3825462. Enzinger FM, Weiss SW. 1996. Soft tissue tumors, 3rd ed. St. Louis, MO: Mosby—​Year Book. Fioretti F, Tavani A, Gallus S, Negri E, Franceschi S, La Vecchia C. 2000. Menstrual and reproductive factors and risk of soft tissue sarcomas. Cancer 88:786–​789. PMID: 10679647. Fletcher CDM, Bridge JA, Hogendoorn CS, Mertens F. 2013. World Health Organization. WHO classification of tumours of soft tissue and bone. Lyon: IARC Press. Forman D, Bray F, Brewster DH, et  al. 2013. Cancer incidence in five continents, Vol. X (electronic version). Lyon:  International Agency for Research on Cancer. Available from www.iarc.fr/​er/​publications/​pdfs-​ online/​epi/​sp 164/​ci5volx_​full.pdf, accessed October 6, 2015. Franceschi S, Serraino D. 1992. Risk factors for adult soft tissue sarcoma in northern Italy. Ann Oncol (Suppl 22) 2:S85–​S88. PMID: 1622876. Frisch M, Biggar RJ, Engels EA, Joedert JJ, for the AIDS-​Cancer Match Registry Study Group. 2001. Association of cancer with AIDS-​related immunosuppression in adults. JAMA 285:1736–​1745. PMID: 11277828. Froehner M, Wirth MP. 2001. Etiologic factors in soft tissue sarcoma. Onkologie 24:139–​142. PMID: 11441293. Gonzales CA, Kogevinas M, Gadea E, et al. 2001. Increase of dioxin blood levels over the last 4 years in the general population in Spain. Epidemiology 12:365. PMID: 11338318. Gramolelli S, Schulz TF. 2015. The role of Kaposi sarcoma-​associated herpesvirus in the pathogenesis of Kaposi sarcoma. J Pathol 235:368–​380. PMID: 25212381. Guo T, Zhang L, Chang NE, Singer S, Maki RG, Antonescu CR. 2010. Consistent MYC and FLT4 amplification in radiation-​induced angiosarcoma but not in other radiation associated atypical vascular lesions. Genes Chromosomes Cancer 50:25–​33. PMCID: PMC3150534. Hameed M. 2014. Molecular diagnosis of soft tissue neoplasia:  clinical applications and recent advances. Expert Rev Mol Diagn 14:961–​977. PMID: 25109831. Hansen MF, Koufos A, Gallie BL, et al. 1985. Osteosarcoma and retinoblastoma: A shared chromosomal mechanism revealing recessive predisposition. Proc Natl Acad Sci USA 82:6216–​6220. PMCID: PMC391023. Hardell L. 1977. Malignant mesenchymal tumours and exposure to phenoxy acids: a clinical observation. Lakartidningen 74:2753–​2754. PMID: 895290. Harris M, Hartley AL, Blair V, et  al. 1991. Sarcomas in North West England:  I.  Histopathological peer review. Br J Cancer 64:315–​320. PMID: 1892759. Hemminki K, Li X. 2001. A population-​based study of familial soft tissue tumors. J Clin Epidemiol 54:411–​416. PMID: 11297891. Hisada M, Garber JE, Fung CY, et al. 1998. Multiple primary cancers in families with Li-​Fraumeni syndrome. J Natl Cancer Inst 90:606–​611. PMID 9554443. Hoover R, Fraumeni JF, Jr. 1973. Risk of cancer in renal transplant recipients. Lancet 2:55–​57. PMID: 4123616.

 837



Soft Tissue Sarcoma

Howlader N, Noone AM, Krapcho M, Miller D, Bishop K, Altekruse SF, Kosary CL, Yu M, Ruhl J, Tatalovich Z, Mariotto A, Lewis DR, Chen HS, Feuer EJ, Cronin KA (eds). 2016. SEER Cancer Statistics Review, 1975–​2013, National Cancer Institute. Bethesda, MD, http://​seer.cancer. gov/​csr/​1975_​2013/​, based on November 2015 SEER data submission, posted to the SEER website April 2016. Hsu TC, Spitz MR, Schantz SP. 1991. Mutagen sensitivity: a biological marker of cancer susceptibility. Cancer Epidemiol Biomarkers Prev 1:83–​89. PMID: 1726967. Hussein, R B, Ludewig B, Kreipe H, Jonigk D. 2014. Clinico-​pathological characteristics of different types of immunodeficiency-​associated smooth muscle tumours. Eur Journal of Cancer 50: 2417–​2424. PMID: 25027306. Hwang S-​J, Lozano G, Amos CI, Strong LC. 2003. Germline p53 mutations in a cohort with childhood sarcoma: sex differences in cancer risk. Am J Hum Genet 72:975–​983. PMID: 12610779. Juzeniene A, Porojnicu AC, Baturaite Z, Lagunova Z, Aksnes L, Bruland ØS, Moan J. 2015. Vitamin D levels and dietary intake among patients with benign soft tissue tumors and sarcomas. Anticancer Res 35:1171–​80. PMID: 25667508. Kang S, Kim HS, Kim W, Kim JH, Kang SH, Han I. 2015. Comorbidity is independently associated with poor outcome in extremity soft tissue sarcoma. Clin Orthop Surg 7:120–​130. PMID: 25729528. PMCID: PMC4329524. King DT, Duffy DM, Hirose FM, et  al. 1979. Lymphangiosarcoma arising from lymphangioma circumscriptum. Arch Dermatol 115:959–​972. PMID: 464625. Krol JJ, Krol VV, Dawkins A, Ganesh HS. 2015. Case 213:  primary splenic angiosarcoma. Radiology 274:298–​303. PMID: 25531483. Laroche-​Clary A, Chaire V, Le Morvan V, et al. 2015. BRACA1 haplotype and clinical benefit of trabectedin in soft-​tissue sarcoma patients. Br J Cancer 112:688–​692. PMCID: PMC4333490. Li FP, Dreyfuss M, Russell TL, Verselis SJ, Hutchinson RJ, Fraumeni JF, Jr. 1997. Molecular epidemiology study of a suspected community cluster of childhood cancers. Med Ped Oncol 28:243–​247. PMID: 9072319. Li FP, Fraumeni JF, Jr. 1969. Soft-​ tissue sarcomas, breast cancer, and other neoplasms:a familial syndrome? Ann Intern Med 71:747–​752. PMID: 5360287. Lynge E, Storm HH, Jensen OM. 1987. The evaluation of trends in soft tissue sarcoma according to diagnostic criteria and consumption of phenoxy herbicides. Cancer 60:1896–​1901. PMID: 3308057. Malkin D, Li F, Strong LC, et al. 1990. Germ line p53 mutations in a familial syndrome of breast cancer, sarcomas, and other neoplasms. Science 250:1233–​1238. PMID: 1978757. Maretty-​Nielson K. 2014. Prognostic factors in soft tissue sarcoma. Dan Med J 61:B4957. PMID: 25370967. Mbulaiteye SM, Parkin DM, Rabkin CS. 2003. Epidemiology of AIDS-​related malignancies: an international perspective. Hematol Oncol Clin North Am 17:673–​696. PMID: 12853650. Miao C, Liu D, Zhang F, et al. 2015. Association of FPGS genetic polymorphisms with primary retroperitoneal liposarcoma. Sci Rep 5:9079. PMID: 25765001. Mitchell G, Ballinger ML, Wong S, et al. 2013. High frequency of germline TP53 mutations in a prospective adult-​onset sarcoma cohort. PLoS One 8:e69026. PMCID: PMC3718831. Molife R, Lorigan P, MacNeil S. 2001. Gender and survival in malignant tumours. Cancer Treatment Rev 27:201–​209. PMID: 11545540. Murphy JD, Ma GL, Baumgartner JM, et  al. 2015. Increased risk of additional cancers among patients with gastrointestinal stromal tumors:  a population-​based study. Cancer 121:2960–​2967. PMCID: PMC4545693. National Cancer Institute–​ PDQ. www.cancer.gov/​types/​soft-​tissue-​sarcoma/​ patient/​adult-​soft-​tissue-​treatment-​pdq#section/​-​-​26. Accessed October 2015. Neuberger M, Grossgut R, Gyimothy J, et al. 2000. Dioxin contamination of feed and food. Lancet 355:1883. PMID: 10866446. Nichols KE, Malkin D, Garber JE, et al. 2001. Germ-​line p53 mutations predispose to a wide spectrum of early-​onset cancers. Cancer Epidemiol Biomarkers Prev 10:83–​87. PMID: 11219776. Ognjanovic S, Olivier M, Bergemann TL, Hainaut P. 2012. Sarcomas in TP53 germline mutation carriers: a review of the IARC TP53 database. Cancer 118:1387–​1396. PMCID: PMC3329663 Parkin DM, Whelan SL, Ferlay J, Teppo L, Thomas D, eds. 2002. Cancer incidence in five continents, Vol. VIII. Lyon, France:  IARC Scientific Publication No. 155. IARC. PMID: 12812229. Piacitelli L, Marlow D, Fingerhut M, Steeland K, Sweeney MH. 2000. A retrospective job exposure matrix for estimating exposure to 2,3,7,8-​ tetrachlorodibenzo-​p-​dioxin. Am J Ind Med 38:28–​39. PMID: 10861764. Ray-​Coquard I, Montesco MC, Coindre JM, et al. 2012. Sarcoma: concordance between initial diagnosis and centralized expert review in a population

837

bases study within three European regions. Ann Oncol 23:2442–​2449. PMCID: PMC3425368. Samartzis D, Nishi N, Cologne J, et al. 2013. Ionizing radiation exposure and the development of soft-​tissue sarcomas in atomic-​bomb survivors. J Bone Joint Surg Am 95:222–​229. PMID: 23389785. Schecter A, ed. 1994. Dioxins and health. New York: Plenum Press. Schwartz RA, Micali G, Nasca MR, Scuderi L. 2008. Kaposi sarcoma: a continuing conundrum. J A Acad Dermatol 59:179–​206. PMID:18638627. SEER. 2015. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database:  Incidence—​SEER 9 Regs Research Data, Nov 2014 Sub (1973–​ 2012) Katrina/​ Rita Population Adjustment—​ Linked to County Attributes—​ Total U.S., 1969–​ 2013 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2015, based on the November 2014 submission. Serraino D, Franceschi S, Talamini R, et al. 1991. Non-​occupational risk factors for adult soft-​tissue sarcoma in northern Italy. Cancer Causes Control 2:157–​164. PMID: 1873445. Shantakumar S, Connelly-​Frost A, Kobayashi MG, Allis R, Li L. 2015. Older soft tissue sarcoma patients experience increased rates of venous thromboembolic events: a retrospective cohort study of SEER-​Medicare data. Clin Sarcoma Res 5:18. PMCID: PMC4515018. Silverberg MJ, Lau B, Achenbach CJ, et  al. 2015. Cumulative incidence of cancer among persons with HIV in North America: a cohort study. Ann Intern Med 163:507–​518. PMCID: PMC4711936. Singer S, Nielsen T, Antonescu CR. 2011. Molecular biology of soft tissue sarcoma. In: DeVita VTJr, Lawrence TS, Rosenberg SA, Cancer: principles and practice of oncology, 9th ed. Philadelphia:  Lippincott Williams & Wilkins, pp. 1522–​1577. Stiller CA, Trama A, Serraino D, et al. 2013. Descriptive epidemiology of sarcomas in Europe:  report from the RARECARE project. Eur J Cancer 49:684–​695. PMID: 23079473. Stratton MR, Moss S, Warren W, et  al. 1990. Mutation of the p53 gene in human soft tissue sarcomas:  association with abnormalities of the RBl gene. Oncogene 5:1297–​1301. PMID: 2216456. Strong LC, Stine M, Norsted TL. 1987. Cancer in survivors of childhood soft tissue sarcoma and their relatives. J Natl Cancer Inst 79:1213–​1220. PMID: 3480372. Strong LC, Williams WR, Ferrell RE, Tainsky MA. 1989. Genetic analysis of childhood sarcoma. Princess Takamatsu Symp 20:51–​157. PMID: 2488230. Szkandera J, Gerger A, Liegl-​Atzwanger B, et al. 2015. The derived neutrophil/​ lymphocyte ratio predicts poor clinical outcome in soft tissue sarcoma patients. Am J Surg 210:111–​115. PMID: 25586599. Taubert H, Meye A, Wurl P. 1996. Prognosis is correlated with p53 mutation type for soft tissue sarcoma patients. Cancer Res 56:4134–​4136. PMID: 8797580. Taubert HW, Bartel F, Kappler M, et al. 2003. Reduced expression of hMSH2 protein is correlated to poor survival for soft tissue sarcoma patients. Cancer 97:2273–​2278. PMID: 12712483. Tavani A, Pregnolato A, Negri E, et  al. 1997. Diet and risk of lymphoid neoplasms and soft tissue sarcomas. Nutrition Cancer 27:256–​260. PMID: 9101555. Tavani A, Soler M, LaVecchia C, Negri E, Gallus S, Franceschi S. 2000. Body weight and risk of soft-​tissue sarcomas. Br J Cancer 81:890–​892. PMID: 10555763. Taylor RS, Barrenrina J, Maki BG. Antonescu CR, Singer S, Ladanyi M. 2011. Advances in sarcoma genomics and new therapeutic targets. Nat Rev Cancer 11:541–​557. PMCID: PMC3361898. Toulmonde M, Le Cesne A, Mendiboure J, et al. 2014. Long-​term recurrence of soft tissue sarcomas: prognostic factors and implications for prolonged follow-​up. Cancer 120:3003–​3006. PMID: 24942887. Tuomisto J and Tuomisto JT. 2012. Is the fear of dioxin cancer more harmful than dioxin? Toxicology Letters 210:338–​344. PMID: 22387160. Varley JM, Evans DGR, Birch JM. 1997. Li-​Fraumeni syndrome: a molecular and clinical review. Br J Cancer 76:1–​14. PMID: 9218725. Vineis P, Zahm SH. 1988. Immunosuppressive effects of dioxin in the development of Kaposi’s sarcoma and non-​Hodgkin’s lymphoma. Lancet 1:55. PMID: 2891916. Vos JG, DeHeer C, Van Loveren. 1997–​1998. Immunotoxic effects of TCDD and toxic equivalency factors. Teratogenesis Carcinog Mutagen 17: 275–​ 284. PMID: 9508737. Vraa S, Keller J, Nielsen OS, et  al. 1998. Prognostic factors in soft tissue sarcomas:  the Aarhus experience. Eur J Cancer 34:  1876–​1882. PMID: 10023309. Wagner P, Alvegård T, Ranstam J, Rydholm A, Vult von Steyern F, Olsson H. 2014. Oral contraceptive use, parity, and constitutional characteristics

83

838

PART IV:  Cancers by Tissue of Origin

in soft tissue sarcoma:  a Swedish population-​based case-​control study 1988–​2009. Cancer Causes Control 25:1167–​1177. PMID: 25034461. Wieczorek-​Godlewska R, Miszewska-​Szyszkowska D, Slubowska K, Ognista-​ Gajda A, Chmura A, Durlik M. 2014. Dramatic recurrence of cancer in a patient who underwent kidney transplantation—​case report. Transplant Proc. 46:2897–​2902. PMID: 25380946. Wong FL, Boice JDJr, Abramson DH, et al. 1997. Cancer incidence after retinoblastoma. Radiation dose and sarcoma risk. JAMA 278:1262–​1267. PMID: 9333268. Wong JR, Morton LM, Tucker MA, et  al. 2014. Risk of subsequent malignant neoplasms in long-​ term hereditary retinoblastoma survivors after chemotherapy and radiotherapy. J Clin Oncol 32:3284–​3290. PMCID: PMC4178525. World Cancer Research Fund/​American Institute for Cancer Research. 2007. Food, nutrition, physical activity and the prevention of cancer: a global perspective. Washington, DC: AICR.

Wu J, Wang X, Lin C, Yu S, Cai L, Gao Z. 2014. Carcinosarcoma of native renal pelvis in recipient after a renal transplant:  a case report. World J Surg Oncol 12:407. PMCID: PMC4320576. Yap J, Chuba PJ, Thomas R, et al. 2002. Sarcoma as a second malignancy after treatment for breast cancer. Int J Radiat Oncol Biol Phys 52:1231–​1237. PMID: 11955733. Youn P, Milano MT, Constine LS, Travis LB. 2014. Long-​term cause-​specific mortality in survivors of adolescent and young adult bone and soft tissue sarcoma: a population-​based study of 28,844 patients. Cancer 120:2334–​ 2342. PMID: 24752471. Yuen NK, Li CS, Monjazeb AM, Borys D, Bold RJ, Canter RJ. 2015. Older age impacts radiotherapy-​related outcomes in soft tissue sarcoma. J Surg Res. 199:499–​505. PMCID: PMC4789295. Zhang D, Ding Y, Wang Z, Wang Y, Zhao G. 2015. Impact of MDM2 gene polymorphism on sarcoma risk. Tumour Biol 36:1791–​1795. PMID: 25366142.

 839

44 Thyroid Cancer CARI M. KITAHARA, ARTHUR B. SCHNEIDER, AND ALINA V. BRENNER

OVERVIEW Thyroid cancer, once considered relatively uncommon in the general population, is now the eighth most commonly diagnosed cancer among women worldwide, and the third most common cancer among women under 45 years of age. The incidence is substantially higher in women than men (3:1 ratio); this differential is highest between ages 15 and 39 and declines with age. Nearly all thyroid cancers derive from the follicular epithelium, and the most common histological type is papillary thyroid cancer (PTC). Incidence of thyroid cancer has been increasing in many countries since the early 1980s. This trend appears to be attributable to diagnostic changes, improvements in the detection and diagnosis of smaller PTCs, and changes in the prevalence of environmental factors. While less common, the incidence of larger, more advanced-​stage PTCs has increased at a similar rate to that of smaller PTCs. Mortality has remained fairly stable over time relative to incidence, though rates in US men have increased slightly over the past two to three decades. To date, ionizing radiation exposure remains the only established modifiable environmental cause of thyroid cancer, though there is increasing evidence that other factors, including obesity, cigarette smoking, hormonal exposures, and certain environmental contaminants, may also play a role. Data from large, prospective cohort studies over the past decade have improved our understanding of thyroid cancer etiology and have helped to generate several new hypotheses. Given recent developments in identifying distinct molecular subtypes of thyroid cancer, future studies that incorporate molecular subtype information in epidemiologic studies of thyroid cancer, in addition to histologic and staging information, could provide greater insight on thyroid cancer as a potentially etiologically heterogeneous disease.

dose–​response association between external radiation exposure and thyroid cancer risk (Hempelmann, 1968). A  series of observational studies on radiation and thyroid cancer followed, including a pooled analysis of case-​control and cohort studies that quantified radiation risk and clearly demonstrated its modification by age at exposure (Ron et al., 1995). Case-​control studies that evaluated other exposures, including benign thyroid diseases, diet, anthropometric factors, and reproductive and hormonal factors, soon followed, including a series of pooled analyses of case-​control studies starting in the mid-​1990s that has been a major source of information about thyroid cancer etiology (Preston-​Martin et al., 2003). Large prospective studies, including pooled analyses of prospective studies, have been published more recently. These studies have generally confirmed the findings from case-​control studies and have identified additional factors that may have influenced thyroid cancer risk. Understanding the etiology of thyroid cancer has been complicated by the fact that an increasing number of thyroid cancers are detected incidentally, either through diagnostic imaging for other, unrelated conditions or diagnostic workup of other thyroid disorders. This is an important source of bias, as exposures associated with a greater opportunity for incidental detection of thyroid cancer may appear as “true” risk factors. Methods to account for detection bias in epidemiologic studies of thyroid cancer are critical. In particular, studies that incorporate information on tumor ascertainment, staging, and molecular markers could substantially improve the understanding of thyroid cancer etiology. This chapter will emphasize the literature published since this topic was reviewed in the previous edition of Cancer Epidemiology and Prevention (Ron and Schneider, 2006).

TUMOR CLASSIFICATION INTRODUCTION The thyroid is a butterfly-​shaped endocrine gland that is located low in the neck surrounding the anterior and lateral sides of the trachea. Thyroid follicular cells produce thyroxine (T4) and triiodothyronine (T3). In the circulation, it is the fraction of these hormones not bound to proteins (free T4 and free T3, also referred to as fT4 and fT3) that is active. Among their many actions, they regulate growth and metabolism and influence brain function, neural development, dentition, and bone development (Demers, 2004). The synthesis of T4 and T3 requires iodide and production of these hormones is regulated by thyroid stimulating hormone (TSH), which is synthesized in and secreted from the anterior pituitary gland. Thyroid parafollicular cells (C-​cells) produce calcitonin, which may play a role in bone metabolism. Nearly all thyroid cancers derive from the follicular epithelium, and the most common type is PTC. The incidence of thyroid cancer, primarily PTC, has increased dramatically in many regions of the world since the early 1980s (Kilfoy et al., 2009). The proposed explanations for this increase have been the subject of much debate (Ito et al., 2013). For most types of thyroid cancer, the prognosis is very good, and the mortality rates for thyroid cancer are much lower relative to incidence. The observation that children diagnosed with thyroid cancer were more likely to have previous irradiation to the head or chest prompted the first observational studies in the 1950s designed to quantify the

A “goiter” is any enlargement of the thyroid gland. A  “nodule” is the broadest term for a mass lesion in the thyroid. In autopsy series, up to about 50% of thyroids are found to have nodules (specifically, 52.5% of 1000 autopsies performed at the Mayo Clinic in the early 1950s) (Mortensen et al., 1955). Most are benign and are not considered to be neoplasms; rather, they are referred to as colloid or hyperplastic nodules. In the Mayo Clinic series, only 28 of 525 nodules (5%) were classified as cancers. Follicular adenomas are considered to be neoplastic, or monoclonal. However, monoclonality is a required, but a not sufficient, criterion since some nodules within a multinodular goiter are the result of clonal expansion (Kopp et al., 1994).

Histological Types In epidemiologic studies, thyroid cancer is typically classified according to histological characteristics. These divisions should be regarded as fluid, as changes in classification over time occur as more data including genomic data are accumulated and analyzed (see further discussion later in this chapter). About 95% of all thyroid cancers (ICD-​10 code C73.9) originate from cells derived from the follicular epithelium, and, of these, the major histologic types are divided into PTCs (ICD-​O-​3 codes 8050, 8260, 8340-​ 8344, 8350, 8450), follicular thyroid cancers (FTCs)

839

840

840

Part IV:  Cancers by Tissue of Origin Table 44–​1.  Distribution of Thyroid Cancer by Histologic Type in Selected Countries, 2003–​2007 Histology %a Country, Region/​Race

No. of Casesb Papillary

Australia, New South Wales Canada China, Beijing Costa Rica Denmark Finland Iceland India, New Delhi Ireland Israel Japan, Hiroshima Korea, Republic of Netherlands Norway Singapore, Chinese Ukraine UK, England UK, Scotland United States c

2946 16,976 1624 1473 811 1670 143 651 556 2874 950 66,772 2149 1056 657 11,227 6764 768 131,434

Follicular

Medullary

Anaplastic

13.1 7.0 3.4 5.8 17.4 11.9 8.4 18.1 25.4 3.6 4.6 3.1 20.0 12.4 13.7 18.1 24.5 24.9 9.7

3.0 1.7 3.4 1.8 6.4 2.8 0.7 4.3 4.9 3.1 0.3 0.7 6.7 3.3 2.4 3.0 5.1 2.6 2.1

1.3 0.9 0.6 1.2 6.5 0.4 2.1 4.0 2.9 1.7 1.2 0.3 5.0 5.0 1.8 1.7 3.9 1.0 1.1

81.4 89.5 84.6 90.1 68.1 84.6 87.4 67.4 65.1 91.4 93.3 92.5 66.5 78.6 80.7 72.6 63.0 65.8 86.2

Totals do not add to 100 because the category “other specified cancers” is not shown. Unspecified cancers, sarcoma, other morphology and unspecified morphology are not included. US National Program of Cancer Registries (NPCR) (42 states) Source: Cancer incidence in five continents, Vol. X (Forman D, 2014). a

b c

12 10 Incidence Rate Ratio

(ICD-​O-​3 codes 8330–​8332, 8335), and anaplastic thyroid cancers (ATCs) (ICD-​O-​3 codes 8004, 8012, 8020–​8021, 8030–​8032) (SEER, 2012). For clinical and etiologic purposes, they often are grouped into well differentiated (PTC and FTC) and poorly differentiated (including ATC). Medullary thyroid cancers (MTCs) (ICD-​O-​3 code 8510) also arise from epithelial cells, but from the calcitonin-​producing C-​cells. There are several non-​epithelial thyroid tumors, all quite rare, including malignant lymphomas and sarcomas (ICD-​O-​3 codes 8800-​9775). There are small differences in the codes aggregated into the larger categories mentioned here between data from Surveillance, Epidemiology, and End Results Program (SEER) and Cancer Incidence in Five Continents, generally involving rare subtypes (Forman et  al., 2014; SEER, 2012). In this section, US data are from SEER and international data are from Cancer Incidence in Five Continents, unless otherwise indicated. The most common histological type of thyroid cancer is, by far, PTC (Table 44–​1), representing between 60% and 95% of all thyroid cancers within individual countries. The proportion of thyroid cancers that are PTCs has increased over time in many countries, including the United States (Aschebrook-​Kilfoy et  al., 2013a; Figure 44–​1). The earlier phase of this increase was a result, in part, from the 1988 change in definition of PTCs to include the follicular variant, but the most important factor contributing to this trend has been the ability to diagnose smaller PTCs, primarily by ultrasonography and other imaging modalities. When iodine supplementation occurs in iodine-​ deficient regions, the proportion of PTCs often increases (Burgess et al., 2000; Lind et al., 2002; Mete and Asa, 2013). Depending on the iodine status of the region, the proportion of thyroid cancers that are FTCs can range from less than 5% (e.g., China [Beijing], Republic of Korea, Japan [Hiroshima], Israel) to over 20% (e.g., Netherlands, United Kingdom) (Table 44–​1). In some areas there may be under-​or over-​reporting of FTC due to the diagnostic difficulty (Cipriani et al., 2015). On the other hand, it is relatively easy to diagnose PTC, and ascertainment is more complete. MTCs and ATCs typically account for 5 cm was 248%, 106%, 113%, and 222%, respectively, among white females from 1988–91 to 2003–05 (Enewold et al., 2009). Thus, discovery of latent disease does not appear to explain all of the observed increase in incidence of PTC in the United States, and changes in the prevalence of etiologic factors over time have been suggested to have also contributed to the increasing incidence (Enewold et al., 2009). Some environmental and lifestyle-​related factors that have been hypothesized to play a role in the increase include obesity, radiation exposure, particularly from medical sources, and environmental pollutants (see later discussion in this chapter for more details). The marked 10-​fold increase in the incidence of thyroid cancer in South Korea over the past 15 years, with an annual percent change of about 24%–​25%, has been suggested to be almost completely a reflection of greater detection and diagnosis of patients with low-​risk PTC (Ahn et al., 2014). During the same period, use of thyroid-​screening ultrasounds and referral to ultrasound-​guided fine-​needle aspiration biopsy simultaneously increased (Oh et al., 2015). Nonetheless, there is evidence that etiologic factors play some role in the increase in thyroid cancer among Korean adolescents and young adults, who are unlikely to participate in thyroid cancer screening, as well as men born after 1950 (Oh et al., 2015). In most countries, thyroid cancer mortality rates have either declined steadily or have remained fairly stable at a low level over the last few decades (La Vecchia et  al., 2015). In the United States, thyroid cancer mortality has remained largely stable in women but

84

844

Part IV:  Cancers by Tissue of Origin Thyroid cancer trends 20

Rate per 100,000 person-years

10

Incidence Total Papillary Follicular Medullary Anaplastic Other/unspec Mortality Total 1

0.1 1990

2000

2010

Calendar year

Figure 44–​2.  Trends in total and histology-​specific thyroid cancer incidence and total thyroid cancer mortality in the United States, 1993–​2012. Rates are age-​adjusted to the 2000 US standard population, and each point represents 4 years. Incidence data are restricted to microscopically confirmed cases and exclude cases identified only by autopsy or death certificates. APC = annual percent change based on rates age-​adjusted to the 2000 US standard population (19 age groups, Census P25-​1130); a calendar year of diagnosis or death. Source: SEER (2015a, 2015b).

has increased slightly in men between 1984 and 2012 (1.3% per year) (Howlader, 2015). Thyroid cancer mortality for men and women combined has increased at a rate of 0.8% per year between 1993 and 2012 (Figure 44–​2).

RISK FACTORS Predisposing Medical Conditions Benign Thyroid Disease

Thyroid cancer is sometimes preceded by other thyroid abnormalities, including endemic and sporadic goiter, benign thyroid nodules, chronic lymphocytic thyroiditis, or Graves’ disease. Of the functional disorders of the thyroid, hypothyroidism (thyroid hormone insufficiency) is the most common (Demers, 2004). When the disorder is intrinsic to the thyroid (most often due to autoimmune chronic thyroiditis, Hashimoto’s disease), it is referred to as “primary hypothyroidism”; if it is caused by a deficiency in the pituitary or hypothalamus it is termed “central hypothyroidism.” Hyperthyroidism, in contrast, is a state of excess thyroid hormone. One major cause of hyperthyroidism, Graves’ disease, is an autoimmune disease characterized by the production of antibodies to the TSH receptor. Another major cause of hyperthyroidism is a multinodular goiter (an enlarged thyroid with multiple nodules) that has developed uncontrolled production of thyroid hormones.

A pooled analysis of 14 case-​control studies with 2725 cases and 4776 controls was conducted to allow a systematic approach to analyzing and interpreting major hypotheses for thyroid cancer etiology (Franceschi et al., 1999; Negri et al., 1999). Large risks were associated with a self-​reported history of goiter (odds ratio [OR] = 5.9; 95% CI:  4.2, 8.1) and benign nodules (OR  =  29.9; 95% CI:  14.5, 62.0) among women; among men, the risk with goiter was even higher (OR  =  38.3; 95% CI:  5.0, 291.2), and the risk with benign nodules was too high to estimate. In a multiethnic case-​control study designed to evaluate reasons for higher incidence rates of thyroid cancer among Asian women living in the San Francisco area, a history of prior goiter or nodules was a significant explanatory factor (Haselkorn et al., 2003). Case control studies, however, are especially prone to recall bias, in which cases may better remember or inaccurately remember previous thyroid diseases than controls. This is less of a problem in prospective studies, although potential ascertainment bias (i.e., one thyroid disorder could draw attention to another) or delayed diagnosis (the preceding condition could actually be an undiagnosed cancer) may not be totally excluded. In a cohort study of 204,964 members of a large health plan, 196 thyroid cancers occurred, and a history of goiter (as well as Asian race, educational attainment, family history, and radiation exposure) was a risk factor (relative risk [RR] = 3.36; 95% CI: 1.82, 6.20) (Iribarren et al., 2001). In another cohort study of 90,713 US radiologic technologists followed prospectively from 1983 through 2006 using periodic surveys, there were 282 incident thyroid cancers. Based on reports in earlier surveys of this population, the hazard ratio (HR) for preceding goiter was 4.1 (95% CI: 2.6, 6.5) for women and 9.3 (95% CI: 1.3, 69.2) for men (Meinhold et al.,

 845

Thyroid Cancer 2010). Perhaps the most convincing data on risk of thyroid cancer with benign thyroid diseases come from several large cohort studies that relied on hospital discharge registries for ascertainment of benign thyroid diseases rather than self-​report (Balasubramaniam et  al., 2012; From et al., 2000; Mellemgaard et al., 1998). In the period 1977–​1991, there were 57,326 patients discharged from Danish hospitals with a diagnosis of goiter, hypothyroidism, or thyrotoxicosis (ICD-​8 codes 240–​242, 244) (Mellemgaard et al., 1998). The subsequent incidence of thyroid cancer was ascertained through the Danish Cancer Registry and compared with thyroid cancer incidence in the general population. The incidence of thyroid cancer was increased among all three patient groups but was substantially higher among goiter patients. Thyroid cancer risk decreased with time following hospital discharge, but the risk remained elevated even 10 years later (Mellemgaard et al., 1998). In a US discharge registry-​based study of about 4.5  million male military veterans admitted to 142 Veteran Affairs (VA) hospitals between 1969 and 1996, nearly 1000 thyroid cancer cases were identified (Balasubramaniam et al., 2012). The risk of thyroid cancer was very high with several benign thyroid diseases preceding diagnosis of thyroid cancer, particularly within 5 years of cancer diagnosis. However, high risks with follicular adenoma (RR  =  28.9; 95% CI:  9.2, 90.2), non-​toxic nodular goiter (RR  =  25.9; 95% CI:  17.9, 38.0), Hashimoto’s disease (RR = 12.9; 95% CI: 4.8, 34.4), and hypothyroidism (RR = 6.0; 95% CI: 3.8, 9.6) were also observed more than 5 years prior to thyroid cancer diagnosis, lessening the probability of ascertainment bias or delayed diagnosis. A difficult question to resolve has been whether autoimmune thyroid diseases (Hashimoto’s or Graves’ disease), often manifesting as hypothyroidism or hyperthyroidism, are risk factors or precursors of thyroid cancer. Published studies tend to be retrospective, variable in patient selection, and usually without controls (e.g., studies using data from thyroidectomy series) (Feldt-​Rasmussen and Rasmussen, 2010). Even in those with controls, the appropriateness is not always optimal. Most, but not all, thyroidectomy studies have concluded that the coexistence of Hashimoto’s disease and Graves’ disease with thyroid cancer is higher than expected; however, these studies are particularly prone to selection bias. Also, thyroidectomy studies must account for the immune infiltrate elicited by, and surrounding, many thyroid cancers. In contrast, a review of population-​based fine needle aspiration biopsy (FNAB) studies did not find an association between Hashimoto thyroiditis and thyroid cancer (Jankovic et al., 2013). Studies of subtle changes in thyroid metabolism and function in relation to risk of thyroid cancer have concerned serum levels of TSH with or without corresponding changes in T4 and T3 level. As TSH is known to increase thyroid cell proliferation, it has long been hypothesized to promote the development of thyroid cancer. In thyroid nodule patients, higher TSH levels are predictive of thyroid malignancy (Boelaert et  al., 2006). However, there is little evidence that higher TSH levels in individuals without thyroid nodules increase their risk of future thyroid cancer. This is partly due to the fact that few epidemiologic studies of thyroid cancer had access to prediagnostic serum samples. One small case-​control study of 43 thyroid cancers and 128 controls selected from individuals who contributed serum samples to a biological bank of Norway did not find a significant difference in serum TSH concentration between cases and controls (Thoresen et al., 1988), while a large study of 357 differentiated thyroid cancer cases and matched controls in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort (Rinaldi et al., 2014) found a significant inverse association for thyroid cancer risk with prediagnostic levels of TSH (OR for the highest versus lowest quartile of TSH 0.56; 95% CI:  0.38, 0.81, P  =  0.001). Investigators in both studies found significantly increased risk with higher serum levels of the thyroid-​ specific protein thyroglobulin (Tg) (Rinaldi et  al., 2014; Thoresen et  al., 1988). The association for Tg in the EPIC study was greater when closer in time to thyroid cancer diagnosis, while the association for TSH did not differ by follow-​up time (Rinaldi et al., 2014). In addition, there was a significantly increased risk of thyroid cancer with higher levels of antibodies to Tg (Rinaldi et al., 2014). These findings have not yet been replicated, to the best of our knowledge. The unexpected finding of higher thyroid cancer risk with lower levels of

845

TSH, if confirmed, illuminates the substantial gap in our understanding of the factors involved in the etiology and natural history of thyroid cancer. The cumulative data suggest that a history of goiter or benign thyroid nodules is one of the strongest and most consistently associated factors with risk of thyroid cancer. Whether this association is causal or through shared genetic susceptibility and/​or environmental risk factors remains to be determined. The evidence concerning associations with autoimmune thyroid diseases or changes in thyroid function are less consistent and challenging to study due to methodological issues. Some of these issues could be addressed in large prospective studies of thyroid cancer with long follow-​up and prediagnostic biological samples, including blood, serum, and urine.

Non-​Thyroid Immune-​Related Conditions

A number of non-​thyroid autoimmune diseases (e.g., systemic lupus erythematosus, Sjogren’s syndrome, and type 1 diabetes), characterized by activation of the immune system against host antigens, have been associated with increased risk of thyroid cancer (Bernatsky et al., 2013; Harding et al., 2015; Mao et al., 2016; Weng et al., 2012; Zhang et al., 2014). A meta-​analysis of seven cohort studies of patients with systemic lupus erythematosus found a pooled standardized incidence ratio (SIR) for thyroid cancer of 2.22 (95% CI: 2.11, 2.34) with no heterogeneity in risk across the studies (Zhang et al., 2014). Another meta-​ analysis of 14 studies of patients with Sjogren’s syndrome reported a comparably elevated, pooled RR for thyroid cancer of 2.58 (95% CI: 1.14, 4.03) (Liang et al., 2014). An Australian national diabetes registry study reported significantly elevated SIR for thyroid cancer with type 1 diabetes in females (Harding et al., 2015). Consistency of the associations for thyroid cancer with systemic lupus erythematosus and Sjogren’s syndrome that, similar to thyroid cancer, are more common in females is intriguing; however, it remains unclear whether these associations are causal or due to co-​occurrence with Hashimoto thyroiditis or Graves’ disease, increased medical surveillance of patients with autoimmune diseases, or other factors. Another spectrum of immune abnormalities, namely heightened immune response to external antigens (e.g., manifest as allergic rhinitis, eczema, or asthma), has not been extensively studied in relation to risk of thyroid cancer, and the available data are inconsistent (Hemminki et al., 2014; Hwang et al., 2012; Ji et al., 2009; Meinhold et al., 2010). Significantly increased SIR for thyroid cancer in patients with allergic rhinitis was found in a Taiwanese study (Hwang et al., 2012), but in a Swedish study the relative risk estimate was close to null (Hemminki et  al., 2014). While the risk of thyroid cancer was significantly increased with self-​reported history of asthma in the US Radiologic Technologists Study (Meinhold et al., 2010), no association was found in a nationwide cohort study in Taiwan (Hwang et al., 2012). An approximate 3-​to 7-​fold increased risk of thyroid cancer has been observed in several studies of organ transplant recipients (Engels et al., 2011; Karamchandani et al., 2010; Mosconi et al., 2011) known to be immunodeficient due to long-​term use of immunosuppressive therapy. It is currently unclear whether this association is attributable to immune suppression, an underlying medical condition common in transplant recipients that is associated with thyroid cancer risk, or some other exposures that are more prevalent in this population, or whether it simply reflects heightened medical surveillance and, thus, greater likelihood of detection and diagnosis of thyroid cancer.

Multiple Primary Cancers

Studies of multiple primary cancers associated with thyroid cancer found that the incidence of thyroid cancer was elevated after most first primary cancers and that the incidence of several second primary cancers was increased after a first thyroid cancer (Hsu et  al., 2014; Hung et al., 2016; Lu et al., 2013; Ronckers et al., 2005; Subramanian et  al., 2007). Cancer sites that demonstrated consistently increased risks with thyroid cancer in both directions include salivary glands, breast, prostate, renal, brain and CNS, and leukemia. Increased medical surveillance undoubtedly accounts for some of the increased risk of second cancers following first cancers, as evidenced by a decline

846

846

Part IV:  Cancers by Tissue of Origin

in risks with longer time since diagnosis of the first cancer (Ronckers et al., 2005). Higher risks of second cancers following radiation treatment or chemotherapy for the first primary cancer have also been observed, though shared genetic susceptibility and exposure to environmental factors might also be important (Hung et al., 2016; Lu et al., 2013; Ronckers et  al., 2005). From this point of view, the positive, bidirectional association is noteworthy between female breast cancer and thyroid cancer among patients without radiation treatment, as well as increased risk of renal cancer not specific to I-​131 treatment for thyroid cancer and increased risk of thyroid cancer among kidney cancer patients receiving no radiation treatment (Ronckers et al., 2005). In the case of breast cancer, the association may point toward the role of female hormones, while in renal cancer toward genetics or thyroid hormones dysfunction.

Ionizing Radiation Exposure to ionizing radiation during childhood is one of the strongest and most consistently associated risk factors for thyroid cancer. This relationship was first recognized by Duffy and Fitzgerald in 1950. They found that an unusually large fraction of their childhood thyroid cancer patients had a history of radiation therapy for benign conditions of the head and neck.

External Radiation

There is much data concerning the risk of thyroid cancer following exposure to external ionizing radiation (e.g., gamma radiation or X-​rays) at different dose levels, particularly received at high-​dose rates. These data derive from studies of the atomic bomb survivors in Hiroshima and Nagasaki exposed to a wide range of doses (mean: 0.18 Gy, range: < 0.1–​3.99 Gy), patients exposed to low-​to-​moderate doses of radiation (mean: 0.24 Gy, range: 0–​28.5 Gy) for treatment of various benign conditions (e.g., thymic enlargement, lymphoid hyperplasia, haemangioma, tinea capitis, etc.) or moderate-​to-​high doses for treatment of cancer (mean:  9.2 Gy, range:  0.1–​76 Gy) (UNSCEAR, 2006; 2013). Here, for the most part, we rely on studies providing quantitative estimates of radiation risk per unit of absorbed dose. A 1995 pooled analysis (a meta-​analysis of individual data) of seven cohort studies reported by Ron et al. was the first comprehensive analysis of external radiation exposure (primarily in the low-​to-​moderate dose range) and thyroid cancer risk (Ron et al., 1995). This analysis included nearly 500 patients with thyroid cancer, most of which were PTCs, and demonstrated a strong, positive association between radiation dose received in childhood (< 15 years) and thyroid cancer (Figure 44–​3). The overall excess relative risk (a measure of cancer risk due to

Fitted Thyroid Cancer Dose-Response

Relative Risk

30

20

10

Age at exposure: < 15: ≥ 15:

0 0

1

2

3

4

5

Dose (Gy)

Figure 44–​3.  Pooled fitted thyroid cancer dose–​response curves from five cohort studies of childhood exposure (< 15 years) and two studies of adult exposure (≥ 15 years). Source: Adapted from Ron et al. (1995).

radiation exposure above the background risk often used in radiation epidemiology, ERR  =  RR -​1), was 7.7 per Gy (95% CI:  2.1, 28.7) (i.e., RR at 1 Gy = 8.7) and linearity best described the dose response down to doses as low as 0.1 Gy. One of the most notable findings was that the ERR per Gy strongly varied by age at exposure, with children exposed under 1 year of age having ERR five times higher than children exposed at 10–​14 years, and with little evidence for a radiation-​ related increase among persons exposed after age 15 years. Several new and updated studies of external radiation and thyroid cancer risk have been published, extending the available evidence. Among these is a study of thyroid cancer incidence among atomic bomb survivors with a follow-​up through 2005 (Furukawa et al., 2013). Based on a linear dose–​response model allowing for independent modification of radiation risk by age at exposure and attained age, the estimated ERR was 1.28 per Gy (95% CI: 0.59, 2.70) for individuals at age 60 after exposure at age 10. The radiation risk decreased sharply with both increasing age at exposure and attained age, and there was little evidence of increased thyroid cancer risk after exposure at 20  years of age or older. Of 371 incident cases included in the analysis, 80% were PTCs. To evaluate the radiation dose–​response relationship at high doses where the data were limited, Veiga et al. (2012) pooled two cohort and two nested case-​control studies of childhood cancer survivors including 16,757 patients and 187 second primary thyroid cancers (> 80% PTCs). The radiation-​related risk increased approximately linearly up to 10 Gy, leveled off at 10–​30 Gy, and then declined, suggesting a cell-​killing effect at very high doses. The investigators found significant negative interaction (i.e., departure from multiplicative interaction) between chemotherapy with alkylating agents or bleomycin and radiation dose. The highest risks were observed in patients treated with either type of chemotherapy without radiation, and decreasing risks with greater radiation doses in patients who had both chemo-​and radiotherapy. As combinations of chemotherapy were common and data limited, it was difficult to separate the effects of individual drugs and type of statistical interaction (additive vs. sub-​multiplicative). The investigators also found that radiation-​related risk decreased with increasing age at exposure, but did not vary with attained age, time since exposure, or number of treatments. Veiga et  al. have conducted a new comprehensive pooling of 12 studies of external radiation in childhood and adolescence (< 20 years of age) with longer follow-​up for studies included in the 1995 analysis and adding cancer survivor studies (Veiga et al., 2016). With 1070 cases of thyroid cancer and 5.3 million person-​years, the current study more than doubled the original 1995 data. As before, there was a strong, significant dose–​response relationship over the entire dose range. The RRs increased supra-​linearly through 2–​4 Gy, linearly between 4–​10 Gy, leveled off between 10–​30 Gy, and declined thereafter. The joint effect between radiation and chemotherapy was consistent with additivity. The fitted RR adjusted for chemotherapy was 6.5 at 1 Gy (95% CI:  5.1, 8.5), more precise, but lower than 8.7 at 1 Gy in the 1995 pooled analysis (Ron et al., 1995). Also, as before, a significant linear dose–​response relationship was evident for doses down to 0.1 Gy and perhaps even lower. The radiogenic effect was observed for both PTC (n = 841) and non-​PTCs (n = 143). While the RRs for non-​PTCs were higher than for PTCs, they were not significantly different. There was a significant, monotonic decrease in radiation risk with increasing age at first exposure and attained age, while the pattern of risk by time since exposure was non-​monotonic, increasing through 20–​30  years after exposure and then declining, although remaining elevated throughout the follow-​up. The RRs at 1 Gy in males and females were not significantly different, and there was no consistent support for a differential effect by number of radiation treatments. Information on risks of thyroid cancer with external low dose-​rate exposures is limited and mainly derives from studies of occupationally exposed workers. However, these are not very informative because thyroid doses are not always available and studies are often based on cancer mortality, thereby missing most cases of thyroid cancer. In the few studies of occupational exposure with incidence data (UNSCEAR, 2006), the ERR per Gy for thyroid cancer was significantly or borderline significantly elevated, being 5.9 (95% CI: 2.5, 9.9) in the Canadian study of occupational exposure (Sont et al., 2001), 1.9 (95% CI: 0.3,

 847

Thyroid Cancer 4.4) in the Chinese study of X-​ray technologists (Wang et al., 2002), and 3.2 (95% CI:  –​0.5, 17.5) in the UK study of radiation workers (Muirhead et  al., 2009). However, because these risk estimates are based on a small number of cases, relate to adult exposure, and could be partially attributed to better medical care and reporting, particularly among medical workers, there remains uncertainty about thyroid cancer risk associated with occupational radiation exposure. In summary, much has been learned about external radiation exposure and the risk of thyroid cancer, particularly PTC; however, a number of questions remain, including the risks associated with very low doses of radiation and low dose-​rate exposures. This is important in view of the increased population exposure to medical diagnostic sources of radiation, attributed largely to the growing use of computed tomography scans (Schonfeld et al., 2011a; Spampinato et al., 2015). More data are also needed concerning the role of genetic background, preexisting thyroid diseases, and other factors as potential modifiers of radiation-​related thyroid cancer risk.

Internal Radiation Internal thyroid radiation exposure includes diagnostic and therapeutic use of radioactive iodine (RAI, iodine-​131 [I-​131] and other radioactive isotopes of iodine) and other thyroid-​targeted isotopes (e.g., 99mTc as pertechnetate), exposure to RAI released into the environment due to nuclear reactor accidents (Three Mile Island, Chernobyl, Fukushima), nuclear weapons testing (Marshall Islands, Nevada, Kazakhstan, etc.) and production (Hanford, Mayak nuclear production complexes).

Medical Exposure to I-​ 131. Studies of medically exposed

patients were the major source of evidence on thyroid cancer risk with I-​131 prior to the Chernobyl accident. A cohort of about 36,000 predominantly adult patients administered diagnostic I-​131 was followed through 1998 (Dickman et  al., 2003). The most common reasons for diagnostic administration of I-​131 were suspected thyroid tumor (32%), hyperthyroidism (42%), hypothyroidism (17%), and hypercalcemia (8%). The average absorbed thyroid dose was 0.94 Gy. Excess cancer cases were observed only among patients who had been previously exposed to external radiotherapy to the neck and whose scan was performed due to suspicion of a thyroid tumor, while there was no dose response among patients without prior exposure to radiation and those who were referred for I-​131 testing for other reasons. Among patients irradiated with I-​131 and no history of external radiotherapy before age 20 (n = 2367), three thyroid cancers were observed, resulting in a SIR of 1.01. Another small study of diagnostic I-​131 exposure was conducted in 1907 German children less than 18 years old, 789 of whom had a past history of I-​131 procedures and 1118 of whom had other thyroid-​related tests without I-​131 (Hahn et al., 2001). The average thyroid dose was 1.0 Gy and the reported RR for thyroid cancer was close to unity (RR = 0.86; 95% CI: 0.14, 5.13). High doses of I-​131 are used to treat hyperthyroidism caused by Graves’ disease or toxic nodular goiter. Dosimetry for these purposes has not been standardized, although administered radioactivity of I-​ 131 between 3 and 8 MBq (80–​220 µCi) per gram of thyroid tissue is considered therapeutically appropriate (Silberstein et al., 2012). The purpose of I-​131 treatment is to destroy hyperactive thyroid tissue, but there is a possibility that it may induce carcinogenesis in the remaining gland. Several cohort studies in Sweden (Holm et al., 1991), England (Franklyn et al., 1999), Finland (Ryodi et al., 2015), and the United States (Hoffman et al., 1982; Ron et al., 1998) investigated the association between thyroid cancer risk and therapeutic amount of I-​131 radioactivity. The results have been inconsistent. The Swedish (Holm et  al., 1991)  and Finnish (Ryodi et  al., 2015)  studies found a non-​ significant increase in incidence of thyroid cancer with SIR of 1.29 (95% CI:  0.76, 2.03) and RR of 1.30 (95% CI:  0.59, 2.83), respectively. In contrast, the SIR and standardized mortality ratio (SMR) in the British study (Franklyn et  al., 1999)  were significantly elevated, 3.25 (95% CI: 1.69, 6.25) and 2.78 (95% CI: 1.16, 6.67), respectively. The largest study thus far, although confined to thyroid cancer death,

847

was conducted in the United States by the Thyrotoxicosis Therapy Follow-​up Study Group (Ron et al., 1998). Among 35,000 hyperthyroid patients treated between 1946 and 1964 (63% with I-​131) and followed through 1990, there were 24 deaths from thyroid cancer among I-​131 treated patients and the corresponding SMR for thyroid cancer was significantly elevated (SMR = 3.94; 95% CI: 2.52, 5.86). However, 50% of the deaths occurred within 4 years of I-​131 treatment and in an internal analysis, the trend with administered I-​131 activity, while positive, was not significant (P = 0.12). This suggests that the underlying thyroid disease, surveillance, or reporting on death certificates may have been responsible for the finding. In an attempt to combine the findings from studies of diagnostic and therapeutic exposure to I-​131, Hieu et al. conducted a meta-​analysis that included all of the European cohorts cited earlier and two small US cohorts that contributed data to the US Thyrotoxicosis Therapy Follow-​up Study, but not the entire Thyrotoxicosis Therapy Study (Hieu et al., 2012). The combined RR for thyroid cancer was significantly elevated (RR  =  1.99; 95% CI:  1.22, 3.26), but if the English study was excluded, the RR became attenuated and was no longer significant (RR = 1.58; 95% CI: 0.91, 2.75). In general, studies of adult I-​131 exposure for diagnostic and therapeutic purposes are reassuring, but a small effect on thyroid cancer incidence and mortality cannot be excluded, and continued follow-​up of medically exposed cohorts is warranted.

Environmental Exposure to I-​131.  The most informative data

concerning the risk of thyroid cancer after environmental exposure to I-​131 derive from populations exposed as the result of the accident at the Chernobyl nuclear power plant on April 26, 1986 (UNSCEAR, 2008). This is in part due to the scale of the accident, amount of I-​131 released, exposure of populations to a wide range of doses including a substantial proportion exposed to high doses, and availability of direct radioactivity measurements taken in subjects soon after the accident, all facilitating detection of radiation-​related excess. While the post-​ Chernobyl studies of thyroid cancer are not limited to epidemiological studies and include clinical, morphological, and molecular studies of radiation-​related thyroid cancer, the focus of the current review is studies with quantitative estimates of radiation risk. The first reports of increased number of pediatric thyroid cancers in Belarus and Ukraine appeared in the early 1990s. Because of the unusually short latency and the intense thyroid screening performed in the area, it was not immediately clear whether these cancers were attributable to exposure to I-​131 from the accident. According to the UNSCEAR 2008 report, over 6800 cases of thyroid cancer among those who were < 18 years at the time of the accident were diagnosed between 1991 and 2005 for the whole of Belarus, Ukraine, and the four most affected regions of the Russian Federation; at least 90% of these were PTCs. Today there is little doubt that a major contributor to the excess in incidence of thyroid cancer was exposure to I-​131 released during the accident. This conclusion is supported by multiple analytic and descriptive epidemiological studies. Prospective cohort studies in Ukraine, Belarus, and the Russian Federation (Brenner et  al., 2011; Ivanov et  al., 2012; Stezhko et  al., 2004; Tronko et  al., 2006; Zablotska et al., 2011) and several case-​control studies in Belarus and the Russian Federation with detailed reconstruction of individual thyroid doses (Astakhova et al., 1998; Cardis et al., 2005; Davis et al., 2004b) found a marked, significant increase in risk of thyroid cancer following childhood exposure to I-​131. The reported RR estimates range between 2.9 and 18.9 per Gy. The results of two cohort studies (Brenner et al., 2011; Tronko et al., 2006; Zablotska et al., 2011) are particularly informative; these studies are subject to less bias because I-​131 doses in these studies were estimated based on direct thyroid radioactivity measurements, and standardized screening of all cohort members was conducted irrespective of dose (Stezhko et  al., 2004). With mean I-​131 thyroid doses of 0.65 Gy in Ukraine and 0.56 Gy in Belarus, the ERR estimates for prevalent thyroid cancer diagnosed during a baseline screening examination 12–​14 years after the accident were 5.25 and 2.15 per Gy, respectively (Tronko et al., 2006; Zablotska et al., 2011), while the ERR for incident thyroid cancer in Ukraine up to 20 years after the accident was 1.91 per Gy (Brenner et al., 2011).

84

848

Part IV:  Cancers by Tissue of Origin

The I-​131 risk estimates from these studies, although somewhat lower, appear compatible to those reported for external radiation (Ron et al., 1995; Veiga et al., 2016). The data concerning effect modifiers of I-​131 risk for thyroid cancer relative to external radiation are more limited and inconsistent. It has been suggested that endemic iodine deficiency in areas affected by the Chernobyl accident could be a modifier of I-​131 risk. Two studies found that the ERR per Gy was two-​to three-​fold higher in areas of severe iodine deficiency (Cardis et al., 2005; Shakhtarin et al., 2003); however, the risk did not significantly vary by prevalence of diffuse goiter, a marker of past iodine deficiency in the Ukrainian cohort (Brenner et al., 2011; Tronko et al., 2006). Some, but not all, studies of thyroid cancer risk in populations exposed to I-​131 from the Chernobyl accident found an age-​at-​exposure effect with higher I-​131-​associated thyroid cancer risks in those exposed at younger compared to older ages during childhood. However, a carefully conducted case-​control study of thyroid cancer among adult cleanup workers from Belarus, the Russian Federation, and Baltic countries found a strong radiation risk with adult exposure (Kesminiene et  al., 2012). Most subjects in the latter study received low doses from external, internal, or combined radiation exposure (median total thyroid dose = 69 mGy). The ERR per 100 mGy of total thyroid dose was 0.38 (95% CI: 0.10, 1.09), that is, RR of 3.8 per Gy. The risk estimates were similar when doses from I-​131 and external radiation were considered separately. These findings, together with several descriptive reports of increased thyroid cancer incidence among cleanup workers relative to the general population (Ostroumova et al., 2014; Rahu et al., 2006), underscore the need for independent replication and better understanding of the potential modifiers of radiation risks with adult exposure (e.g., iodine deficiency, exposure to chemicals, etc.). Other studies of populations exposed to internal (or combined with external) radiation and thyroid cancer risk have been conducted. Individuals living on certain atolls of the Marshall Islands were exposed to fallout from a nuclear test explosion in 1954 and some subsequently developed thyroid tumors, including cancers (Takahashi et al., 2003). However, their radiation exposure came from a combination of I-​131, other more rapidly decaying isotopes of iodine, and external gamma radiation (Bouville et al., 2010; Simon et al., 2010). An epidemiologic study of children exposed to nuclear fallout from weapons testing at the Nevada Test Site found a significant association between all thyroid nodules and dose, but there were too few malignant cases to estimate the risk for thyroid cancer separately (Lyon et al., 2006). In the Hanford Thyroid Disease Study, 3441 people who were born in the vicinity of the Hanford nuclear plant (in the State of Washington) between 1940 and 1946 were thoroughly examined, including by ultrasound imaging in 1992–1997 (Davis et al., 2004a). Among participants who lived near the Hanford site during the years of atmospheric emissions, the mean and median thyroid doses were 0.19 and 0.10 Gy, respectively. No evidence of an increased risk for benign or malignant thyroid neoplasms associated with childhood exposure to the atmospheric releases of RAI from the Hanford nuclear site was found. The apparent differences between findings of studies of thyroid cancer conducted in populations exposed due to the Chernobyl accident and studies of other environmentally exposed populations may include different levels and types of exposure (e.g., contribution of short-​lived isotopes of iodine, protraction of exposure over different periods of time), uncertainty in dose estimates, limited sample size, different intake of dietary iodine, and genetic background. The concern about thyroid cancer risk following environmental exposure to I-​131 has been greatly heightened following the Fukushima nuclear reactor accident, initiated by a strong earthquake on March 11, 2011, and the subsequent tsunami. Although the amount of I-​131 released into the environment from the Fukushima accident was about one-​tenth of that from the Chernobyl accident, because of timely evacuation, shielding, and food control, exposure to the public was substantially reduced. Even so, to determine if there is an increased risk of thyroid cancer and to address public concern, the local Fukushima government launched in October 2011 the Thyroid Ultrasound Examination Survey (Yamashita and Radiation Medical Science Center for the Fukushima Health Management, 2016). This survey is directed at 360,000 residents who were between 0 and 18 years on March 11, 2011. By March 2014, about 300,000 individuals were screened for the first time. Of

113 individuals diagnosed with malignancy or suspected malignancy by FNAB, 99 underwent surgery and PTC was confirmed in 95 cases (Suzuki, 2016a; Tsuda et al., 2015). These data have been a source of controversy. Tsuda et al., using baseline survey data (October 2011–​March 2014) supplemented with partially completed second survey data (April 2014–​December 2014), concluded that the observed number of thyroid cancer cases is, in comparison to population data, in excess of expected and unlikely to be explained by screening, suggesting a radiation effect (Tsuda et al., 2015). A series of letters disputed this interpretation, largely based on shortcomings in the analysis (Jorgensen, 2016; Suzuki, 2016b; Wakeford et al., 2016). Also, using the same data as Tsuda et al. (2015), Suzuki has argued that cancers identified in the baseline survey are unlikely to be due to radiation exposure because of extremely low doses, lack of difference in the prevalence of malignant cases according to geographic areas with different levels of radiation exposure, short latency (≤ 4 years), the fact that the cases cluster among the oldest children at the time of the accident, do not include solid variant PTCs, and have low frequency of RET/​PTC rearrangements, the type of genetic alteration previously linked to I-​131 exposure from the Chernobyl accident (Suzuki, 2016a). In the future, data from subsequent ultrasound examination surveys, coupled with improved dose estimates, should allow re-​evaluation of the potential contribution of radiation exposure and/​or other factors to the observed high rates of thyroid cancer.

Comparison of External and Internal Radiation

Recent human studies, collectively, suggest that differences in the carcinogenic effects of external and internal I-​131 radiation may be less than originally thought (Doniach, 1957; Lindsay et  al., 1957). The strongest reason to believe that the effects might be comparable comes from studies of thyroid cancer in individuals exposed to I-​131 from the Chernobyl accident during childhood. However, this does not completely agree with the findings in patients exposed to I-​131 for medical reasons and people exposed to I-​131 around the Hanford nuclear facility. To make a proper comparison across different studies and populations, the radiation risk estimates need to be standardized by age at exposure and attained age (or time since exposure), as these are known to have a strong modifying effect for external radiation. Because most patients included in studies of the medical uses of I-​131 have been adults, the results cannot be directly extrapolated to children. In addition, internal I-​131 exposure typically results in a range of doses to different areas of the thyroid due to functional heterogeneity of thyroid follicles, whereas external radiation exposure tends to be more homogeneous, making it difficult to compare the two types of radiation. Also, small studies or studies of populations exposed to low doses of radiation might be unable to detect radiation-​related increase of thyroid cancer risk due to low statistical power. In summary, the current evidence suggests that thyroid cancer risk with external and internal radiation exposure might be comparable; however, this conclusion remains to be evaluated further, for example, in a pooled analysis of populations exposed to external and/​or internal I-​131 radiation, taking into account differences in age at exposure, attained age, iodine intake, and other relevant characteristics.

Non-​Ionizing Radiation Extremely Low-​Frequency Magnetic Fields

Few observational studies have examined the association between non-​ionizing radiation exposure and risk of thyroid cancer, and, to date, no convincing associations have been observed. A  Swedish study that followed nearly 3 million employed men and women from 1971 through 1989 found no relationship between occupational exposure to extremely low-​frequency magnetic fields and thyroid cancer risk, though elevated risks were observed for women occupationally exposed to high levels of ionizing radiation (Lope et al., 2006).

Ultraviolet Radiation

A large prospective study of men and women older than 50 years in the United States, the NIH-​AARP Diet and Health Study, showed a nonlinear protective association between ambient ultraviolet radiation

 849

Thyroid Cancer exposure and thyroid cancer risk (Lin et al., 2012). However, the study did not account for individual susceptibility to ultraviolet radiation exposure (e.g., skin, eye, or hair color) or personal behaviors (such as use of sunscreen or protective clothing), and the possible underlying biological mechanisms are not well understood.

Environmental Exposures Other Than Radiation Thyroid cancer incidence tends to be higher in areas of volcanic activity, including Hawaii, Iceland, Melanesia, and French Polynesia. In Sicily, thyroid cancer incidence was higher in the province of Catania, which includes the volcanic area of Mount Etna, compared with adjacent non-​volcanic areas (Pellegriti et  al., 2009). Furthermore, levels of boron, iron, manganese, and vanadium were higher in the drinking water of Catania compared with the rest of Sicily (Pellegriti et al., 2009); whether these elements have a direct role in thyroid cancer development is not known.

Chemical (Man-​Made) Exposures A wide range of chemicals have been shown to alter thyroid hormone synthesis in animal and other experimental studies, but the applicability of these findings to human populations remains unclear, and epidemiologic data are scarce. Epidemiologic studies of occupational exposure to these chemicals offer opportunities to study populations at relatively high levels of a particular exposure. A  systematic review of epidemiologic studies of thyroid cancer risk in relation to specific occupations did not find any consistent evidence of a dose–​response relationship for pesticide exposures, although many of the studies were based on small numbers of cases (Aschebrook-​Kilfoy et al., 2014). Among female spouses of men employed as pesticide applicators in the US Agricultural Health Study cohort, use of organophosphate insecticides, specifically malathion, was associated with an increased risk of thyroid cancer as well as breast and ovarian cancer, which is consistent with the hypothesis that organophosphates have endocrine-​disrupting properties in addition to other more established carcinogenic properties (Lerro et al., 2015); however, this finding was based only on 22 thyroid cancer cases. The same systematic review did not reveal any other consistent findings with regard to specific exposures in the workplace, apart from ionizing radiation (Aschebrook-​Kilfoy et al., 2014). However, in a study of female textile workers in Shanghai, occupational exposure for 10 or more years to benzene (HR  =  6.43; 95% CI:  1.08, 38), organic or inorganic gases (HR = 7.65; 95% CI: 1.14, 51), and formaldehyde (HR = 8.33; 95% CI: 1.16, 60) were strongly associated with an increased risk of thyroid cancer, though each of these findings was based on only two exposed cases (Wong et  al., 2006). Administrative workers had a modest increased risk (HR = 1.56; 95% CI: 1.08, 2.25). No other textile industry exposures were clearly associated with risk (Wong et al., 2006). Polybrominated diphenyl ethers (PBDEs), widely used as flame retardants in a range of commercial and household products in the United States since the 1970s, have been hypothesized to influence thyroid cancer risk due to their effects on thyroid hormone homeostasis (Aschebrook-​Kilfoy et al., 2015). However, no association was found for prediagnostic serum levels of PBDEs and thyroid cancer risk in a case-​control study of 104 cases and 208 matched controls nested in the large prospective US cohort study of men and women (Aschebrook-​Kilfoy et al., 2015). Overall, there is a need for additional studies of specific chemical exposures and risk of thyroid cancer. The few studies to date have generally reported small numbers of incident thyroid cancers and/​or have lacked detailed exposure measurements.

Lifestyle Factors Obesity/​Weight Gain and Metabolic Syndrome

A number of epidemiologic studies have focused on a possible relationship between obesity and thyroid cancer risk. The increasing attention on this topic is largely due to the dramatic increase in both obesity

849

prevalence in the United States and several other countries worldwide over the past four decades (Flegal et  al., 2010)  and the established relationship of obesity with a number of other cancer sites (Renehan et al., 2008). A systematic review and meta-​analysis found that body mass index (BMI), weight, waist and hip circumference, and waist-​to-​ hip ratio were each positively associated with thyroid cancer risk in a dose–​response manner (Schmid et al., 2015). The meta-​analysis also showed positive associations for PTC, FTC, and ATC risks, but inverse associations for MTC risk. Largely consistent with these findings were results from an international pooled analysis of 22 prospective studies that found a positive association between BMI and risk of total thyroid cancer among men (HR [per 5 kg/​m2] = 1.17: 95% CI: 1.06, 1.28) and a slightly weaker positive association among women (HR [per 5 kg/​ m2]  =  1.04; 95% CI:  1.00, 1.09) (Kitahara et  al., 2016). Compared to men with BMI in the normal-​weight range (18.5–​24.9  kg/​m2), obese men had a 35% increased risk of thyroid cancer (HR  =  1.35; 95% CI:  1.07, 1.7), and overweight men had a 23% increased risk (HR  =  1.23; 95% CI:  1.02, 1.47). Similar to the meta-​analysis, the pooled analysis showed positive associations for PTC, FTC, and ATC risks and inverse associations for MTC risk; however, there were some differences in these patterns by sex. For instance, BMI was significantly positively associated with risk of PTC only in men (HR [per 5 kg/​m2] = 1.15; 95% CI: 1.02, 1.29) and risk of ATC only in women (HR [per 5 kg/​m2] = 1.66; 95% CI: 1.23, 2.23). Greater young-​adult BMI and adulthood weight gain (i.e., net weight gain between young adulthood and study baseline) were both positively associated with risk of PTC in men, and weight gain was positively associated with ATC risk in women. Greater waist circumference was also associated with a modest increase in total thyroid cancer risk (HR [per 5 cm] = 1.03; 95% CI: 1.01, 1.05) with no clear differences by histologic type of thyroid cancer. In a subset of the population with death certificate information, higher BMI values in young adulthood and at baseline and larger baseline waist circumferences were clearly associated with an increased risk of thyroid cancer mortality; for the same per-​unit increase, HRs were stronger in magnitude for thyroid cancer mortality compared with thyroid cancer incidence. A 5-​kg/​m2 increase in BMI, for instance, was associated with a 29% increased risk of thyroid cancer mortality (HR  =  1.29; 95% CI:  1.07, 1.55). Overall, findings from the pooled analysis provide some evidence that greater adiposity is associated with an increased risk of more aggressive forms of thyroid cancer. Findings from some (Choi et al., 2015; Kim et al., 2013; Liu et al., 2015; Tresallet et al., 2014), but not all (Kwon et al., 2015), cross-​sectional studies have also suggested that PTC patients with higher BMI tend to have more aggressive clinical and pathologic tumor features than patients with lower BMI. A registry-​linkage study of more than 320,000 schoolchildren in Copenhagen, Denmark, with recorded measurements of height and weight during ages 7–​13 showed that greater values of BMI at each age of measurement were positively associated with risk of adult thyroid cancer (Kitahara et  al., 2014). Associations of childhood BMI with adult thyroid cancer were stronger, but not significantly different, for men versus women. Using BMI at age 10 as an example, a one-​standard deviation increase in BMI was associated with a 14% increased risk of thyroid cancer in women (HR = 1.14; 95% CI: 0.96, 1.35) and 21% increased risk of thyroid cancer in men (HR = 1.21; 95% CI:  0.91, 1.60). Childhood BMI was more strongly associated with risk of adult PTC compared with FTC. For the same per-​unit increase, adolescent BMI was more strongly associated with adult thyroid cancer risk than adult BMI (the latter based on published findings from other studies). Furthermore, adolescent BMI was more strongly associated with thyroid cancers diagnosed at younger versus older adult ages. Based on these findings, the authors of this study hypothesized that factors related to body size in early life may be similarly, if not more, important for thyroid cancer development than factors related to adult body size. Some studies have investigated whether other components of the metabolic syndrome, including measures of insulin resistance, type 2 diabetes, triglycerides, and blood pressure, are associated with thyroid cancer risk independent of BMI or other measures of adiposity. In contrast with obesity, the available evidence to date does not suggest a clear association between type 2 diabetes and risk of thyroid

850

850

Part IV:  Cancers by Tissue of Origin

cancer. No association with type 2 diabetes was found in an Australian national diabetes registry study, despite an observed positive association for type 1 diabetes (Harding et al., 2015). Similarly, no association was observed for a self-​reported history of diabetes and risk of thyroid cancer in a pooled analysis of prospective US cohorts (Kitahara et al., 2012b) or in a study of male veterans admitted to VA hospitals after allowing for increased surveillance within 5 years of documented diagnosis of diabetes (Balasubramaniam et  al., 2012). Even though these latter studies could not discriminate between type 1 and type 2 diabetes, most of the patients were likely to have type 2 diabetes as this is more common, particularly at older ages. In a case-​control study using data from the UK Clinical Practice Research Datalink, neither a diabetes diagnosis nor greater duration of diabetes was associated with risk of thyroid cancer (Becker et al., 2015). The UK study showed that thyroid cancer cases were more likely to be long-​term users of metformin versus controls, while no associations were observed for use of other antidiabetic drugs (Becker et al., 2015). In a study based on electronically recorded insurance data in Taiwan, no associations with thyroid cancer risk were observed in relation to a diagnosis of diabetes, longer duration of diabetes, or use of most types of antidiabetic drugs, including metformin (OR = 0.70; 95% CI: 0.42, 1.16) (Tseng, 2012). This study did, however, find an increased risk of thyroid cancer associated with use of sulfonylureas (OR = 1.88; 95% CI: 1.20, 2.95). Using a larger sample of Taiwanese patients with type 2 diabetes, longer duration and greater cumulative dose of metformin were associated with a reduced risk of thyroid cancer in dose-​dependent manner (Tseng, 2014). However, the latter studies, which relied on electronic medical record data, did not have information on BMI, smoking, alcohol drinking, or other factors to control for confounding. A large pooled study of seven cohorts from Norway, Austria, and Sweden (Me-​Can) found that serum glucose levels were positively associated with thyroid cancer risk in men and inversely associated with risk in women after adjusting for age, BMI, and smoking status. No associations were observed for serum cholesterol levels, triglyceride levels, blood pressure, or a combined metabolic syndrome score based on a combination of these factors, glucose levels, and BMI (Almquist et al., 2011). In contrast, among a wide range of metabolic factors examined, a cohort study of older persons in Iceland found an increased risk of thyroid cancer in men who had higher serum triglycerides and an increased risk of thyroid cancer in women who had higher diastolic blood pressure or higher creatinine levels, after adjusting for other metabolic syndrome components (Tulinius et al., 1997). Despite the relatively consistent findings from epidemiologic studies on the association between excess adiposity and thyroid cancer, a causal relationship has not yet been established. In a review article, Arnold et al. (2015) estimated that, worldwide, 19% of thyroid cancers (n = 11,615) in men and 9% in women (n = 18,108) were attributable to a high BMI (≥ 25 kg/​m2). However, some of the attributable risk due to obesity may reflect a greater likelihood of thyroid cancer detection and diagnosis in obese versus normal-​weight individuals. The proportion of thyroid cancer risks attributable to a causal influence of obesity on thyroid cancer risk remains to be determined, but may be aided by studies that distinguish between screen-​detected thyroid cancers and those that are detected due to symptoms or palpation, as these are more likely to be larger or to otherwise have more aggressive features (and, thus, more likely to be attributable to genetic or environmental determinants). Although the few studies of specific obesity-​related biomarkers and thyroid cancer risk have not been entirely consistent, the mechanism underlying the observed obesity and thyroid cancer association could be complex, involving an interplay between the thyroid hormone and insulin-​like growth factor (IGF)-​I axis, insulin resistance, and local and systemic inflammation (Pazaitou-​Panayiotou et al., 2013). Studies that evaluate thyroid cancer risk following intentional weight loss could provide additional etiologic insight, but such studies are currently lacking.

Additional Anthropometric Factors

Taller attained height has been positively associated with thyroid cancer risk in most case-​control and prospective studies (Jing et al., 2015). An international pooled analysis of 22 prospective studies found an

overall positive association between height and thyroid cancer in both women (HR per 5 cm = 1.08; 95% CI: 1.04, 1.12) and men (HR per 5  cm  =  1.05, 95% CI:  1.00, 1.11) (Kitahara et  al., 2016). Height in women, but not in men, was positively associated with thyroid cancer mortality (Kitahara et al., 2016). In the Danish record-​linkage study of schoolchildren with recorded measurements of height and weight measured during ages 7–​ 13, greater values of height at each age of measurement were positively associated with risk of adult thyroid cancer (Kitahara et  al., 2014). Using height at age 10 as an example, a one standard deviation increase in height was associated with a 19% increased risk of thyroid cancer in women (HR = 1.19; 95% CI: 1.02, 1.39) and 33% increased risk in men (HR = 1.33; 95% CI: 1.03, 1.71). In general, associations were stronger, but not significantly different, for men versus women. In women, but not men, the association for height became attenuated slightly with increasing age at measurement. For the same per-​unit increase, these associations were stronger than has been observed in other studies of adult height and thyroid cancer risk. Furthermore, in this study, adolescent height was more strongly associated with thyroid cancer diagnosed at younger versus older ages. These findings are in agreement with data from EPIC, which was the first study to measure leg length and evaluate its association with thyroid cancer risk (Rinaldi et al., 2012). This study showed that longer leg length (standing height minus sitting height) was associated with an increased risk of PTC in women and all differentiated thyroid cancer in men (Rinaldi et  al., 2012). Compared to trunk length (and, thus, total attained height), leg length is more strongly influenced by growth and thyroid hormone levels, as well as nutritional status, in childhood. Therefore, the recent findings for adolescent height and leg length suggest that factors related to growth in early life, particularly before puberty, may be particularly important clues in understanding the etiology of thyroid cancer. In addition to childhood and adolescent growth, growth factor pathways involving IGF-​I levels at young ages may also mediate associations observed between larger birth size and thyroid cancer risk (Crump et  al., 2015; Kitahara et  al., 2014). Growth hormone (GH) and IGF-​I levels have also been hypothesized to play a role in thyroid carcinogenesis, as some studies of patients with acromegaly, a condition characterized by excess production of GH by the pituitary, have indicated a higher than expected prevalence of thyroid nodules and thyroid cancer (Wolinski et al., 2014), though these findings are controversial (Dabrowska et al., 2014). In a nested case-​control study using data from the large EPIC cohort, investigators found a positive association between IGF-​I concentrations and risk of differentiated thyroid cancer risk (Schmidt et al., 2014). The OR for a doubling in IGF-​I concentration was 1.48 (95% CI:  1.06, 2.08, P  =  0.02). This association was stable over follow-​up time. However, as IGF-​I levels were generally measured in older adults, it remains unclear whether IGF-​I levels earlier in life have a similar or stronger impact on thyroid cancer development.

Diet

Recent case-​control and cohort studies on diet and thyroid cancer have focused mainly on dietary iodine, which has long been considered to play a role in thyroid cancer development (Ron and Schneider, 2006). Newer directions include evaluating the role of total energy and macronutrient intakes, selenium and other micronutrient intakes, dietary patterns in middle to older adulthood, and dietary intakes of foods and nutrients in adolescence. When iodine supplementation occurs in iodine-​deficient regions, overall thyroid cancer incidence generally increases to a similar extent as observed in iodine stable regions; however, the proportion of PTCs often increases, while the proportion of FTCs and ATCs decreases (Burgess et al., 2000; Harach et al., 2002; Lind et al., 2002). In addition, animal studies have demonstrated tumor-​promoting effects of iodine deficiency, and weaker promotional effects of iodine excess (Zimmermann and Galetti, 2015). Whether findings from animal experiments can be generalized to humans is not certain, however. Furthermore, the few case-​control studies on iodine intake and thyroid cancer risk have provided mixed results. For instance, a reduction in

 851

Thyroid Cancer risk of thyroid cancer with higher estimated intakes of iodine based on intakes of fish and seafood was observed in a case-​control study in French Polynesia (Clero et al., 2012). In contrast, a case-​control study in California found a reduction in risk with greater consumption of iodine from multivitamins but no association with increasing intake of iodine from iodine-​ containing foods (Horn-​ Ross et  al., 2001). Two other case-​control studies, one in Hawaii and another in New Caledonia, found no association between estimated dietary iodine and thyroid cancer risk (Kolonel et al., 1990; Truong et al., 2010). In addition to the usual biases specific to case-​control studies, including differential recall and selection biases, both case-​control and prospective studies on this topic are limited due to the inability to accurately estimate total iodine intake based on self-​reported intakes of individual foods. Estimating iodine consumption from iodized salt is particularly challenging (Zimmermann and Galetti, 2015). One study found no association between concentration of iodine measured in toenails and risk of PTC (Horn-​Ross et al., 2001), but toenails have limited utility as a biomarker of iodine intake. Other biomarkers of iodine, such as urine, reflect only short-​term intakes. Cruciferous vegetables contain thioglucosides that can be degraded to form goitrogens, which are naturally occurring substances that disrupt thyroid function and have been shown in animal studies to influence the development of goiter and thyroid cancer (Dal Maso et  al., 2009). Cruciferous vegetables also contain other constituents with anti-​carcinogenetic properties. However, few observational studies, to date, have shown an association between intake of cruciferous vegetables and thyroid cancer risk (Dal Maso et  al., 2009). A  large, international pooled analysis of case-​control studies found that high, compared to low, intake of cruciferous vegetables was associated with a non-​significant reduced risk (RR = 0.9; 95% CI: 0.8, 1.11), while a high versus low intake of non-​cruciferous vegetables was associated with a significant reduced risk (RR = 0.8; 95% CI: 0.7, 1.0) (Bosetti et al., 2001). In a case-​control study in New Caledonia (Truong et al., 2010), which did not find an association between dietary iodine and thyroid cancer risk, a higher intake of cruciferous vegetables was associated with increased thyroid cancer risk only among women with iodine intakes of less than 96 μg/​day (OR for third versus first tertile = 1.86; 95% CI: 1.01, 3.43). Thus, an effect of goitrogens on thyroid cancer development may be more pronounced in iodine-​deficient populations. In addition to iodine, other micronutrients, particularly selenium, have been hypothesized to influence thyroid cancer development due to their essential role in thyroid hormone synthesis (Duntas, 2006). Selenium also has antioxidant properties that have been hypothesized to protect the thyroid gland from oxidative stress, for instance, from autoimmune thyroid disease (Duntas, 2006). However, fingernail selenium was not associated with thyroid cancer risk in a case-​control study in French Polynesia based on post-​diagnostic fingernail samples (Ren et al., 2014). Data from the NIH-​AARP Diet and Health Study, a large prospective US cohort of older men and women (ages 50–​71 at baseline), showed no associations between thyroid cancer and intakes of selenium (highest versus lowest quintile, HR = 1.23; 95% CI: 0.92, 1.65; P trend = 0.26) or most other micronutrients apart from a positive association observed with greater intake of vitamin C (highest versus lowest quintile, HR = 1.34; 95% CI: 1.02, 1.76; P trend < 0.01) (O’Grady et al., 2014). Using data from the same large cohort study, a higher intake of flavan-​3-​ols was associated with a reduced risk of thyroid cancer (highest versus lowest quintile, HR = 0.70; 95% CI: 0.55, 0.91; P trend  =  0.03), a higher intake of flavanones was associated with an increased risk (highest versus lowest quintile, HR = 1.50; 95% CI: 1.14, 1.96; P trend = 0.004), and no associations were observed for other classes of flavonoids or total flavonoids (Xiao et al., 2014). Overall, few consistent associations have emerged in studies of micronutrients, including selenium, and thyroid cancer risk. Nitrate is a common drinking water contaminant, particularly in agricultural areas where nitrogen fertilizers are used, and a natural component of foods such as leafy green and root vegetables (Ward et al., 2010). Nitrate may alter thyroid function by competitively inhibiting iodine uptake by the thyroid (Tajtakova et al., 2006; Tonacchera et al., 2004). In a prospective study of nearly 22,000 older women in

851

Iowa, increasing mean nitrate concentrations in the public water supplies were associated with greater thyroid cancer risk (> 2.46 versus < 0.36 mg/​L, RR  =  2.18,; 95% CI:  0.83, 5.76) (Ward et  al., 2010). Furthermore, 5 or more years of use of public water supplies with nitrate levels > 5 mg/​L (versus 0 years) was associated with a 2.59-​ fold increased risk (95% CI:  1.09, 6.19). Higher intakes of nitrate from dietary sources were also associated with an increased risk of thyroid cancer (highest vs. lowest quartile of intake, RR = 2.85; 95% CI: 1.00, 8.11; P trend = 0.05) (Ward et al., 2010). In the NIH-​AARP study, higher intakes of nitrates from dietary sources were shown to be positively associated with PTC and FTC risk in men, but no associations were observed in women (Kilfoy et al., 2011). In addition, higher intakes of dietary nitrite (which are found mainly in cured meats and which can react with dietary amines to form carcinogenic nitrosamines) were only associated with FTC risk in men (Kilfoy et al., 2011). In a prospective study of more than 73,000 women in Shanghai, ages 40–​70 at baseline, a positive association was observed for greater dietary nitrite and thyroid cancer risk, driven largely by consumption of nitrite from processed meat (Aschebrook-​ Kilfoy et  al., 2013b); however, no association was observed between dietary nitrate and thyroid cancer risk. Thus, the data from prospective studies examining intakes of nitrates and nitrites suggest that they may influence thyroid cancer risk, but additional studies are needed to confirm this. Data from observational studies on macronutrients and thyroid cancer risk are scarce. In one large prospective cohort study (EPIC), greater total energy intake and intake of polyunsaturated fats was associated with increased risk of differentiated thyroid cancer overall (Zamora-​Ros et  al., 2015b). Greater starch intake and higher glycemic index were associated with increased risk among participants with BMI values of 25 or greater, and greater sugar intake was associated with increased risk among those with BMI values < 25. These findings provide some support for a role of insulin resistance in thyroid cancer development. Neither tea nor coffee intake has been consistently linked with thyroid cancer risk (Hashibe et al., 2015; Mack et al., 2003). To date, most of the epidemiologic studies of diet and thyroid cancer risk rely on dietary intakes captured in middle to older adulthood. Few studies have evaluated the role of diet in adolescence or young adulthood. In the NIH-​AARP Diet and Health Study, participants were asked to recall dietary intakes of certain foods in adolescence and 10  years ago (Braganza et  al., 2015). Adolescent intakes of certain foods, particularly greater chicken/​turkey and sweet baked goods, and lower butter/​margarine were associated with increased risk of thyroid cancer. In men, but not women, greater intake of canned tuna in adolescence and intake of broccoli in mid-​life were associated with increased risk. While additional studies should be conducted to confirm or refute these findings, studies having reasonably accurate data on diet in childhood or adolescence and sufficiently large sample sizes to evaluate risk factors for thyroid cancer are, unfortunately, quite limited.

Alcohol

While no association was observed between alcohol intake and thyroid cancer risk in a large pooled analysis of case-​control studies after adjusting for smoking (Mack et al., 2003), data from several, but not all (Kabat et al., 2012c), prospective studies, including a pooled analysis of US-​based cohorts (Kitahara et al., 2012a), the EPIC cohort (Sen et  al., 2015), and the Million Women Study in the United Kingdom (Allen et  al., 2009), have provided reasonably consistent evidence supporting an inverse, dose–​ response association between alcohol intake and thyroid cancer risk that persists after adjusting for cigarette smoking and other potential confounders. In the pooled analysis of prospective studies, for example, the inverse association for alcohol intake was strongest among never smokers (HR = 0.81; 95% CI: 0.67, 0.97; HRs for former and current smokers were 0.89, 95% CI: 0.82=, 0.98 and 0.90, 95% CI:  0.76, 1.06, respectively). In EPIC, participants who consumed ≥ 15 versus 0.1–​4.9 grams of alcohol per day had a 23% reduced risk of differentiated thyroid cancer (HR = 0.77; 95% CI: 0.60, 0.98) (Sen et al., 2015); similar findings were observed

852

852

Part IV:  Cancers by Tissue of Origin

for different types of alcohol (wine, beer, spirits) and when evaluating lifetime alcohol consumption rather than alcohol consumption at baseline. The inverse association was also strongest in never smokers, though no significant effect modification by smoking was observed. Potential underlying biological mechanisms remain speculative; however, alcohol consumption has been inversely associated with thyroid enlargement, nodularity, and fT3 and positively associated with serum TSH independent of age, sex, and cigarette smoking, and thus has been hypothesized to have a direct influence on thyroid function (Knudsen et al., 2001).

Physical Activity

Although data are limited, physical activity level has not been consistently associated with thyroid cancer risk. A pooled analysis of prospective US-​based studies found, unexpectedly, an elevated risk for the highest versus lowest level of physical activity after adjusting for BMI and other factors (HR = 1.18; 95% CI: 1.00, 1.39), though this result may reflect heightened medical surveillance in the more highly active group (Kitahara et  al., 2012b). A  separate large prospective study, the Women’s Health Initiative, showed no clear association between higher levels of physical activity, defined by greater metabolic equivalent (MET) hours of physical activity per week, and risk of PTC or FTC (Kabat et al., 2012a).

Smoking

In 2003, data from the international pooled analysis of case-​control studies were published, showing a reduced risk of differentiated thyroid cancer in current smokers (OR = 0.6; 95% CI: 0.6, 0.7), but in not former smokers (OR  =  0.9; 95% CI:  0.8, 1.1), and a dose–​response association with greater duration and frequency of smoking (Mack et al., 2003). These findings have since been confirmed in prospective studies (Kabat et al., 2012c) and a pooled analysis of five prospective US-​based studies (Kitahara et  al., 2012a). In the pooled analysis of prospective studies, current versus never smoking was associated with a 32% decline in risk (HR = 0.68; 95% CI: 0.55, 0.85). Alcohol intake modified the association of current smoking and thyroid cancer risk, whereby the largest decline in risk was observed among non-​drinkers (HR = 0.46; 95% CI: 0.29, 0.74; P interaction = 0.004). Among both former and current smokers, thyroid cancer risk declined with greater smoking intensity, duration, and number of pack-​years (Kitahara et al., 2012a). The decline in cigarette smoking prevalence in the United States and several other countries over recent decades follows a similar pattern with the rising incidence of thyroid cancer during the same time period (Kitahara et al., 2012a). This inverse association may be explained by lower levels of TSH, T3, and T4, reduced prevalence of serum thyroid autoantibodies, increased risk of Graves’ hyperthyroidism, and reduced estrogen in smokers compared with non-​smokers (Wiersinga, 2013). Whether thyroid cancer incidence trends are directly attributable to, or caused by, the declining prevalence of cigarette smokers in the United States and elsewhere around the world remains unclear, however.

Prescription and Over-​the-​Counter Medications

A Taiwanese case-​control study using electronic prescription data found a positive association between statin use and thyroid cancer risk (OR  =  1.39; 95% CI:  1.08, 1.78) (Hung et  al., 2015). The higher OR for statin users versus non-​users was significant in women (OR = 1.43; 95% CI: 1.07, 1.90), but not in men (OR = 1.28; 95% CI:  0.75, 2.17). When statin users were divided according to regular or irregular use, thyroid cancer risk was significantly elevated only in the regular users (OR = 1.40; 95% CI: 1.05, 1.86). The OR for irregular users was 1.35 (95% CI: 0.88, 2.07). Although a prior history of hyperlipidemia was associated with an increased risk of thyroid cancer overall (OR = 1.30; 95% CI: 1.03, 1.65), no association was observed between prior hyperlipidemia and thyroid cancer risk after excluding statin users; however, it was not possible to completely isolate the associations for hyperlipidemia and statin use, as most statin users had a history of hyperlipidemia, and the severity of hyperlipidemia may have been greater in the statin users. Also, these associations may have been confounded by other factors associated

with hyperlipidemia and/​or statin use, including BMI, which were not available in the study. As described earlier, electronic medical record linkage studies have shown inconsistent associations between metformin use and thyroid cancer risk (Becker et  al., 2015; Tseng, 2012; 2014). One of these studies, conducted in Taiwan, found an increased risk of thyroid cancer in users of sulfonylureas (OR  =  1.88; 95% CI:  1.20, 2.95), and reduced risks in users of aspirin (OR = 0.77; 95% CI: 0.63, 0.94) and of NSAIDs (OR = 0.10; 95% CI: 0.08, 0.13) (Tseng, 2012). In a pooled analysis of three prospective studies, no association was observed for self-​reported use of aspirin or non-​aspirin NSAIDs 1  year prior to baseline and subsequent risk of thyroid cancer after controlling for study, sex, race/​ethnicity, weight, smoking status, and alcohol intake (Patel et al., 2015). HRs for regular use of aspirin and non-​aspirin NSAIDs were 1.06 (95% CI: 0.82, 1.39) and 1.14 (95% CI: 0.84, 1.55), respectively.

Reproductive and Hormonal Factors Considering the higher incidence of thyroid cancer in women compared with men, particularly during the reproductive ages, sex steroid hormones have been hypothesized to play a role in thyroid cancer etiology. Nonetheless, “classical” reproductive and hormonal factors, such as age at menarche and menopause, parity, and use of oral contraceptives and menopausal hormone replacement therapy, which have been associated with other female-​ specific or female-​ predominant cancers (e.g., breast, ovarian, endometrial), have not been consistently associated with risk of thyroid cancer in women. The earliest studies were largely case-​control in design and yielded largely inconsistent results for most reproductive and hormonal factors, raising the concern that some of the results were influenced by recall and selection biases. Several case-​control and prospective studies have been published in recent years, with some of these studies providing new information about the role of sex steroid hormone levels on the development of thyroid cancer. A meta-​analysis of 10 prospective, 12 case-​control, and one pooled analysis of case-​control studies found a 9% increased risk of thyroid cancer in parous versus nulliparous women (RR = 1.09; 95% CI: 1.03, 1.15) (Zhu et al., 2015). Compared with nulliparous women, the risk of thyroid cancer was significantly increased for women with two (RR = 1.11; 95% CI: 1.01, 1.22) and three (RR = 1.16; 95% CI: 1.01, 1.33) live births (Zhu et al., 2015). However, when stratified by study design, the positive association between parity and thyroid cancer was observed in the case-​control, but not prospective, studies. Not included in the meta-​analysis, but somewhat consistent with these findings, were results from the large prospective Prostate, Lung, Colorectal, and Ovarian Cancer Screening Study (PLCO), which showed a 72% increased risk of thyroid cancer for women with ≥ 5 live births compared with 1–​2 live births; however, no consistent pattern of increasing risk with increasing number of births was observed (Braganza et al., 2014). Also not included in the meta-​analysis, and not in agreement with the conclusions of the meta-​analysis, were results from the prospective US Radiologic Technologists Study (USRT). They showed a slight, non-​significant decreasing risk with increasing number of live births (≥ 3 births versus nulliparity, HR = 0.87; 95% CI: 0.59, 1.29; P trend  =  0.61) (Meinhold et  al., 2010). A  cancer-​registry linkage study in Finland showed a 2.33-​fold (95% CI: 1.59, 3.29) increased risk of thyroid cancer for women with ≥ 10 births compared to the general population (Hognas et al., 2014). Whether an association of a relatively large number of live births and thyroid cancer risk reflects pregnancy as an etiologically relevant exposure versus heightened medical surveillance during multiple pregnancies or another phenomenon unrelated to etiology is not clear. The timing of thyroid cancer diagnosis in relation to pregnancy may be more relevant than the number of pregnancies, as thyroid cancer risk has been shown to be elevated shortly after pregnancy (Andersson et  al., 2015). In the EPIC cohort, women with a recent pregnancy (≤ 5  years before baseline) had a statistically significant 3.87-​fold increased risk of differentiated thyroid cancer risk during follow-​up

 853

Thyroid Cancer compared with women whose last pregnancy was more than 5 years before baseline (Zamora-​Ros et  al., 2015a); however, this finding was based on small numbers. A similar finding was observed in the California Teacher’s Cohort (last pregnancy ≤ 5 versus > 5 years before baseline: RR = 2.28; 95% CI: 1.16, 4.45) (Horn-​Ross et al., 2011). Among other pregnancy-​related factors that have been examined in relation to thyroid cancer risk in women, no clear associations have emerged for studies of age at first full-​term pregnancy (Braganza et al., 2014; Horn-​Ross et  al., 2011; Kabat et  al., 2012b; Navarro Silvera et al., 2005; Schonfeld et al., 2011b; Wong et al., 2006; Zamora-​Ros et al., 2015a), or studies examining risks after miscarriage, stillbirths, or abortions (Horn-​Ross et al., 2011; Kabat et al., 2012b; Kohler et al., 2015; Wong et al., 2006; Zamora-​Ros et al., 2015a). In a Nordic birth and cancer registry linkage study, hyperemesis gravidarum, an early complication of pregnancy characterized by severe nausea and vomiting and potentially related to altered pregnancy hormone levels and subsequent immune diseases (Jorgensen et al., 2012), was associated with an increased risk of thyroid cancer in women after their pregnancy (RR = 1.45; 95% CI: 1.06, 1.99), and risk increased with more than one hyperemetic pregnancy (RR = 1.80; 95% CI:  1.23, 2.63) (Vandraas et  al., 2015a). Among offspring of mothers diagnosed with hyperemesis gravidarum during pregnancy, thyroid cancer risk was non-​significantly elevated (RR = 2.21; 95% CI: 0.73, 6.71); however, results were based on small numbers of thyroid cancer cases (4 exposed and 339 unexposed) (Vandraas et al., 2015b). A large Swedish study of nearly 3.6  million live births between 1973 and 2008 showed that high fetal growth, defined using birth weight standardized to gestational age and sex, was associated with an increased risk of thyroid cancer in the mothers following their pregnancy (per standard deviation: 1.05; 95% CI: 1.01, 1.09) (Crump et al., 2015). Similarly, risk of thyroid cancer in mothers increased by 13% for each 1000 g increase in offspring birth weight. These results suggest a possible role of elevated umbilical cord levels of IGF-​I on thyroid cancer development, potentially due to the pro-​carcinogenic properties of IGF-​ I, including the inhibition of cellular apoptosis (Pollak et al., 2004). Prolonged breastfeeding delays ovulation, suppresses gonadotropins and production of estradiol, and has been associated with a reduced risk of breast and ovarian cancers (Chowdhury et al., 2015). The few prospective studies on breastfeeding and thyroid cancer have shown non-​significant reduced risks of thyroid cancer with greater duration of breastfeeding (Kabat et al., 2012b; Zamora-​Ros et al., 2015a), though there is limited information to directly evaluate an underlying role for delayed ovulation or suppressed hormone levels. A self-​reported history of infertility has been positively associated with thyroid cancer risk in some (Zamora-​Ros et al., 2015a), but not all (Braganza et al., 2014), prospective studies. It is not entirely clear whether a link between infertility and thyroid cancer reflects an underlying thyroid disorder, which could play a role etiologically or simply could increase the likelihood of thyroid cancer detection. Alternatively, use of fertility drugs could influence thyroid cancer development through a hormone-​driven mechanism. In support of the latter hypothesis, higher doses (> 2250 versus 1–​900 mg, HR = 1.96; 95% CI: 0.92, 4.17) and receipt of more cycles of clomiphene citrate (12 or more versus less than 6, HR = 1.77; 95% CI: 0.51, 6.12) were associated with non-​significant increased risk of thyroid cancer in a cohort of US women treated for fertility issues (Brinton et al., 2015). An earlier Danish cohort of more than 50,000 women referred to hospitals and clinics with fertility problems showed that clomiphene (RR = 2.29; 95% CI: 1.08, 4.82) and progesterone (RR = 10.14; 95% CI: 1.93, 53.34) treatment were associated with increased risk of thyroid cancer, primarily in parous women, whereas no association was observed after use of gonadotrophins, hCG, or GnRH. However, no associations were observed by number of cycles or years since last use, and results were based on small numbers of exposed cases (Hannibal et al., 2008). Younger ages at menarche and older ages at natural menopause, which are considered as indicators of greater lifetime exposure to endogenous estrogen, have been associated with thyroid cancer risk in women in some, but not all, studies. In the PLCO study, both factors

853

were associated with an increased risk of total thyroid cancer and PTC, as were a greater number of years between menarche and menopause (among women who had experienced natural menopause) and greater estimated lifetime number of ovulatory cycles, which additionally accounts for time in which women were taking oral contraceptives or were pregnant (Braganza et al., 2014). There was some indication of a U-​shaped relationship for age at menarche and thyroid cancer risk in the USRT study, with elevated risks observed for ages at menarche younger than 12 and ≥ 16 years (Meinhold et al., 2010). Like the USRT, the California Teacher’s Study included both pre-​and postmenopausal women at baseline; in this study, a menarche age of 14 or older was associated with increased risk of PTC, but only if diagnosed before age 45 (Horn-​Ross et al., 2011). Other large prospective studies have not shown clear associations for ages at menarche or menopause (Kabat et al., 2012b; Navarro Silvera et al., 2005; Schonfeld et  al., 2011b; Wong et  al., 2006; Zamora-​Ros et  al., 2015a). In the California Teacher’s Study, longer menstrual cycle length and irregular cycles were associated with modestly elevated risks of PTC (Horn-​ Ross et al., 2011). Oral contraceptive use (ever versus never) has not been consistently associated with thyroid cancer risk in prospective studies (Braganza et al., 2014; Horn-​Ross et al., 2011; Meinhold et al., 2010; Navarro Silvera et  al., 2005; Zamora-​Ros et  al., 2015a). Current use of oral contraceptives at baseline recruitment in the EPIC cohort was associated with a significant reduced risk of differentiated thyroid cancer (RR  =  0.48; 95% CI:  0.25, 0.92), though this association may have been confounded by menopausal status (Zamora-​Ros et  al., 2015a). Studies of longer versus shorter duration of oral contraceptive use have been inconsistent, with some studies finding inverse associations (Schonfeld et al., 2011b; Zamora-​Ros et al., 2015a) and some finding no association (Horn-​Ross et al., 2011; Kabat et al., 2012b; Meinhold et al., 2010; Navarro Silvera et al., 2005; Wong et al., 2006). Results from prospective studies of menopausal hormone therapy use and thyroid cancer risk have been inconsistent, with studies showing positive (Schonfeld et al., 2011b; Zamora-​Ros et al., 2015a) and null (Horn-​Ross et  al., 2011; Kabat et  al., 2012b; Meinhold et  al., 2010; Navarro Silvera et al., 2005) associations. No clear patterns have emerged when evaluating risks according to duration of use or after separating use (Kabat et al., 2012b) according to type of menopausal hormone therapy (estrogen-​ only versus estrogen-​ plus-​ progestin) (Horn-​Ross et al., 2011; Kabat et al., 2012b). Some large prospective studies have found elevated risks of total thyroid cancer and PTC associated with a history of hysterectomy (Kohler et  al., 2015), bilateral oophorectomy (Kabat et  al., 2012b), and surgical versus natural menopause (Kabat et al., 2012b; Zamora-​ Ros et  al., 2015a). Other prospective studies that evaluated thyroid cancer risk of surgical versus natural menopause, however, showed no associations (Braganza et al., 2014; Meinhold et al., 2010; Schonfeld et al., 2011b). An elevated risk of thyroid cancer has been observed in women with history of benign breast disease (Meinhold et  al., 2010; Schonfeld et  al., 2011b), although not observed consistently (Braganza et  al., 2014), in women with a history of uterine fibroids or ovarian cysts (Braganza et al., 2014), and in women with a history of breast cancer (Nielsen et al., 2016). These associations may reflect shared genetic, immune-​related, or hormonal etiologies for thyroid cancer and other conditions that exclusively or predominantly affect women. A limitation of some of the prospective studies described in the preceding is that enrollment and subsequent follow-​up for thyroid cancer typically began at or around the ages that most women reach menopause (Braganza et  al., 2014; Kabat et  al., 2012b; Schonfeld et  al., 2011b; Zamora-​ Ros et  al., 2015a). Thus, prospective studies that recruit and follow women at much younger ages are needed to help understand whether endogenous and exogenous sex steroid hormone levels influence the develop of thyroid cancer well before the onset of menopause. To summarize, the hypothesis that sex steroid hormone levels influence thyroid cancer development has generally not been supported by studies of “classical” reproductive and hormonal characteristics and risk of thyroid cancer in women. However, many of these studies

854

854

Part IV:  Cancers by Tissue of Origin

are limited by the type of exposure information collected and the age range of the participants under study. Studies evaluating risks of thyroid cancer in women associated with other, less-​studied indicators of sex steroid hormone levels during or following pregnancy (e.g., infertility treatment, pregnancy complications, such as hyperemesis gravidarum, birth weight of offspring, breastfeeding) or throughout life (estimated number of ovulatory cycles) provide some support that sex steroid hormones play a direct or indirect role in the development of thyroid cancer. However, some of these associations may be explained by increased medical surveillance of thyroid disorders or other benign thyroid conditions. For instance, women of reproductive ages, particularly those who have infertility issues, and women who make more frequent visits to their physician around the time of menopause may be more likely to undergo a thyroid exam or have their thyroid hormone levels measured.

In utero Exposures Studying exposures occurring in utero in relation to thyroid cancer risk in adulthood poses many challenges, namely the ability to accurately capture relevant exposures and the relatively long interval from birth to thyroid cancer diagnosis. Hence, there is a paucity of information on possible risk factors occurring during this period of life. Thyroid cancer risk following in utero exposure to I-​131 was evaluated in a cohort of 2582 subjects who were in utero at the time of the Chernobyl accident, but the data were limited due to the limited size of the available sample. The study showed an elevated, but non-​ significant, risk of thyroid cancer (excess odds ratio per gray of 11.66) based on seven cases of thyroid carcinoma and one case of Hurthle cell neoplasm during screening (Hatch et al., 2009). As mentioned earlier, risk of thyroid cancer in the offspring of mothers diagnosed with hyperemesis gravidarum was non-​ significantly elevated compared to mothers without this diagnosis (Vandraas et al., 2015b). Little information is available regarding risk of thyroid cancer in relation to in utero exposures resulting from other pregnancy complications, maternal nutrition status, fetal or maternal iodine status, or exposure to tobacco smoke or other environmental contaminants.

Genetic Susceptibility In the large majority of cases, follicular cell-​derived thyroid cancers (non-​medullary thyroid cancer; NMTC) are sporadic. In some cases, there appears to be a familial component (5%–​15% depending on the definition). In rare instances, about 5%, the familial pattern is part of a hereditary cancer syndrome. The hereditary cancer syndromes associated with thyroid cancer include Cowden syndrome, familial adenomatous polyposis (FAP) including Gardner syndrome, Carney complex, Werner syndrome, and DICER1 syndrome. With the exception of Werner syndrome, which follows an autosomal recessive inheritance mode, other syndromes follow an autosomal dominant mode. The genes responsible for these syndromes are known; the Cowden syndrome is related to germline mutations in PTEN gene, FAP is attributed to mutations in APC gene, the Carney complex to PRKAR1A gene, and Werner syndrome to WRN gene. When NMTC occurs in two or more first-​or second-​degree relatives within a family not known to have a hereditary cancer syndrome, it is referred to as familial non-​medullary thyroid cancer (FNMTC). Often, but not always, it follows an autosomal dominant pattern. Similar to sporadic thyroid cancer, PTC is the most common histologic type in FNMTC but, unlike sporadic cancer, it tends to occur at younger age (Moses et  al., 2011) and have a more aggressive course. Evidence concerning familial risk for NMTC originally derived from population-​based family registry studies, twin studies, and observational epidemiological studies. It has been reported that the relative risk for developing thyroid cancer given a family history in first-​or second-​ degree relatives ranges between 2 and 10 (Xu et al., 2012). Estimate of genetic heritability might be as high as 50% (Czene et al., 2002), being one of the highest of all cancers, along with testicular cancer and multiple myeloma. While several linkage studies of FMNTC

families identified promising loci on chromosomes 14q31, 1q21, 2q21, and 19p13, to date no high-​penetrance protein-​coding genes have been identified in these regions. One recent report concerning the role of HABP2 G534E germline mutation on 10q25 identified in a seven-​member family with FNMTC (Gara et al., 2015) has not been confirmed in several follow-​up studies (Alzahrani et al., 2016; Tomsic et al., 2016). Overall, linkage and candidate gene studies suggest that most of the genetic thyroid cancer risk in the affected families is due to multiple common variants with low penetrance or rare variants with moderate penetrance (Landa and Robledo, 2011). Low-​penetrance genes have been evaluated in genome-​wide association studies (GWAS). Several GWAS have consistently linked locus 9q22.33, particularly rs965513 located near a thyroid-​specific transcription factor 1 gene (FOXE1), with differentiated thyroid cancer risk (Gudmundsson et al., 2009, 2012). The association with this allele and several other FOXE1 alleles was replicated in different ethnic populations, populations exposed and unexposed to ionizing radiation, and in FNMTC (Pereira et  al., 2015). Other associations have been observed for variants on 14q13.3, 2q35, 8p12 (Gudmundsson et  al., 2009), 10q26.12 (Mancikova et  al., 2015), and 6q14.1 (Mancikova et al., 2015). In addition, rs965513, rs966423 on 2q35, rs2439302 on 8p12, and rs116909374 on 14q13.3 variants have been associated with TSH levels (Gudmundsson et al., 2012), and the 9q22.33 variant has been associated with low fT4 and high fT3 (Gudmundsson et al., 2009). Variants on 2q35 and 14q13.3 have also been associated with elevated fT4 (Gudmundsson et al., 2012). Based on genetic variants identified and replicated thus far, either in GWAS or candidate gene studies, it appears that at least for sporadic NMTC these have an additive effect and collectively explain a relatively small part of thyroid cancer risk. The discrepancy between heritability of thyroid cancer estimated from epidemiological studies and accounted for by available genetic data suggests that predisposition to NMTC is complex and that most genetic variants remain to be discovered. In contrast to differentiated thyroid cancer, very little is known about genetic susceptibility to anaplastic thyroid cancer. This is mainly due to its rarity and high lethality. In one of the largest population-​based studies of thyroid cancer in 63,495 first-​degree relatives of 11,206 NMTC patients identified through cancer registries in five Nordic countries, no single familial case of concordant anaplastic carcinoma was found (Fallah et al., 2013). However, high familial risk was seen with discordant thyroid cancers (i.e., different histological types in the proband and other family members). When a relative had anaplastic thyroid carcinoma, risk of PTC in females was increased 4-​fold, while risk of FTC in males was increased 10-​fold. These findings suggest the role of shared genetic susceptibility and/​or environmental factors in the etiology of different histological types of NMTC. One way of testing this hypothesis is by evaluating whether any of the known germline susceptibility variants for PTC and FTC are also associated with the risk of anaplastic thyroid carcinoma. The hereditary forms of MTC account for about 25% of all MTCs (Nose, 2011; Pacini et al., 2010). The hereditary forms include multiple endocrine neoplasia type 2 (MEN2), comprising MEN2A (95%) and MEN2B (5%), and familial MTC (FMTC). The latter is now recognized by the American Thyroid Association (ATA) Task Force as a MEN2A variant rather than a distinct variant (Wells et al., 2015). Both MEN2A and MEN2B syndromes are associated with MTC and C-​cell hyperplasia in 100% of cases, while MEN2A is associated with pheochromocytoma in 30%–​60% of cases and with parathyroid hyperplasia in 10%–​30% of cases; and MEN2B syndrome is associated with marfanoid habitus, mucosal neuromas, ganglioneurmatosis, and with pheochromocytoma in 50% of cases (Nose, 2011). Nearly 98% of MEN2 families carry activating germline mutations in RET proto-​oncogene inherited in autosomal dominant mode (Pacini et al., 2010). RET gene is located on chromosome 10q11.2; it has 21 exons distributed over 60 Kb. It encodes for a membrane bound receptor that, with a co-​receptor, has affinity for glial-​derived nerve factor. Upon binding, it activates the tyrosine kinase transduction pathway. It is normally expressed in C-​cells but not follicular thyroid cells. In the majority of MEN2A patients, germline RET mutations cluster in exon 10 (codons 609, 611, 618, or 620)  or 11 (codon

 85

Thyroid Cancer 634) corresponding to the cysteine-​rich extracellular domain of RET. In very rare families with typical MEN2A syndrome, no identifiable RET mutations are found. Approximately 95% of patients with MEN2B syndrome have RET mutations in exon 16 (codon M918T) and less than 5% have mutations in exon 15 (codon A883F) corresponding to intracellular tyrosine kinase domain of RET. The specific site of germline RET mutations has been correlated with phenotypic differences and clinical aggressiveness of MTC, as evident by the development of MTC in early age and presentation with metastatic disease (Nose, 2011; Pacini et al., 2010). In contrast to hereditary MTC, susceptibility to sporadic MTC remains poorly understood. One meta-​analysis of published association studies of sporadic MTC showed suggestive association with RET S836S variant (rs800862) and significant association with an intronic RET variant rs2565206 (Figlioli et al., 2013). It should be noted that up to 7% of patients with sporadic MTC might actually have hereditary disease (Wells et  al., 2015). Therefore, it is recommended that any patient with sporadic MTC should undergo genetic counseling and DNA testing to detect germline mutations in RET gene. If found, first-​degree relatives of such patients should also be offered genetic counseling and genetic testing. To summarize, the genetic basis of several hereditary syndromes including NMTC and MTC is well understood and explained by high-​ penetrance loci. However, much more remains to be learned about genetic variants accounting for susceptibility to FNMTC, sporadic NMTC, and sporadic MTC. The current evidence points to the role of yet to be identified multiple common mild penetrance or rare moderate penetrance alleles.

OPPORTUNITIES FOR PREVENTION As discussed earlier, radiation is the best studied and the strongest modifiable risk factor for thyroid cancer. However, there is increasing evidence suggesting that obesity and weight gain throughout the life course are associated with an elevated risk of thyroid cancer. While modest weight loss among overweight and obese individuals is reasonably attainable, it remains unclear if thyroid cancer risk declines with intentional weight loss. Maintenance of a healthy weight from a young age may be a more effective means for prevention of thyroid cancer, among a wide range of other obesity-​related health consequences. Other factors, such as sex, preexisting thyroid nodules, and reproductive factors, either cannot be modified or are difficult to change. Avoidance of unnecessary radiation exposure, especially in children, is the most potent and available method to reduce the risk of developing thyroid cancer. The “Image Gently Campaign” has taken a leading role in raising awareness of the need to reduce radiation exposure to the minimum, the ALARA (As Low As Reasonably Achievable) principle, in diagnostic, dental, and therapeutic settings (Goske et al., 2014; Law et al., 2014). As mentioned earlier, remediation of iodine deficiency changes the balance between FTC and PTC. Reducing FTC incidence reduces the cancer type that is more likely to spread hematogenously to distant sites. Despite efforts at mitigation, nuclear power plant accidents remain a possibility. The thyroid is at special risk because radioactive iodine is a large component of the released radioactivity, and the gland has the ability to concentrate it. Evacuation and control of the food chain, as occurred after the Fukushima accident, are the principal methods of reducing exposure (Yamashita, 2014). However, when evacuations cannot be initiated rapidly enough, timely ingestion of potassium iodide can reduce the thyroid dose considerably. The US Centers for Disease Control and Prevention and the World Health Organization have specific (slightly different) guidelines, taking into account age and projected level of exposure (US Food and Drug Administration, 2001; World Health Organization, 1999). Early detection of cases can be achieved by ultrasound screening. However, screening of asymptomatic people without any risk factors for thyroid cancer is generally discouraged due to the high rates of false positives (benign nodules) and the fact that most small thyroid cancers do not progress. Nevertheless, screening has been very

855

prevalent in some countries, such as Korea, although it has recently abated (Ahn et al., 2014; Minamimoto et al., 2014). If screening is to be used at all, it should be reserved for individuals at the highest risk. The need for screening is clearest for members of families at risk for MTC, principally by measuring calcitonin and, as needed, imaging. In other settings, screening refers to the use of ultrasound to detect thyroid nodules, some of which would be expected to be malignant. This most clearly applies to patients with familial genetic syndromes such as Gardner’s syndrome and Cowden’s disease (Rowland and Moley, 2015). The recommendation for screening also applies to members of families with two first-​degree relatives or multiple cases of thyroid cancer. Some go further and advocate screening if any two members of a family have thyroid cancer or even all first-​degree relatives of any patient with thyroid cancer, but it is not clear how widely this is done (Rowland and Moley, 2015). Similarly, screening beyond routine palpation has not been adopted by most guidelines for patients who were exposed to therapeutic radiation for cancer as children. This guideline may be re-​evaluated, in part as a result of confirmation that the risk continues to increase with dose into the range used in childhood cancer treatments (Mulder, 2013; Veiga et al., 2012).

SUMMARY AND FUTURE RESEARCH The incidence of thyroid cancer, once considered a relatively uncommon malignancy, has been reported to have increased substantially over the past few decades in nearly every country with long-​term, high-​quality registry data. Much of this increase is due to an increase in PTC, the most common histologic type. Several lines of evidence support that the rising incidence of PTC in many parts of the world is explained in part by the greater opportunity and ability to detect and diagnose small thyroid cancers, while some of the increase also appears to be attributable to environmental or lifestyle-​related factors. While there is substantial variability in background incidence rates of thyroid cancer worldwide, it has been difficult to use international data to identify potential environmental factors for thyroid cancer because the economic status of countries and the related penetration of imaging modalities accounts for much of these differences. The relatively high incidence of thyroid cancer in women compared with men, particularly between the ages at puberty and menopause, remains largely unexplained. Recent studies on “classical” reproductive and hormonal factors have not provided any consistent evidence for a role of sex steroid hormones in the etiology of this disease. Long-​ term prospective studies of thyroid cancer with direct (pre-​diagnostic concentrations) or indirect indicators of sex steroid hormones and other biomarkers at potentially etiologically relevant time periods, such as pregnancy, could help to elucidate the mechanisms by which women are more susceptible to thyroid cancer than men. Other potential explanations for the sex differential need to be explored as well. In general, most case-​control and cohort studies evaluate exposures in mid-​to late adulthood, which may not capture the period in the natural history of the disease when certain exposures have the greatest effect. More data are needed on risks associated with exposures occurring early in life and in adolescence to better understand whether or not this is an etiologically relevant time period. Despite advancements in our understanding of the natural history and etiology of thyroid cancer, a number of major questions remain largely unanswered. For instance, an inverse, as opposed to the expected positive, relationship observed between circulating TSH levels and differentiated thyroid cancer risk is perplexing, as TSH has been hypothesized to mediate many of the observed associations for environmental and lifestyle-​related factors (e.g., smoking, BMI, diet, reproductive factors) and thyroid cancer risk. The associations of thyroid hormone and TSH levels should be examined in other studies and populations (preferably younger at the time of measurement) to better understand the role of TSH and thyroid function in the development of thyroid cancer. The relative contributions of ionizing radiation and other known or suspected modifiable risk factors to the changing incidence of thyroid cancer remain to be determined, though it is unlikely that ionizing

856

856

Part IV:  Cancers by Tissue of Origin

radiation has played a major role, considering the declining temporal trends observed for radiation-​related molecular subtypes of PTC (e.g., RET/​PTC), together with the rising proportion of PTCs with point mutations, such as RAS and BRAF, that are not associated with radiation exposure. The rise in obesity prevalence and decline in smoking are two potentially important factors, as greater adiposity and cigarette smoking have been consistently positively and inversely associated with thyroid cancer risk in epidemiologic studies, respectively. Whether thyroid cancer incidence trends have been impacted directly by these factors remains unclear, as a causal relationship for obesity and smoking has not been established. There are a limited number of studies evaluating other environmental factors, such as chemical pollutants, that could also play a role. Large case-​control and prospective studies, including pooled analyses of these studies, have afforded the opportunity to evaluate risk factors for thyroid cancer by histologic types. Few studies have further evaluated risk factors by stage at diagnosis, which would help to distinguish tumors that are indolent and are often identified incidentally versus those with greater aggressive potential. Given recent developments in identifying distinct molecular subtypes of thyroid cancer, future studies that incorporate molecular subtype information in epidemiologic studies of thyroid cancer, in addition to histologic and staging information, could provide greater insight on thyroid cancer as a potentially etiologically heterogeneous disease.

ACKNOWLEDGMENTS We are grateful for Dr.  Susan Devesa and David Check of the Biostatistics Branch and Dr.  Hyeyeun Lim of the Radiation Epidemiology Branch of the Division of Cancer Epidemiology and Genetics, National Cancer Institute, for figure development and assistance with the analysis of SEER and US mortality data. References Ahn HS, Kim HJ, and Welch HG. 2014. Korea’s thyroid-​ cancer “epidemic”:  screening and overdiagnosis. N Engl J Med, 371(19), 1765–​ 1767. PMID: 25372084. Allen NE, Beral V, Casabonne D, et  al. 2009. Moderate alcohol intake and cancer incidence in women. J Natl Cancer Inst, 101(5), 296–​ 305. PMID: 19244173. Almquist M, Johansen D, Bjorge T, et al. 2011. Metabolic factors and risk of thyroid cancer in the Metabolic syndrome and Cancer project (Me-​Can). Cancer Causes Control, 22(5), 743–​751. PMID: 21380729. Alzahrani AS, Murugan AK, Qasem E, and Al-​Hindi HN. 2016. HABP2 gene mutations do not cause familial or sporadic non-​medullary thyroid cancer in a highly inbred Middle Eastern population. Thyroid, 26(5), 667–​671. PMID: 26906432. Andersson TM, Johansson AL, Fredriksson I, and Lambe M. 2015. Cancer during pregnancy and the postpartum period: a population-​based study. Cancer, 121(12), 2072–​2077. PMID: 25737403. Arnold M, Pandeya N, Byrnes G, et al. 2015. Global burden of cancer attributable to high body-​mass index in 2012: a population-​based study. Lancet Oncol, 16(1), 36–​46. PMCID: PMC4314462. Aschebrook-​Kilfoy B, DellaValle CT, Purdue M, et al. 2015. Polybrominated diphenyl ethers and thyroid cancer risk in the Prostate, Colorectal, Lung, and Ovarian Cancer Screening Trial cohort. Am J Epidemiol, 181(11), 883–​888. PMCID: PMC4445391. Aschebrook-​Kilfoy B, Grogan RH, Ward MH, Kaplan E, and Devesa SS. 2013a. Follicular thyroid cancer incidence patterns in the United States, 1980–​2009. Thyroid, 23(8), 1015–​1021. PMCID: PMC3752506. Aschebrook-​Kilfoy B, Shu XO, Gao YT, et al. 2013b. Thyroid cancer risk and dietary nitrate and nitrite intake in the Shanghai women’s health study. Int J Cancer, 132(4), 897–​904. PMCID: PMC3443521. Aschebrook-​Kilfoy B, Ward MH, Della Valle CT, and Friesen MC. 2014. Occupation and thyroid cancer. Occup Environ Med, 71(5), 366–​380. PMCID: PMC4059396. Astakhova LN, Anspaugh LR, Beebe GW, et  al. 1998. Chernobyl-​ related thyroid cancer in children of Belarus: a case-​control study. Radiat Res, 150(3), 349–​356. PMID: 9728663. Balasubramaniam S, Ron E, Gridley G, Schneider AB, and Brenner AV. 2012. Association between benign thyroid and endocrine disorders and subsequent risk of thyroid cancer among 4.5 million U.S. male veterans. J Clin Endocrinol Metab, 97(8), 2661–​2669. PMCID: PMC3410263.

Baloch Z, LiVolsi VA, and Tondon R. 2013. Aggressive variants of follicular cell derived thyroid carcinoma; the so called “real thyroid carcinomas.” J Clin Pathol, 66(9), 733–​743. PMID: 23626010. Becker C, Jick SS, Meier CR, and Bodmer M. 2015. No evidence for a decreased risk of thyroid cancer in association with use of metformin or other antidiabetic drugs: a case-​control study. BMC Cancer, 15, 719. PMCID: PMC4609106. Bernatsky S, Ramsey-​Goldman R, Labrecque J, et  al. 2013. Cancer risk in systemic lupus:  an updated international multi-​centre cohort study. J Autoimmun, 42, 130–​135. PMCID: PMC3646904. Boelaert K, Horacek J, Holder RL, et al. 2006. Serum thyrotropin concentration as a novel predictor of malignancy in thyroid nodules investigated by fine-​needle aspiration. J Clin Endocrinol Metab, 91(11), 4295–​4301. PMID: 16868053. Bosetti C, Kolonel L, Negri E, et al. 2001. A pooled analysis of case-​control studies of thyroid cancer. VI. Fish and shellfish consumption. Cancer Causes Control, 12(4), 375–​382. PMID: 11456234. Bouville A, Beck HL, and Simon SL. 2010. Doses from external irradiation to Marshall Islanders from Bikini and Enewetak nuclear weapons tests. Health Phys, 99(2), 143–​156. PMID: 20622549. Braganza MZ, de Gonzalez AB, Schonfeld SJ, et  al. 2014. Benign breast and gynecologic conditions, reproductive and hormonal factors, and risk of thyroid cancer. Cancer Prev Res (Phila), 7(4), 418–​425. PMCID: PMC3976437. Braganza MZ, Potischman N, Park Y, et al. 2015. Adolescent and mid-​life diet and subsequent risk of thyroid cancer in the NIH-​AARP diet and health study. Int J Cancer, 137(10), 2413–​2423. PMID: 25974060. Brenner AV, Tronko MD, Hatch M, et al. 2011. I-​131 dose response for incident thyroid cancers in Ukraine related to the Chornobyl accident. Environ Health Perspect, 119(7), 933–​939. PMCID: PMC3222994. Brinton LA, Moghissi KS, Scoccia B, et  al. 2015. Effects of fertility drugs on cancers other than breast and gynecologic malignancies. Fertil Steril, 104(4), 980–​988. PMCID: PMC4592810. Burgess JR, Dwyer T, McArdle K, Tucker P, and Shugg D. 2000. The changing incidence and spectrum of thyroid carcinoma in Tasmania (1978–​1998) during a transition from iodine sufficiency to iodine deficiency. J Clin Endocrinol Metab, 85(4), 1513–​1517. PMID: 10770190. Cancer Genome Atlas Research Network. 2014. Integrated genomic characterization of papillary thyroid carcinoma. Cell, 159(3), 676–​ 690. PMCID: PMC4243044. Cappola AR, and Mandel SJ. 2015. Improving the long-​term management of benign thyroid nodules. JAMA, 313(9), 903–​904. PMID: 25734731. Cardis E, Kesminiene A, Ivanov V, et  al. 2005. Risk of thyroid cancer after exposure to 131I in childhood. J Natl Cancer Inst, 97(10), 724–​732. PMID: 15900042. Carling T, Ocal IT, and Udelsman R. 2007. Special variants of differentiated thyroid cancer: does it alter the extent of surgery versus well-​differentiated thyroid cancer? World J Surg, 31(5), 916–​923. PMID: 17345120. Choi JS, Kim EK, Moon HJ, and Kwak JY. 2015. Higher body mass index may be a predictor of extrathyroidal extension in patients with papillary thyroid microcarcinoma. Endocrine, 48(1), 264–​271. PMID: 24858734. Chowdhury R, Sinha B, Sankar MJ, et  al. 2015. Breastfeeding and maternal health outcomes: a systematic review and meta-​analysis. Acta Paediatr, 104(467), 96–​113. PMCID: PMC4670483. Cipriani NA, Nagar S, Kaplan SP, et  al. 2015. Follicular thyroid carcinoma:  how have histologic diagnoses changed in the last half-​century and what are the prognostic implications? Thyroid, 25(11), 1209–​1216. PMID: 26440366. Clero E, Doyon F, Chungue V, et al. 2012. Dietary iodine and thyroid cancer risk in French Polynesia: a case-​control study. Thyroid, 22(4), 422–​429. PMID: 22280227. Cornett WR, Sharma AK, Day TA, et al. 2007. Anaplastic thyroid carcinoma: an overview. Curr Oncol Rep, 9(2), 152–​158. PMID: 17288883. Crump C, Sundquist J, Sieh W, Winkleby MA, and Sundquist K. 2015. Fetal growth and subsequent maternal risk of thyroid cancer. Int J Cancer, 138(5), 1085–​1093. PMID: 26379007. Czene K, Lichtenstein P, and Hemminki K. 2002. Environmental and heritable causes of cancer among 9.6 million individuals in the Swedish Family-​ Cancer Database. Int J Cancer, 99(2), 260–​266. PMID: 11979442. Dabrowska AM, Tarach JS, Kurowska M, and Nowakowski A. 2014. Thyroid diseases in patients with acromegaly. Arch Med Sci, 10(4), 837–​845. PMCID: PMC4175760. Dal Maso L, Bosetti C, La Vecchia C, and Franceschi S. 2009. Risk factors for thyroid cancer: an epidemiological review focused on nutritional factors. Cancer Causes Control, 20(1), 75–​86. PMID: 18766448. Davies L, and Welch HG. 2006. Increasing incidence of thyroid cancer in the United States, 1973–​2002. JAMA, 295(18), 2164–​2167. PMID: 16684987.

 857

Thyroid Cancer Davis S, Kopecky KJ, Hamilton TE, Onstad L, and Hanford Thyroid Disease Study Team. 2004a. Thyroid neoplasia, autoimmune thyroiditis, and hypothyroidism in persons exposed to iodine 131 from the hanford nuclear site. JAMA, 292(21), 2600–​2613. PMID: 15572718. Davis S, Stepanenko V, Rivkind N, et al. 2004b. Risk of thyroid cancer in the Bryansk Oblast of the Russian Federation after the Chernobyl Power Station accident. Radiat Res, 162(3), 241–​248. PMID: 15332999. Demers LM. 2004. Thyroid disease: pathophysiology and diagnosis. Clin Lab Med, 24(1), 19–​28. PMID: 15157555. Dickman PW, Holm LE, Lundell G, Boice JD, Jr., and Hall P. 2003. Thyroid cancer risk after thyroid examination with 131I: a population-​based cohort study in Sweden. Int J Cancer, 106(4), 580–​587. PMID: 12845656. Doniach I. 1957. Comparison of the carcinogenic effect of x-​ irradiation with radioactive iodine on the rat’s thyroid. Br J Cancer, 11(1), 67–​76. PMCID: PMC2073934. Duffy BJ, Jr., and Fitzgerald PJ. 1950. Cancer of the thyroid in children:  a report of 28 cases. J Clin Endocrinol Metab, 10(10), 1296–​ 1308. PMID: 14794754. Duntas LH. 2006. The role of selenium in thyroid autoimmunity and cancer. Thyroid, 16(5), 455–​460. PMID: 16756467. Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A, eds. 2010. AJCC cancer staging manual (7th ed). New York: Springer. Enewold L, Zhu K, Ron E, et  al. 2009. Rising thyroid cancer incidence in the United States by demographic and tumor characteristics, 1980–​2005. Cancer Epidemiol Biomarkers Prev, 18(3), 784–​ 791. PMCID: PMC2676561. Engels EA, Pfeiffer RM, Fraumeni JF, Jr., et al. 2011. Spectrum of cancer risk among US solid organ transplant recipients. JAMA, 306(17), 1891–​1901. PMCID: PMC3310893. Fallah M, Pukkala E, Tryggvadottir L, et al. 2013. Risk of thyroid cancer in first-​degree relatives of patients with non-​medullary thyroid cancer by histology type and age at diagnosis: a joint study from five Nordic countries. J Med Genet, 50(6), 373–​382. PMID: 23585692. Feldt-​Rasmussen U, and Rasmussen AK. 2010. Autoimmunity in differentiated thyroid cancer: significance and related clinical problems. Hormones (Athens), 9(2), 109–​117. PMID: 20687394. Ferlay J, Soerjomataram I, Dikshit R, et al. 2015. Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012. Int J Cancer, 136(5), E359–​386. PMID: 25220842. Ferlay J, Soerjomataram I, Ervik M, Dikshit R, Eser S, Mathers C, Rebelo M, Parkin DM, Forman D, Bray F. 2013. GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality Worldwide: IARC CancerBase No. 11 [Internet]. Available from: http://​globocan.iarc.fr. Accessed on October 16, 2015. Figlioli G, Landi S, Romei C, Elisei R, and Gemignani F. 2013. Medullary thyroid carcinoma (MTC) and RET proto-​oncogene: mutation spectrum in the familial cases and a meta-​analysis of studies on the sporadic form. Mutat Res, 752(1), 36–​44. PMID: 23059849. Flegal KM, Carroll MD, Ogden CL, and Curtin LR. 2010. Prevalence and trends in obesity among US adults, 1999–​2008. JAMA, 303(3), 235–​241. PMID: 20071471. Forman D BF, Brewster DH, Gombe Mbalawa C, Kohler B, Piñeros M, Steliarova-​Foucher E, Swaminathan R, Ferlay J. 2014. Cancer incidence in five continents, IARC (ed.), (Vol. X). Lyon, France. Franceschi S, Preston-​Martin S, Dal Maso L, et  al. 1999. A pooled analysis of case-​control studies of thyroid cancer. IV. Benign thyroid diseases. Cancer Causes Control, 10(6), 583–​595. PMID: 10616827. Franklyn JA, Maisonneuve P, Sheppard M, Betteridge J, and Boyle P. 1999. Cancer incidence and mortality after radioiodine treatment for hyperthyroidism: a population-​based cohort study. Lancet, 353(9170), 2111–​2115. PMID: 10382695. From G, Mellemgaard A, Knudsen N, Jorgensen T, and Perrild H. 2000. Review of thyroid cancer cases among patients with previous benign thyroid disorders. Thyroid, 10(8), 697–​700. PMID: 11014315. Furukawa K, Preston D, Funamoto S, et al. 2013. Long-​term trend of thyroid cancer risk among Japanese atomic-​bomb survivors: 60 years after exposure. Int J Cancer, 132(5), 1222–​1226. PMCID: PMC3910094. Ganly I, Ricarte Filho J, Eng S, et  al. 2013. Genomic dissection of Hurthle cell carcinoma reveals a unique class of thyroid malignancy. J Clin Endocrinol Metab, 98(5), E962–​972. PMID: 23543667. Gara SK, Jia L, Merino MJ, et al. 2015. Germline HABP2 mutation causing familial nonmedullary thyroid cancer. N Engl J Med, 373(5), 448–​455. PMCID: PMC4562406. Goske MJ, Frush DP, Brink JA, et al. 2014. Curbing potential radiation-​induced cancer risks in oncologic imaging: perspectives from the “image gently” and “image wisely” campaigns. Oncology (Williston Park), 28(3), 232–​ 238, 243. PMID: 24855732.

857

Gudmundsson J, Sulem P, Gudbjartsson DF, et al. 2009. Common variants on 9q22.33 and 14q13.3 predispose to thyroid cancer in European populations. Nat Genet, 41(4), 460–​464. PMCID: PMC3664837. Gudmundsson J, Sulem P, Gudbjartsson DF, et  al. 2012. Discovery of common variants associated with low TSH levels and thyroid cancer risk. Nat Genet, 44(3), 319–​322. PMCID: PMC3655412. Hahn K, Schnell-​Inderst P, Grosche B, and Holm LE. 2001. Thyroid cancer after diagnostic administration of iodine-​131 in childhood. Radiat Res, 156(1), 61–​70. PMID: 11418074. Hannibal CG, Jensen A, Sharif H, and Kjaer SK. 2008. Risk of thyroid cancer after exposure to fertility drugs: results from a large Danish cohort study. Hum Reprod, 23(2), 451–​456. PMID: 18065402. Harach HR, Escalante DA, and Day ES. 2002. Thyroid cancer and thyroiditis in Salta, Argentina: a 40-​yr study in relation to iodine prophylaxis. Endocr Pathol, 13(3), 175–​181. PMID: 12446916. Harari A, Li N, and Yeh MW. 2014. Racial and socioeconomic disparities in presentation and outcomes of well-​differentiated thyroid cancer. J Clin Endocrinol Metab, 99(1), 133–​141. PMCID: PMC3879674. Harding JL, Shaw JE, Peeters A, Cartensen B, and Magliano DJ. 2015. Cancer risk among people with type 1 and type 2 diabetes:  disentangling true associations, detection bias, and reverse causation. Diabetes Care, 38(2), 264–​270. PMID: 25488912. Haselkorn T, Stewart SL, and Horn-​Ross PL. 2003. Why are thyroid cancer rates so high in southeast asian women living in the United States? The Bay Area Thyroid Cancer Study. Cancer Epidemiol Biomarkers Prev, 12(2), 144–​150. PMID: 12582024. Hashibe M, Galeone C, Buys SS, et al. 2015. Coffee, tea, caffeine intake, and the risk of cancer in the PLCO cohort. Br J Cancer, 113(5), 809–​816. PMID: 26291054. Hatch M, Brenner A, Bogdanova T, et al. 2009. A screening study of thyroid cancer and other thyroid diseases among individuals exposed in utero to iodine-​131 from Chernobyl fallout. J Clin Endocrinol Metab, 94(3), 899–​ 906. PMCID: PMC2681280. Hedinger C, Williams ED, Sobin LH. 1988. Histological typing of thyroid tumors (2nd ed.). Berlin; Heidelberg; New York: Springer. Hedinger C, Williams ED, and Sobin LH. 1989. The WHO histological classification of thyroid tumors: a commentary on the second edition. Cancer, 63(5), 908–​911. PMID: 2914297. Hemminki K, Forsti A, Fallah M, et al. 2014. Risk of cancer in patients with medically diagnosed hay fever or allergic rhinitis. Int J Cancer, 135(10), 2397–​2403. PMID: 24692097. Hempelmann LH. 1968. Risk of thyroid neoplasms after irradiation in childhood. Studies of populations exposed to radiation in childhood show a dose response over a wide dose range. Science, 160(3824), 159–​163. PMID: 5642565. Hieu TT, Russell AW, Cuneo R, et al. 2012. Cancer risk after medical exposure to radioactive iodine in benign thyroid diseases: a meta-​analysis. Endocr Relat Cancer, 19(5), 645–​655. PMID: 22851687. Hoffman DA, McConahey WM, Fraumeni JF, Jr., and Kurland LT. 1982. Cancer incidence following treatment of hyperthyroidism. Int J Epidemiol, 11(3), 218–​224. PMID: 7129736. Hognas E, Kauppila A, Pukkala E, and Tapanainen JS. 2014. Cancer risk in women with 10 or more deliveries. Obstet Gynecol, 123(4), 811–​816. PMID: 24785609. Holm LE, Hall P, Wiklund K, et  al. 1991. Cancer risk after iodine-​131 therapy for hyperthyroidism. J Natl Cancer Inst, 83(15), 1072–​ 1077. PMID: 1875414. Horn-​Ross PL, Canchola AJ, Ma H, Reynolds P, and Bernstein L. 2011. Hormonal factors and the risk of papillary thyroid cancer in the California Teachers Study cohort. Cancer Epidemiol Biomarkers Prev, 20(8), 1751–​ 1759. PMCID: PMC3288117. Horn-​Ross PL, Morris JS, Lee M, et  al. 2001. Iodine and thyroid cancer risk among women in a multiethnic population:  the Bay Area Thyroid Cancer Study. Cancer Epidemiol Biomarkers Prev, 10(9), 979–​ 985. PMID: 11535551. Howlader N, Noone AM, Krapcho M, Garshell J, Miller D, Altekruse SF, Kosary CL, Yu M, Ruhl J, Tatalovich Z, Mariotto A, Lewis DR, Chen HS, Feuer EJ, and Cronin KA. 2015. SEER cancer statistics review, 1975–​ 2012. Available from: http://​seer.cancer.gov/​csr/​1975_​2012/​. Accessed on December 1, 2015.: Hsu CH, Huang CL, Hsu YH, et al. 2014. Co-​occurrence of second primary malignancy in patients with thyroid cancer. QJM, 107(8), 643–​ 648. PMID: 24623860. Hung MH, Liu CJ, Teng CJ, et al. 2016. Risk of second non-​breast primary cancer in male and female breast cancer patients: a population-​based cohort study. PLoS One, 11(2), e0148597. PMCID: PMC4760946.

85

858

Part IV:  Cancers by Tissue of Origin

Hung SH, Lin HC, and Chung SD. 2015. Statin use and thyroid cancer:  a population-​based case-​control study. Clin Endocrinol (Oxf), 83(1), 111–​ 116. PMID: 25074346. Hwang CY, Chen YJ, Lin MW, et al. 2012. Cancer risk in patients with allergic rhinitis, asthma and atopic dermatitis: a nationwide cohort study in Taiwan. Int J Cancer, 130(5), 1160–​1167. PMID: 21455988. Iribarren C, Haselkorn T, Tekawa IS, and Friedman GD. 2001. Cohort study of thyroid cancer in a San Francisco Bay area population. Int J Cancer, 93(5), 745–​750. PMID: 11477590. Ito Y, and Miyauchi A. 2015. Nonoperative management of low-​risk differentiated thyroid carcinoma. Curr Opin Oncol, 27(1), 15–​20. PMID: 25340592. Ito Y, Nikiforov YE, Schlumberger M, and Vigneri R. 2013. Increasing incidence of thyroid cancer:  controversies explored. Nat Rev Endocrinol, 9(3), 178–​184. PMID: 23358352. Ivanov VK, Kashcheev VV, Chekin SY, et al. 2012. Radiation-​epidemiological studies of thyroid cancer incidence in Russia after the Chernobyl accident (estimation of radiation risks, 1991–​2008 follow-​up period). Radiat Prot Dosimetry, 151(3), 489–​499. PMID: 22416255. Jankovic B, Le KT, and Hershman JM. 2013. Clinical Review:  Hashimoto’s thyroiditis and papillary thyroid carcinoma: is there a correlation? J Clin Endocrinol Metab, 98(2), 474–​482. PMID: 23293329. Ji J, Shu X, Li X, et al. 2009. Cancer risk in hospitalised asthma patients. Br J Cancer, 100(5), 829–​833. PMCID: PMC2653753. Jing Z, Hou X, Liu Y, et  al. 2015. Association between height and thyroid cancer risk: a meta-​analysis of prospective cohort studies. Int J Cancer, 137(6), 1484–​1490. PMID: 25693727. Jorgensen KT, Nielsen NM, Pedersen BV, Jacobsen S, and Frisch M. 2012. Hyperemesis, gestational hypertensive disorders, pregnancy losses and risk of autoimmune diseases in a Danish population-​based cohort. J Autoimmun, 38(2–​3), J120–​128. PMID: 22226784. Jorgensen TJ. 2016. Re: Thyroid cancer among young people in Fukushima. Epidemiology, 27(3), e17. PMID: 26841060. Kabat GC, Kim MY, Thomson CA, et al. 2012a. Anthropometric factors and physical activity and risk of thyroid cancer in postmenopausal women. Cancer Causes Control, 23(3), 421–​430. PMID: 22212611. Kabat GC, Kim MY, Wactawski-​Wende J, et al. 2012b. Menstrual and reproductive factors, exogenous hormone use, and risk of thyroid carcinoma in postmenopausal women. Cancer Causes Control, 23(12), 2031–​2040. PMID: 23090034. Kabat GC, Kim MY, Wactawski-​Wende J, and Rohan TE. 2012c. Smoking and alcohol consumption in relation to risk of thyroid cancer in postmenopausal women. Cancer Epidemiol, 36(4), 335–​340. PMID: 22525339. Karamchandani D, Arias-​Amaya R, Donaldson N, Gilbert J, and Schulte KM. 2010. Thyroid cancer and renal transplantation: a meta-​analysis. Endocr Relat Cancer, 17(1), 159–​167. PMID: 19942714. Kesminiene A, Evrard AS, Ivanov VK, et al. 2012. Risk of thyroid cancer among chernobyl liquidators. Radiat Res, 178(5), 425–​436. PMID: 22998226. Kilfoy BA, Zhang Y, Park Y, et al. 2011. Dietary nitrate and nitrite and the risk of thyroid cancer in the NIH-​AARP Diet and Health Study. Int J Cancer, 129(1), 160–​172. PMCID: PMC3016446. Kilfoy BA, Zheng T, Holford TR, et al. 2009. International patterns and trends in thyroid cancer incidence, 1973–​2002. Cancer Causes Control, 20(5), 525–​531. PMCID: PMC2788231. Kim HJ, Kim NK, Choi JH, et al. 2013. Associations between body mass index and clinico-​pathological characteristics of papillary thyroid cancer. Clin Endocrinol (Oxf), 78(1), 134–​140. PMID: 22812676. Kitahara CM, Gamborg M, Berrington de Gonzalez A, Sorensen TI, and Baker JL. 2014. Childhood height and body mass index were associated with risk of adult thyroid cancer in a large cohort study. Cancer Res, 74(1), 235–​242. PMCID: PMC3891884. Kitahara CM, Linet MS, Beane Freeman LE, et al. 2012a. Cigarette smoking, alcohol intake, and thyroid cancer risk: a pooled analysis of five prospective studies in the United States. Cancer Causes Control, 23(10), 1615–​ 1624. PMCID: PMC3511822. Kitahara CM, McCullough M, Franceschi S, et al. 2016. Anthropometric factors and thyroid cancer risk by histological subtype: pooled analysis of 22 prospective studies. Thyroid, 26(2), 306–​318. PMID: 26756356. Kitahara CM, Platz EA, Beane Freeman LE et al. 2012b. Physical activity, diabetes, and thyroid cancer risk: a pooled analysis of five prospective studies. Cancer Causes Control [Epub ahead of print]. PMCID: PMC3586378. Knudsen N, Bulow I, Laurberg P, et al. 2001. Alcohol consumption is associated with reduced prevalence of goitre and solitary thyroid nodules. Clin Endocrinol (Oxf), 55(1), 41–​46. PMID: 11453951. Kohler BA, Sherman RL, Howlader N, et al. 2015. Annual Report to the Nation on the Status of Cancer, 1975–​2011, Featuring incidence of breast cancer subtypes by race/​ethnicity, poverty, and state. J Natl Cancer Inst, 107(6), djv048. PMID: 25825511.

Kolonel LN, Hankin JH, Wilkens LR, Fukunaga FH, and Hinds MW. 1990. An epidemiologic study of thyroid cancer in Hawaii. Cancer Causes Control, 1(3), 223–​234. PMID: 2102295. Kopp P, Kimura ET, Aeschimann S, et  al. 1994. Polyclonal and monoclonal thyroid nodules coexist within human multinodular goiters. J Clin Endocrinol Metab, 79(1), 134–​139. PMID: 7517946. Kwon H, Kim M, Choi YM, et al. 2015. Lack of associations between body mass index and clinical outcomes in patients with papillary thyroid carcinoma. Endocrinol Metab (Seoul), 30(3), 305–​311. PMCID: PMC4595355. La Vecchia C, Malvezzi M, Bosetti C, et  al. 2015. Thyroid cancer mortality and incidence: a global overview. Int J Cancer, 136(9), 2187–​2195. PMID: 25284703. Landa I, and Robledo M. 2011. Association studies in thyroid cancer susceptibility:  are we on the right track? J Mol Endocrinol, 47(1), R43–​58. PMID: 21830323. Law CS, Douglass JM, Farman AG, et al. 2014. The image gently in dentistry campaign: partnering with parents to promote the responsible use of x-​rays in pediatric dentistry. Pediatr Dent, 36(7), 458–​459. PMID: 25514072. Lerro CC, Koutros S, Andreotti G, et  al. 2015. Organophosphate insecticide use and cancer incidence among spouses of pesticide applicators in the Agricultural Health Study. Occup Environ Med, 72(10), 736–​744. PMID: 26150671. Liang Y, Yang Z, Qin B, and Zhong R. 2014. Primary Sjogren’s syndrome and malignancy risk: a systematic review and meta-​analysis. Ann Rheum Dis, 73(6), 1151–​1156. PMID: 23687261. Lin SW, Wheeler DC, Park Y, et al. 2012. Prospective study of ultraviolet radiation exposure and risk of cancer in the United States. Int J Cancer, 131(6), E1015–​1023. PMCID: PMC3402606. Lind P, Kumnig G, Heinisch M, et  al. 2002. Iodine supplementation in Austria: methods and results. Thyroid, 12(10), 903–​907. PMID: 12487773. Lindsay S, Potter GD, and Chaikoff IL. 1957. Thyroid neoplasms in the rat: a comparison of naturally occurring and I131-​induced tumors. Cancer Res, 17(3), 183–​189. PMID: 13413859. Liu X, Qu S, Liu R, et al. 2014. TERT promoter mutations and their association with BRAF V600E mutation and aggressive clinicopathological characteristics of thyroid cancer. J Clin Endocrinol Metab, 99(6), E1130–​1136. PMID: 24617711. Liu Z, Maimaiti Y, Yu P, et al. 2015. Correlation between body mass index and clinicopathological features of papillary thyroid microcarcinoma. Int J Clin Exp Med, 8(9), 16472–​16479. PMCID: PMC4659061. LiVolsi VA, Abrosimov AA, Bogdanova T, et al. 2011. The Chernobyl thyroid cancer experience:  pathology. Clin Oncol (R Coll Radiol), 23(4), 261–​ 267. PMID: 21333507. LiVolsi VA, and Asa SL. 1994. The demise of follicular carcinoma of the thyroid gland. Thyroid, 4(2), 233–​236. PMID: 7920009. Lope V, Perez-​Gomez B, Aragones N, et al. 2006. Occupational exposure to ionizing radiation and electromagnetic fields in relation to the risk of thyroid cancer in Sweden. Scand J Work Environ Health, 32(4), 276–​284. PMID: 16932825. Lu CH, Lee KD, Chen PT, et al. 2013. Second primary malignancies following thyroid cancer:  a population-​based study in Taiwan. Eur J Endocrinol, 169(5), 577–​585. PMID: 23939917. Lyon JL, Alder SC, Stone MB, et  al. 2006. Thyroid disease associated with exposure to the Nevada nuclear weapons test site radiation: a reevaluation based on corrected dosimetry and examination data. Epidemiology, 17(6), 604–​614. PMID: 17028502. Mack WJ, Preston-​Martin S, Dal Maso L, et  al. 2003. A pooled analysis of case-​control studies of thyroid cancer: cigarette smoking and consumption of alcohol, coffee, and tea. Cancer Causes Control, 14(8), 773–​785. PMID: 14674742. Mancikova V, Cruz R, Inglada-​Perez L, et  al. 2015. Thyroid cancer GWAS identifies 10q26.12 and 6q14.1 as novel susceptibility loci and reveals genetic heterogeneity among populations. Int J Cancer, 137(8), 1870–​ 1878. PMID: 25855579. Mao S, Shen H, and Zhang J. 2016. Systemic lupus erythematosus and malignancies risk. J Cancer Res Clin Oncol, 142(1), 253–​262. PMID: 26319223. McFadden DG, Dias-​Santagata D, Sadow PM, et  al. 2014. Identification of oncogenic mutations and gene fusions in the follicular variant of papillary thyroid carcinoma. J Clin Endocrinol Metab, 99(11), E2457–​2462. PMCID: PMC4223441. Meinhold CL, Ron E, Schonfeld SJ, et al. 2010. Nonradiation risk factors for thyroid cancer in the US Radiologic Technologists Study. Am J Epidemiol, 171(2), 242–​252. PMCID: PMC3290908. Mellemgaard A, From G, Jorgensen T, et al. 1998. Cancer risk in individuals with benign thyroid disorders. Thyroid, 8(9), 751–​754. PMID: 9777744. Mete O, and Asa SL. 2013. Precursor lesions of endocrine system neoplasms. Pathology, 45(3), 316–​330. PMID: 23478233.

 859

Thyroid Cancer Minamimoto R, Senda M, Jinnouchi S, et al. 2014. Detection of thyroid cancer by an FDG-​PET cancer screening program: a Japanese nation-​wide survey. Anticancer Res, 34(8), 4439–​4445. PMID: 25075083. Mortensen JD, Woolner LB, and Bennett WA. 1955. Gross and microscopic findings in clinically normal thyroid glands. J Clin Endocrinol Metab, 15(10), 1270–​1280. PMID: 13263417. Mosconi G, Stalteri L, Centofanti F, et  al. 2011. Incidence of cancer in kidney transplantation waiting list patients:  a single center experience. Transplant Proc, 43(4), 1003–​1005. PMID: 21620036. Moses W, Weng J, and Kebebew E. 2011. Prevalence, clinicopathologic features, and somatic genetic mutation profile in familial versus sporadic nonmedullary thyroid cancer. Thyroid, 21(4), 367–​371. PMCID: PMC3070337. Muirhead CR, O’Hagan JA, Haylock RG, et  al. 2009. Mortality and cancer incidence following occupational radiation exposure: third analysis of the National Registry for Radiation Workers. Br J Cancer, 100(1), 206–​212. PMCID: PMC2634664. Mulder R. 2013. Kick off expert panel thyroid cancer surveillance. Available from:  http://​www.ighg.org/​kick-​off-​expert-​panel-​thyroid-​cancer-​surveillance/​ . Accessed on March 1, 2016. Navarro Silvera SA, Miller AB, and Rohan TE. 2005. Risk factors for thyroid cancer: a prospective cohort study. Int J Cancer, 116(3), 433–​438. PMID: 15818623. Negri E, Ron E, Franceschi S, et al. 1999. A pooled analysis of case-​control studies of thyroid cancer. I.  Methods. Cancer Causes Control, 10(2), 131–​142. PMID: 10231162. Nielsen SM, White MG, Hong S, et al. 2016. The breast-​thyroid cancer link: a systematic review and meta-​ analysis. Cancer Epidemiol Biomarkers Prev, 25(2), 231–​238. PMCID: PMC4770576. Nikiforov YE, Rowland JM, Bove KE, Monforte-​Munoz H, and Fagin JA. 1997. Distinct pattern of ret oncogene rearrangements in morphological variants of radiation-​induced and sporadic thyroid papillary carcinomas in children. Cancer Res, 57(9), 1690–​1694. PMID: 9135009. Nikiforov YE, Seethala RR, Tallini G, et al. 2016. Nomenclature revision for encapsulated follicular variant of papillary thyroid carcinoma: a paradigm shift to reduce overtreatment of indolent tumors. JAMA Oncol, 2(8), 1023-1029, PMID: 27078145. Nikiforova MN, Lynch RA, Biddinger PW, et al. 2003. RAS point mutations and PAX8-​ PPAR gamma rearrangement in thyroid tumors:  evidence for distinct molecular pathways in thyroid follicular carcinoma. J Clin Endocrinol Metab, 88(5), 2318–​2326. PMID: 12727991. Nose V. 2011. Familial thyroid cancer: a review. Mod Pathol, 24 Suppl 2, S19–​ 33. PMID: 21455198. O’Grady TJ, Kitahara CM, DiRienzo AG, and Gates MA. 2014. The association between selenium and other micronutrients and thyroid cancer incidence in the NIH-​AARP Diet and Health Study. PLoS One, 9(10), e110886. PMCID: PMC4203851. Oh CM, Jung KW, Won YJ et  al. 2015. Age-​period-​cohort analysis of thyroid cancer incidence in Korea. Cancer Res Treat, 47(3), 362–​ 369. PMCID: PMC4506117. Ostroumova E, Gudzenko N, Brenner A, et al. 2014. Thyroid cancer incidence in Chornobyl liquidators in Ukraine:  SIR analysis, 1986–​2010. Eur J Epidemiol, 29(5), 337–​342. PMID: 24705729. Pacini F, Castagna MG, Cipri C, and Schlumberger M. 2010. Medullary thyroid carcinoma. Clin Oncol (R Coll Radiol), 22(6), 475–​485. PMID: 20627492. Patel D, Kitahara CM, Park Y, et al. 2015. Thyroid cancer and nonsteroidal anti-​ inflammatory drug use: a pooled analysis of patients older than 40 years of age. Thyroid, 25(12), 1355–​1362. PMCID: PMC4684667. Pazaitou-​Panayiotou K, Polyzos SA, and Mantzoros CS. 2013. Obesity and thyroid cancer: epidemiologic associations and underlying mechanisms. Obes Rev, 14(12), 1006–​1022. PMID: 24034423. Pellegriti G, De Vathaire F, Scollo C, et al. 2009. Papillary thyroid cancer incidence in the volcanic area of Sicily. J Natl Cancer Inst, 101(22), 1575–​ 1583. PMID: 19893009. Pereira JS, da Silva JG, Tomaz RA, et al. 2015. Identification of a novel germline FOXE1 variant in patients with familial non-​medullary thyroid carcinoma (FNMTC). Endocrine, 49(1), 204–​214. PMID: 25381600. Pollak MN, Schernhammer ES, and Hankinson SE. 2004. Insulin-​ like growth factors and neoplasia. Nat Rev Cancer, 4(7), 505–​ 518. PMID: 15229476. Preston-​Martin S, Franceschi S, Ron E, and Negri E. 2003. Thyroid cancer pooled analysis from 14 case-​control studies:  what have we learned? Cancer Causes Control, 14(8), 787–​789. PMID: 14674743. Rahu M, Rahu K, Auvinen A, et  al. 2006. Cancer risk among Chernobyl cleanup workers in Estonia and Latvia, 1986–​1998. Int J Cancer, 119(1), 162–​168. PMID: 16432838. Ren Y, Kitahara CM, Berrington de Gonzalez A, et al. 2014. Lack of association between fingernail selenium and thyroid cancer risk: a case-​control

859

study in French Polynesia. Asian Pac J Cancer Prev, 15(13), 5187–​5194. PMID: 25040973. Renehan AG, Tyson M, Egger M, Heller RF, and Zwahlen M. 2008. Body-​mass index and incidence of cancer:  a systematic review and meta-​analysis of prospective observational studies. Lancet, 371(9612), 569–​ 578. PMID: 18280327. Rinaldi S, Lise M, Clavel-​Chapelon F, et al. 2012. Body size and risk of differentiated thyroid carcinomas: findings from the EPIC study. Int J Cancer, 131(6), E1004–​1014. PMID: 22511178. Rinaldi S, Plummer M, Biessy C, et  al. 2014. Thyroid-​stimulating hormone, thyroglobulin, and thyroid hormones and risk of differentiated thyroid carcinoma:  the EPIC study. J Natl Cancer Inst, 106(6), dju097. PMID: 24824312. Ron E, Doody MM, Becker DV, et al. 1998. Cancer mortality following treatment for adult hyperthyroidism. Cooperative Thyrotoxicosis Therapy Follow-​up Study Group. JAMA, 280(4), 347–​355. PMID: 9686552. Ron E, Lubin JH, Shore RE, et al. 1995. Thyroid cancer after exposure to external radiation: a pooled analysis of seven studies. Radiat Res, 141(3), 259–​ 277. PMID: 7871153. Ron E, Schneider AB. 2006. Thyroid cancer. In: Schottenfeld D, Fraumeni JF, Jr. (eds.), Cancer epidemiology and prevention (3rd ed.). New York: Oxford University Press. Ronckers CM, McCarron P, and Ron E. 2005. Thyroid cancer and multiple primary tumors in the SEER cancer registries. Int J Cancer, 117(2), 281–​ 288. PMID: 15880372. Rowland KJ, and Moley JF. 2015. Hereditary thyroid cancer syndromes and genetic testing. J Surg Oncol, 111(1), 51–​60. PMID: 25351655. Ryodi E, Metso S, Jaatinen P, et al. 2015. Cancer incidence and mortality in patients treated either with RAI or thyroidectomy for hyperthyroidism. J Clin Endocrinol Metab, 100(10), 3710–​3717. PMID: 26262435. Schmid D, Ricci C, Behrens G, and Leitzmann MF. 2015. Adiposity and risk of thyroid cancer: a systematic review and meta-​analysis. Obes Rev, 16(12), 1042–​1054. PMID: 26365757. Schmidt JA, Allen NE, Almquist M, et  al. 2014. Insulin-​like growth factor-​i and risk of differentiated thyroid carcinoma in the European prospective investigation into cancer and nutrition. Cancer Epidemiol Biomarkers Prev, 23(6), 976–​985. PMCID: PMC4046912. Schonfeld SJ, Lee C, and Berrington de Gonzalez A. 2011a. Medical exposure to radiation and thyroid cancer. Clin Oncol (R Coll Radiol), 23(4), 244–​ 250. PMID: 21296564. Schonfeld SJ, Ron E, Kitahara CM, et  al. 2011b. Hormonal and reproductive factors and risk of postmenopausal thyroid cancer in the NIH-​ AARP Diet and Health Study. Cancer Epidemiol, 35(6), e85–​ 90. PMCID: PMC3215902. SEER. 2012. ICD-​ O-​ 3 SEER site/​ histology validation list. Available from:  http://​seer.cancer.gov/​icd-​o-​3/​sitetype.icdo3.d20121205.pdf. Accessed on May 1, 2016. SEER. 2015. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov). Research Data (1973–​ 2012) from National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch. Sen A, Tsilidis KK, Allen NE, et al. 2015. Baseline and lifetime alcohol consumption and risk of differentiated thyroid carcinoma in the EPIC study. Br J Cancer, 113(5), 840–​847. PMID: 26313664. Shakhtarin VV, Tsyb AF, Stepanenko VF, et al. 2003. Iodine deficiency, radiation dose, and the risk of thyroid cancer among children and adolescents in the Bryansk region of Russia following the Chernobyl power station accident. Int J Epidemiol, 32(4), 584–​591. PMID: 12913034. Silberstein EB, Alavi A, Balon HR, et al. 2012. The SNMMI practice guideline for therapy of thyroid disease with 131I 3.0. J Nucl Med, 53(10), 1633–​ 1651. PMID: 22787108. Simon SL, Bouville A, Land CE, and Beck HL. 2010. Radiation doses and cancer risks in the Marshall Islands associated with exposure to radioactive fallout from Bikini and Enewetak nuclear weapons tests: summary. Health Phys, 99(2), 105–​123. PMCID: PMC4042840. Sont WN, Zielinski JM, Ashmore JP, et al. 2001. First analysis of cancer incidence and occupational radiation exposure based on the National Dose Registry of Canada. Am J Epidemiol, 153(4), 309–​318. PMID: 11207146. Spampinato MV, Tipnis S, Tavernier J, and Huda W. 2015. Thyroid doses and risk to paediatric patients undergoing neck CT examinations. Eur Radiol, 25(7), 1883–​1890. PMID: 25638220. Stezhko VA, Buglova EE, Danilova LI, et al. 2004. A cohort study of thyroid cancer and other thyroid diseases after the Chornobyl accident: objectives, design and methods. Radiat Res, 161(4), 481–​492. PMID: 15038762. Subramanian S, Goldstein DP, Parlea L, et  al. 2007. Second primary malignancy risk in thyroid cancer survivors:  a systematic review and meta-​ analysis. Thyroid, 17(12), 1277–​1288. PMID: 18020916.

860

860

Part IV:  Cancers by Tissue of Origin

Sugishita Y, Kammori M, Yamada O, et al. 2014. Biological differential diagnosis of follicular thyroid tumor and Hurthle cell tumor on the basis of telomere length and hTERT expression. Ann Surg Oncol, 21(7), 2318–​ 2325. PMID: 24562933. Suzuki S. 2016a. Childhood and adolescent thyroid cancer in Fukushima after the Fukushima Daiichi nuclear power plant accident:  5  years on. Clin Oncol (R Coll Radiol), 28(4), 263–​271. PMID: 26822892. Suzuki S. 2016b. Re:  Thyroid cancer among young people in Fukushima. Epidemiology, 27(3), e19. PMID: 26849640. Tajtakova M, Semanova Z, Tomkova Z, et al. 2006. Increased thyroid volume and frequency of thyroid disorders signs in schoolchildren from nitrate polluted area. Chemosphere, 62(4), 559–​564. PMID: 16095667. Takahashi T, Schoemaker MJ, Trott KR, et al. 2003. The relationship of thyroid cancer with radiation exposure from nuclear weapon testing in the Marshall Islands. J Epidemiol, 13(2), 99–​107. PMID: 12675119. Thoresen SO, Myking O, Glattre E, et  al. 1988. Serum thyroglobulin as a preclinical tumour marker in subgroups of thyroid cancer. Br J Cancer, 57(1), 105–​108. PMCID: PMC2246681. Tomsic J, Fultz R, Liyanarachchi S, et  al. 2016. HABP2 G534E variant in papillary thyroid carcinoma. PLoS One, 11(1), e0146315. PMCID: PMC4706330. Tonacchera M, Pinchera A, Dimida A, et  al. 2004. Relative potencies and additivity of perchlorate, thiocyanate, nitrate, and iodide on the inhibition of radioactive iodide uptake by the human sodium iodide symporter. Thyroid, 14(12), 1012–​1019. PMID: 15650353. Tresallet C, Seman M, Tissier F, et al. 2014. The incidence of papillary thyroid carcinoma and outcomes in operative patients according to their body mass indices. Surgery, 156(5), 1145–​1152. PMID: 24878452. Tronko MD, Howe GR, Bogdanova TI, et al. 2006. A cohort study of thyroid cancer and other thyroid diseases after the chornobyl accident:  thyroid cancer in Ukraine detected during first screening. J Natl Cancer Inst, 98(13), 897–​903. PMID: 16818853. Truong T, Baron-​Dubourdieu D, Rougier Y, and Guenel P. 2010. Role of dietary iodine and cruciferous vegetables in thyroid cancer: a countrywide case-​ control study in New Caledonia. Cancer Causes Control, 21(8), 1183–​ 1192. PMCID: PMC3496161. Tseng CH. 2012. Thyroid cancer risk is not increased in diabetic patients. PLoS One, 7(12), e53096. PMCID: PMC3531389. Tseng CH. 2014. Metformin reduces thyroid cancer risk in Taiwanese patients with type 2 diabetes. PLoS One, 9(10), e109852. PMCID: PMC4193839. Tsuda T, Tokinobu A, Yamamoto E, and Suzuki E. 2015. Thyroid cancer detection by ultrasound among residents ages 18 years and younger in Fukushima, Japan: 2011 to 2014. Epidemiology, 27(3), 316–​322. PMID: 26441345. Tulinius H, Sigfusson N, Sigvaldason H, Bjarnadottir K, and Tryggvadottir L. 1997. Risk factors for malignant diseases: a cohort study on a population of 22,946 Icelanders. Cancer Epidemiol Biomarkers Prev, 6(11), 863–​ 873. PMID: 9367058. UNSCEAR. 2006. UNSCEAR: effects of ionizing radiation (Vol. I). New York: United Nations. UNSCEAR. 2008. United Nations Scientific Committee on the Effects of Atomic Radiation:  Sources and effects of ionizing radiation (Vol. II). New York: United Nations. UNSCEAR. 2013. Sources, effects and risks of ionizing radiation. New York: United Nations. US Food and Drug Administration. 2001. Guidance:  Potassium iodide as a thyroid blocking agent in radiation emergencies. Rockville, MD: US Department of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and Research. Vandraas KF, Grjibovski AM, Stoer NC, et al. 2015a. Hyperemesis gravidarum and maternal cancer risk: a scandinavian nested case-​control study. Int J Cancer, 137(5), 1209–​1216. PMID: 25665163. Vandraas KF, Vikanes AV, Stoer NC, et al. 2015b. Hyperemesis gravidarum and risk of cancer in offspring:  a Scandinavian registry-​based nested case-​ control study. BMC Cancer, 15, 398. PMCID: PMC4430929. Veiga LH, Holmberg E, Anderson H, et al. 2016. Thyroid cancer after childhood exposure to external radiation:  an updated pooled analysis of 12 studies. Radiat Res, 185(5), 473–​484. PMCID: PMC4893786. Veiga LH, Lubin JH, Anderson H, et al. 2012. A pooled analysis of thyroid cancer incidence following radiotherapy for childhood cancer. Radiat Res, 178(4), 365–​376. PMCID: PMC3488851.

Wagle N, Grabiner BC, Van Allen EM, et  al. 2014. Response and acquired resistance to everolimus in anaplastic thyroid cancer. N Engl J Med, 371(15), 1426–​1433. PMCID: PMC4564868. Wakeford R, Auvinen A, Gent RN, et al. 2016. Re: Thyroid cancer among young people in Fukushima. Epidemiology, 27(3), e:20–​21. PMID: 26841059. Wang JX, Zhang LA, Li BX, et al. 2002. Cancer incidence and risk estimation among medical x-​ray workers in China, 1950–​1995. Health Phys, 82(4), 455–​466. PMID: 11906134. Ward MH, Kilfoy BA, Weyer PJ, et  al. 2010. Nitrate intake and the risk of thyroid cancer and thyroid disease. Epidemiology, 21(3), 389–​ 395. PMCID: PMC2879161. Wells SA, Jr., Asa SL, Dralle H, et  al. 2015. Revised American Thyroid Association guidelines for the management of medullary thyroid carcinoma. Thyroid, 25(6), 567–​610. PMCID: PMC4490627. Weng MY, Huang YT, Liu MF, and Lu TH. 2012. Incidence of cancer in a nationwide population cohort of 7852 patients with primary Sjogren’s syndrome in Taiwan. Ann Rheum Dis, 71(4), 524–​527. PMID: 22072014. Wiersinga WM. 2013. Smoking and thyroid. Clin Endocrinol (Oxf), 79(2), 145–​151. PMID: 23581474. Williams D. 2002. Cancer after nuclear fallout:  lessons from the Chernobyl accident. Nat Rev Cancer, 2(7), 543–​549. PMID: 12094241. Williams D. 2015. Thyroid growth and cancer. Eur Thyroid J, 4(3), 164–​173. PMCID: PMC4637514. Wolinski K, Czarnywojtek A, and Ruchala M. 2014. Risk of thyroid nodular disease and thyroid cancer in patients with acromegaly: meta-​analysis and systematic review. PLoS One, 9(2), e88787. PMCID: PMC3925168. Wong EY, Ray R, Gao DL, et  al. 2006. Reproductive history, occupational exposures, and thyroid cancer risk among women textile workers in Shanghai, China. Int Arch Occup Environ Health, 79(3), 251–​258. PMID: 16220287. World Health Organization. 1999. Guidelines for iodine prophylaxis following radiation accidents: update. Genevea: WHO. Xiao Q, Park Y, Hollenbeck AR, and Kitahara CM. 2014. Dietary flavonoid intake and thyroid cancer risk in the NIH-​AARP diet and health study. Cancer Epidemiol Biomarkers Prev, 23(6), 1102–​1108. PMCID: PMC4047159. Xing M, Liu R, Liu X, et al. 2014. BRAF V600E and TERT promoter mutations cooperatively identify the most aggressive papillary thyroid cancer with highest recurrence. J Clin Oncol, 32(25), 2718–​2726. PMID: 25024077. Xu L, Li G, Wei Q, El-​Naggar AK, and Sturgis EM. 2012. Family history of cancer and risk of sporadic differentiated thyroid carcinoma. Cancer, 118(5), 1228–​1235. PMCID: PMC3208119. Yamashita S. 2014. Tenth Warren K. Sinclair keynote address: the Fukushima nuclear power plant accident and comprehensive health risk management. Health Phys, 106(2), 166–​180. PMID: 24378490. Yamashita S, and Radiation Medical Science Center for the Fukushima Health Management S. 2016. Comprehensive health risk management after the Fukushima nuclear power plant accident. Clin Oncol (R Coll Radiol), 28(4), 255–​262. PMID: 26817782. Yu GP, Li JC, Branovan D, McCormick S, and Schantz SP. 2010. Thyroid cancer incidence and survival in the national cancer institute surveillance, epidemiology, and end results race/​ethnicity groups. Thyroid, 20(5), 465–​ 473. PMID: 20384488. Zablotska LB, Ron E, Rozhko AV, et al. 2011. Thyroid cancer risk in Belarus among children and adolescents exposed to radioiodine after the Chornobyl accident. Br J Cancer, 104(1), 181–​187. PMCID: PMC3039791. Zamora-​Ros R, Rinaldi S, Biessy C, et al. 2015a. Reproductive and menstrual factors and risk of differentiated thyroid carcinoma: the EPIC study. Int J Cancer, 136(5), 1218–​1227. PMID: 25041790. Zamora-​Ros R, Rinaldi S, Tsilidis KK, et al. 2015b. Energy and macronutrient intake and risk of differentiated thyroid carcinoma in the European Prospective Investigation into Cancer and Nutrition study. Int J Cancer, 138(1), 65–​73. PMID: 26190646. Zhang M, Li XM, Wang GS, et  al. 2014. Thyroid cancer in systemic lupus erythematosus: a meta analysis. Int J Clin Exp Pathol, 7(9), 6270–​6273. PMCID: PMC4203250. Zhu J, Zhu X, Tu C, et al. 2015. Parity and thyroid cancer risk: a meta-​analysis of epidemiological studies. Cancer Med, 5(4), 739–​752. PMID: 26714593. Zimmermann MB, and Galetti V. 2015. Iodine intake as a risk factor for thyroid cancer:  a comprehensive review of animal and human studies. Thyroid Res, 8, 8. PMCID: PMC4490680.

 861

45 Breast Cancer LOUISE A. BRINTON, MIA M. GAUDET, AND GRETCHEN L. GIERACH

OVERVIEW Breast cancer is the most frequently diagnosed cancer in women worldwide, with annual estimates of 1.7 million newly diagnosed cases and 522,000 deaths. Although more breast cancers are diagnosed in economically developed than developing countries, the reverse is true for mortality, reflecting limited screening and less effective treatments in such areas. Breast cancer incidence has been on the rise in the United States for many years. In recent years this is restricted to certain subgroups, while abroad there have been continued generalized increases, likely reflecting the adoption of more Westernized lifestyles. Breast cancer is widely recognized as being hormonally influenced, with most of the established risk factors believed to reflect the influence of cumulative exposure of the breast to the stimulatory effects of ovarian hormones—​leading to increased cellular proliferation, which in turn can result in genetic errors during cell division. To better understand the role of endogenous hormones as etiologic factors, studies have taken advantage of recent laboratory advances, including efforts to measure hormone metabolites. Although epidemiologically breast cancer has been one of the most intensively studied tumors, only about half of the disease occurrence is explained by well-​established risk factors. Most of the identified risk factors are not readily modifiable, leading to a need for additional research to better understand etiologic processes and to identify factors for which interventions might be feasible. During the past decade, there has been increasing appreciation that breast cancer is not one disease, but rather a collection of divergent diseases that have different biologic, clinical, and prognostic characteristics. Attention to such heterogeneity is essential for accurate risk prediction, as well as for advancing our understanding of biological processes and developing effective prevention modalities.

NORMAL BREAST DEVELOPMENT The breast is composed of adipose and glandular tissue and the surrounding stroma. The glandular tissue is a network of ducts that branch from the nipple to the terminal duct lobular units (TDLUs) (Figure 45–1) (Ali and Coombes, 2002). The TDLU contains a small segment of terminal duct and a cluster of acini (or alveoli), which are the secretory units of the breast. The mammary epithelial cells lining the glandular tissue are composed of two main cellular lineages: luminal cells that surround a central lumen and myoepithelial cells that are located in a basal position adjacent to the basement membrane. The collagen-​rich stroma includes the surrounding connective tissue, blood vessels, and lymph vessels (Russo and Russo, 2004). The mammary epithelium undergoes remodeling during distinct developmental stages (Russo and Russo, 2004). A female newborn has primitive terminal end buds that form the network of primary and secondary branches, which develop during puberty under the influence of ovarian hormones. In the adult gland, tertiary branching and small alveolar bud formation occur during each menstrual cycle. Before the first pregnancy, the breast epithelium is thought to be particularly susceptible to carcinogens because the TDLUs are not fully differentiated. The breast during pregnancy undergoes extensive expansion of the TDLUs; late in pregnancy, the epithelial cells differentiate to allow for lactation. During weaning, the TDLUs revert back to an earlier pre-​pregnancy state or completely disappear in a process called

post-​lactational involution. A similar process of involution also occurs as part of aging (Figueroa et al., 2014). Age-​related involution appears to start around the third decade of age and accelerates during the menopausal transition (Russo and Russo, 2004).

TUMOR CLASSIFICATION Breast cancer is a heterogeneous disease at the morphological and molecular levels. The fourth edition of the World Health Organization (WHO) Classification of Tumors of the Breast defines 21 distinct histological types based on cell morphology, growth, and architecture patterns (World Health Organization, 2012). Approximately 95% of invasive breast cancers diagnosed in the United States are adenocarcinomas. They originate in the epithelial tissue of the TDLUs and infiltrate the surrounding stroma. The majority of adenocarcinomas (75%–​80%) are classified as “invasive carcinoma of no special type” (NST) because they do not display sufficient histopathologic characteristics to warrant classification into the “special” subtypes of breast cancer. NST tumors prior to 2012 were labeled as “ductal carcinomas, not otherwise specified” (World Health Organization, 2003); however, the reference to “ductal” has been dropped because the term suggests that these tumors, as opposed to the other subtypes, specifically arise from the TDLUs. The “special” breast cancer subtypes are defined by their distinct morphology and clinical behavior. Lobular cancer, which comprises the largest special group and accounts for 5%–​15% of all invasive cancers, usually displays a distinctive pattern of single-​file cells that are not disruptive to the normal tissue architecture. Less frequently diagnosed histological subtypes include medullary (< 2% of diagnoses in the US), mucinous (1%–​4%), and tubular (< 2%) subtypes, among others (each < 1% of diagnoses) (World Health Organization, 2012). Histopathologic subtypes can be further classified by molecular characteristics. As the NST is the largest and most heterogeneous group of breast cancers (and these tumors have been the basis of most molecular studies), molecular characteristics are useful to discriminate unique clinical and biological subgroups, whereas the special histopathology tumor subtypes, with the exception of lobular cancers, tend to have more homogeneous molecular characteristics (Weigelt et  al., 2010). Since the 1970s, breast tumors have been classified at the molecular level using immunohistochemical (IHC) stains for estrogen receptor (ER) and progesterone receptor (PR) status. Tumors that express ER (ER+) or PR (PR+), collectively referred to as hormone receptor positive (HR+) tumors, are likely to respond to endocrine therapy, and have been shown to have different risk factor profiles than ER-​negative (ER–​) and PR-​negative (PR–​) tumors (as discussed in more detail throughout the chapter). In the 1990s, human epidermal growth factor receptor-​2 (HER2) status also began to be used to identify a subgroup of breast tumors with poor prognosis. The monoclonal antibody trastuzumab was developed as an adjuvant therapy to target the amplification and/​or overexpression of the HER2 gene (also known as ERBB2). Global gene expression profiling of tumors has revealed further heterogeneity at the molecular level. At least four “intrinsic” subtypes (luminal A, luminal B, HER2-​enriched, and basal-​like tumors) have been identified. With some important exceptions, the intrinsic subtypes recapitulate the historical subtypes defined by IHC staining

861

862

862

PART IV:  CANCERS BY TISSUE OF ORIGIN (a)

Figure  45–1. Anatomy of the human mammary gland. (a) Anatomy of the human mammary gland. Each mammary gland contains 15–20 lobes, each lobe containing a series of branched ducts that drain into the nipple. (b) Each duct is lined with a layer of epithelial cells, responsible for milk production. These are surrounded by an outer layer of myoepithelial cells with contractile properties. The glandular ducts are embedded in fibroblast stroma. (c) This structure breaks down in breast cancer, resulting in an epithelial cell mass. (b) and (c) are immunostained using antibodies to the estrogen receptor showing that only a small proportion of epithelial cells are ER positive in the normal breast. Source: Ali and Coombes, 2002.

Clavicle

(b) Skin

Rib

Epithelial cell

Suspensory ligaments

Muscle

Adipose tissue Stroma

Fibroblast

Loose connective tissue

Myoepithelial cell

(c) Nipple

Lobe

for ER, PR, and HER2 (Table 45–1). Although the subtypes differ in clinical presentation, response to therapy, and prognosis, the biological and etiological inferences are still an active area of research (Norum et  al., 2014). In the United States, the luminal A  subtype comprises the largest proportion of tumors (73%) (Howlader et al., 2014). The proportion of molecular subtypes differs in other countries due to differences in screening and distributions of age and other breast cancer risk factors (Carvalho et al., 2014). Luminal A tumors are usually low grade and are characterized at the molecular level by strong expression of the genes encoding ER (ESR1), PR (PGR), and other genes regulated by ER; by overexpression of cyclin D; by low expression of proliferation-​associated genes; and by lack of expression of HER2. They have a relatively low mutation rate, but phosphatidyl inositol-​ 4,5-​bisphosphate 3-​kinase catalytic subunit-​α (PIK3CA), GATA3, and mitogen-​activated protein kinase-​1 (MAPK1) are commonly mutated (Norum et al., 2014). Similar to luminal A tumors, luminal B cancers express proliferation genes as well as ER and PR, albeit at much lower levels. They also are distinguished from luminal A  tumors by overexpression of HER2, higher rates of proliferation, and higher grades at diagnosis (Ades et  al., 2014). They are more often aneuploidic than luminal A tumors, and display mutations in TP53 and PIK3CA genes (Norum

Alveolus

Duct

et al., 2014). The proportion of luminal B tumors is about 10% in the United States (Howlader et al., 2014). Luminal A and basal-​like subtypes are the most contrasted groups at every level, including etiology, clinical presentation, response to treatment, and prognosis. Basal-​like tumors present with high grades, necrosis, prominent lymphocytic infiltrates, and pushing borders. On the molecular level, basal-​like tumors share more similarity with tumors arising in the basal layer of the epidermis, such as squamous carcinomas of the lung or head and neck as well as epithelial ovarian tumors. They express cytokeratins, epidermal growth factor receptor (EGFR), and other genes commonly expressed in basal/​myoepithelial cells, and do not express ER, PR, or HER2 (collectively referred to as “triple negative”). They are frequently aneuploidic with complex genomic rearrangements and have somatic mutations in TP53. BRCA1 mutation carriers are more likely to be diagnosed with basal-​ like tumors than other subtypes, suggesting a strong role of double-​ stranded DNA repair mechanisms in their development (Norum et al., 2014). In the United States, the proportion of basal-​like cancer diagnosed is 12% (Howlader et al., 2014). The proportion of HER2-​ enriched tumors is nearly 5% in the United States (Howlader et  al., 2014). The HER2-​enriched tumors are characterized by overexpression of HER2 and other genes in the

Table 45–1.  Clinical and Molecular Features and Immunohistochemical Definitions for the Common Intrinsic Subtypes of Breast Cancer Gene Expression Intrinsic Subtype

Grade

ER

PR

Luminal A

Low

High

Some, high

None

Luminal epithelial genes, cyclin D1

Luminal B

High

Low

Low or none

Some, overexpression

Ki-​67

Basal-​like

High

None

None

None

HER2 enriched

High

None

None

High

HER2

Expression of Other Genes

Immuno-​ Histochemical Definition

Somatic Mutations

Chromosomal Aberrations

ER+ and/​or PR+, HER2–​, Ki-​67 low ER+ and/​or PR+, Ki-​67 high

PIK3CA, GATA3, MAPK3 TP53 and PIK3CA, RB1, MAPK

Diploid with whole arm aberrations

Basal cytokeratins, EGFR

ER–​, PR–​, HER2–​ (triple negative)

TP53

Genes on 17q22

ER–​, HER2+ (amplified or overexpressed)

TP53

Whole arm aberrations and complex rearrangements Aneuploid

Focal high-​level amplifications

Histological Special Type Lobular, tubular, mucinous, neuroendocrine Lobular, micropapillary Secretory, adenoid cystic, medullary, metaplastic, acinic cell Lobular, apocrine, micropapillary

Abbreviations: EGFR = epidermal growth factor receptor; ER = estrogen receptor; GATA3 = GATA binding protein 3; HER2 = human epidermal growth factor receptor; MAPK = mitogen-​activated protein kinase; PIK3CA = phosphatidylinositol-​4,5-​bisphospate 3-​kinase catalytic subunit α; PR = progesterone receptor; TP53 = tumor protein p53. Adapted from Norum et al. (2014); Weigelt et al. (2010).

 863

Breast Cancer same chromosomal region. They are high grade at diagnosis and lack expression of ER and PR (Norum et al., 2014). Although lobular cancers encompass all intrinsic molecular subtypes, luminal A tumors predominate (Dieci et al., 2014). The distinct histopathology of lobular cancer is the result of the loss of cellular membrane expression of E-​ cadherin (encoded by CDH1), which causes the dysregulation of cell-​cell adhesion properties (McCart Reed et al., 2015). The rare histological subtypes are more homogeneous at the transcriptomic level than invasive carcinoma of NST, and each subtype clusters into only one or two molecular subtypes (Weigelt et al., 2010). Tubular and mucinous carcinomas are typically characterized as luminal tumors. Adenoid cystic, medullary, and metaplastic carcinomas display the basal-​like phenotype. Metaplastic tumors are also found in the recently characterized claudin-​low subtype. Apocrine carcinomas cluster in the HER2-​enriched and molecular apocrine subtypes. Micropapillary carcinomas display characteristics of luminal and HER2-​enriched tumors (Weigelt et al., 2010). While these tumor subgroups have been widely accepted in clinical and research settings, they are not highly reproducible (Bombonati and Sgroi, 2011). Gene expression definitions of subgroups are dependent on the analytical method used, although most clusters result in similar ability to predict outcomes. It is likely that broad biological processes (e.g., cell proliferation) are driving each of the tumor subtypes, and these processes can be characterized by the expression of a number of different genes. Furthermore, although a gene expression array of a minimized set of genes for breast cancer is commercially available (e.g., PAM50), its use currently is not widespread in clinical and most research settings. While surrogate definitions of these subtypes based on the results of IHC staining for ER, PR, HER2, and the basal markers are widely used in clinical and research settings, there are issues with accuracy due to problems with marker staining (Welsh et  al., 2011) and precision (Jenkins et al., 2014). Future research that integrates gene expression data with other molecular profiling techniques (e.g., somatic mutations, copy number, epigenetic alterations) is likely to result in deeper understanding of subtypes in breast and other anatomical tissues (Dieci et al., 2014). Intratumoral heterogeneity presents further complexities in the identification of the molecular subtype of a tumor, and itself might also account for variability in prognosis, treatment response, and factors related to tumor initiation, promotion, and progression (Norum et al., 2014). Intratumoral heterogeneity is the coexistence of subclones of cancer cells that differ in histology, genetic sequence, epigenetic patterns, or protein expression. The heterogeneity might occur spatially or temporally. Subpopulations of tumor cells develop when changes in the genetic sequence or epigenetic patterns provide a selective advantage in the surrounding microenvironment. It is unclear whether the establishment and maintenance of tumor heterogeneity are due, in part, to different cells of origin (Martelotto et al., 2014).

CELL OF ORIGIN Two theories of breast carcinogenesis currently are debated in the literature (Visvader and Stingl, 2014):  the sporadic clonal evolution model (stochastic) and the cancer stem cell (hierarchical) theories. The sporadic clonal evolution model posits that any breast epithelial cell can be the target of random mutations. The cells with advantageous genetic and epigenetic alterations are selected over time to contribute to tumorigenesis. In contrast, the cancer stem cell theory postulates that stem cells, which either gave rise to the tumor or were acquired by a subpopulation of cells within the tumor, maintain tumorigenesis through their capacity for self-​renewal and differentiation. The remaining bulk of tumor cells have limited proliferative potential. In support of the cancer stem cell theory, researchers have proved the existence of mammary stem cells with multilineage differentiation and self-​renewing capabilities that are responsible for the dynamic nature of the mammary epithelium throughout the life course (Visvader and Stingl, 2014). Although the lineage of mammary stem cells in the adult breast is an active area of research, it currently is thought that

863

the mammary stem cell compartment contains long-​term and short-​ term repopulating cells. These cells give rise to committed progenitor cells for the myoepithelial and luminal (ductal and alveolar) epithelial lineages. The epithelial cellular hierarchy appears to be critical to understanding the different cells of origin for each molecular subtype. There is strong evidence to support that the claudin-​low molecular subtype is derived from mammary stem cells, while the basal-​like subtype is derived from luminal progenitor cells. It is suspected, but unconfirmed at this time, that luminal A, luminal B, and HER2-​enriched tumors develop from more mature luminal progenitor cells (Visvader and Stingl, 2014). It is expected that the deeper understanding of cellular hierarchy in the normal breast tissue will be an active area of research and will lead to a better understanding of the carcinogenesis of breast cancer molecular subtypes.

PRECANCERS OR PRECURSOR LESIONS While about a quarter of diagnostic breast biopsies in the United States yield invasive diagnoses, the vast majority of biopsy-​based diagnoses range from benign to preinvasive disease (Weaver et  al., 2006). Benign lesions and in situ carcinomas (lobular carcinoma in  situ [LCIS] and ductal carcinoma in situ [DCIS]) are a morphologically and biologically heterogeneous group of lesions associated with varying degrees of subsequent breast cancer risk (Morrow et al., 2015). A classification scheme for benign breast disease (BBD), proposed by Dupont and Page (1985) and endorsed by the College of American Pathologists (Fitzgibbons et  al., 1998), categorizes BBD lesions into three clinically relevant groups:  nonproliferative, proliferative without atypia, and atypical hyperplasia (Hartmann et  al., 2005) (Figure 45–2). Large epidemiologic cohorts of patients diagnosed with BBD have established a relation between the histologic classification of BBD and breast cancer risk (Morrow et al., 2015). In a retrospective cohort of 9087 women diagnosed with BBD at the Mayo Clinic who were followed for a median of 15 years, the relative risks (RRs) in comparison to the general population were 1.27 (95% confidence interval [CI]: 1.15–​1.41) for nonproliferative lesions, 1.88 (1.66–​2.12) for proliferative changes without atypia, and 4.24 (3.26–​5.41) for atypia (Hartmann et  al., 2005). Results from a nested case-​control study of BBD and breast cancer risk in the Nurses’ Health Study (NHS) also demonstrated that risks were highest among those with atypical hyperplasia: compared with women who had nonproliferative lesions, the ORs associated with proliferative lesions without atypia and atypical hyperplasia diagnoses were 1.62 (95% CI: 1.21–​2.18) and 4.04 (2.76–​5.92), respectively (Collins et al., 2006). Atypical epithelial hyperplasia encompasses two histologically distinct lesions—​atypical ductal hyperplasia and atypical lobular hyperplasia—​and both lesions are associated with approximately a 4-​fold increased risk of breast cancer (Morrow et al., 2015). In a 2014 update from the Mayo BBD Cohort, 698 women with atypical hyperplasia were followed for an average of 12.5  years, and subsequent breast cancers occurred with a 2:1 ratio in the ipsilateral compared with the contralateral breast. This ipsilateral predominance was marked in the first 5  years, suggesting that atypical hyperplasia lesions are in fact cancer precursors in some women (Hartmann et al., 2014). However, most women with atypical hyperplasia do not develop breast cancer (Morrow et al., 2015). In the Mayo BBD Cohort, the 20-​year cumulative incidence of DCIS or invasive breast cancer among women diagnosed with atypical hyperplasia was 21% (95% CI:  14%–​28%) (Degnim et  al., 2007). Thus, identifying factors that modify cancer risk associated with these lesions is of great interest. Younger age or premenopausal status at diagnosis of atypical hyperplasia, multiple foci of atypia (Hartmann et al., 2014), and reduced TDLU involution in women with BBD (Baer et  al., 2009; Milanese et  al., 2006)  (see later discussion) have all been associated with increased breast cancer risk. Although a positive family history of breast cancer was initially reported to increase the RR of breast cancer among women with atypical hyperplasia (Dupont and Page, 1985), this has not been

864

864

PART IV:  CANCERS BY TISSUE OF ORIGIN (a)

(b)

(c)

(d)

(e)

(f)

Figure 45–2.  Histopathological appearance of benign breast disease. Panel (a) shows nonproliferative fibrocystic changes: the architecture of the terminal duct lobular unit is distorted by the formation of microcysts, associated with interlobular fibrosis. Panel (b) shows proliferative hyperplasia without atypia. This is adenosis, a distinctive form of hyperplasia characterized by the proliferation of lobular acini, forming crowded gland-like structures. For comparison, a normal lobule is on the left side. Panel (c) also shows proliferative hyperplasia without atypia. This is moderate ductal hyperplasia, which is characterized by a duct that is partially distended by hyperplastic epithelium within the lumen. Panel (d) again shows proliferative hyperplasia without atypia, but this is florid ductal hyperplasia: the involved duct is greatly expanded by a crowded, jumbled-appearing epithelial proliferation. Panel (e) shows atypical ductal hyperplasia: these proliferations are characterized by a combination of architectural complexity with partially formed secondary lumens and mild nuclear hyperchromasia in the epithelial-cell population. Panel (f) shows atypical lobular hyperplasia: monotonous cells fill the lumens of partially distended acini in this terminal duct lobular unit. Source: Hartmann et al. (2005).

confirmed in more recent investigations (Collins et al., 2006; Hartmann et al., 2014). While atypical ductal and lobular hyperplasia have been recognized as breast cancer risk factors since the 1980s, flat epithelial atypia, an alteration of breast lobules, has more recently been recognized by the WHO Working Group on the Pathology and Genetics of Tumors of the Breast and Female Genital Organs (World Health Organization, 2003). The natural history of flat epithelial atypia is not well understood, but emerging epidemiologic evidence suggests that the breast cancer risk associated with flat epithelial atypia is not as high as that observed with other atypical breast lesions (Morrow et al., 2015). A recent study from the Mayo BBD Cohort found that flat epithelial atypia was accompanied by atypical hyperplasia about half of the time, and the RR of breast cancer associated with isolated flat epithelial atypia (RR = 2.75, 95% CI: 1.76–​4.10) was comparable to the risk associated with proliferative disease without atypia (Said et al., 2014). LCIS and DCIS are both associated with increased risk of subsequent breast cancer. DCIS is considered to be a non-​obligate precursor lesion to invasive breast cancer (Sherman et al., 2014), whereas LCIS is generally thought to be a more general marker of risk (Morrow et al., 2015); however, molecular studies suggesting a clonal link with invasive lobular carcinoma (Venkitaraman, 2010)  have renewed interest in LCIS as a non-​obligate precursor lesion as well as a risk indicator. Whereas the incidence of DCIS has increased dramatically with mammographic screening, manifesting generally as clustered calcifications (Virnig et al., 2010), LCIS is typically an incidental finding in breast biopsies, as it lacks clinical manifestations, such as a lump or other changes to the breast, and may not be detectable by screening mammography (Morrow et al., 2015). The RR of invasive breast cancer subsequent to a diagnosis of LCIS ranges from ~7–​10, and the RRs from more contemporary studies are similar to those reported in the 1970s (Morrow et al., 2015). The lack of untreated cohorts with DCIS is a challenge for calculating the RR of invasive breast cancer associated with DCIS. Among women diagnosed with DCIS who are treated with excision only, the risk of subsequent invasive breast cancer appears to exceed that seen with a prior LCIS diagnosis (Morrow et al., 2015). In addition, the risk of invasive

breast cancer is highest in the same breast as the initial DCIS diagnosis, and the cancer often arises in the same breast quadrant (Erbas et al., 2006). While the evidence for DCIS as an invasive precursor is strong, DCIS is not an obligate precursor; in autopsy studies, the prevalence of undiagnosed DCIS is estimated to be around 9% (Welch and Black, 1997). A  recent systematic review has shown that DCIS and invasive breast cancer share similar risk factor associations, including elevated mammographic density, family history of breast cancer, and history of benign breast disease, supporting the idea that they share a common etiology (Virnig et al., 2010). As these high-​risk lesions are diagnosed more frequently in the current era of broad mammography screening programs, individualized risk prediction is needed. This includes the identification of molecular biomarkers that predict progression to invasive carcinoma, given that such efforts to date have been unsuccessful (Allred, 2011). Until predictors of invasive carcinoma risk are found, high-​risk precursor lesions pose a dual problem of overdiagnosis (and over-​treatment) among some women and failure of early detection (or undertreatment) among others.

DESCRIPTIVE EPIDEMIOLOGY Demographic Factors Age Unlike cancers at other sites, which generally show a log-​linear relationship with age, the age incidence curve for breast cancer is log-​ linear only until about age 50, after which time the slope begins to flatten out. The change point around age 50 is referred to as the “Clemmesen’s hook.” To account for the peculiar shape of the breast cancer age-​specific rate curve, Pike et al. (1983) proposed a concept of “breast tissue age” to reflect biological rather than chronological age based on key reproductive events. Risk factors accelerated breast cancer incidence rates, while protective factors retarded rates, culminating in a pause or inflection near menopause. Others have refined the Pike model, but none of these models incorporates the concept of breast cancer heterogeneity, nor can these models fully explain the distinct

 865

865

Breast Cancer age-​specific incidence rate patterns by histopathologic and/​or molecular subtypes (e.g., the different patterns for ER–​and ER+ cancers). Age-​specific rates for ER–​cancers rise rapidly until age 50 years, and then plateau, decline, or fall. In contrast, age-​specific rates for ER+ cancers rise continuously with advancing age, though more slowly after 50 years of age. Alternatively, the distinct ER–​and ER+ age-​specific incidence rate patterns suggest that breast cancer overall consists of not one type but rather a mixture of two main types of cancer, with a different incidence rate pattern for each type. The first type is premenopausal with peak occurrence early in life (similar to ER–​cancers). The second type is postmenopausal with peak incidence later in life (similar to ER+ cancers). Clemmesen’s hook can then be viewed as the confluence or superimposition of falling early-​onset ER–​and rising late-​ onset ER+ breast cancers. Notably, and somewhat paradoxically, the falling ER–​incidence rates near age 50 years suggest that menopause (and by implication, exposures that occur early in reproductive life) have a greater impact on ER–​than ER+ cancers. Further evidence for two main breast cancer types is provided by the corresponding age distribution at diagnosis for breast cancer overall, which demonstrates a bimodal pattern with modal or peak ages near 50 and 70 years (Anderson et al., 2014). Bimodality is of interest because it implies heterogeneity in an otherwise homogeneous population. ER–​and ER+ cancers also have bimodal age distributions at diagnosis. ER–​cancers have a bimodal pattern with a dominant early mode near age 50 years and minor mode near age 70 years, whereas ER+ cancers have a bimodal pattern with a dominant mode near age 70 years and minor mode near age 50 years. In recent analyses that considered not only ER and PR but also HER2 (which has recently been incorporated into Surveillance Epidemiology and End Results [SEER] data), patients with triple

HR+/HER2–

negative, HR+/​HER2+ and HR–​/​HER2+ breast cancers were 10%–​ 30% less likely to be diagnosed at older ages compared with HR+/​ HER2–​patients. Notably, incidence rates for HR+/​HER2–​tumors peaked at 75–​79  years of age, while triple negative cancers peaked prior to age 70 years; in contrast, HER2–​overexpressing tumors (both HR+/​HER2+ and HR–​/​HER2+) showed less dramatic increases in incidence with age than either of the other tumor subtypes (Howlader et al., 2014).

Sex

Although breast cancer is predominately a female disease, it does occur rarely among men, who exhibit an incidence rate that is approximately 1/​ 100th that of females. Recent investigations have demonstrated that risk factors among men appear relatively similar to those among women (Brinton et al., 2014a), including showing a strong relation with higher levels of endogenous estrogens (Brinton et al., 2015). Because of the rarity of breast cancer in males, this chapter will focus on the disease that occurs much more commonly among females, where there has been extensive study of patterns of disease and risk factors.

Race, Ethnicity, and Socioeconomic Status

Breast cancer incidence rates show significant heterogeneity by race and ethnicity. The latest available SEER-18 statistics in the United States (SEER, 2015) show that, in general, among non-​Hispanics rates are higher for white than black women (respective incidence rates per 100,000 of 135.3 vs. 125.0), but the reverse is true at younger ages, where there is a crossover and higher rates among blacks (Jatoi and Anderson, 2010). Among Hispanics, there are also differential rates between whites and blacks (respective rates of 95.7 vs. 55.8). In addition, rates are relatively low among Asian/​Pacific Islanders (94.9) and American Indian/​Alaskan Natives (61.1).

HR+/HER2+

Triple-negative

HR–/HER2+

350

350

350

350

300

300

300

300

NH white NH black

250

200

200

200

200

150

150

150

150

100

100

100

100

50

50

50

50

0

0

0

0 20–24 25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64 65–69 70–74 75–79 80–84

250

20–24 25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64 65–69 70–74 75–79 80–84

250

20–24 25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64 65–69 70–74 75–79 80–84

250

20–24 25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64 65–69 70–74 75–79 80–84

Rate per 100,000 US women

NH API Hispanic

Age at diagnosis (years)

Figure 45–3.  Average annual incidence of breast cancer per 100,000 women by age group for different subtypes of breast cancer. Age-specific incidence rates of breast cancer subtypes by race/ethnicity, Surveillance, Epidemiology, and End Results 18, excluding Alaska, 2010. Abbreviations: API = Asian Pacific Islander; HER = human epidermal growth factor; HR = hormone receptor; NH = non-Hispanic. Source: Howlader et al. (2014).

86

866

PART IV:  CANCERS BY TISSUE OF ORIGIN

Racial and ethnic heterogeneity has also been observed for different subtypes of breast cancer, with non-​Hispanic white women having the highest incidence rates of the HR+/​HER2–​subtype, and non-​Hispanic black women having the highest rates of triple negative cancers. Compared with women with the HR+/​HER2–​subtype, triple negative patients are more likely to be non-​Hispanic black and Hispanic, HR+/​HER2+ patients are more likely to be Asian/​Pacific Islander, and HR–​/H ​ ER2+ patients are more likely to be non-​Hispanic black, Asian/​ Pacific Islander, and Hispanic (Howlader et al., 2014) (Figure 45–3). These ethnic differences do not appear to be explained by socioeconomic strata differences (Sineshaw et  al., 2014). However, it is increasingly being recognized that distributions across racial and ethnic groups are complex, with considerable diversity of subtypes even within populations, such as Asians (Parise and Caggiano, 2014), as well as within other population subgroups. Breast cancer incidence has been noted to be highest among single women and women of higher socioeconomic status, presumably reflecting, at least in part, the influence of more prevalent risk factors (such as fewer children/​delays in childbearing and greater exposure to menopausal hormone therapy) and greater access to screening. Such differences in prevalence of risk factors may also partially explain the racial and ethnic differences.

GEOGRAPHIC VARIATION The United States has one of the highest incidence rates of breast cancer in the world. Although there is some geographic variation within the country (Siegel et  al., 2015), it is small in comparison to international variation, with much of the national fluctuations presumably due to differences in the prevalence of established breast cancer risk factors.

International Patterns of Incidence and Mortality Breast cancer is the most frequently diagnosed cancer and the leading cause of cancer death worldwide, with an estimated 1.7  million cases diagnosed during 2012 (comprising 25% of all cancers) (Torre et al., 2015). More developed countries account for about one-​half of all breast cancer cases. Incidence rates vary more than 4-​fold across world regions, with annual standardized incidence rates ranging from 27 per 100,000 in Middle Africa and Eastern Asia to 96 in Western Europe. Rates are generally high in North America, Australia/​New Zealand, and northern and western Europe; intermediate in central and eastern Europe, and the Caribbean; and low in most of Africa and Asia (Figure 45–4). Breast cancer ranks as the fifth cause of death from cancer overall (522,000 deaths). While it is the most frequent cause of death in women in less developed regions (324,000 deaths, 14.3% of total), it is now the second cause of cancer death, after lung cancer, in more developed regions (198,000 deaths, 15.4%) (Ferlay et al., 2015). Mortality differences across countries are not as substantial as incidence differences because of the more favorable survival of breast cancer in developed regions, with rates ranging from 6 per 100,000 in eastern Asia to 20 per 100,000 in western Africa. In many developing areas, including Africa, large proportions of women present with late stage tumors, for which treatment modalities are ineffective (Brinton et al., 2014b); further, in many of these countries, treatments such as radiotherapy and chemotherapy are not always available. Thus, a major challenge for many developing countries is in educating women to seek medical assistance at the first signs of symptoms to assure that the most effective treatment modalities can be utilized. Limited screening in many countries (largely reflecting an absence of mammography facilities) is a further impediment to early detection.

Migrant Studies An interesting facet of breast cancer is that incidence rates change when individuals migrate from low incidence countries to high

incidence countries. Among Asian women, it has been documented that when they migrate to the United States that within two to three generations their rates increase to or become even greater than US whites (Hoover, 2012). This pattern of change supports a more important role for environmental than genetic factors. Although initially women in the West were noted to have breast cancer incidence rates 4-​to 6-​fold higher than women in certain Asian countries, more recently—​in response to the “Westernization” of certain countries—​this difference dropped to 2-​to 3-​fold. In a cohort of women from Shanghai, it has been demonstrated that a 2.8-​fold difference in incidence with US whites decreased to a 1.4-​fold difference after adjustment for a variety of breast cancer risk factors (Linos et al., 2008). Given changes in migration patterns to the United States, attention has also focused on how breast cancer incidence rates change among Hispanics when they leave their host countries to reside in the United States. Analyses indicate that breast cancer risk is 50% lower in foreign-​born Hispanics than US-​born Hispanics (John et al., 2005). Among long-​ term foreign-​ born residents, risk was lower among Hispanics who moved to the United States at age 20 or older and those who mainly spoke Spanish. Further, the difference between third-​or higher-​generation Hispanics and recent migrants from rural areas was approximately 6-​fold in postmenopausal women and 4-​fold in premenopausal women. Adjustment for breast cancer risk factors attenuated these relationships.

TEMPORAL TRENDS In the United States, incidence rates of breast cancer have been gradually increasing. Between 1980 and 1987, a time when there was increased uptake of mammography, breast cancer incidence rates increased rapidly (Figure 45–5), followed by a slower rate of increase between 1987 and 2002 (Siegel et  al., 2015), when rates declined sharply—​presumably in response to declining rates of menopausal hormone therapy (MHT) usage following publication of findings from the Women’s Health Initiative (WHI) clinical trial linking estrogen plus progestin therapy to increases in the development of breast cancer (Ravdin et al., 2007). Since that time, incidence rates have been more stable, although with some upswings for older (≥ 50  years) women, particularly blacks. Recent projections (Rosenberg et  al., 2015)  indicate that the total number of new tumors in the United States should rise from 283,000 in 2011 to 444,000 by 2030. This largely reflects proportional increases in older women and for ER+ in situ cancers; in contrast, the proportion of ER–​tumors is expected to decrease. Increasing rates of ER+ tumors are primarily thought to reflect changing prevalences in breast cancer risk factors, including greater delays in childbirth and more obesity; reasons for declining rates of ER–​tumors are less clear. Globally, breast cancer incidence has been continually increasing. Between 1990 and 2013, incidence rates have increased 17% globally:  46% in developing countries and 8% in developed countries (Global Burden of Disease Cancer, 2015). Some of the most rapid increases in incidence have occurred in South America, Africa, and Asia (Torre et al., 2015), although the absence of standardized cancer registration in many countries leads to questions as to the accuracy of estimates. These increases are believed to be the result of changing reproductive patterns, increasing obesity, decreasing physical activity, and some breast cancer screening activity. The convergence toward the risk profile of Western countries has resulted in a narrowing of the international gap in breast cancer incidence. In the United States, mortality from breast cancer increased during the 1930s, remained relatively stable from the 1940s through the 1970s, increased 0.4% per year from 1975 to 1990, and then decreased 36% from 1990 to 2012. The decrease occurred in both younger and older women, but since 2007 the breast cancer death rate has been level among women younger than 50 years of age. These trends are difficult to interpret given that they reflect combined effects of changes in incidence (e.g., ER–​rates have recently fallen), variations in screening practices, and effectiveness of treatment. In recent modeling efforts

 867

64.8+ 45.8 10 versus < 1 mg/​L (Ose et al., 2015c). Most individual studies were unable to evaluate associations by tumor subtype, although in one study, CRP was more strongly associated with risk of serous tumors among obese women (Ose et  al., 2015c). Results for other inflammatory markers have been less consistent, although some studies reported suggestive

Study

First Author, Year (Reference No.)

associations for IL-​6, TNFα, and TNFα receptor 2 (Clendenen et al., 2011; Ose et al., 2015c; Poole et al., 2013a; Trabert et al., 2014b).

Dietary Intake Coffee/​Caffeine

Caffeine may increase ovarian cancer risk through several mechanisms. It has been hypothesized to affect not only the metabolism of DNA precursors, but the structure and function of DNA as well. In addition, it may affect hormone levels hypothesized to play an etiologic role in ovarian cancer (Kotsopoulos et  al., 2009; Kuper et  al., 2000). Two prospective studies have assessed total caffeine intake and ovarian cancer (Lueth et al., 2008; Tworoger et al., 2008a). In one study, no association was seen, while in the second a significant trend of decreasing ovarian cancer risk with increasing caffeine intake was observed, although significant associations were not observed for any of the individual categories of intake (Tworoger et al., 2008a). The 2014 World Cancer Research Fund (WCRF) Report concluded that there was limited and inconsistent evidence for the association between coffee consumption and ovarian cancer (WCRF, 2014). The majority of prospective cohort studies have observed no association between coffee consumption and ovarian cancer (Braem et al., 2012; Larsson and Wolk, 2005; Steevens et al., 2007). Only two prospective cohort studies have reported significant results for coffee intake. In the Nurses’ Health Study, there was a significant inverse trend across caffeinated coffee categories; however, again none of the individual categories was significant (Tworoger et al., 2008a). Additionally, the Iowa’s Women Health Study reported a significant increased risk of ovarian cancer with drinking five or more cups/​day of caffeinated coffee versus non-​drinkers, but observed no association for total coffee or decaffeinated coffee (Lueth et al., 2008).

Tea

Flavonoids found in tea, beverages including fruits, vegetables, wine, and other foods may have chemopreventive effects through the regulation of multiple inflammation pathways (Lee et  al., 2011; Neergheen et al., 2010; Vanden Berghe, 2012). An early prospective study observed a suggestive decreased risk of ovarian cancer for two flavonoid subclasses, kaempferol and luteolin intake (Gates et  al., 2007), although another smaller prospective study did not observe any

No. of No. of Cases Controls

RR (95% Cl)

Nested Case-Control Analyses CLUE I, CLUE II, Columbia MO, Guernsey Phase I, Guernsey Phase II

McSorley, 2007 (16)

166

334

1.72 (1.06, 2.77)

NSHDS, NYUWHS, ORDET

Lundin, 2009 (17)

237

427

1.15 (0.74. 1.79)

FMC

Toriola, 2011 (18)

170

170

1.62 (0.93, 2.83)

NHS, NHS II

Present study

263

518

1.15 (0.74, 1.77)

Subtotal (I2 = 0%, 95%, Cl: 0, 100)

1.35 (1.06, 1.71)

Cohort Analysis WHS

Present study

159

27,746

1.38 (0.88, 2.18)

Overall (I2 = 0%, 95%, Cl: 0, 100)

1.35 (1.10, 1.67) 0.5

1

1.5 2 Relative Risk

Figure 46–​4.  Meta-​analysis of CRP and ovarian cancer risk in five cohort studies. Source: Poole et al. (2013).

2.5

3

894

894

PART IV:  Cancers by Tissue of Origin

association with flavonoids or related foods (Wang et al., 2009). An update of the Gates study reported no association for total flavonoids but did note a modest protective effect in the top versus bottom 20% of intake for both flavonols (HR = 0.76; 95% CI: 0.59–​0.98) and flavanones (HR = 0.79; 95% CI: 0.63–​1.00), although no dose–​response relationship was observed (Cassidy et  al., 2014). Additionally, the association with flavanone intake was stronger for serous invasive and poorly differentiated tumors compared to non-​serous tumors (Cassidy et al., 2014). When considering specific flavonoid-​rich foods, the only significant association was observed for tea intake of >1 cup/​day versus < 1 cup/​week (HR = 0.69; 95% CI: 0.52–​0.93), although there were suggestive inverse associations for celery and oranges/​orange juice. Interestingly, a recent meta-​analysis including data from five prospective cohort studies and seven case-​control studies reported similar results, with significant associations for isoflavones (RR = 0.67; 95% CI: 0.50–​0.92) and flavonols (RR = 0.68; 95% CI: 0.58–​0.80) as well as a suggestive association for flavones (RR  =  0.86; 95% CI:  0.71–​ 1.03) (Hua et al., 2016). Despite the suggestion of a decreased risk of ovarian cancer with certain subclasses of flavonoid intake and tea discussed in the preceding, the 2014 World Cancer Research Fund report concluded that there is limited and inconsistent evidence for the association between tea and ovarian cancer risk (WCRF, 2014). In the majority of prospective studies and one recent meta-​analysis, no significant association was reported between tea consumption and ovarian cancer (Braem et al., 2012; Steevens et  al., 2007; Zhang et  al., 2015), although a meta-​ analysis specifically examining green tea consumption did observe an inverse relationship (Gao et al., 2013).

Alcohol

Consumption of alcohol does not appear to be associated with ovarian cancer (Genkinger et al., 2006c; Kelemen et al., 2013; Yan-​Hong et al., 2015). A recent meta-​analysis of 13 prospective cohort studies and a pooled analysis of 12 case-​control studies both reported no association of alcohol consumption overall or by type with ovarian cancer risk overall or for specific histologic types (Kelemen et al., 2013; Yan-​ Hong et al., 2015).

Lactose/​Galactose

Early studies on lactose/​galactose consumption and ovarian cancer risk were conflicting, but overall there was a suggestion of a positive association (Hankinson and Danforth, 2006). Dietary lactose and galactose are hypothesized to influence ovarian cancer risk through either direct toxicity to oocytes or stimulating secretion of gonadotropins (Cramer et  al., 1989). However, recent research has observed no association between the intake of dairy products and ovarian cancer, except for the suggestion of an increased risk for very high lactose intake (e.g., RR, > = 30 vs. < 10 g/​d = 1.19; 95% CI: 1.01–​1.40) (Genkinger et al., 2006a; Liu et al., 2015; Merritt et al., 2013a).

Vitamins

Multivitamin intake in adulthood has not been associated with ovarian cancer risk (Koushik et al., 2015; Neuhouser et al., 2009). In the same pooled analysis of 10 prospective cohort studies, no association was observed for intakes of vitamins A, C, and E or folate for ovarian cancer overall or by subtype (Koushik et al., 2015). Additionally, dietary intake of major carotenoids was not associated with ovarian cancer (Koushik et al., 2006).

Fat

Dietary fat has been hypothesized to alter ovarian cancer risk by increasing circulating estrogen levels, which increases cell proliferation (Hill et al., 1971). Additionally, animal studies have reported that the female offspring of mice fed high-​fat diets in pregnancy had more reproductive tumors than the offspring of mice fed low-​fat diets (Walker, 1990). However, results from studies in humans are inconclusive. In the only randomized controlled trial, a low-​fat /​ high–​fruit, vegetable, and grain dietary intervention compared with usual diet was associated with a statistically significant reduction in ovarian cancer risk in the last 4 years of follow-​up (HR = 0.50;

95% CI: 0.38–​0.96); however, this was based on relatively few cases and the dietary intervention was not specific to fat intake (Prentice et  al., 2007). Among observational studies, a pooled analysis of 12 cohorts reported no association between fat intake and ovarian cancer risk, although there were marginally significant associations comparing the highest versus lowest decile of animal (RR = 1.23, 95% CI:  0.98–​ 1.55) and saturated fat intake (RR  =  1.29; 95% CI:  1.01–​1.66) (Genkinger et  al., 2006b). Recent individual prospective studies generally have observed associations with some types of fat, although the type is inconsistent across studies (Blank et al., 2012; Gilsing et al., 2010; Merritt et al., 2014). However, two studies reported a modest positive association for polyunsaturated fat (Blank et al., 2012; Merritt et al., 2014). Finally, a recent meta-​ analysis observed a very modest association between ovarian cancer and intake of red and processed meats, although the association was not significant when considering these types of meats separately (Wallin et al., 2011).

Diet Patterns

Many studies have attempted to link intakes of individual foods, food groups, or nutrients and the risk of ovarian cancer (reviewed in Crane et  al., 2014). However, foods are consumed in combination, and their combined effect on disease risk cannot be predicted from individual effects. Three prospective studies have assessed dietary patterns and ovarian cancer risk (Chang et al., 2007; Romaguera et al., 2012; Xie et al., 2014). No association was noted between the three diet quality scores and ovarian cancer risk in one study (Xie et  al., 2014). Additionally, no association was observed with ovarian cancer for adherence to the World Cancer Research Fund and American Institute of Cancer Research recommendations, which also include physical activity and weight management (Romaguera et al., 2012). In contrast, the California Teachers Study (n = 311 cases) identified five dietary patterns: (1) plant-​based, (2) high protein /​high fat, (3) high carbohydrate, (4) ethnic, and (5) salad /​wine (Chang et al., 2008). The plant-​based diet was positively associated with risk (RR, highest vs. lowest quintile  =  1.65; p-​trend  =  0.03); however, this was attributed to residual confounding as women with high plant intake had more known ovarian cancer risk factors (e.g., PMH use). The other patterns were not associated with risk (Chang et al., 2007).

Body Size A recent meta-​ analysis of 17 prospective cohort studies reported a modest positive association between body mass index (BMI) and ovarian cancer (Beral et  al., 2012b). Each 5  kg/​m2 increase in BMI was associated with a 3% (95% CI: 0%–​6%) increase in ovarian cancer risk (Beral et al., 2012b). There did not appear to be heterogeneity across subtypes of ovarian cancer, except for a stronger positive association for serous borderline compared to serous invasive tumors (Beral et al., 2012b). In contrast, a pooled analysis of 12 prospective cohort studies observed no association between BMI and ovarian cancer (HR, per 4kg/​m2 increase:  1.01; 95% CI:  0.95–​1.07) (Schouten et al., 2008). Again, there was no difference by histology; however, the association did differ by menopausal status with a stronger positive association for premenopausal compared to postmenopausal ovarian cancer (HR:  1.12; 95% CI:  0.96–​1.31 vs. HR:  1.02; 95% CI:  0.95–​ 1.08; p-​interaction = 0.07) (Schouten et al., 2008). Two large prospective cohort studies reported suggestive or significant heterogeneity between tumor subtypes with a stronger association for endometrioid tumors (Gates et al., 2009; Yang et al., 2012). Results for other measures of adiposity, including waist and hip circumference and waist-​ to-​hip ratio, have been inconclusive (Beral et  al., 2012b; Schouten et al., 2008). Height has been consistently, positively associated with ovarian cancer risk. In a large collaborative analysis, each 5  cm increase in height was associated with an 8% (95% CI: 6%–​10%) increased risk of ovarian cancer (Beral et  al., 2012b). No difference by histologic subtype was observed (Beral et  al., 2012b; Schouten et  al., 2008). Height is not thought to affect ovarian cancer development directly but

 895

Ovarian Cancer instead through an indirect route such as increased exposure to growth hormones, although timing of these exposures may matter, as analyses of adulthood insulin-​like growth factor I (IGF-​I) have been conflicting (Ose et al., 2015b; Tworoger et al., 2007b).

Physical Activity The research on physical activity and ovarian cancer has been very inconsistent, with case-​control studies reporting inverse or null associations and prospective studies generally reporting positive or null associations (Cannioto and Moysich, 2015). The literature’s inconsistency may reflect differences in the effects of physical activity by frequency, duration, intensity, and point in the life span, as well as the infrequent examination of the combination of occupational and leisure-​time activity. One recent prospective study has assessed leisure-​time physical activity with respect to intensity and points in the life span (Huang T et  al., 2015). Total cumulative average physical activity measured as metabolic equivalent hours per week (MET-​hrs/​week) had a modest J-​shaped relationship with ovarian cancer risk. Women with ≥ 27 MET-​ hrs/​ week had a significant increased risk of ovarian cancer (HR = 1.26; 95% CI: 1.02, 1.55) compared to women with 3–​9 MET-​ hrs/​week. Additionally, there was a suggestion of a positive association among women with < 3 MET-​hrs/​week compared to women with 3–​9 MET-​hrs/​week (HR = 1.19; 95% CI: 0.94, 1.52). However, both of these associations were observed only for premenopausal physical activity (Huang T et al., 2015). Plausible mechanisms exist to explain how physical activity might increase or decrease risk. For instance, very vigorous physical activity (e.g., running marathons) may decrease the frequency of ovulation, thereby decreasing the risk of ovarian cancer. However, vigorous physical activity may also decrease estrogen levels, which increases gonadotropin levels and could increase the risk of ovarian cancer (Cannioto and Moysich, 2015). Notably, most women exercise at a more moderate level, which may improve fertility and increase ovulation (Bonen, 1992; Rich-​Edwards et  al., 2002; Rogol et al., 1992), potentially explaining the modest increased risk observed in premenopausal women in one study. While clarification of the association between physical activity and ovarian cancer will not influence public health recommendations for exercise, a better understanding of the relationship might yield important information about the underlying etiology of ovarian cancer. A  pooled analysis of prospective cohort studies may now be warranted to increase sample sizes.

Surgical Interventions The most effective risk-​reducing surgical intervention is a bilateral salpingo-​oophorectomy (BSO), which is commonly offered to women with a BRCA1 or BRCA2 germline mutation. A 2009 meta-​analysis including three studies of BRCA+ women reported a 79% (95% CI: 61%–​88%) decreased risk of ovarian cancer among women with versus without BSO (Rebbeck et  al., 2009). Two studies assessed the effects of hysterectomy plus concurrent BSO versus hysterectomy only in average-​risk women (Jacoby et  al., 2011; Parker et al., 2009). Both studies observed more than a 90% decrease in risk with BSO. It is important to note that although BSO dramatically reduces the risk of ovarian cancer, there are substantial side effects from BSO, including early menopause, cardiovascular disease, and increased overall mortality (Parker, 2010; Parker et al., 2009, 2013), which preclude it from being used in the general population for prevention. The majority of studies of unilateral oophorectomy have observed a decreased risk of ovarian cancer (Chiaffarino et  al., 2005; Rice et  al., 2014; Rossing et  al., 2008). In the only prospective study among the general population, women with a unilateral oophorectomy were 30% (95% CI:  9%–​4 7%) less likely to develop ovarian cancer compared to women with both ovaries intact (Rice et al., 2014).

895

As stated in the section on pathogenesis in this chapter, it is now postulated that the more aggressive serous tumors arise within the fallopian tubes; thus removal of the fallopian tubes has been hypothesized to reduce the risk of ovarian cancer. In a prospective Swedish health registry study, salpingectomy was associated with a lower risk of ovarian cancer (HR = 0.65; 95% CI: 0.52–​0.81) (Falconer et al., 2015). The risk reduction was greater when both tubes were removed compared to one (HR = 0.35; 95% CI: 0.17–​0.73 vs. HR = 0.71; 95% CI: 0.56–​0.91, respectively), although this was based on a small number of cases. Results were generally similar in a second study (Madsen et al., 2015). Neither study had a sufficient number of exposed cases to assess the association by histologic subtype. A meta-​analysis including the two preceding studies plus a case-​control study, observed an OR of 0.51 (95% CI:  0.35–​0.75) for bilateral salpingectomy (Yoon et al., 2016). Despite these promising results, it is important to bear in mind that the women in this study were undergoing salpingectomy for medical reasons, and it is unknown at this point what the long-​term effects of salpingectomy may be in low-​risk women with no medical indications (Poole et al., 2015). Although the literature has been inconsistent, overall hysterectomy has been associated with a modest decreased risk of ovarian cancer (Jordan et  al., 2013; Rice et  al., 2012, 2014). A meta-​analysis of 24 studies, reported a 26% (95% CI: 15%–​36%) decreased risk of ovarian cancer among women with a hysterectomy compared to women without (Rice et  al., 2012). A later meta-​analysis reported similar results, although more recent case-​control studies showed an increased risk of ovarian cancer with hysterectomy, which may be due to increasing age at hysterectomy (Jordan et al., 2013). Additionally, postmenopausal HT use may influence the observed association as hysterectomized women are generally prescribed estrogen-​only therapy, which may have a stronger positive association with ovarian cancer than estrogen plus progestin HT. Finally, one study noted significant heterogeneity by histologic subtype for ovarian cancer and hysterectomy, with inverse associations for low-​ grade serous, mucinous, endometrioid, and clear cell tumors only (Merritt et al., 2013b). A meta-​analysis of 30 studies reported a significant decreased risk of ovarian cancer associated with tubal ligation (HR = 0.70; 95% CI: 0.64, 0.75) (Rice et  al., 2012). The association with tubal ligation is stronger for non-​serous tumors, particularly for endometrioid and clear cell tumors (HR = 0.41; 95% CI: 0.28, 0.60) (Merritt et al., 2013b; Rice et al., 2012, 2014). Additionally, younger women (< 35 years at tubal ligation) have a stronger inverse association between tubal ligation and ovarian cancer compared to older women (Rice et al., 2012, 2014). Tubal ligation and hysterectomy have been hypothesized to reduce the risk of ovarian cancer by blocking the flow of carcinogens and endometriosis from the uterus to the ovary. As discussed in the endometriosis section, endometriosis is most strongly associated with risk of endometrioid and clear cell tumors.

Chemical Agents In 2009, the International Agency for Research on Cancer (IARC) concluded that there was sufficient evidence for a causal relationship between asbestos exposure and ovarian cancer (Straif et  al., 2009). A  2011 meta-​analysis supported this conclusion, reporting a 77% (95% CI: 37%–​128%) increased risk of ovarian cancer mortality with asbestos exposure (Camargo et al., 2011). Interest in talc was stimulated by its similarity to asbestos. Previous studies have noted a positive association between talc powder use and risk of ovarian cancer (Hankinson and Danforth, 2006). A large pooled analysis of eight case-​control studies reported an increased risk of ovarian cancer with genital powder use (OR = 1.24; 95% CI: 1.15–​ 1.33) (Terry et al., 2013). Risk was increased for all subtypes, with the exception of mucinous tumors. Among genital powder users, no dose-​ response relationship was observed for number of lifetime applications (Terry et al., 2013). Data from prospective studies is much more limited, and the results have not been consistent (Gates et al., 2008; Houghton et al., 2014).

896

896

PART IV:  Cancers by Tissue of Origin

Acrylamide was classified as a “probable carcinogen” to humans in 1994 by IARC. A recent meta-​analysis of two prospective cohorts, one case-​control and one case-​cohort study observed no overall association between dietary intake of acrylamide and ovarian cancer risk (Pelucchi et  al., 2015). Additionally, a recent study including over 300,000 women observed no association between dietary acrylamide intake and ovarian cancer risk (HR  =  1.02; 95% CI:  0.96, 1.09 per 10 ug/​day intake) (Obon-​Santacana et al., 2015). No differences were observed by histologic subtype or by smoking status (Obon-​Santacana et al., 2015). One possible explanation for the lack of association is that dietary intake does not accurately reflect bioavailable acrylamide. However, two studies to date have assessed biomarkers of acrylamide and observed no association between acrylamide adducts in blood samples and subsequent ovarian cancer risk (Obon-​Santacana et  al., 2016; Xie et al., 2013).

Smoking Prior studies reported mixed results for the association between smoking and ovarian cancer (Hankinson and Danforth, 2006); however, recent studies suggest that smoking may be associated with specific subtypes of ovarian cancer. The Collaborative Group on Epidemiologic Studies of Ovarian Cancer reported a slight increased risk of ovarian cancer for current smoking versus never smoking (RR = 1.06; 95% CI: 1.01, 1.11) (Beral et al., 2012a). This increased risk appeared to be driven by the substantial increase in risk for mucinous tumors with current smoking (Beral et  al., 2012a; Faber et al., 2013b; Gram et al., 2012). Current smokers were 79% (95% CI:  60%–​100%) more likely to develop mucinous ovarian tumors compared to never smokers (Beral et al., 2012a). Further, this association appeared to be stronger for borderline as opposed to invasive tumors. Additionally, there was a suggestion of a decreased risk of endometrioid and clear cell tumors with current smoking (Beral et al., 2012a; Faber et al., 2013b). Biologically, there is reason to believe that mucinous and non-​ mucinous tumors may have different etiologies (Terry et  al., 2003). Researchers have noted that mucinous cells are similar to those of the cervix and intestine, sites where smoking has been implicated as a factor in cancer development (Marchbanks et al., 2000). Additionally, some mucinous ovarian tumors may actually be metastases from the gastrointestinal tract (Seidman et al., 2003).

Radiation The majority of previous studies on radiation, particularly among atomic bomb survivors and women treated with radiation for other conditions, observed a suggestion of an increased risk of ovarian cancer (Hankinson and Danforth, 2006). An updated analysis of atomic bomb survivors in Japan observed a higher risk of ovarian cancer in the highest radiation exposure category compared to the lowest; however, this difference was not statistically significant (Preston et al., 2007). Some, but not all, ecologic studies have noted that women living in Northern climates have higher incidence and/​or mortality rates of ovarian cancer, suggesting that cutaneous exposure to ultraviolet-​ B radiation, leading to increased production of vitamin D, may be important in ovarian cancer development (Garland et al., 2006; Grant, 2006; 2007, 2014). However, epidemiologic studies of vitamin D and ovarian cancer have been inconsistent (Cook et al., 2010; Lin et al., 2012; Merritt et al., 2013a; Prescott et al., 2013). A 2010 systematic review concluded that there was no consistent evidence that vitamin D impacted ovarian cancer incidence or mortality (Cook et al., 2010). In the majority of studies, neither dietary intake of vitamin D nor plasma levels of 25-​(OH)D3 were associated with ovarian cancer risk overall. However, the association between plasma 25-​(OH)D3 and ovarian cancer risk may vary by BMI (Cook et al., 2010). Among overweight and obese women, a significant trend of decreasing ovarian cancer risk with increasing plasma 25-​(OH)D3 levels was observed (RR: 0.39; 95% CI: 0.16–​0.93 for quartile 4 vs. quartile 1) (Tworoger et al., 2007c). Similar results among women with a BMI of ≥ 25 kg/​m2 were reported

in a pooled analysis of seven prospective cohort studies (Zheng et al., 2010), which included the initial study (Tworoger et al., 2007c). Several additional approaches to measuring vitamin D exposure have been reported in relation to ovarian cancer risk. A score that predicted plasma 25-​(OH)D3 levels was created using race, repeated measures over time of dietary vitamin D intake, ultraviolet-​B flux, supplemental vitamin D, and lifestyle factors such as BMI and alcohol consumption. This predicted score was not associated with ovarian cancer risk (Prescott et  al., 2013). In addition, several studies have evaluated genetic variation in the vitamin D receptor (VDR). While some studies have identified associations, the specific genetic variant associated with risk has not been consistent across studies (Dai et al., 2015; Li S et al., 2014; Liu et al., 2013; Mun et al., 2015; Qin et al., 2013; Song and Lee, 2013; Xu et al., 2013, 2014).

HOST FACTORS Reproductive Factors Age at Menarche, Age at Menopause, and Ovulatory Years

Overall, age at menarche and age at menopause do not appear to be strongly associated with ovarian cancer (Hankinson and Danforth, 2006), although recent data suggest the associations again may vary by histologic subtype. In the European Prospective Investigation into Cancer and Nutrition (EPIC), later age at menarche was associated with a significantly decreased risk of clear cell ovarian tumors (HR = 0.40; 95% CI: 0.16–​0.98 for ≥ 15 years vs. ≤ 13 years) but was not associated with serous, endometrioid, or mucinous tumors (Fortner et al., 2015). Associations between age at menopause and ovarian cancer also have been inconsistent. Two prospective cohort studies have reported null results, while two other prospective studies observed an increased risk with increasing age at menopause (Braem et al., 2010; Fortner et al., 2015; Gates et al., 2009; Yang et al., 2012). In the EPIC cohort, there was a significant trend of increasing ovarian cancer risk with increasing age at menopause (HR  =  1.62; 95% CI:  1.21–​2.17 for ≥ 55 years vs. ≤ 48 years) (Fortner et al., 2015). Additionally, two of the prospective cohort studies observed significant heterogeneity across histologic subtypes, with stronger positive associations for clear cell and endometrioid tumors compared to serous and mucinous tumors (Fortner et al., 2015; Gates et al., 2009). In the Nurses’ Health Study (NHS) and NHS II cohorts, each 1-​year increase in age at menopause was associated with a 13% increased risk of endometrioid tumors (95% CI: 1.04, 1.22) compared to a non-​significant 2% (95% CI: 0.99, 1.06) increased risk of serous tumors and a non-​ significant 1% (95% CI:  0.93, 1.10) increased risk of mucinous tumors (Gates et al., 2009). It has been proposed that at least some types of ovarian cancer result from repeated trauma of the ovary epithelium caused by ovulation. If so, younger age at menarche or later age at menopause might increase risk of ovarian cancer by increasing the number of ovulations. This hypothesis is supported by prospective cohort studies reporting an increased risk of ovarian cancer with increasing number of ovulatory years, (i.e., the number of years a woman has ovulatory cycles during her lifetime) (Braem et  al., 2010; Fortner et  al., 2015; Gates et  al., 2009). In the NHS and NHS II cohorts, each 1-​year increase in ovulatory years was associated with a 7% (95% CI:  5%–​8%) increased risk of ovarian cancer (Gates et  al., 2009). The association between ovulatory years and ovarian cancer appears to be stronger for serous, endometrioid, and clear cell tumors, and weak or null for mucinous tumors (Fortner et al., 2015; Gates et al., 2009).

Parity, Age at First Birth, Age at Last Birth

Parity has been consistently associated with an approximately 30%–​ 40% decreased risk of ovarian cancer (Braem et  al., 2010; Fortner et al., 2015; Gates et al., 2009; Yang et al., 2012). Increasing number of full-​term pregnancies is associated with a decreasing risk of ovarian cancer, with a 6% (95% CI: 1%–​11%) decrease in risk per child among

 897

Ovarian Cancer parous women in two large cohorts (Gates et al., 2009). Although data remain limited, early evidence suggests that the inverse association between parity and ovarian cancer is strongest for endometrioid and clear cell tumors and null for mucinous tumors (Gates et  al., 2009; Tung, 2003; Yang et al., 2012). In 1994, Adami et al. hypothesized that pregnancy might be protective because it clears cells that have undergone malignant transformation from the ovaries, suggesting that late age at pregnancy might exert greater protection than an earlier age (Adami et  al., 1994). However, studies of the association between timing of pregnancies and ovarian cancer have not been consistent. One large population-​ based case-​control study observed no association between age at last birth or years since last pregnancy and ovarian cancer risk (Tung, 2003). Additionally, no clear trend was observed in the AARP prospective cohort study for age at first birth and ovarian cancer (Yang et  al., 2012). In contrast, a recent analysis in the Swedish Family Cancer-​Database observed a decreased risk of ovarian cancer with both an increasing age at first birth (RR: 0.61; 0.44, 0.86 for 40+ years vs. ≤ 19 years) and with increasing age at last birth (RR: 0.67; 0.62–​ 0.72 for 40+ years vs. ≤ 29 years) (Bevier et al., 2011). However, the authors were unable to adjust for important ovarian cancer confounders, particularly OC use.

Lactation

Three recent meta-​analyses largely based on the same studies have all reported a significant 20%–​30% decreased risk of ovarian cancer associated with ever breastfeeding among parous women (Feng et al., 2014; Li D-​P et al., 2014; Luan et al., 2013). A stronger inverse association was observed for case-​control versus cohort studies (Li D-​P et al., 2014; Luan et al., 2013). Among the five cohort studies that have assessed breastfeeding, parous women who ever breastfed had a 12% (95% CI: 1%–​22%) decreased risk of ovarian cancer compared to parous women who never breastfed (Luan et al., 2013). Increasing duration of breastfeeding has been associated with a stronger protective association (Li D-​P et al., 2014), for example, each month of breastfeeding was associated with a 2% (95% CI: 0%–​3%) decreased risk of ovarian cancer (Danforth et al., 2007a). Only a few studies have assessed breastfeeding in relation to histologic subtype. Including both case-​ control and cohort studies, breastfeeding appeared to reduce risk of all ovarian cancer subtypes, although the association was potentially stronger for endometrioid or clear cell tumors (Li D-​P et  al., 2014; Luan et  al., 2013). A  pooled analysis of prospective cohort studies is needed for a more powerful assessment of this issue.

Infertility and Fertility Drugs

Previous research reported mixed results for infertility and ovarian cancer, as disentangling the effects of nulliparity and infertility is challenging (Hankinson and Danforth, 2006). However, more recent studies have suggested a modest increased risk of ovarian cancer for female-​factor infertility (Brinton et  al., 2004; Rossing et  al., 2004; Tworoger et al., 2007a), particularly for endometriosis (Brinton et al., 2004). In the Nurses’ Health Study, female factor infertility was associated with a 36% (95% CI: 7%–​75%) increased risk of ovarian cancer (Tworoger et  al., 2007a). However, women who gave birth with assisted reproductive technology (ART) in Norway were not at an increased risk of ovarian cancer compared to women who did not use ART (Reigstad et al., 2015). Literature on the association between fertility drugs and ovarian cancer risk has been somewhat inconsistent (Bjornholt et al., 2015; Trabert et al., 2013; van Leeuwen et al., 2011). In a large retrospective cohort study at five US infertility clinics, no association was observed with ever use of either clomiphene citrate or gonadotropins and ovarian cancer, although in a subgroup analysis, a 3-​to 4-​ fold increased risk was observed for clomiphene citrate use among women who remained nulligravid (RR:  3.63; 95% CI:  1.36–​9.72) (Trabert et al., 2013). Additionally, there is a suggestion that longer use of in vitro fertilization and progesterone may increase the risk of ovarian cancer (Bjornholt et al., 2015; van Leeuwen et al., 2011). A  2014 review concluded that there was no overall association

897

between fertility drugs and ovarian cancer risk; however, several questions remained regarding specific subgroups of women, such as those who remain nulligravid after treatment (Diergaarde and Kurta, 2014).

Endometriosis

Approximately 5%–​10% of premenopausal women have endometriosis, a benign gynecological condition that can lead to chronic pelvic pain, infertility, and dysmenorrhea (Bulun, 2009; Somigliana et al., 2006). In a pooled analysis of 13 case-​control studies, endometriosis was associated with an increased risk of developing invasive ovarian cancer (OR  =  1.46; 95% CI:  1.31–​1.63) (Pearce et  al., 2012). There is clear variation in this association by histologic subtype of the ovarian tumor. Endometriosis is associated with an increased risk of clear cell, endometrioid, and low-​grade serous tumors, but has not been associated with mucinous or high-​grade serous tumors. Women with endometriosis have a 3-​fold increased risk of clear cell ovarian cancer compared to women without endometriosis (OR = 3.05; 95% CI: 2.43–​3.84) (Pearce et al., 2012). Endometriosis is estimated to be associated with 30%–​55% of clear cell tumors and 30%–​40% of endometrioid tumors (Acien et al., 2015; Brinton, 2005; Wang et al., 2013). It is thought that some endometrial implants in the peritoneal cavity undergo mutations that lead to the development of “atypical” endometriosis and may lead to ovarian cancer (Pavone and Lyttle, 2015).

Pelvic Inflammatory Disease and Sexually Transmitted Infections

Chronic pelvic inflammatory disease (PID) develops from sexually transmitted infections and increases ovarian inflammation and circulating CRP levels (Bychkov, 1990; Reljic and Gorisek, 1998). Studies of PID and ovarian cancer risk have been mixed, with some (Lin et  al., 2011; Rasmussen et  al., 2013; Risch and Howe, 1995; Shu et al., 1989), but not all (Ness et al., 2000; Parazzini et al., 1996; Rowlands et  al., 2011), reporting a positive association. However, PID has not been uniformly defined, potentially leading to misclassification in some studies. Chlamydia is the most common cause of PID, in addition to gonorrhea (Barrett and Taylor, 2005). Chlamydia inhibits apoptosis, reduces cell adhesion, and increases proliferation (Alibek et  al., 2012; Chumduri et  al., 2013), and it causes fallopian tube inflammation and infertility (Bebear and de Barbeyrac, 2009; Land et al., 2010; Manavi, 2006), which have been linked to ovarian cancer risk (Brinton et al., 2005; Tworoger et al., 2007a). Chlamydia has been detected in ovarian tumors but not in normal tissue (Idahl et al., 2011; Shanmughapriya et al., 2012). Among women in Oahu, Hawaii, there was a non-​significant increased risk of ovarian cancer associated with higher levels of chlamydia antibodies (OR = 1.9 95% CI:  0.9, 3.9 for ≥ 0.40 vs. < 0.10 OD units) (Ness 2003). However, in an analysis among women in Pennsylvania, Ohio and New  York, a 40% (95% CI:  10%–​60%) decreased risk of ovarian cancer was observed for women with the highest titers compared to the lowest (Ness 2008). In a third study, chlamydia infection was reported to be associated with ovarian cancers that likely arose in the fallopian tube (Idahl et al., 2011). However, because chemotherapy may alter antibody production (Maxwell and Maher, 1992), it will be important to evaluate this association prospectively.

Family history Women with a first-​degree family history of ovarian cancer have an approximately 2-​to 3-​fold increased risk of developing ovarian cancer (Soegaard et al., 2009; Walker et al., 2002). The relative risk among women whose mother had ovarian cancer was 3.04 (95% CI: 1.73–​5.34) and with at least one sister with ovarian cancer was 2.93 (95% CI:  1.36–​6.28) (Walker et  al., 2002). Although based on small numbers, women with a family history of ovarian cancer appear to be more likely to develop serous tumors compared to mucinous or endometrioid tumors (Chiaffarino et  al., 2007; Soegaard et al., 2009).

89

898

PART IV:  Cancers by Tissue of Origin

Genetic Predisposition /​Major Gene Mutations Approximately 10%–​12% of women with ovarian cancer have germline mutations in BRCA1 or BRCA2 genes and a further 2%–​3% of women are from families with hereditary nonpolyposis colorectal cancer (HNPCC) (Nelson et  al., 2014; Søgaard et  al., 2006; Taylor and Mutch, 2006). Over 60% of ovarian tumors among women with BRCA mutations are serous tumors (Pal et al., 2005). Recent research has revealed new germline mutations that may affect the development of ovarian cancer. In a case series of 281 women with ovarian cancer, 48 women with peritoneal cancer, and 31 women with fallopian tube cancer, 10 genes accounted for 6% of all cases with germline mutations, including BRIP1, CHEK2, PALB2, RAD51C, TP53, and BARD1 (Walsh et  al., 2011). Additionally, two rare syndromes, Li-​ Fraumeni and Peutz-​Jeghers, associated with mutations in TP53 and STK11/​LKB1 genes, respectively, are associated with an increased risk of ovarian cancer (Lancaster et al., 2015). The population frequency of BRCA1 and BRCA2 mutations varies substantially. In the general population, an estimated prevalence is 1 in 280 (range 1 in 112 to 1 in 441). In the Ashkenazi Jewish population, approximately 1 in 40 women are carriers; most of these women have one or more specific founder mutations in BRCA1 or BRCA2 (i.e., 185delAG and 5382insC in BRCA1 and 6174delT in BRCA2) (Boyd, 2003). Estimates of the probability of developing ovarian cancer among BRCA1 or BRCA2 mutation carriers range from 35%–​ 50% for BRCA1 and 10%–​30% for BRCA2 (Antoniou et  al., 2003; Boyd, 2003).

Gene Polymorphisms and Ovarian Cancer Risk in Normal-​Risk Women Among women with a first-​degree family history of ovarian cancer, approximately 24% is estimated to be attributable to BRCA1 and BRCA2 mutations (Jervis et al., 2014), suggesting that multiple low penetrance genes may influence familial ovarian cancer risk. In the last 6 years, substantial research has assessed low penetrance genes and ovarian cancer risk through genome-​wide association studies (GWAS). To date, 18 loci have been associated with ovarian cancer risk (Bojesen et  al., 2013; Bolton et  al., 2010; Goode et  al., 2010; Kuchenbaecker et al., 2015; Permuth-​Wey et al., 2013; Pharoah et al., 2013; Shen et al., 2013; Song et al., 2009). Interestingly, many of the loci have stronger associations with serous compared to non-​serous tumors (Goode et  al., 2010; Kuchenbaecker et  al., 2015; Pharoah et al., 2013; Song et al., 2009), although this may be due to the substantially higher sample size for this tumor subtype. Three single nucleotide polymorphisms (SNPs) within the TERT locus have been associated with ovarian cancer (Bojesen et  al., 2013). Specifically, SNP rs7705526 is associated with longer telomeres and higher risk of borderline ovarian cancer (p  =  1.3 x 10–​15). Additionally, SNPs rs10069690 and rs2242652 increase the risk of invasive ovarian cancer (p  =  1.3 x 10–​11) but are not associated with telomere length (Bojesen et al., 2013).

Endogenous Hormones Earlier studies of androgens (e.g., testosterone) and ovarian cancer were inconsistent and based on small sample sizes (Hankinson and Danforth, 2006). More recent studies with larger sample sizes have observed no association between circulating androgen levels and ovarian cancer risk (Ose et al., 2015a; Tworoger et al., 2008b). There was a suggestion of a decreased risk for serous tumors with higher levels of androstenedione (OR = 0.79; 95% CI: 0.64, 0.97 for a doubling of androstenedione). This association appeared to differ by grade with a positive association for low-​grade serous tumors (OR = 1.99) and an inverse association for high-​grade serous tumors (OR  =  0.75) (Ose et  al., 2015a). One of the main characteristics of polycystic ovarian syndrome (PCOS) is hyperandrogenism. A  2015 meta-​analysis included three case-​control studies of PCOS and ovarian cancer and observed a non-​significant increased risk (OR = 1.41; 95% CI: 0.93,

2.15) (Barry et al., 2014). The association appeared stronger among younger women, although this subanalysis was based on limited data (Barry et al., 2014). In contrast, two recent registry studies noted no association between PCOS and ovarian cancer risk (Gottschau et al., 2015; Shen et al., 2015). Although a positive association between insulin-​like growth factor-​I (IGF-​I) or related binding proteins (IGFBP-​3 and IGFBP-​2) and ovarian cancer has been hypothesized, current prospective studies are not supportive (Ose et  al., 2015b; Peeters et  al., 2007; Tworoger et  al., 2007b). An inverse association between IGF-​I and ovarian cancer was noted in a nested case-​control study within three cohorts (OR = 0.56; 95% CI: 0.32–​0.97 for top vs. bottom quartiles), although there was no clear dose–​response relationship (Tworoger et al., 2007b). The largest prospective cohort study of IGF-​I and ovarian cancer to date did not observe an association between IGF-​I and ovarian cancer overall or by tumor characteristics (Ose et al., 2015b). One mechanistic hypothesis underlying the inverse association between parity and ovarian cancer suggests that changes in the hormonal milieu during pregnancy reduce later risk. In the Finnish and Swedish Maternity cohorts, higher levels of androgens were associated with an increased risk of borderline serous and mucinous tumors (Schock et  al., 2014b). A  doubling of testosterone and 17-​ hydroxyprogesterone (17-​OHP) was associated with a 41% (95% CI:  8%–​86%) and 46% (95% CI:  8%–​97%) increased risk of borderline serous tumors, respectively. Additionally, testosterone and androstenedione were associated with a 50% and 56%, respectively, higher risk of borderline mucinous tumors. Increased concentrations of estradiol were associated with an increased risk of endometrioid tumors (OR = 1.89; 95% CI: 1.20–​2.98 for doubling of endometrioid concentration). Finally, there was no association between progesterone and sex hormone binding globulin and ovarian cancer (Schock et al., 2014b). Within the same cohorts, IGF-​I concentrations during pregnancy were associated with a non-​significant decreased risk of invasive (OR  =  0.79; 0.62–​1.02 for tertile 3 vs. 1)  and endometrioid tumors (OR = 0.55; 95% CI: 0.28–​1.07 for tertile 3 vs. 1). The association was stronger for invasive tumors among women aged < 55 years at diagnosis (Schock et al., 2015).

Other Potential Risk Factors Animal models have strongly suggested that chronic stress and related stress hormones, including catecholamines, increase tumor metastasis, tumor volume, and angiogenesis (Sood et  al., 2006; Thaker et al., 2006, 2007). Further, in ovarian cancer patients, dysregulated diurnal cortisol rhythms were associated with higher inflammatory markers in ascites as well as worse survival (Schrepf et al., 2015). Relatively little epidemiologic research has been conducted on measures of chronic stress and distress with risk of ovarian cancer. A meta-​analysis of prospective studies considering any measure of stress with total cancer observed modest associations of 6%–​20% increased risk of cancer, with stronger associations for studies with large sample sizes and long follow-​up. Specifically for ovarian cancer, two recent studies observed significant and suggestive positive associations for depression (Huang T et al., 2015) (HR = 1.30, 95% CI: 1.05–​1.60) and high levels of anxiety (Poole et  al., 2016) (HR = 1.14; 95% CI: 0.96–​1.36). Such work has led to a hypothesis that ß-​blockers, which block catecholamine activity, may reduce ovarian cancer risk. However, a recent study did not observe an association for ß-​blocker use with ovarian cancer risk (Huang et al., 2016). Interestingly, there was an increased risk for use of diuretics, which was hypothesized to possibly be due to alterations in uric acid, as hypertension itself has not been associated with risk (Bjorge et al., 2011; Huang et al., 2016; Soler et al., 1999). Other novel factors that may influence risk include telomere length, which was been associated with ovarian cancer in some, but not all, case-​control studies (Martinez-​Delgado et al., 2012; Mirabello et al., 2010; Terry et al., 2012), markers of ovarian reserve (e.g., anti-​Mullerian hormone) (Schock et al., 2014a), and aspects of the human microbiome (Babic et al., 2015; Chase et al., 2015).

 89

Ovarian Cancer

EPIDEMIOLOGIC APPROACHES TO UNDERSTANDING TUMOR HETEROGENEITY One of the key challenges in studying ovarian cancer etiology is the extensive heterogeneity of tumor subtypes. As discussed previously, it has become increasingly clear that ovarian cancer is not one disease, but rather a constellation of diseases defined by their cell of origin (e.g., ovary epithelium, fallopian tube secretory cells, endometriosis cells) and mutational profile (e.g., p53 mutations) (National Academies of Sciences, 2016). Due to the relative rarity of ovarian cancer, most individual studies have not had sufficient power to examine associations by specific subtypes. One important question is how to best define tumor subtypes. While many studies have used histology as a surrogate for different tumor development pathways, fewer studies have incorporated grade, which has relatively modest inter-​and intra-​rater reliability and often is not reported on pathology reports or in tumor registries (Matsuno et al., 2013). Newer research supports the use of immunohistochemical (IHC) evaluation of tumor tissue expression of WT1, p16, and p53 (reflecting p53 mutation status [Soslow, 2008]) to more accurately identify high-​grade serous tumors (Kobel et al., 2010; Kobel et al., 2013; Malpica et al., 2007), the most aggressive subtype. For example, some tumors classified as endometrioid by pathologists are found to be high-​grade serous carcinomas by IHC. Because grade is not reliably reported and tissue is often hard to obtain in large-​scale epidemiologic studies, other metrics for classifying tumors to examine risk factor associations have been examined, including how rapidly fatal the tumor is as a measure of tumor aggressiveness (Poole et al., 2013b) and tumor dominance/​sidedness as a surrogate for origin at the ovary (tumors dominant on one side or the other) or fallopian tube (tumors spread throughout the peritoneal cavity) (Kotsopoulos et  al., 2013). Based on these studies, it has become clear that there are distinct differences in risk factor profiles for various tumor types (Table 46–​1), suggesting that prediagnosis exposures can differentially drive tumor development and influence aggressiveness. Importantly, substantial effort is now being devoted to the development of research consortia to study ovarian cancer. The inclusion of multiple studies, with harmonized risk factor and outcome data, can substantially increase power to assess associations, particularly for the rarer subtypes. The Ovarian Cancer Association Consortium was initially developed to evaluate genetic factors related to ovarian cancer, and has reported specific genetic variants associated with specific histologic subtypes (e.g., Chornokur et al., 2015; Earp et al., 2014; Johnatty et al., 2010; Kelemen et al., 2015). Further, this consortium and the Collaborative Group on Epidemiological Studies of Ovarian Cancer Consortium have leveraged these data sets to examine traditional epidemiologic risk factors by histology, with substantial power to assess even the rarest subtypes (e.g., Beral et al., 2008, 2012a, 2012b, 2015; Pearce et  al., 2012; Sieh et  al., 2013; Terry et  al., 2013). Similarly, the Diet and Cancer Pooling Project has brought together data on dietary factors using validated questionnaires from 12 prospective studies (Genkinger et al., 2006a, 2006b, 2006c; Koushik et al., 2005, 2006, 2015; Schouten et al., 2008). More recently, the Ovarian Cancer Cohort Consortium published a comprehensive analysis of 14 known or putative risk factors for ovarian cancer by histologic subtype. This Table 46–​1.  Risk Factors that are Differentially Associated by Metrics of Ovarian Tumor Heterogeneity Histology Age Parity Tubal ligation Endometriosis Age at menopause Ovulatory years Hormone therapy Smoking

Dominance

Aggressiveness

Age Parity Tubal ligation Endometriosis IUD use

Age Parity Oral contraceptive use Menopausal status Ovulatory years Hormone therapy

899

approach considered unstructured hierarchical clustering to examine similarities and differences across histologic subtypes and further stratified serous tumors by grade (Figure 46–5). Importantly, this study illustrated that most established ovarian cancer risk factors are more strongly associated with the rarer, less aggressive subtypes (i.e., endometrioid, clear cell, low-​grade serous), but relatively few associations were noted for high-​grade serous disease, which comprises 60%–​65% of cases and has poor outcomes. Cumulatively, this research highlights the power that such consortial analyses have to understand etiology and direct future research efforts.

PREVENTIVE MEASURES Primary Prevention and Risk Prediction For women at high risk of ovarian cancer due to BRCA mutations, surgical intervention is the main form of primary prevention. Due to the impact of bilateral oophorectomy on other disease outcomes (e.g., increased cardiovascular disease), this procedure is not indicated for the general population. Other opportunities for primary prevention are limited, as few modifiable factors have been established for ovarian cancer. Oral contraceptive use, breastfeeding among parous women, not using hormone therapy, and tubal ligation are well-​confirmed, potentially modifiable factors. However, because uptake of most of these factors requires complex decision-​making (e.g., tubal ligation is an invasive medical procedure), these are not recommended exclusively for the prevention of ovarian cancer (ACS, 2015a; Walker et al., 2015). With the understanding that a large portion of serous tumors arise from the distal end of the fallopian tube, there is increased interest in removing the tubes as a mode for prevention. Both the Society for Gynecologic Oncology (Oncology, 2013) and the American Congress of Obstetricians and Gynecologists (ACOG, 2015) have recommended that physicians discuss bilateral salpingectomy as an option at time of permanent sterilization (in lieu of tubal ligation) or hysterectomy without oophorectomy. Additionally, one of the key recommendations of the National Academy of Medicine report was that “[c]‌linicians, researchers, and funding organizations should focus on quantifying the risk-​benefit balance of nonsurgical and surgical prevention strategies for specific subtypes and at-​risk populations” (National Academies of Sciences, 2016). Further, efforts have been made to develop risk prediction models to identify women at high risk of ovarian cancer, which could help identify women for whom surgical prevention may provide a net benefit. While multiple known ovarian cancer risk factors were included, the predictive capability of these models based on current knowledge was low (area under the curve ranged from 0.59 to 0.64) (Hartge et al., 1994; Li et al., 2015; Pfeiffer et al., 2013; Rosner et al., 2005) (Table 46–​2). This, in combination with the fact that a large proportion of ovarian cancers are diagnosed at a late stage, and hence have poor survival, has led to considerable interest in developing screening tests for early detection of ovarian cancer.

Screening and Early Detection Three screening trials have been conducted to evaluate combinations of existing tests for early ovarian cancer detection. In the United States, the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial randomized women to receive either screening with transvaginal ultrasound (TVU) and CA-​125 measurements or usual care. Of 28,816 women who received at least one test, 4.7% had an abnormal TVU and 1.4% had an abnormal CA-​125 level. Twenty-​nine tumors were identified, of which nine were borderline and 20 were invasive. The positive predictive value for invasive cancer was 3.7% for an abnormal CA-​125, 1.0% for an abnormal TVU, and 23.5% if both tests were abnormal (Buys et al., 2005). The mortality rate ratio between the intervention screening arm and the usual care group was 1.18 (95% CI:  0.82–​1.71), showing no reduction in mortality with screening (Koushik et al., 2005).

90

900

PART IV:  Cancers by Tissue of Origin

Figure 46–​5.  Clustering of risk factor associations by histologic subtype. Unsupervised hierarchical clustering of the (a) four subtypes and (b) including the serous subtype subdivided into well, moderately, and poorly differentiated carcinomas using the beta estimates, complete linkage, and an uncentered correlation similarity metric. The categories used in the cluster analysis were ever vs. never parous, ever vs. never OC use, ever vs. never tubal ligation, ever vs. never endometriosis, age at menarche >15 years vs. ≤ 11 years, age at menopause < 40 years vs. 50–​55 years, ever vs. never menopausal HT use, ever vs. never hysterectomy, family history of breast cancer (yes vs. no), family history of ovarian cancer (yes vs. no), BMI > 35 vs. 20–​25, height (per 5-​cm increase) and ever vs. never smoking. The color scale shows the range of ß values for each exposure.

A screening trial in Japan randomized women to receive annual TVU and serum CA-​125 tests or usual care. Among 41,688 screened women, 27 ovarian cancers were detected, and among 40,779 control women, 32 ovarian cancers were detected. The proportion of stage I ovarian cancer cases was higher in the screen group compared to the control group (63% vs. 38%, respectively); however, the difference did not reach statistical significance (Kobayashi et al., 2008). The most recent screening trial, the United Kingdom Collaborative Trial of Ovarian Cancer Screening, randomized women to one of three

arms: (1) no treatment; (2) annual CA-​125 screening with TVU as a second-​line test (multimodal screening (MMS) model); or (3) annual TVU alone (ultrasound screening [USS] model). Of 50,078 women who underwent MMS, 0.3% required clinical evaluation and 0.2% underwent surgery. Of the 48,230 in the USS intervention, 3.9% required clinical evaluation and 1.8% underwent surgery. In total, 87 ovarian and tubal cancers were detected, 42 in the MSS group and 45 in the USS group. For invasive cancers there was no difference in the stage distribution between the MMS and USS groups. For

Table 46–​2.  Existing Risk Prediction Models for Ovarian Cancer. Risk-​Prediction Models for Ovarian Cancer in Average-​Risk Populations Reference

Study Used to Develop Model

Validation of Model

Risk Factors Included in Risk Prediction Model Parity, OC use, family history of breast or ovarian cancer Estimated ovulatory years (incorporates parity and OC), duration of menopause, tubal ligation Parity, OC use, family history of breast or ovarian cancer, menopausal hormone use Parity, OC use, menopausal status, age at menopause, menopausal hormone use, unilateral oophorectomy, BMI

Hartge et al., 1994

Seven case-​control studies

No

Rosner et al., 2005

Nurses’ Health Study I and II

Yes

Pfeiffer et al., 2013

PLCO sceening trial, NIH-​AARP Diet and Health Study European Prospective Investigation into Cancer and Nutrition

Yes

Li et al., 2015

Yes

Note: AUC = area under the curve; NIH = National Institutes of Health; PLCO = prostate, lung, colorectal, and overian. a An AUC of 0.5 predicts outcomes no better than chance (generally an AUC greater than 0.7 may be considered for clinical use).

AUCa None provided 0.60 0.59 0.64

 901

Ovarian Cancer invasive ovarian and tubal cancers, both the MMS and USS interventions produced fairly high sensitivity (89.5% and 75.0%, respectively) and specificity (99.8% and 98.2%, respectively). However, the positive predictive value of both inventions was low, particularly for the USS intervention (35.1% for MMS and 2.8% for USS) (Menon et  al., 2009). While there was a suggestion of a mortality reduction (MMS: 15% reduction; 95% CI: –​30 –​3; p = 0.10; USS: 11% reduction; 95% CI: –​27 –​7; p = 0.21) overall, the reduction appeared to be larger in later follow-​up (Jacobs et al., 2016). Finally, von Nagell et  al. (2007) assessed the efficacy of annual TVU screening among 25,327 women. At baseline, TVU had a sensitivity of 85%, specificity of 98.7%, and positive predictive value of 14.01%. The women were followed up for death from ovarian cancer and after 107,276 screening-​years, seven ovarian cancer deaths occurred in the annual screening group and three ovarian cancer deaths occurred in the non-​compliant group, suggesting that TVU screening did not impact on ovarian cancer mortality (van Nagell et al., 2007). In 2012, the US Preventive Services Task Force reaffirmed their recommendation against the use of ovarian cancer screening, including transvaginal ultrasounds and cancer antigen-​125 (CA-​125) blood tests, among women at average risk (Barton and Lin, April 2012). Clearly, additional work in this important area is needed.

FUTURE DIRECTIONS Substantial progress has been made over the last decade in determining a number of lifestyle, reproductive, and genetic factors that cause ovarian cancer. Several biologic pathways (e.g., inflammation) have garnered substantial support, although many questions remain. Examining the function of genes identified in GWA studies as well as identification of putative precursor lesions will enhance understanding of etiology. Given the increasing evidence that risk factor associations can vary by histologic subtype of ovarian cancer, epidemiologic studies will need to routinely address associations by tumor subtype, or by other molecular characteristics of the tumor. To do so with sufficient statistical power will require extremely large studies, or pooling of multiple individual studies. Large cohorts, with biologic samples (e.g., blood or urinary) to allow focused studies of risk or early detection biomarkers, coupled with tumor tissue, should allow continued progress, particularly for improving risk prediction models to identify women who could benefit from targeted prevention measures. Evaluations of lifestyle factors that can prolong survival in women with ovarian cancer also are needed. References Acien P, Velasco I, Acien M, Capello C, and Vela P. 2015. Epithelial ovarian cancers and endometriosis. Gynecol Obstet Invest, 79(2), 126–​135. ACOG. 2015. Salpingectomy for ovarian cancer prevention. Committee Opinion. Number 620. Available from:  http://​www.acog.org/​Resources-​ And-​Publications/​Committee-​Opinions/​Committee-​on-​Gynecologic-​ Practice/​Salpingectomy-​for-​Ovarian-​Cancer-​Prevention#here. ACS. 2015a. Can ovarian cancer be prevented? Available from:  www.cancer.org/​cancer/​ovariancancer/​detailedguide/​ovarian-​cancer-​prevention Accessed September 28, 2015. ACS. 2015b. Cancer facts & ­figures 2015. Atlanta: American Cancer Society. Adami HO, Hsieh CC, Lambe M, et al. 1994. Parity, age at first childbirth, and risk of ovarian cancer. Lancet, 344(8932), 1250–​1254. Alibek K, Karatayeva N, and Bekniyazov I. 2012. The role of infectious agents in urogenital cancers. Infec Agents Cancer, 7(1), 35. PMCID: PMC3626724. Anderson GL, Judd HL, Kaunitz AM, et al. 2003. Effects of estrogen plus progestin on gynecologic cancers and associated diagnostic procedures: The Women’s Health Initiative Randomized Trial. JAMA, 290, 1739–​1748. Antoniou A, Pharaoh PD, Narod S, et  al. 2003. Average risks of breast and ovarian cancer associated with BRCA1 or BRCA2 mutations detected in case series unselected for family history: a combined analysis of 22 studies. Am J Hum Genet, 72, 1117–​1130. Anuradha S, Webb PM, Blomfield P et al. 2014. Survival of Australian women with invasive epithelial ovarian cancer: a population-​based study. Med J Aust, 201(5), 283–​288.

901

Baandrup L, Faber MT, Christensen J, et  al. 2013. Nonsteroidal anti-​ inflammatory drugs and risk of ovarian cancer:  systematic review and meta-​analysis of observational studies. Acta Obstet Gynecol Scand, 92(3), 245–​255. PMID: 23240575. Babic A, Poole EM, Terry KL, et  al. 2015. Periodontal bone loss and risk of epithelial ovarian cancer. Cancer Causes Control, 26(6), 941–​947. PMCID: PMC4439281. Barrett S, and Taylor C. 2005. A review on pelvic inflammatory disease. Int J STD AIDS, 16(11), 715–​720; quiz 721. PMID: 16303062. Barry JA, Azizia MM, and Hardiman PJ. 2014. Risk of endometrial, ovarian and breast cancer in women with polycystic ovary syndrome: a systematic review and meta-​analysis. Human Reproduct Update, 20(5), 748–​758. PMCID: PMC4326303. Barton MB, and Lin K. April 2012. Screening for ovarian cancer: evidence update for the U.S. Preventive Services Task Force Reaffirmation Recommendation Statement. Rockville, MD: Agency for Healthcare Research and Quality. Bebear C, and de Barbeyrac B. 2009. Genital Chlamydia trachomatis infections. Clin Microbiol Infec, 15(1), 4–​10. PMID: 19220334. Bell D, Berchuck A, Birrer M, et al. 2011. Integrated genomic analyses of ovarian carcinoma. Nature, 474(7353), 609–​615. PMCID: PMC3163504. Beral V, Bull D, Green J et al. 2007. Ovarian cancer and hormone replacement therapy in the Million Women Study. Lancet, 369, 1703–​1710. Beral V, Doll R, Hermon C, et  al. 2008. Ovarian cancer and oral contraceptives:  collaborative reanalysis of data from 45 epidemiological studies including 23 257 women with ovarian cancer and 87 303 controls. Lancet, 371(9609), 303–​314. Beral V, Gaitskell K, Hermon C, et  al. 2012a. Ovarian cancer and smoking: individual participant meta-​analysis including 28 114 women with ovarian cancer from 51 epidemiological studies. Lancet Oncol, 13(9), 946–​956. Beral V, Gaitskell K, Hermon C, et  al. 2015. Menopausal hormone use and ovarian cancer risk: individual participant meta-​analysis of 52 epidemiological studies. Lancet, 385(9980), 1835–​1842. Beral V, Hermon C, Peto R, et al. 2012b. Ovarian cancer and body size: individual participant meta-​analysis including 25,157 women with ovarian cancer from 47 epidemiological studies. PLoS Med, 9(4), e1001200. PMCID: PMC3317899. Bevier M, Sundquist J, and Hemminki K. 2011. Does the time interval between first and last birth influence the risk of endometrial and ovarian cancer? Eur J Cancer, 47(4), 586–​591. PMID: 21055917. Bjorge T, Lukanova A, Tretli S et al. 2011. Metabolic risk factors and ovarian cancer in the Metabolic Syndrome and Cancer project. Int J Epidemiol, 40(6), 1667–​1677. PMID: 21984693. Bjornholt SM, Kjaer SK, Nielsen TS, and Jensen A. 2015. Risk for borderline ovarian tumours after exposure to fertility drugs: results of a population-​ based cohort study. Hum Reprod, 30(1), 222–​231. PMID: 25376453. Blank MM, Wentzensen N, Murphy MA, Hollenbeck A, and Park Y. 2012. Dietary fat intake and risk of ovarian cancer in the NIH-​AARP Diet and Health Study. Br J Cancer, 106(3), 596–​602. Bojesen SE, Pooley KA, Johnatty SE, et al. 2013. Multiple independent variants at the TERT locus are associated with telomere length and risks of breast and ovarian cancer. Nat Genet, 45(4), 371–​384, 384e371-​372. PMCID: PMC3670748. Bolton KL, Tyrer J, Song H, et al. 2010. Common variants at 19p13 are associated with susceptibility to ovarian cancer. Nat Genet, 42(10), 880–​884. PMCID: PMC3125495. Bonen A. 1992. Recreational exercise does not impair menstrual cycles: a prospective study. Int J Sports Med, 13, 110–​120. Bosetti C, Rosato V, Gallus S, Cuzick J, and La Vecchia C. 2012. Aspirin and cancer risk: a quantitative review to 2011. Ann Oncol, 23(6), 1403–​1415. PMID: 22517822. Boyd J. 2003. Specific keynote:  hereditary ovarian cancer:  what we know. Gynecol Oncol, 88, S8–​10. Braaten T, Weiderpass E, and Lund E. 2009. Socioeconomic differences in cancer survival: the Norwegian Women and Cancer Study. BMC Public Health, 9, 178. PMCID: PMC2702382. Braem MG, Onland-​Moret NC, Schouten LJ, et  al. 2012. Coffee and tea consumption and the risk of ovarian cancer:  a prospective cohort study and updated meta-​analysis. Am J Clin Nutr, 95(5), 1172–​1181. PMID: 22440851. Braem MG, Onland-​Moret NC, van den Brandt PA, et al. 2010. Reproductive and hormonal factors in association with ovarian cancer in the Netherlands cohort study. Am J Epidemiol, 172(10), 1181–​ 1189. PMCID: PMC2970782. Brewer KC, Peterson CE, Davis FG, et  al. 2015. The influence of neighborhood socioeconomic status and race on survival from ovarian cancer: a population-​ based analysis of Cook County, Illinois. Ann Epidemiol, 25(8), 556–​563. PMCID: PMC4494782.

902

902

PART IV:  Cancers by Tissue of Origin

Brinton LA. 2005. Relationship of benign gynecologic diseases to subsequent risk of ovarian and uterine tumors. Cancer Epidemiol Biomarkers Prev, 14(12), 2929–​2935. Brinton LA, Lamb EJ, Moghissi KS, et  al. 2004. Ovarian cancer risk associated with varying causes of infertility. Fertil Steril, 82(2), 405–​414. PMID: 15302291. Brinton LA, Westhoff CL, Scoccia B, et  al. 2005. Causes of infertility as predictors of subsequent cancer risk. Epidemiology, 16(4), 500–​ 507. PMID: 15951668. Bristow RE, Chang J, Ziogas A, et al. 2015. Sociodemographic disparities in advanced ovarian cancer survival and adherence to treatment guidelines. Obstet Gynecol, 125(4), 833–​842. PMID: 25751200. Bristow RE, Powell MA, Al-​Hammadi N et al. 2013. Disparities in ovarian cancer care quality and survival according to race and socioeconomic status. J Natl Cancer Inst, 105(11), 823–​832. PMCID: PMC3691950. Bulun SE. 2009. Endometriosis. New Engl J Med, 360, 268–​279. Buys SS, Partridge E, Greene MH, et  al. 2005. Ovarian cancer screening in the Prostate, Lung, Colorectal and Ovarian (PLCO) cancer screening trial: findings from the initial screen of a randomized trial. Am J Obstet Gynecol, 193(5), 1630–​1639. PMID: 16260202. Bychkov V. 1990. Ovarian pathology in chronic pelvic inflammatory disease. Gynecol Obstet Inves, 30(1), 31–​33. PMID: 2227609. Camargo MC, Stayner LT, Straif K, et  al. 2011. Occupational exposure to asbestos and ovarian cancer:  a meta-​ analysis. Environ Health Persp, 119(9), 1211–​1217. PMCID: PMC3230399. Cannioto RA, and Moysich KB. 2015. Epithelial ovarian cancer and recreational physical activity:  a review of the epidemiological literature and implications for exercise prescription. Gynecol Oncol, 137(3), 559–​573. PMCID: PMC4447575. Cassidy A, Huang T, Rice MS, Rimm EB, and Tworoger SS. 2014. Intake of dietary flavonoids and risk of epithelial ovarian cancer. Am J Clin Nutr, 100(5), 1344–​1351. PMCID: PMC4196485. Chang ET, Lee VS, Canchola AJ, et al. 2007. Diet and risk of ovarian cancer in the California Teachers Study cohort. Am J Epidemiol, 165(7), 802–​813. PMCID: PMC2093945. Chang ET, Lee VS, Canchola AJ, et al. 2008. Dietary patterns and risk of ovarian cancer in the California Teachers Study cohort. Nutr Cancer, 60(3), 285–​291. PMCID: PMC2365491. Chase D, Goulder A, Zenhausern F, Monk B, and Herbst-​Kralovetz M. 2015. The vaginal and gastrointestinal microbiomes in gynecologic cancers: a review of applications in etiology, symptoms and treatment. Gynecol Oncol, 138(1), 190–​200. PMID: 25957158. Chiaffarino F, Parazzini F, Bosetti C, et al. 2007. Risk factors for ovarian cancer histotypes. Eur J Cancer, 43(7), 1208–​1213. PMID: 17376671. Chiaffarino F, Parazzini F, Decarli A, et al. 2005. Hysterectomy with or without unilateral oophorectomy and risk of ovarian cancer. Gynecol Oncol, 97(2), 318–​322. PMID: 15863124. Choi M, FUller CD, Thomas CRJ, and Wang SJ. 2008. Conditional survival in ovarian cancer:  results from the SEER dataset 1988–​2001. Gynecol Oncol, 109(2), 203–​209. Chornokur G, Lin HY, Tyrer JP, et  al. 2015. Common Genetic Variation In Cellular Transport Genes and Epithelial Ovarian Cancer (EOC) Risk. PLoS One, 10(6), e0128106. PMCID: PMC4474865. Chumduri C, Gurumurthy RK, Zadora PK, Mi Y, and Meyer TF. 2013. Chlamydia infection promotes host DNA damage and proliferation but impairs the DNA damage response. Cell Host Microbe, 13(6), 746–​758. PMID: 23768498. Clendenen TV, Lundin E, Zeleniuch-​Jacquotte A, et  al. 2011. Circulating inflammation markers and risk of epithelial ovarian cancer. Cancer Epidemiol Biomarkers Prev, 20(5), 799–​810. PMCID: PMC3089656. Cook LS, Neilson HK, Lorenzetti DL, and Lee RC. 2010. A systematic literature review of vitamin D and ovarian cancer. Am J Obstet Gynecol, 203(1), 70 e71–​78. PMID: 20227054. Cramer DW, Harlow BL, Willett WC, et  al. 1989. Galactose consumption and metabolizm in relation to the risk of ovarian cancer. Lancet, 2(8654), 66–​71. Cramer DW, Titus-​Ernstoff L, McKolanis JR, et al. 2005. Conditions associated with antibodies against the tumor-​associated antigen MUC1 and their relationship to risk for ovarian cancer. Cancer Epidemiol Biomarkers Prev, 14(5), 1125–​1131. PMID: 15894662. Crane TE, Khulpateea BR, Alberts DS, Basen-​Engquist K, and Thomson CA. 2014. Dietary intake and ovarian cancer risk:  a systematic review. Cancer Epidemiol Biomarkers Prev, 23(2), 255–​ 273. PMCID: PMC4077283. Crum CP. 1999. Ovaries. In: Cotran RS, and Tucker C (Eds.), Pathologic basis of disease. Philadelphia: W. B. Saunders.

Crum CP, McKeon FD, and Xian W. 2012. BRCA, the oviduct, and the space and time continuum of pelvic serous carcinogenesis. Int J Gynecol Cancer, 22(Suppl 1), S29–​34. Dai ZM, Fei YL, Zhang WG, et al. 2015. Association of vitamin D receptor Cdx-​2 polymorphism with cancer risk: a meta-​analysis. Medicine (Balt), 94(33), e1370. PMCID: PMC4616440. Danforth KN, Tworoger SS, Hecht JL, et al. 2007a. Breastfeeding and risk of ovarian cancer in two prospective cohorts. Cancer Causes Control, 18(5), 517–​523. PMID: 17450440. Danforth KN, Tworoger SS, Hecht JL, et  al. 2007b. A prospective study of postmenopausal hormone use and ovarian cancer risk. Br J Cancer, 96(1), 151–​156. PMCID: PMC2360221. Diergaarde B, and Kurta ML. 2014. Use of fertility drugs and risk of ovarian cancer. Curr Opin Obstet Gynecol, 26(3), 125–​129. PMCID: PMC4217689. Earp MA, Kelemen LE, Magliocco AM, et  al. 2014. Genome-​wide association study of subtype-​specific epithelial ovarian cancer risk alleles using pooled DNA. Hum Genet, 133(5), 481–​497. PMCID: PMC4063682. Faber MT, Jensen A, Frederiksen K, et al. 2013a. Oral contraceptive use and impact of cumulative intake of estrogen and progestin on risk of ovarian cancer. Cancer Causes Control, 24(12), 2197–​2206. PMID: 24077761. Faber MT, Kjaer SK, Dehlendorff C, et al. 2013b. Cigarette smoking and risk of ovarian cancer: a pooled analysis of 21 case-​control studies. Cancer Causes Control, 24(5), 989–​1004. PMCID: PMC3818570. Falconer H, Yin L, Gronberg H, and Altman D. 2015. Ovarian cancer risk after salpingectomy: a nationwide population-​based study. J Natl Cancer Inst, 107(2). PMID: 25628372. Feng LP, Chen HL, and Shen MY. 2014. Breastfeeding and the risk of ovarian cancer: a meta-​analysis. J Midwifery Womens Health, 59(4), 428–​437. Ferlay J, Soerjomataram I, Dikshit R et al. 2015. Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012. Int J Cancer, 136(5), E359–​386. PMID: 25220842. Folkins AK, Jarboe EA, Saleemuddin A, et al. 2008. A candidate precursor to pelvic serous cancer (p53 signature) and its prevalence in ovaries and fallopian tubes from women with BRCA mutations. Gynecol Oncol, 109(2), 168–​173. PMCID: PMC2746001. Fortner RT, Ose J, Merritt MA, et al. 2015. Reproductive and hormone-​related risk factors for epithelial ovarian cancer by histologic pathways, invasiveness and histologic subtypes: results from the EPIC cohort. Int J Cancer. PMID: 25656413. Fuh KC, Shin JY, Kapp DS, et  al. 2015. Survival differences of Asian and Caucasian epithelial ovarian cancer patients in the United States. Gynecol Oncol, 136(3), 491–​497. PMID: 25455734. Fukunaga M, Nomura K, Ishikawa E, and Ushigome S. 1997. Ovarian atypical endometriosis: its close association with malignant epithelial tumours. Histopathology, 20, 249–​255. Gao M, Ma W, Chen XB, et al. 2013. Meta-​analysis of green tea drinking and the prevalence of gynecological tumors in women. Asia Pac J Public Health, 25(4 Suppl), 43S–​48S. PMID: 23858521. Garland CF, Mohr SB, Gorham ED, Grant WB, and Garland FC. 2006. Role of ultraviolet B irradiance and vitamin D in prevention of ovarian cancer. Am J Prev Med, 31(6), 512–​514. PMID: 17169713. Gates MA, Rosner BA, Hecht JL, and Tworoger SS. 2009. Risk factors for epithelial ovarian cancer by histologic subtype. Am J Epidemiol, 171(1), 45–​53. PMCID: PMC2796984. Gates MA, Tworoger SS, Hecht JL, et al. 2007. A prospective study of dietary flavonoid intake and incidence of epithelial ovarian cancer. Int J Cancer, 121(10), 2225–​2232. PMID: 17471564. Gates MA, Tworoger SS, Terry KL, et  al. 2008. Talc use, variants of the GSTM1, GSTT1, and NAT2 genes, and risk of epithelial ovarian cancer. Cancer Epidemiol Biomarkers Prev, 17(9), 2436–​ 2444. PMCID: PMC2630413. Genkinger JM, Hunter DJ, Spiegelman D, et  al. 2006a. Dairy products and ovarian cancer: a pooled analysis of 12 cohort studies. Cancer Epidemiol Biomarkers Prev, 15(2), 364–​372. PMID: 16492930. Genkinger JM, Hunter DJ, Spiegelman D, et al. 2006b. A pooled analysis of 12 cohort studies of dietary fat, cholesterol and egg intake and ovarian cancer. Cancer Causes Control 17(3), 273–​285. PMID: 16489535. Genkinger JM, Hunter DJ, Spiegelman D, et  al. 2006c. Alcohol intake and ovarian cancer risk: a pooled analysis of 10 cohort studies. Br J Cancer, 94(5), 757–​762. PMCID: PMC2361197. Gilsing AM, Weijenberg MP, Goldbohm RA, van den Brandt PA, and Schouten LJ. 2010. Consumption of dietary fat and meat and risk of ovarian cancer in the Netherlands Cohort Study. Am J Clin Nutr, 93(1), 118–​126. Goff BA, Mandel LS, Melancon CH, and Muntz HG. 2004. Frequency of symptoms of ovarian cancer in women presenting to primary care clinics. JAMA, 291(22), 2705–​2712.

 903

Ovarian Cancer Goode EL, Chenevix-​Trench G, Song H et al. 2010. A genome-​wide association study identifies susceptibility loci for ovarian cancer at 2q31 and 8q24. Nat Genet, 42(10), 874–​879. PMCID: PMC3020231. Gottschau M, Kjaer SK, Jensen A, Munk C, and Mellemkjaer L. 2015. Risk of cancer among women with polycystic ovary syndrome: a Danish cohort study. Gynecol Oncol, 136(1), 99–​103. PMID: 25451694. Gram IT, Lukanova A, Brill I, et al. 2012. Cigarette smoking and risk of histological subtypes of epithelial ovarian cancer in the EPIC cohort study. Int J Cancer, 130(9), 2204–​2210. PMID: 21678398. Grant WB. 2006. The likely role of vitamin D from solar ultraviolet-​B irradiance in increasing cancer survival. Anticancer Res, 26, 2605–​2614. Grant WB. 2007. A meta-​analysis of second cancers after a diagnosis of nonmelanoma skin cancer: additional evidence that solar ultraviolet-​B irradiance reduces the risk of internal cancers. J Ster Biochem Mol Biol, 103(3–​5), 668–​674. PMID: 17208438. Grant WB. 2014. An ecological study of cancer mortality rates in the United States with respect to solar ultraviolet-​B doses, smoking, alcohol consumption, and urban/​ rural residence. Dermato-​Endocrinology, 2(2), 68–​76. Hankinson SE, and Danforth KN. 2006. Ovarian cancer. In:  Schottenfeld D, and Fraumeni JF (Eds.), Cancer epidemiology and prevention (3rd ed.). Oxford; New York: Oxford University Press. Harrison ML, Jameson C, and Gore ME. 2008. Mucinous ovarian cancer. Int J Gynecol Cancer, 18(2), 209–​214. PMID: 17624989. Hartge P, Whittemore AS, Itnyre J et al. 1994. Rates and risks of ovarian cancer in subgroups of white women in the United States. Obstet Gynecol, 84, 760–​764. Hildebrand JS, Gapstur SM, Feigelson HS, et al. 2010. Postmenopausal hormone use and incident ovarian cancer: associations differ by regimen. Int J Cancer, 127(12), 2928–​2935. PMID: 21351271. Hill MJ, Goodard R, and Williams REO. 1971. Gut bacteria and aetiology of cancer of the breast. Lancet, 2(7722), 472–​473. Houghton SC, Reeves KW, Hankinson SE, et al. 2014. Perineal powder use and risk of ovarian cancer. J Natl Cancer Inst, 106(9). PMCID: PMC4200058. Howlader N, Noone AM, Krapcho M, et  al. 2015. SEER Cancer Statistics Review, 1975–​2012, National Cancer Institute. Bethesda, MD, http://​seer. cancer.gov/​csr/​1975_​2012/​, based on November 2014 SEER data submission, posted to the SEER website, April 2015. Hua X, Yu L, You R, et  al. 2016. Association among dietary flavonoids, flavonoid subclasses and ovarian cancer risk:  a meta-​analysis. PLoS One, 11(3), e0151134. PMCID: PMC4784737. Huang T, Poole EM, Eliassen AH, et al. 2016. Hypertension, use of antihypertensive medications, and risk of epithelial ovarian cancer. Int J Cancer. 139(2), 291–​299. PMID: 26934358. Huang T, Poole EM, Okereke OI, et al. 2015. Depression and risk of epithelial ovarian cancer:  Results from two large prospective cohort studies. Gynecol Oncol, 139(3), 481–​486. PMCID: PMC4679423. Huang Z, Gao Y, Wen W, et al. 2015. Contraceptive methods and ovarian cancer risk among Chinese women: a report from the Shanghai Women’s Health Study. Int J Cancer, 137(3), 607–​614. PMCID: PMC4437849. IARC. 1997. IARC Scientific Publications:  social inequalities and cancer. Kogevinas M, Pearce N, Susser M, and Boffetta P (Eds.), Vol. 138. Lyon: International Agency for Research on Cancer. IARC. 2002. Cancer incidence in five continents, Vol. VIII. Parkin DM, Whelan SL, Ferlay J, Teppo L, and Thomas DB (Eds.). Lyon:  International Agency for Research on Cancer. Ibfelt EH, Dalton SO, Hogdall C, et al. 2015. Do stage of disease, comorbidity or access to treatment explain socioeconomic differences in survival after ovarian cancer? A cohort study among Danish women diagnosed 2005–​ 2010. Cancer Epidemiol, 39(3), 353–​359. PMID: 25841586. Idahl A, Lundin E, Jurstrand M, et  al. 2011. Chlamydia trachomatis and Mycoplasma genitalium plasma antibodies in relation to epithelial ovarian tumors. Infec Dis Obstet Gynecol, 2011, 824627. PMCID: PMC3147007. Jacobs IJ, Menon U, Ryan A, et  al. 2016. Ovarian cancer screening and mortality in the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS):  a randomised controlled trial. Lancet, 387(10022), 945–​ 956. PMCID: PMC4779792. Jacoby VL, Grady D, Wactawski-​Wende J, et al. 2011. Oophorectomy vs ovarian conservation with hysterectomy. Arch Intern Med, 171(8), 760–​768. Jarboe EA, Folkins AK, Drapkin R, et al. 2008. Tubal and ovarian pathways to pelvic epithelial cancer:  a pathological perspective. Histopathology, 53(2), 127–​138. PMID: 18298580. Jensen KE, Hannibal CG, Nielsen A, et al. 2008. Social inequality and incidence of and survival from cancer of the female genital organs in a population-​based study in Denmark, 1994–​2003. Eur J Cancer, 44(14), 2003–​2017. PMID: 18662869.

903

Jervis S, Song H, Lee A, et al. 2014. Ovarian cancer familial relative risks by tumour subtypes and by known ovarian cancer genetic susceptibility variants. J Med Genet, 51(2), 108–​113. PMID: 24277755. Ji J, Forsti A, Sundquist J, Lenner P, and Hemminki K. 2008. Survival in ovarian cancer patients by histology and family history. Acta Oncol, 47(6), 1133–​1139. PMID: 18607880. Johnatty SE, Beesley J, Chen X, et al. 2010. Evaluation of candidate stromal epithelial cross-​talk genes identifies association between risk of serous ovarian cancer and TERT, a cancer susceptibility “hot-​spot.” PLoS Genet, 6(7), e1001016. PMCID: PMC2900295. Jordan SJ, Nagle CM, Coory MD, et al. 2013. Has the association between hysterectomy and ovarian cancer changed over time? A  systematic review and meta-​ analysis. Eur J Cancer, 49(17), 3638–​ 3647. PMID: 23890943. Kelemen LE, Bandera EV, Terry KL, et al. 2013. Recent alcohol consumption and risk of incident ovarian carcinoma: a pooled analysis of 5,342 cases and 10,358 controls from the Ovarian Cancer Association Consortium. BMC Cancer, 13, 28. PMCID: PMC3568733. Kelemen LE, Lawrenson K, Tyrer J, et al. 2015. Genome-​wide significant risk associations for mucinous ovarian carcinoma. Nat Genet, 47(8), 888–​897. PMCID: PMC4520768. Kindelberger DW, Lee Y, Miron A, et al. 2007. Intraepithelial carcinoma of the fimbria and pelvic serous carcinoma: evidence for a causal relationship. Am J Surg Pathol, 31, 161–​169. Kobayashi H, Yamada Y, Sado T, et al. 2008. A randomized study of screening for ovarian cancer: a multicenter study in Japan. Int J Gynecol Cancer, 18(3), 414–​420. PMID: 17645503. Kobel M, Kalloger SE, Baker PM, et al. 2010. Diagnosis of ovarian carcinoma cell type is highly reproducible: a transcanadian study. Am J Surg Pathol, 34(7), 984–​993. PMID: 20505499. Kobel M, Kalloger SE, Lee S, et al. 2013. Biomarker-​based ovarian carcinoma typing:  a histologic investigation in the ovarian tumor tissue analysis consortium. Cancer Epidemiol Biomarkers Prev, 22(10), 1677–​1686. PMCID: PMC3955399. Kosary CL. 2007. Cancer of the ovary. In: Ries LAG, Young JL, Keel GE, Eisner MP, Lin YD, and Horner M-​J (Eds.), SEER Survival Monograph: cancer survival among adults:  U.S. SEER Program, 1988–​2001, patient and tumor characteristics. National Cancer Institute, SEER Program, NIH Pub. No. 07-​6215, Bethesda, MD. Kotsopoulos J, Lubinski J, Gronwald J, et al. 2014. Factors influencing ovulation and the risk of ovarian cancer in BRCA1 and BRCA2 mutation carriers. Int J Cancer. PMCID: PMC4458227. Kotsopoulos J, Terry KL, Poole EM, et al. 2013. Ovarian cancer risk factors by tumor dominance, a surrogate for cell of origin. Int J Cancer, 133(3), 730–​739. PMCID: PMC3690361. Kotsopoulos J, Vitonis AF, Terry KL, et  al. 2009. Coffee intake, variants in genes involved in caffeine metabolism, and the risk of epithelial ovarian cancer. Cancer Causes Control, 20(3), 335–​344. PMCID: PMC2692932. Koushik A, Hunter DJ, Spiegelman D, et al. 2005. Fruits and vegetables and ovarian cancer risk in a pooled analysis of 12 cohort studies. Cancer Epidemiol Biomarkers Prev, 14(9), 2160–​2167. PMID: 16172226. Koushik A, Hunter DJ, Spiegelman D, et al. 2006. Intake of the major carotenoids and the risk of epithelial ovarian cancer in a pooled analysis of 10 cohort studies. Int J Cancer, 119(9), 2148–​2154. PMID: 16823847. Koushik A, Wang M, Anderson KE, et al. 2015. Intake of vitamins A, C, and E and folate and the risk of ovarian cancer in a pooled analysis of 10 cohort studies. Cancer Causes Control, 26(9), 1315–​1327. PMID: 26169298. Kuchenbaecker KB, Ramus SJ, Tyrer J, et al. 2015. Identification of six new susceptibility loci for invasive epithelial ovarian cancer. Nat Genet, 47(2), 164–​171. PMCID: PMC4445140. Kuper H, Titus-​Ernstoff L, Harlow BL, and Cramer D. 2000. Population based study of coffee, alcohol and tobacco use and risk of ovarian cancer. Int J Cancer, 88, 313–​318. Kurman RJ, and Shih IM. 2008. Pathogenesis of ovarian cancer: lessons from morphology and molecular biology and their clinical implications. Int J Gynecol Pathol, 27(2), 151–​160. PMCID: PMC2794425. Lancaster JM, Powell CB, Chen LM, Richardson DL, and Committee SGOCP. 2015. Society of Gynecologic Oncology statement on risk assessment for inherited gynecologic cancer predispositions. Gynecol Oncol, 136(1), 3–​ 7. PMID: 25238946. Land JA, Van Bergen JE, Morre SA, and Postma MJ. 2010. Epidemiology of Chlamydia trachomatis infection in women and the cost-​effectiveness of screening. Human Reprod Update, 16(2), 189–​204. PMID: 19828674. Larsson SC, and Wolk A. 2005. Coffee consumption is not associated with ovarian cancer incidence. Cancer Epidemiol Biomarkers Prev, 14(9), 2273–​2274. PMID: 16172244.

904

904

PART IV:  Cancers by Tissue of Origin

Le GM, Gomez SL, Clarke CA, Glaser SL, and West DW. 2002. Cancer incidence patterns among Vietnamese in the United States and Ha Noi, Vietnam. Int J Cancer, 102(4), 412–​417. PMID: 12402312. Lee KR, and Young RH. 2003. The distinction between primary and metastatic mucinous carcinomas of the ovary. Am J Surg Pathol, 27(3), 281–​292. Lee KW, Bode AM, and Dong Z. 2011. Molecular targets of phytochemicals for cancer prevention. Nature Rev Cancer, 11(3), 211–​218. PMID: 21326325. Lee Y, Miron A, Drapkin R, et al. 2007. A candidate precursor to serous carcinoma that originates in the distal fallopian tube. J Pathol, 211(1), 26–​35. PMID: 17117391. Li D-​P, Du C, Zhang Z-​M, et al. 2014. Breastfeeding and ovarian cancer risk: a systematic review and meta-​analysis of 40 epidemiological studies. Asian Pacific J Cancer Prev, 15(12), 4829–​4837. Li K, Husing A, Fortner RT, et  al. 2015. An epidemiologic risk prediction model for ovarian cancer in Europe: the EPIC study. Br J Cancer, 112(7), 1257–​1265. PMCID: PMC4385951. Li S, Xu H, Li SC, Qi XQ, and Sun WJ. 2014. Vitamin D receptor rs2228570 polymorphism and susceptibly to ovarian cancer:  a meta-​ analysis. Tumour Biol, 35(2), 1319–​1322. PMID: 24136742. Lin HW, Tu YY, Lin SY, et al. 2011. Risk of ovarian cancer in women with pelvic inflammatory disease: a population-​based study. Lancet Oncol, 12(9), 900–​904. PMID: 21835693. Lin SW, Wheeler DC, Park Y, et al. 2012. Prospective study of ultraviolet radiation exposure and risk of cancer in the United States. Int J Cancer, 131(6), E1015–​E1023. Liu J, Tang W, Sang L, et  al. 2015. Milk, yogurt, and lactose intake and ovarian cancer risk:  a meta-​ analysis. Nutr Cancer, 67(1), 68–​ 72. PMID: 25298278. Liu Y, Li C, Chen P, et  al. 2013. Polymorphisms in the vitamin D Receptor (VDR) and the risk of ovarian cancer: a meta-​analysis. PLoS One, 8(6), e66716. PMCID: PMC3691226. Lu Y, Cuellar-​Partida G, Painter JN, et  al. 2015. Shared genetics underlying epidemiological association between endometriosis and ovarian cancer. Human Mol Genet, 24(20), 5955–​5964. PMCID: PMC4581608. Luan NN, Wu QJ, Gong TT, et  al. 2013. Breastfeeding and ovarian cancer risk:  a meta-​analysis of epidemiologic studies. Am J Clin Nutr, 98(4), 1020–​1031. PMCID: PMC3778857. Lueth NA, Anderson KE, Harnack LJ, Fulkerson JA, and Robien K. 2008. Coffee and caffeine intake and the risk of ovarian cancer:  the Iowa Women’s Health Study. Cancer Causes Control, 19(10), 1365–​1372. PMCID: PMC2581636. Lurie G, Thompson P, McDuffie KE, et al. 2007. Association of estrogen and progestin potency of oral contraceptives with ovarian carcinoma risk. Obstet Gynecol, 109, 597–​607. Madsen C, Baandrup L, Dehlendorff C, and Kjaer SK. 2015. Tubal ligation and salpingectomy and the risk of epithelial ovarian cancer and borderline ovarian tumors:  a nationwide case-​control study. Acta Obstet Gynecol Scand, 94(1), 86–​94. PMID: 25256594. Malpica A, Deavers MT, Tornos C, et al. 2007. Interobserver and intraobserver variability of a two-​tier system for grading ovarian serous carcinoma. Am J Surg Pathol, 31(8), 1168–​1174. PMID: 17667538. Manavi K. 2006. A review on infection with Chlamydia trachomatis:  best practice & research. Clin Obstet Gyn, 20(6), 941–​ 951. PMID: 16934531. Marchbanks PA, Wilson H, Bastos E, et al. 2000. Cigarette smoking and epithelial ovarian cancer by histologic type. Obstet Gynecol, 95, 255–​260. Martinez-​Delgado B, Yanowsky K, Inglada-​Perez L, et al. 2012. Shorter telomere length is associated with increased ovarian cancer risk in both familial and sporadic cases. J Med Genet, 49(5), 341–​344. PMID: 22493152. Matsuno RK, Sherman ME, Visvanathan K, et al. 2013. Agreement for tumor grade of ovarian carcinoma: analysis of archival tissues from the surveillance, epidemiology, and end results residual tissue repository. Cancer Causes Control, 24(4), 749–​757. PMCID: PMC4000689. Maxwell MB, and Maher KE. 1992. Chemotherapy-​induced myelosuppression. Semin Oncol Nurs, 8(2), 113–​123. PMID: 1621002. McCluggage WG. 2011. Morphological subtypes of ovarian carcinoma:  a review with emphasis on new developments and pathogenesis. Pathology, 43(5), 420–​432. McConechy MK, Anglesio MS, Kalloger SE, et  al. 2011. Subtype-​specific mutation of PPP2R1A in endometrial and ovarian carcinomas. J Pathol, 223(5), 567–​573. PMID: 21381030. Menon U, Gentry-​Maharaj A, Hallett R, et al. 2009. Sensitivity and specificity of multimodal and ultrasound screening for ovarian cancer, and stage distribution of detected cancers: results of the prevalence screen of the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS). Lancet Oncol, 10, 327–​340.

Merritt MA, Cramer DW, Vitonis AF, Titus LJ, and Terry KL. 2013a. Dairy foods and nutrients in relation to risk of ovarian cancer and major histological subtypes. Int J Cancer, 132(5), 1114–​1124. PMCID: PMC3495088. Merritt MA, De Pari M, Vitonis AF, et al. 2013b. Reproductive characteristics in relation to ovarian cancer risk by histologic pathways. Hum Reprod, 28(5), 1406–​1417. PMCID: PMC3627336. Merritt MA, Green AC, Nagle CM, and Webb PM. 2008. Talcum powder, chronic pelvic inflammation and NSAIDs in relation to risk of epithelial ovarian cancer. Int J Cancer, 122(1), 170–​176. PMID: 17721999. Merritt MA, Riboli E, Weiderpass E, et  al. 2014. Dietary fat intake and risk of epithelial ovarian cancer in the European Prospective Investigation into Cancer and Nutrition. Cancer Epidemiol, 38(5), 528–​ 537. PMID: 25155210. Mirabello L, Garcia-​Closas M, Cawthon R, et al. 2010. Leukocyte telomere length in a population-​based case-​control study of ovarian cancer: a pilot study. Cancer Causes Control, 21(1), 77–​82. PMCID: PMC3130499. Moorman PG, Palmieri RT, Akushevich L, Berchuck A, and Schildkraut JM. 2009. Ovarian cancer risk factors in African-​American and white women. Am J Epidemiol, 170(5), 598–​606. PMCID: PMC2732987. Mun MJ, Kim TH, Hwang JY, and Jang WC. 2015. Vitamin D receptor gene polymorphisms and the risk of female reproductive cancers:  a meta-​ analysis. Maturitas, 81(2), 256–​265. Murphy MA, Trabert B, Yang HP, et al. 2012. Non-​steroidal anti-​inflammatory drug use and ovarian cancer risk: findings from the NIH-​AARP Diet and Health Study and systematic review. Cancer Causes Control, 23(11), 1839–​1852. PMCID: PMC3469773. National Academies of Sciences Engineering, and Medicine. 2016. Ovarian cancers: evolving paradigms in research and care. Washington, DC: The National Academies Press. Neergheen VS, Bahorun T, Taylor EW, Jen LS, and Aruoma OI. 2010. Targeting specific cell signaling transduction pathways by dietary and medicinal phytochemicals in cancer chemoprevention. Toxicology, 278(2), 229–​ 241. PMID: 19850100. Nelson HD, Pappas M, Zakher B, et al. 2014. Risk assessment, genetic counseling, and genetic testing for BRCA-​related cancer in women:  a systematic review to update the U.S. Preventive Services Task Force Recommendation. Ann Intern Med, 160, 255–​266. Ness RB, Dodge RC, Edwards RP, Baker JA, and Moysich KB. 2011. Contraception methods, beyond oral contraceptives and tubal ligation, and risk of ovarian cancer. Ann Epidemiol, 21(3), 188–​ 196. PMCID: PMC3052991. Ness RB, Grisso JA, Cottreau C et al. 2000. Factors related to inflammation of the ovarian epithelium and risk of ovarian cancer. Epidemiology, 11(2), 111–​117. PMID: 11021606. Ness RB, Grisso JA, Vergona R, et al. 2001. Oral contraceptives, other methods of contraception, and risk reduction for ovarian cancer. Epidemiology, 12(3), 307–​312. PMID: 11337604. Neuhouser ML, Wassertheil-​Smoller S, Thomson C, et al. 2009. Multivitamin use and risk of cancer and cardiovascular disease in the Women’s Health Initiative cohorts. Arch Intern Med, 169(3), 294–​ 304. PMCID: PMC3868488. Ni X, Ma J, Zhao Y, Wang Y, and Wang S. 2013. Meta-​analysis on the association between non-​steroidal anti-​inflammatory drug use and ovarian cancer. Br J Clin Pharmacol, 75(1), 26–​35. PMCID: PMC3555044. Obon-​Santacana M, Lujan-​Barroso L, Travis RC, et  al. 2016. Acrylamide and glycidamide hemoglobin adducts and epithelial ovarian cancer:  a nested case-​control study in nonsmoking postmenopausal women from the EPIC Cohort. Cancer Epidemiol Biomarkers Prev, 25(1), 127–​134. PMID: 26598536. Obon-​Santacana M, Peeters PH, Freisling H, et  al. 2015. Dietary intake of acrylamide and epithelial ovarian cancer risk in the european prospective investigation into cancer and nutrition (EPIC) cohort. Cancer Epidemiol Biomarkers Prev, 24(1), 291–​297. PMCID: PMC4295892. Society of Gynecologic Oncology 2013. SGO Clinical Practice Statement: Salpingectomy for Ovarian Cancer Prevention. Available from: https:// ​ w ww.sgo.org/ ​ c linical- ​ p ractice/ ​ g uidelines/ ​ s go- ​ c linical-​ practice-statement-​s alpingectomy-​f or-​o varian-​c ancer-​p revention. Accessed October 1, 2013. Ose J, Fortner RT, Rinaldi S, et al. 2015a. Endogenous androgens and risk of epithelial invasive ovarian cancer by tumor characteristics in the European Prospective Investigation into Cancer and Nutrition. Int J Cancer, 136(2), 399–​410. PMID: 24890047. Ose J, Fortner RT, Schock H, et al. 2015b. Insulin-​like growth factor I and risk of epithelial invasive ovarian cancer by tumour characteristics: results from the EPIC cohort. Br J Cancer, 112(1), 162–​ 166. PMCID: PMC4453611.

 905

Ovarian Cancer Ose J, Schock H, Tjonneland A, et al. 2015c. Inflammatory markers and risk of epithelial ovarian cancer by tumor subtypes: The EPIC Cohort. Cancer Epidemiol Biomarkers Prev, 24(6), 951–​961. PMCID: PMC4454588. Pal T, Permuth-​Wey J, Betts JA, et al. 2005. BRCA1 and BRCA2 mutations account for a large proportion of ovarian carcinoma cases. Cancer, 104(12), 2807–​2816. PMID: 16284991. Parazzini F, La Vecchia C, Negri E, et al. 1996. Pelvic inflammatory disease and risk of ovarian cancer. Cancer Epidemiol Biomarkers Prev, 5(8), 667–​ 669. PMID: 8824371. Parker WH. 2010. Bilateral oophorectomy versus ovarian conservation: effects on long-​term women’s health. J Minim Invasive Gynecol, 17(2), 161–​166. Parker WH, Broder MS, Chang E, et al. 2009. Ovarian conservation at the time of hysterectomy and long-​term health outcomes in the nurses’ health study. Obstet Gynecol, 113(5), 1027–​1037. PMCID: PMC3791619. Parker WH, Feskanich D, Broder MS, et al. 2013. Long-​term mortality associated with oophorectomy compared with ovarian conservation in the nurses’ health study. Obstet Gynecol, 121(4), 709–​716. PMCID: PMC4254662. Pavone ME, and Lyttle BM. 2015. Endometriosis and ovarian cancer:  links, risks, and challenges faced. Int J Womens Health, 7, 663–​ 672. PMCID: PMC4494101. Pearce CL, Templeman C, Rossing MA, et al. 2012. Association between endometriosis and risk of histological subtypes of ovarian cancer:  a pooled analysis of case-​control studies. Lancet Oncol, 13, 385–​394. Peeters PH, Lukanova A, Allen N, et al. 2007. Serum IGF-​I, its major binding protein (IGFBP-​3) and epithelial ovarian cancer risk:  the European Prospective Investigation into Cancer and Nutrition (EPIC). Endocr-​Relat Cancer, 14(1), 81–​90. PMID: 17395977. Pelucchi C, Bosetti C, Galeone C, and La Vecchia C. 2015. Dietary acrylamide and cancer risk: an updated meta-​analysis. Int J Cancer, 136(12), 2912–​ 2922. PMID: 25403648. Permuth-​Wey J, Lawrenson K, Shen HC, et al. 2013. Identification and molecular characterization of a new ovarian cancer susceptibility locus at 17q21.31. Nat Commun, 4, 1627. PMCID: PMC3709460. Permuth-​Wey J, and Sellers TA. 2009. Chapter 20: Epidemiology of ovarian cancer. In: Verma M (Ed.), Methods of molecular biology, cancer epidemiology, Vol. 472. Totowa, NJ: Humana Press. Pfeiffer RM, Park Y, Kreimer AR, et al. 2013. Risk prediction for breast, endometrial, and ovarian cancer in white women aged 50 y or older: derivation and validation from population-​based cohort studies. PLoS Med, 10(7), e1001492. PMCID: PMC3728034. Pharoah PD, Tsai YY, Ramus SJ, et al. 2013. GWAS meta-​analysis and replication identifies three new susceptibility loci for ovarian cancer. Nat Genet, 45(4), 362–​370, 370e361-​362. PMCID: PMC3693183. Pike MC, Pearce CL, Peters R, et al. 2004. Hormonal factors and the risk of invasive ovarian cancer:  a population-​based case-​control study. Fertil Steril, 82(1), 186–​195. PMID: 15237010. Pinheiro SP, Tworoger SS, Cramer DW, Rosner BA, and Hankinson SE. 2009. Use of nonsteroidal antiinflammatory agents and incidence of ovarian cancer in 2 large prospective cohorts. Am J Epidemiol, 169(11), 1378–​ 1387. PMCID: PMC2727251. Poole EM, Kubzansky LD, Sood AK, Okereke OI, and Tworoger SS. 2016. A prospective study of phobic anxiety, risk of ovarian cancer, and survival among patients. Cancer Causes Control, 27(5), 661–​ 668. PMID: 27023470. Poole EM, Lee IM, Ridker PM, et al. 2013a. A prospective study of circulating C-​reactive protein, interleukin-​6, and tumor necrosis factor alpha receptor 2 levels and risk of ovarian cancer. Am J Epidemiol, 178(8), 1256–​1264. PMCID: PMC3792725. Poole EM, Merritt MA, Jordan SJ, et  al. 2013b. Hormonal and reproductive risk factors for epithelial ovarian cancer by tumor aggressiveness. Cancer Epidemiol Biomarkers Prev, 22(3), 429–​437. PMCID: PMC3596470. Poole EM, Rice MS, Crum CP, and Tworoger SS. 2015. Salpingectomy as a potential ovarian cancer risk-​reducing procedure. J Natl Cancer Inst, 107(2), dju490. Prentice RL, Thomson CA, Caan B et  al. 2007. Low-​fat dietary pattern and cancer incidence in the Women’s Health Initiative Dietary Modification Randomized Controlled Trial. J Natl Cancer Inst, 99(20), 1534–​1543. PMCID: PMC2670850. Prescott J, Bertrand KA, Poole EM, Rosner BA, and Tworoger SS. 2013. Surrogates of long-​term vitamin d exposure and ovarian cancer risk in two prospective cohort studies. Cancers (Basel), 5(4), 1577–​1600. PMCID: PMC3875955. Preston DL, Ron E, Tokuoka S, et  al. 2007. Solid cancer incidence in atomic bomb survivors: 1958–​1998. Radiat Res, 168(1), 1–​64. PMID: 17722996. Prizment AE, Folsom AR, and Anderson KE. 2010. Nonsteroidal anti-​ inflammatory drugs and risk for ovarian and endometrial cancers in the

905

Iowa Women’s Health Study. Cancer Epidemiol Biomarkers Prev, 19(2), 435–​442. PMCID: PMC2820129. Qin X, Lu Y, Qin A, et al. 2013. Vitamin D receptor Bsmcapital I, Ukrainian polymorphism and ovarian cancer risk:  a meta-​analysis. Int J Gynecol Cancer, 23(7), 1178–​1183. PMID: 23873178. Raja FA, Chopra N, and Ledermann JA. 2012. Optimal first-​line treatment in ovarian cancer. Ann Oncol, 23 Suppl 10, x118–​127. PMID: 22987945. Rasmussen CB, Faber MT, Jensen A, et al. 2013. Pelvic inflammatory disease and risk of invasive ovarian cancer and ovarian borderline tumors. Cancer Causes Control, 24(7), 1459–​1464. PMID: 23615817. Rebbeck TR, Kauff ND, and Domchek SM. 2009. Meta-​analysis of risk reduction estimates associated with risk-​reducing salpingo-​oophorectomy in BRCA1 or BRCA2 mutation carriers. J Natl Cancer Inst, 101(2), 80–​87. PMCID: PMC2639318. Reigstad MM, Larsen IK, Myklebust TA, et al. 2015. Cancer risk among parous women following assisted reproductive technology. Hum Reprod, 30(8), 1952–​1963. PMCID: PMC4507330. Reljic M, and Gorisek B. 1998. C-​reactive protein and the treatment of pelvic inflammatory disease. Int J Gyn Obstet, 60(2), 143–​150. PMID: 9509952. Rice MS, Hankinson SE, and Tworoger SS. 2014. Tubal ligation, hysterectomy, unilateral oophorectomy, and risk of ovarian cancer in the Nurses’ Health Studies. Fertil Steril, 102(1), 192–​198 e193. PMCID: PMC4074555. Rice MS, Murphy MA, and Tworoger SS. 2012. Tubal ligation, hysterectomy and ovarian cancer: A meta-​analysis. J Ovar Res, 5, 13. Rich-​Edwards JW, Spiegelman D, Garland M, et  al. 2002. Physical activity, body mass index, and ovulatory disorder infertility. Epidemiology, 13, 184–​190. Risch HA, and Howe GR. 1995. Pelvic inflammatory disease and the risk of epithelial ovarian cancer. Cancer Epidemiol Biomarkers Prev, 4(5), 447–​ 451. PMID: 7549798. Rogol AD, Weltman A, Weltman JY, et al. 1992. Durability of the reproductive axis in eumenorrheic women during 1 yr of endurance training. J Appl Physiol (1985), 72, 1571–​1580. Romaguera D, Vergnaud AC, Peeters PH, et  al. 2012. Is concordance with World Cancer Research Fund/​American Institute for Cancer Research guidelines for cancer prevention related to subsequent risk of cancer? Results from the EPIC study. Am J Clin Nutr, 96(1), 150–​163. PMID: 22592101. Romero IL, Gordon IO, Jagadeeswaran S, et al. 2009. Effects of oral contraceptives or a gonadotropin-​releasing hormone agonist on ovarian carcinogenesis in genetically engineered mice. Cancer Prev Res, 2(9), 792–​799. PMCID: PMC2758654. Rosner BA, Colditz GA, Webb PM, and Hankinson SE. 2005. Mathematical models of ovarian cancer incidence. Epidemiology, 16(4), 508–​515. Rossing MA, Cushing-​Haugen KL, Wicklund KG, Doherty JA, and Weiss NS. 2008. Risk of epithelial ovarian cancer in relation to benign ovarian conditions and ovarian surgery. Cancer Causes Control, 19(10), 1357–​1364. PMCID: PMC2751585. Rossing MA, Tang MT, Flagg EW, Weiss LK, and Wicklund KG. 2004. A case-​control study of ovarian cancer in relation to infertility and the use of ovulation-​inducing drugs. Am J Epidemiol, 160(11), 1070–​1078. PMID: 15561986. Rowlands IJ, Nagle CM, Spurdle AB, and Webb PM. 2011. Gynecological conditions and the risk of endometrial cancer. Gynecol Oncol, 123(3), 537–​541. PMID: 21925719. Royar J, Becher H, and Chang-​Claude J. 2001. Low-​dose oral contraceptives: protective effect on ovarian cancer risk. Int J Cancer (Pred Oncol), 95, 370–​374. Sainz de la Cuesta R, Eichhorn JH, Rice LW, et al. 1996. Histologic transformation of benign endometriosis to early epithelial ovarian cancer. Gynecol Oncol, 60, 238–​244. Sanderson M, Williams MA, Weiss NS, Hendrix NW, and Chauhan SP. 2000. Oral contraceptives and epithelial ovarian cancer. Does dose matter? J Reprod Med, 45, 720–​726. Schildkraut JM, Calingaert B, Marchbanks PA, Moorman PG, and Rodiguez GC. 2002. Impact of progestin and estrogen potency in oral contraceptives on ovarian cancer risk. J Natl Cancer Inst, 94, 32–​38. Schock H, Fortner RT, Surcel HM, et  al. 2015. Early pregnancy IGF-​I and placental GH and risk of epithelial ovarian cancer: a nested case-​control study. Int J Cancer, 137(2), 439–​447. PMCID: PMC4428944. Schock H, Lundin E, Vaarasmaki M, et al. 2014a. Anti-​Mullerian hormone and risk of invasive serous ovarian cancer. Cancer Causes Control, 25(5), 583–​589. PMID: 24562905. Schock H, Surcel HM, Zeleniuch-​Jacquotte A, et al. 2014b. Early pregnancy sex steroids and maternal risk of epithelial ovarian cancer. Endocr-​Relat Cancer, 21(6), 831–​844. PMCID: PMC4282682.

906

906

PART IV:  Cancers by Tissue of Origin

Schouten LJ, Rivera C, Hunter DJ, et al. 2008. Height, body mass index, and ovarian cancer: a pooled analysis of 12 cohort studies. Cancer Epidemiol Biomarkers Prev, 17(4), 902–​912. PMCID: PMC2572258. Schrepf A, Thaker PH, Goodheart MJ, et al. 2015. Diurnal cortisol and survival in epithelial ovarian cancer. Psychoneuroendocrinology, 53, 256–​267. PMCID: PMC4440672. Seidman JD, Kurman RJ, and Ronnett BM. 2003. Pirmary and metastatic mucinous adenocarcinomas in the ovaries: incidence in routine practice with a new approach to improve intraoperative diagnosis. Am J Surg Pathol, 27, 985–​993. Setiawan VW, Matsuno RK, Lurie G, et  al. 2012. Use of nonsteroidal anti-​ inflammatory drugs and risk of ovarian and endometrial cancer:  the Multiethnic Cohort. Cancer Epidemiol Biomarkers Prev, 21(9), 1441–​ 1449. PMCID: PMC3436975. Shanmughapriya S, Senthilkumar G, Vinodhini K, et al. 2012. Viral and bacterial aetiologies of epithelial ovarian cancer. Eur J Clin Microbiol, 31(9), 2311–​2317. PMID: 22402815. Shen CC, Yang AC, Hung JH, Hu LY, and Tsai SJ. 2015. A nationwide population-​based retrospective cohort study of the risk of uterine, ovarian and breast cancer in women with polycystic ovary syndrome. Oncologist, 20(1), 45–​49. PMCID: PMC4294614. Shen H, Fridley BL, Song H, et al. 2013. Epigenetic analysis leads to identification of HNF1B as a subtype-​specific susceptibility gene for ovarian cancer. Nat Commun, 4, 1628. PMCID: PMC3848248. Shih I-​M, and Kurman RJ. 2004. Ovarian tumorigenesis. Am J Pathol, 164(5), 1511–​1518. Shih IM, Panuganti PK, Kuo KT, et al. 2011. Somatic mutations of PPP2R1A in ovarian and uterine carcinomas. Am J Pathol, 178(4), 1442–​1447. PMCID: PMC3078449. Shu XO, Brinton LA, Gao YT, and Yuan JM. 1989. Population-​based case-​ control study of ovarian cancer in Shanghai. Cancer Res, 49(13), 3670–​ 3674. PMID: 2731180. Sieh W, Salvador S, McGuire V, et al. 2013. Tubal ligation and risk of ovarian cancer subtypes:  a pooled analysis of case-​control studies. Int J Epidemiol, 42(2), 579–​589. PMCID: PMC3619957. Singh SD, Ryerson AB, Wu M, and Kaur JS. 2014. Ovarian and uterine cancer incidence and mortality in American Indian and Alaska Native women, United States, 1999–​2009. Am J Public Health, 104, S423–​S431. Smith HO, Berwick M, Verschraegen CF, et al. 2006. Incidence and survival rates for female malignant germ cell tumors. Obstet Gynecol, 107(5), 1075–​1085. Soegaard M, Frederiksen K, Jensen A, et  al. 2009. Risk of ovarian cancer in women with first-​degree relatives with cancer. Acta Obstet Gynecol Scand, 88(4), 449–​456. PMID: 19266357. Søgaard M, Krüger Kjær S, and Gayther S. 2006. Ovarian cancer and genetic susceptibility in relation to the BRCA1 and BRCA2 genes: occurrence, clinical importance and intervention. Acta Obstet Gynecol Scand, 85(1), 93–​105. Soler M, Chatenoud L, Negri E, et al. 1999. Hypertension and hormone-​related neoplasms in women. Hypertension, 34(2), 320–​325. PMID: 10454461. Soliman NF, and Evans AJ. 2004. Malignancy arising in residual endometriosis following hysterectomy and hormone replacement therapy. J Br Menopause Soc, 10(3), 123–​124. Somigliana E, Vigano P, Parazzini F, et al. 2006. Association between endometriosis and cancer: a comprehensive review and a critical analysis of clinical and epidemiological evidence. Gynecol Oncol, 101(2), 331–​341. PMID: 16473398. Song GG, and Lee YH. 2013. Vitamin D receptor Fokl, Bsml, Apal, and Taql polymorphorisms and susceptibility to ovarian cancer:  a meta-​analysis. Immunol Invest, 42(7), 661–​672. Song H, Ramus SJ, Tyrer J, et al. 2009. A genome-​wide association study identifies a new ovarian cancer susceptibility locus on 9p22.2. Nat Genet, 41(9), 996–​1000. PMCID: PMC2844110. Sood AK, Bhatty R, Kamat AA et  al. 2006. Stress hormone-​mediated invasion of ovarian cancer cells. Clin Cancer Res, 12(2), 369–​ 375. PMCID: PMC3141061. Sopik V, Iqbal J, Rosen B, and Narod SA. 2015a. Why have ovarian cancer mortality rates declined? Part I. Incidence. Gynecol Oncol, 138(3), 741–​ 749. PMID: 26080287. Sopik V, Iqbal J, Rosen B, and Narod SA. 2015b. Why have ovarian cancer mortality rates declined? Part II. Case-​ fatality. Gynecol Oncol. PMID: 26080288. Soslow RA. 2008. Histologic subtypes of ovarian carcinoma: an overview. Int J Gynecol Pathol, 27(2), 161–​174. PMID: 18317227. Steevens J, Schouten LJ, Verhage BA, Goldbohm RA, and van den Brandt PA. 2007. Tea and coffee drinking and ovarian cancer risk: results from the Netherlands Cohort Study and a meta-​analysis. Br J Cancer, 97(9), 1291–​ 1294. PMCID: PMC2360476.

Straif K, Benbrahim-​ Tallaa L, Baan R, et  al. 2009. A review of human carcinogens—​ Part C:  metals, arsenic, dusts, and fibres. Lancet, 10, 453–​454. Taylor N, and Mutch DG. 2006. Gynecologic manifestations of hereditary nonpolyposis colorectal cancer: from inherited to sporadic disease. Oncology (Williston Park), 20(1), 85–​94. Terry KL, Karageorgi S, Shvetsov YB, et al. 2013. Genital powder use and risk of ovarian cancer: a pooled analysis of 8,525 cases and 9,859 controls. Cancer Prev Res, 6(8), 811–​821. PMCID: PMC3766843. Terry KL, Tworoger SS, Vitonis AF, et  al. 2012. Telomere length and genetic variation in telomere maintenance genes in relation to ovarian cancer risk. Cancer Epidemiol Biomarkers Prev, 21(3), 504–​512. PMCID: PMC3297716. Terry PD, Miller AB, JOnes JG, and Rohan TE. 2003. Cigarette smoking and risk of invasive epithelial ovarian cancer in a prospective cohort study. Eur J Cancer, 39, 1157–​1164. Thaker PH, Han LY, Kamat AA, et  al. 2006. Chronic stress promotes tumor growth and angiogenesis in a mouse model of ovarian carcinoma. Nat Med, 12(8), 939–​944. PMID: 16862152. Thaker PH, Lutgendorf SK, and Sood AK. 2007. The neuroendocrine impact of chronic stress on cancer. Cell Cycle, 6(4), 430–​433. PMID: 17312398. Tingulstad S, Skjeldestad FE, Halvorsen TB, and Hagen B. 2003. Survival and prognostic factors in patients with ovarian cancer. Obstet Gynecol, 101(5 Pt 1), 885–​891. Trabert B, Lamb EJ, Scoccia B, et  al. 2013. Ovulation-​inducing drugs and ovarian cancer risk:  results from an extended follow-​ up of a large United States infertility cohort. Fertil Steril, 100(6), 1660–​ 1666. PMCID: PMC3873340. Trabert B, Ness RB, Lo-​Ciganic WH, et  al. 2014a. Aspirin, nonaspirin nonsteroidal anti-​ inflammatory drug, and acetaminophen use and risk of invasive epithelial ovarian cancer:  a pooled analysis in the Ovarian Cancer Association Consortium. J Natl Cancer Inst, 106(2), djt431. PMCID: PMC3924755. Trabert B, Pinto L, Hartge P, et al. 2014b. Pre-​diagnostic serum levels of inflammation markers and risk of ovarian cancer in the prostate, lung, colorectal and ovarian cancer (PLCO) screening trial. Gynecol Oncol, 135(2), 297–​ 304. PMCID: PMC4254357. Tsilidis KK, Allen NE, Key TJ, et  al. 2011. Menopausal hormone therapy and risk of ovarian cancer in the European prospective investigation into cancer and nutrition. Cancer Causes Control, 22(8), 1075–​1084. PMID: 21637986. Tung KH. 2003. Reproductive factors and epithelial ovarian cancer risk by histologic type: a multiethnic case-​control study. Am J Epidemiol, 158(7), 629–​638. Tworoger SS, Fairfield KM, Colditz GA, Rosner BA, and Hankinson SE. 2007a. Association of oral contraceptive use, other contraceptive methods, and infertility with ovarian cancer risk. Am J Epidemiol, 166(8), 894–​901. PMID: 17656616. Tworoger SS, Gertig DM, Gates MA, Hecht JL, and Hankinson SE. 2008a. Caffeine, alcohol, smoking, and the risk of incident epithelial ovarian cancer. Cancer, 112(5), 1169–​1177. PMID: 18213613. Tworoger SS, Lee IM, Buring JE, and Hankinson SE. 2008b. Plasma androgen concentrations and risk of incident ovarian cancer. Am J Epidemiol, 167(2), 211–​218. PMID: 17982156. Tworoger SS, Lee IM, Buring JE, Pollak MN, and Hankinson SE. 2007b. Insulin-​like growth factors and ovarian cancer risk: a nested case-​control study in three cohorts. Cancer Epidemiol Biomarkers Prev, 16(8), 1691–​ 1695. PMID: 17684148. Tworoger SS, Lee IM, Buring JE, et al. 2007c. Plasma 25-​hydroxyvitamin D and 1,25-​dihydroxyvitamin D and risk of incident ovarian cancer. Cancer Epidemiol Biomarkers Prev, 16(4), 783–​788. PMID: 17416771. USCS. 2014. U.S. Cancer Statistics Working Group. United States Cancer Statistics:  1999–​ 2011 Incidence and Mortality Web-​ based Report. Atlanta:  US Department of Health and Human Services, Centers for Disease Control and Prevention and National Cancer Institute. Available at: www.cdc.gov/​uscs. van Leeuwen FE, Klip H, Mooij TM, et al. 2011. Risk of borderline and invasive ovarian tumours after ovarian stimulation for in vitro fertilization in a large Dutch cohort. Hum Reprod, 26(12), 3456–​3465. PMCID: PMC3212878. van Nagell JR, Jr., DePriest PD, Ueland FR, et al. 2007. Ovarian cancer screening with annual transvaginal sonography:  findings of 25,000 women screened. Cancer, 109(9), 1887–​1896. PMID: 17373668. Vanden Berghe W. 2012. Epigenetic impact of dietary polyphenols in cancer chemoprevention:  lifelong remodeling of our epigenomes. Pharmacol Res, 65(6), 565–​576. PMID: 22465217. Veras E, Mao TL, Ayhan A, et al. 2009. Cystic and adeonfibromatous clear cell carcinomas of the ovary. Am J Surg Pathol, 33, 844–​853.

 907

Ovarian Cancer Walker BE. 1990. Tumors in female offspring of control and diethylstilbestrol-​ exposed mice fed high-​fat diets. J Natl Cancer Inst, 82, 50–​54. Walker GR, Schlesselman JJ, and Ness RB. 2002. Family history of cancer, oral contraceptive use, and ovarian cancer risk. Am J Obstet Gynecol, 186(1), 8–​14. Walker JL, Powell CB, Chen LM, et al. 2015. Society of Gynecologic Oncology recommendations for the prevention of ovarian cancer. Cancer, 121(13), 2108–​2120. PMID: 25820366. Wallin A, Orsini N, and Wolk A. 2011. Red and processed meat consumption and risk of ovarian cancer: a dose-​response meta-​analysis of prospective studies. Br J Cancer, 104(7), 1196–​1201. PMCID: PMC3068494. Walsh T, Casadei S, Lee MK, et al. 2011. Mutations in 12 genes for inherited ovarian, fallopian tube, and peritoneal carcinoma identified by massively parallel sequencing. PNAS, 108(44), 18032–​18037. Wang L, Lee IM, Zhang SM, et al. 2009. Dietary intake of selected flavonols, flavones, and flavonoid-​rich foods and risk of cancer in middle-​aged and older women. Am J Clin Nutr, 89(3), 905–​912. PMCID: PMC2667658. Wang S, Qiu L, Lang JH, et  al. 2013. Clinical analysis of ovarian epithelial carcinoma with coexisting pelvic endometriosis. Am J Obstet Gynecol, 208(5), 413 e411–​415. PMID: 23220508. WCRF. 2014. World Cancer Research Fund/​American Institute for Cancer Research. Continuous Update Project Report. Food, Nutrition, Physical Activity, and the Prevention of Ovarian Cancer. Available at http://​www. dietandcancerreport.org/​cup/​cup_​resources.php. Weiderpass E, Oh JK, Algeri S, and Bellocco R. 2014. Socioeconomic status and epithelial ovarian cancer survival in Sweden. Cancer Causes Control, 25(8), 1063–​1073. PMID: 24906473. Winter WE, 3rd, Maxwell GL, Tian C, et al. 2007. Prognostic factors for stage III epithelial ovarian cancer:  a Gynecologic Oncology Group Study. J Clin Oncol, 25(24), 3621–​3627. PMID: 17704411. World Health Organization (WHO). 2016. Globocan 2012: Estimated Cancer Incidence Mortality and Prevalence Worldwide in 2012. Available from: http://​globocan.iarc.fr/​Pages/​bar_​site_​sel.aspx. Wu AH, Pearce CL, Tseng CC, and Pike MC. 2015. African Americans and Hispanics remain at lower risk of ovarian cancer than non-​Hispanic whites after considering nongenetic risk factors and oophorectomy rates. Cancer Epidemiol Biomarkers Prev, 24(7), 1094–​1100. PMCID: PMC4490941.

907

Xie J, Poole EM, Terry KL, et al. 2014. A prospective cohort study of dietary indices and incidence of epithelial ovarian cancer. J Ovar Res, 7, 112. PMCID: PMC4263215. Xie J, Terry KL, Poole EM, et al. 2013. Acrylamide hemoglobin adduct levels and ovarian cancer risk: a nested case-​control study. Cancer Epidemiol Biomarkers Prev, 22(4), 653–​660. PMCID: PMC3617048. Xu H, Li S, Qiu JQ, et al. 2013. The VDR gene FokI polymorphism and ovarian cancer risk. Tumour Biol, 34(6), 3309–​3316. PMID: 24078452. Xu Y, He B, Pan Y, et al. 2014. Systematic review and meta-​analysis on vitamin D receptor polymorphisms and cancer risk. Tumour Biol, 35(5), 4135–​4169. Yan-​Hong H, Jing L, Hong L, et al. 2015. Association between alcohol consumption and the risk of ovarian cancer:  a meta-​analysis of prospective observational studies. BMC Public Health, 15, 223. PMCID: PMC4415339. Yang HP, Anderson WF, Rosenberg PS, et  al. 2013. Ovarian cancer incidence trends in relation to changing patterns of menopausal hormone therapy use in the United States. J Clin Oncol, 31(17), 2146–​2151. PMCID: PMC3731982. Yang HP, Trabert B, Murphy MA, et  al. 2012. Ovarian cancer risk factors by histologic subtypes in the NIH-​AARP Diet and Health Study. Int J Cancer, 131(4), 938–​948. PMCID: PMC3505848. Yoon SH, Kim SN, Shim SH, Kang SB, and Lee SJ. 2016. Bilateral salpingectomy can reduce the risk of ovarian cancer in the general population: a meta-​analysis. Eur J Cancer, 55, 38–​46. PMID: 26773418. Zaino R, Whitney C, Brady MF, et  al. 2001. Simultaneously detected endometrial and ovarian carcinomas—​a prospective clinicopathologic study of 74 cases: a gynecologic oncology group study. Gynecol Oncol, 83(2), 355–​362. PMID: 11606097. Zhang YF, Xu Q, Lu J, et al. 2015. Tea consumption and the incidence of cancer: a systematic review and meta-​analysis of prospective observational studies. Eur J Cancer Prev, 24(4), 353–​362. PMID: 25370683. Zheng W, Danforth KN, Tworoger SS, et al. 2010. Circulating 25-​hydroxyvitamin D and risk of epithelial ovarian cancer:  Cohort Consortium Vitamin D Pooling Project of Rarer Cancers. Am J Epidemiol, 172(1), 70–​80. PMCID: PMC2892541.

908

 90

47 Endometrial Cancer LINDA S. COOK, ANGELA L. W. MEISNER, AND NOEL S. WEISS

OVERVIEW Endometrial cancer is rare among women of reproductive age, but in older women can occur at an annual age-​adjusted rate of up to 50–​100 per 100,000. The incidence varies more than 5-​fold across regions of the world, with the rates generally being highest in North America and Europe. Endometrial cancer can be classified into two broad histologic groups:  the more common type I  tumors (e.g., endometrioid adenocarcinoma), which have a relatively good prognosis (case-​fatality in the neighborhood of 20%); and the less common type II tumors (e.g., serous carcinoma), which have a poorer prognosis. The endometrium is a hormone responsive tissue, and there is a large body of evidence to support a hormonal basis for the development of malignancy in this organ. Specifically, exposure to high levels of circulating estrogens increases endometrial cancer risk, especially for type I cancer, whereas exposure to progestogens reduces risk. The actions of many known or suspected factors that increase endometrial cancer risk (for example, obesity and certain medical conditions such as diabetes) and factors that decrease endometrial cancer risk (such as increasing parity, oral contraceptive [OC] use, and cigarette smoking) may be explained at least in part by their influence on estrogen and progestogen levels/​ activity. The genetic and the genetic-​environmental combined basis of endometrial cancer susceptibility remains uncertain, apart from the increased risk associated with inherited defects in DNA mismatch repair, or Lynch syndrome, that accounts for 3%–​5% of all endometrial cancer. Future research areas include a clearer understanding of etiologic heterogeneity and the etiologic effects of closely related risk factors, such body mass index (BMI), physical activity, metabolic syndrome/​diabetes, and hypertension. Additionally, continued research in hormonal therapy to identify hormones and combinations of hormones that minimize any increase in the occurrence of endometrial cancer while not increasing the risk of other long-​term deleterious health outcomes, such as breast cancer, is needed.

TUMOR MODEL A dualistic model of endometrial cancer carcinogenesis exists that is broadly grouped into type I and type II cancer (Bokhman, 1983). This classification generally is aligned with the histology of the tumors:  type I  cancers with the lower grade endometrioid tumors and type II with the high-​grade endometrioid and non-​endometrioid tumors (e.g., serous, clear cell). Type I tumors represent 80%–​85% of all endometrial cancers and often arise in the setting of endometrial hyperplasia. Etiologically, exposure to high levels of circulating estrogens increases the risk for this type of endometrial cancer, whereas exposure to progestogens reduces risk. As noted earlier, the actions of many known or suspected factors that increase endometrial cancer risk (e.g., obesity and certain medical conditions such as diabetes) and factors that decrease endometrial cancer risk (e.g., increasing parity, OC use, and cigarette smoking) may be explained at least in part by their influence on estrogen and progestogen levels/​activity. Although type I cancers are more likely than type II cancers to have certain molecular/​genetic alterations such as PTEN inactivation, KRAS mutations, and microsatellite instability (MSI), these alterations are not exclusively found in type I tumors.

Less is known regarding the more rare type II cancers, but they are thought to arise in an atrophic environment without a precursor lesion. These tumors have been described as estrogen-​independent, but a recent pooled analysis suggests that the type II cancers may have hormonal etiologies as well (Setiawan et al., 2013). The frequency of molecular/​genetic alterations such as p53 mutations and p16 inactivation are more common in type II than in type I cancers, but the overlap in many molecular/​genetic alterations suggests that there is more heterogeneity in endometrial cancer than can be accurately represented by a dualistic model. One example of such heterogeneity is the MSI that occurs in approximately 30% of endometrial cancers among women without known genetic syndromes, commonly referred to as sporadic MSI (Amankwah et al., 2013b; Basil et al., 2000; Duggan et al., 1994; MacDonald et al., 2000). It is unknown at present if this represents an unidentified genetic syndrome, an unidentified extension of Lynch syndrome, or is completely related to lifestyle and environmental exposures. In a case-​control study, OC use was associated with a risk reduction for MSI cancer, but not microsatellite stable (MSS) cancer (Amankwah et al., 2013b), and higher BMI (> 30 kg/​m2) was associated with a 2-​fold greater risk in MSI cancer than the lower elevation in risk noted for MSS cancer (Amankwah et al., 2013a). These results suggest that these molecular signatures have specific etiologies. A more complete understanding of molecular and genetic characteristics with respect to the underlying biology and risk associations may be useful in establishing a more comprehensive tumor model for endometrial cancer.

PATTERNS OF INCIDENCE AND MORTALITY There are a number of challenges complicating the interpretation of incidence and mortality rates of this disease across populations, over time, and by age. The body, or corpus, of the uterus has several types of tissue: the endometrium (the inner mucosal layer); the myometrium (the thick, middle muscular layer); and the serosa (the thin external coat). The exact location of tumors in the uterine corpus may not always specified in available records, and thus uterine corpus cancer is often used as a proxy for endometrial cancer when analyzing data based on those records. Also, deaths from cancers of the cervix occasionally are included in the general category of uterine cancer deaths, and while this would only be a small fraction of the uterine cancer deaths in countries such as the United States (Szekely et al., 1978), this is a larger problem in countries where cervical cancer mortality is relatively high, such as in eastern Africa (Ferlay et al., 2013). And finally, the lack of correction for the percentage of women with hysterectomies who are no longer at risk for developing endometrial cancer will lead to an underestimate of both the incidence and mortality, especially in countries where hysterectomy is common.

Incidence The incidence of endometrial cancer over the past six decades was estimated in the United States using the International Classification of Diseases of Oncology (ICD-​O-​3) (WHO, 2000)  endometrial specific site and histology codes in the Surveillance, Epidemiology, and

909

910

910

PART IV:  Cancers by Tissue of Origin 35

ICD-O-1

ICD-O-3

Rate per 100,000

30 25 20 15 10 5

1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

0

Year at Diagnosis White (1973–1991)/Non-Hispanic White (1992–2013)

Black

Hispanic White

Figure 47–​1.  Age-​adjusted (US 2000 Standard Population) incidence rates of endometrial cancer in the United States, 1973–​2013. In years 1973–​1991, diagnoses were restricted to “corpus uteri” (ICD-​O-​1). In years 1992–​2013, diagnoses were restricted to the endometrium (code C54.1, ICD-​O-​3). For all years (1973–​2013), malignant histologic types were restricted to 8140, 8210, 8255, 8260, 8261, 8262, 8263, 8310, 8323, 8380, 8381, 8382, 8383, 8441, 8460, 8461, 8470, 8471, 8480, and 8481. The nine SEER Program Registries are Atlanta (Metropolitan), Connecticut, Detroit (Metropolitan), Hawaii, Iowa, New Mexico, San Francisco-​Oakland SMSA, Seattle (Puget Sound), and Utah. Source: Surveillance, Epidemiology, and End Results (SEER) Program

(www.seer.cancer.gov) SEER*Stat Database:  Incidence—​SEER 9 Regs Research Data, Nov 2015 Sub (1973–​2013) —​ Linked to County Attributes—​Total U.S., 1969–​2014 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2016, based on the November 2015 submission.

End Results Program database (SEER, 2016) (Figure 47–​1). Incidence began to rise in the 1960s and reached a peak in the mid-​1970s among white women, particularly among those who were postmenopausal (Weiss et  al., 1976). The magnitude of this increase is likely to be an underestimate of the actual increase, because the rate of hysterectomy for reasons other than cancer rose rapidly during the same period, and the rates have not been corrected for this change (Dicker et  al., 1982; Howe, 1984; Koepsell et  al., 1980; Lyon and Gardner, 1977). Following the peak in the mid-​1970s, the incidence of endometrial cancer steadily declined until leveling off in the 1990s. This increase and then decline in incidence parallels patterns of postmenopausal estrogen-​ only (E-​ only) use (Austin and Roe, 1982). With strong evidence of an association between E-​only use and endometrial cancer, the US Food and Drug Administration issued a warning to physicians in 1976 (Food and Drug Administration, 1976). E-​only use declined through both a reduction in use of hormone therapy altogether and to an increase in use of combined estrogen and progestogen (EP) hormone therapy (Austin and Roe, 1982; Gruber and Luciani, 1986; Kennedy et al., 1985; Ross et al., 1988; Standeven et al., 1986; Wysowski et al., 1995). In contrast, black women in the United States did not experience an increase in incidence over the same time period, perhaps due to a lower prevalence of hormone therapy use (Hall et al., 2005), and their incidence was relatively stable through the beginning of the 1990s. In 1992, new coding in the SEER Program made it possible to distinguish non-​Hispanic white (NHW) women from Hispanic white (HW) women (previously they were all categorized as white women) (Figure 47–​1). Black women (annual percent change [APC] = 3.0) and HW women (APC = 1.6) have experienced a steady increase in endometrial cancer incidence since 1992 (Figure 47–​1), as have American Indian/​Alaska Native (APC = 3.2) and Asian/​Pacific Islander women (APC = 2.3) (data not shown). Incidence in NHWs has been more variable, but also appears to be on an increasing trajectory since the middle of the first decade of the 2000s. Among black women 50 years of age and older, incidence that is corrected for hysterectomy also shows

consistent increases from 1992 to 2008, both for the less aggressive type I endometrial cancer (APC = 2.3) and the more aggressive type II cancers (APC = 5.1) (Jamison et al., 2013). The increasing incidence in all these women may be due in part to the increasing prevalence of obese and overweight women in the United States (see Chapter 20 in this volume), as this is a strong risk factor for endometrial cancer. Additionally, increases since the first decade of the 2000s may reflect a reduction in EP menopause-​related hormone use. Some types of combined hormone use can reduce endometrial cancer risk (see later discussion), but following reports in 2002 of the other serious health outcomes positively associated with EP use (e.g., stroke, breast cancer) (Rossouw et al., 2002), EP use dramatically declined in the United States (Jewett et al., 2014). The incidence during the same time periods in other regions of the world are quite variable based on location (Figure 47–​2). For example, in Mozambique, age-​ adjusted incidence has increased from 2.0/​ 100,000 to 3.9/​ 100,000 between 1991 and 2008 (Lorenzoni et  al., 2015); in Slovenia, age-​adjusted incidence has remained stable at approximately 21.0/​100,000 between 1992 and 2008 (Zadnik and Krajc, 2016); and in Iran, age-​adjusted incidence has decreased from 2.3/​100,000 to 1.7/​100,000 between 2004 to 2008 (Arab and Noghabaei, 2014). In the United States, the incidence of endometrial cancer rises rapidly in late reproductive life, generally peaks between 60 and 70 years of age, and then declines in later life (Figure 47–​3). NHW women have the highest age-​specific incidence after 50 years of age, and American Indians/​Alaska Native as well as Asian/​Pacific Islander women have the lowest. The peak incidence rate for black women is about 5 years of age older than the peak for other women.

Mortality Over the past four decades in the United States, the age-​standardized mortality rate due to uterine corpus cancer has remained relatively low (< 5 per 100,000) (Figure 47–​4). Mortality peaked among white women

 91

911

Endometrial Cancer

Incidence rates of uterine corpus cancer 13.9 + 8.4–13.9 4.9–8.4 3.0–4.9 < 3.0 No Data

Figure 47–​2.  Annual age-​adjusted (World Standard) incidence rates of uterine corpus cancer expressed per 100,000. Source: Ferlay J, Soerjomataram I, Ervik M, Dikshit R, Eser S, Mathers C, Rebelo M, Parkin DM, Forman D, Bray, F.  GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality Worldwide: IARC CancerBase No. 11 [Internet]. Lyon, France: International Agency for Research on Cancer; 2013. Available from: http://​globocan.iarc. fr, accessed on 7/​1/​2016.

in the early 1980s, most likely a direct result of the peak incidence a few years earlier, and has steadily declined since that time. Similarly, since the early 1990s HW women have experienced a decreasing trend in mortality rates. Some of this may be due to increases in hysterectomy that occurred in the United States until 2002 (Wright et  al., 2013), although since that time, the median number of hysterectomies performed per hospital in the United States has declined by more than 40% (Wright et al., 2013). In contrast, black women, who experience the highest mortality rates, have not experienced a similar mortality

decrease, with recent trends that suggest a possible increase in the mortality rate. Likewise, age-​standardized mortality rates from uterine cancer also show a downward trend in both the European Union and selected regions of Asia. From 1970 to 2005, EU countries generally have noted decreased mortality in countries such as Spain and Greece that started with lower mortality (< 3.0/​100,000) compared to countries such as Austria, Hungary, and Romania that started with higher mortality (> 6.0/​ 100,000) (Weiderpass et  al., 2014). In contrast, mortality

100 90 Rate per 100,000

80 70 60 50 40 30 20 10 85+

80–84

75–79

70–74

65–69

60–64

55–59

50–54

45–49

40–44

35–39

30–34

25–29

20–24

15–19

0–14

0

Age at Diagnosis Non-Hispanic White Hispanic White Asian or Pacific Islander

Black American Indian/Alaskan Native

Figure 47–​3.  Average annual age-​specific incidence rates of endometrial cancer by race/​ethnicity, 2009–​2013. Diagnoses were restricted to the endometrium (code C54.1, ICD-​O-​3) and malignant histologic types were restricted to 8140, 8210, 8255, 8260, 8261, 8262, 8263, 8310, 8323, 8380, 8381, 8382, 8383, 8441, 8460, 8461, 8470, 8471, 8480, and 8481. Source: Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat

Database: Incidence -​SEER 9 Regs Research Data, Nov 2015 Sub (1973-​2013) -​Linked To County Attributes -​Total U.S., 1969-​2014 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2016, based on the November 2015 submission.

912

912

PART IV:  Cancers by Tissue of Origin 4.0 3.5

Rate per 100,000

3.0 2.5 2.0 1.5 1.0

2012

2010

2008

2006

2004

2002

2000

1998

1996

1994

1992

1990

1989

1987

1985

1983

1981

1979

1977

1975

0.0

1973

0.5

Year at Diagnosis White (1973–1989)/Non-Hispanic White (1990–2013)

Black

Hispanic White

Figure 47–​4.  Average annual age-​adjusted (US 2000 Standard Population) mortality rates of cancers of the corpus uteri by race/​ethnicity, 1973–​2013. In years 1973–​2013, diagnoses were restricted to “corpus uteri”. The nine SEER Program Registries are Atlanta (Metropolitan), Connecticut, Detroit (Metropolitan), Hawaii, Iowa, New Mexico, San Francisco-​Oakland SMSA, Seattle (Puget Sound), and Utah. Source: Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Mortality -​All COD, Aggregated With State, Total U.S. (1969-​2013 , National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2016. Underlying mortality data provided by NCHS (www.cdc.gov/​nchs).

has been consistent at roughly 3.5/​100,000 in Sweden and Latvia (Weiderpass et al., 2014). In Hong Kong, Japan, Korea, and Singapore, mortality from all uterine cancer, which includes mortality from both the cervix and corpus uteri, has also declined between 1990 and 2009 (Lee et al., 2014). However, a sub-​analysis of the Korean data showed an increase in corpus cancer mortality with an APC of 6.7 from 1995 to 2009 (Lee et al., 2014) when additional efforts were made to distinguish between cervix cancer mortality and corpus cancer mortality. These results highlight the challenges that can occur with international comparisons and trends over time when cervical cancer mortality is relatively high and mixed with corpus cancer mortality.

SURVIVAL/​MORTALITY AMONG WOMEN WITH ENDOMETRIAL CANCER Relative to many other cancers, the observed 5-​year survival following a diagnosis of endometrial cancer is relatively high at approximately 80% overall. Higher stage, more aggressive histology (e.g., type II cancer), higher grade, and non-​guideline concordant treatment are all correlated with poorer survival (Colombo et al., 2011). For example, 5-​year survival is roughly 90% for small, localized FIGO stage IA tumors but falls to less than 20% for the metastasized FIGO stage IVB tumors (American Joint Commission on Cancer [AJCC], 2010). Factors other than tumor and treatment characteristics also influence survival. Survival is poorer in blacks than whites, with relative 5-​year survival 15%–​20% lower in blacks (SEER, 2016). Even after adjustment for stage, histology, grade, and form of treatment, black women have relatively poorer survival (Long et al., 2013). Survival following a diagnosis of endometrial cancer may also differ by risk factors. For low-​grade endometrioid cases, BMI greater than 30 kg/​m2 has been associated with an increased risk for endometrial cancer–​specific mortality (Felix et al., 2015b). In a meta-​analysis of international data, increasing BMI at diagnosis of endometrial cancer was associated with increasing all-​cause mortality (Secord

et al., 2016), although other studies suggest that diabetes or metabolic syndrome at diagnosis better predicts mortality (Liao et al., 2014; Ni et al., 2015) or that diabetes may be more highly related to poorer survival than obesity (Lindemann et al., 2015). It is challenging to disentangle these two comorbid conditions given their high correlation with each other. Nulliparous women may have a poorer survival than parous women (Hachisuga et al., 2001; Salvesen et al., 1998), although this might be restricted to women with non-​endometrioid histologies (Felix et al., 2015b). A number of studies have reported that E-​only therapy is associated with less aggressive endometrial tumors (e.g., lower stage, lower grade, less myometrial invasion) (Chapman et al., 1996; Collins et al., 1980; Felix et al., 2015b; Mittal and Barwick, 1993; Nyholm et al., 1993; Orgeas et al., 2009; Robboy and Bradley, 1979; Shapiro et al., 1998), which suggests a better prognosis. Among studies that directly evaluated survival or mortality, most reported lower mortality (Chu et al., 1982; Collins et al., 1980; Elwood and Boyes, 1980; Orgeas et al., 2009; Paganini-​Hill, et al., 1989; Petitti, et al., 1987; Robboy and Bradley, 1979; Schwartzbaum et al., 1987; Smith et al., 1981), although some reported no difference (Felix et al., 2015b; Schairer et al., 1997) in E-​only users relative to non-​users. There is less information about EP therapy and survival; one study found no decrease in relative survival with EP use prior to diagnosis (Orgeas et al., 2009), but another found the suggestion of lower mortality with EP use (Felix et al., 2015b). There is also no strong evidence to suggest that women with hereditary non-polyposis colorectal cancer (HNPCC)/Lynch syndrome have a better or worse prognosis than sporadic endometrial cancer cases (Boks et al., 2002).

ENDOGENOUS AND EXOGENOUS HORMONES Endogenous Estrogen Estrogens are the primary stimulants of endometrial proliferation. Because sustained proliferative signaling is one of the hallmark traits

 913

Endometrial Cancer that can lead to malignant transformation (Hanahan and Weinberg, 2011), it is plausible that estrogens can play a causal role in the development of endometrial carcinoma. Several prospective studies or nested case-​control studies have assessed hormone levels from blood samples prior to endometrial cancer diagnosis. In postmenopausal women, higher circulating levels of estradiol (Brinton et al., 2016; Dossus et al., 2013; Luhn et al., 2013; Lukanova et al., 2004; Zeleniuch-​ Jacquotte et al., 2001) or estrone (Allen et al., 2008; Brinton et al., 2016; Lukanova et al., 2004; Zeleniuch-​Jacquotte et al., 2001) were associated with 2-​to 4-​fold elevations in endometrial cancer risk after adjustment for BMI and exogenous hormone use (or restriction to non-​ hormone users at blood draw). Consistent with these findings, risk factors for endometrial cancer include medical conditions that are known to result in relatively high endogenous estrogen levels. For example, women with estrogen-​secreting ovarian tumors (e.g., granulosa-​theca cell tumors) had a high prevalence of endometrial hyperplasia and endometrial carcinoma at the time of oophorectomy, estimated as 25% and 7%, respectively, in a large multicenter trial in Italy (Ottolina et al., 2015). Similarly, polycystic ovary syndrome (PCOS), characterized by chronic anovulation resulting in unopposed estrogen action in premenopausal women and high androstenedione levels metabolized peripherally by the enzyme aromatase to estrone in postmenopausal women (Siiteri and MacDonald, 1973), has been linked with endometrial cancer. A recent meta-​analysis estimated that women with PCOS have an elevated risk for endometrial cancer (odds ratio [OR] = 2.8; 95% CI: 1.3, 6.0), with an even stronger association for those under the age of 54 years (OR = 4.1; 95% CI: 2.4, 6.8) (Barry et al., 2014). And finally, obesity, an established risk factor for endometrial cancer (WCRF/​AICR, 2013), leads to higher endogenous estrogen levels, as well as decreased circulating levels of sex hormone–​binding globulin (Liedtke et al., 2012). Because anthropometric characteristics and physical activity can influence endometrial cancer risk by means of etiologic pathways beyond endogenous estrogen, these risk factors will be discussed in more detail later.

Endogenous Progesterone The surge of luteal progesterone just prior to ovulation each month in premenopausal women arrests endometrial proliferation, promotes secretory differentiation of the endometrium, and initiates endometrial sloughing in the absence of fertilization (King and Whitehead, 1983). These events decrease the likelihood of developing pre-​cancer lesions, because cell differentiation alone can arrest or reverse neoplastic cellular transformation (Pitot, 1986). Thus, low endogenous progesterone levels, particularly when coupled with relatively high estrogen levels, are generally associated with higher rates of endometrial cancer. The peak endometrial cancer incidence occurs at ages in the late 60s (Figure 47–​3), after a menopausal period of estrogen production through peripheral conversion of adrenal androgens without any cyclic ovarian progesterone production or substantial peripheral production of progesterone. Similarly, women with the PCOS (discussed previously) not only have excessive production of estrogen, but also lack cyclic progesterone secretion (Farhi et  al., 1986). Obesity in premenopausal women is associated with chronic anovulation and decreased progesterone levels (Kaaks and Lukanova, 2002). In 175 ovulatory premenopausal women evaluated for a tubal ligation study, baseline progesterone levels were lower in those who weighed 140–​ 270 pounds (12.4 ng/​ml) than in those weighing 90–​140 pounds (15.2 ng/​ml) (Westhoff et al., 1996).

Exogenous Estrogen E-​Only and Endometrial Hyperplasia

Women given E-​only have an increased incidence of adenomatous hyperplasia of the endometrium. For example, 2 years after initiation of treatment in randomized trials (Furness et al., 2012; Roberts et al., 2014) there was a dose-​dependent increase in risk of endometrial hyperplasia, ranging from a doubling of risk at low doses

913

(e.g., < 0.50 mg conjugated equine estrogens, ≤ 0.1 mg 17β estradiol, etc.) to more than a 10-​fold increase at moderate (e.g., 0.625 mg conjugated equine estrogen, 1.5–​2 mg 17β estradiol) and high doses (e.g., 1.25 mg conjugated equine estrogen, 4 mg 17β estradiol, etc.). Hyperplasia frequently coexists with endometrial cancer in estrogen users and has been observed to precede the appearance of carcinoma in individual women (Deligdisch and Holinka, 1987; Gusberg and Hall, 1961; Pettersson et al., 1985). Among women with hyperplasia not treated with a progestin, endometrial cancer has been estimated to occur in about 1% of those with the simple/​non-​atypical form after an average follow-​up of 15.2 years, 3%–​9% of those with the complex/​ non-​atypical form after an average follow-​up of 13.5 years, 7%–​8% of those with the simple/​atypical form after an average follow-​up of 11.4 years, and 20%–​29% of those with the complex/​atypical form after an average follow-​up of 4.1 years (Kurman et al., 1985).

E-​Only and Endometrial Cancer

Virtually all studies of the question have reported that peri-​and postmenopausal E-​only use is positively associated with the incidence of endometrial cancer. Detailed reviews of E-​only and endometrial cancer can be found elsewhere (Grady and Ernster, 1996; Grady et  al., 1995; Herrington and Weiss, 1993). In the interest of brevity, the following discussion will provide only a summary of selected aspects of E-​only use. Most studies have found an elevation in risk beginning after 1–​ 5 years of use, with risk elevations increasing further with increasing duration of use (Green et al., 1996; Rubin et al., 1990). Risk decreases after cessation of E-​only, but an elevated risk relative to non-​users may persist for 10 years or more (Grady et al., 1995; Lacey et al., 2005). Use of all types of oral E-​only, including but not limited to estriol, stilbestrol, ethinyl estradiol, and conjugated equine estrogens, the type of estrogen most commonly prescribed in the United States, are associated with an increased risk (Cushing et al., 1998; Grady et al., 1995; Morch et al., 2016; Persson et al., 1989; Weiderpass et al., 1999a). The risk of endometrial cancer appears to be elevated with all commonly prescribed dosages of estrogens (0.3 mg–​1.25 mg per day of conjugated equine estrogens or the equivalent amount of other estrogens) (Cushing et al., 1998). Among users of E-​ only vaginal creams/​ gels, pessaries/​ tablets, or rings, results are mixed but are suggestive of, at most, a small increase risk for endometrial cancer. Among 541 women using low-​ dose 17β-​estradiol vaginal tablets, the incidence of hyperplasia was 0.52% in 1 year, similar to overall background incidence, suggesting little impact on cancer occurrence (Simon et al., 2010). Three studies observed an elevation in endometrial cancer risk with vaginal E-​only (Kelsey et  al., 1982; Morch et  al., 2016; Weiderpass et  al., 1999c), including one that found an elevation for both type I and type II cancer (Morch et  al., 2016), whereas two smaller studies did not (Brinton et al., 1993; Gray et al., 1977). Historically, exclusive use of vaginal E-​only without systemic therapy during other periods of time has been rare, and variation in results between studies may be due to the low frequency of use, as well as the manner of adjustment for use of hormones administered by other means. When E-​only use is considered within subgroups of other risk factors, such as parity, OC use, BMI, and so on, almost every study shows an elevated risk for endometrial cancer in each subgroup with E-​only use (e.g., Brinton et al., 1993; Kelsey et al., 1982; Rubin et al., 1990; Weiss and Sayvetz, 1980).

Exogenous Progestins, Estrogens Plus Progestins (EP), and Selected Estrogen Receptor Modulators (SERMs) Most endometrial hyperplastic lesions can be successfully treated with synthetic progestogens (i.e., progestins) (Thom et  al., 1979). One study of 185 women with hyperplasia found that regression of complex hyperplasia without atypia was common whether a progestin had or had not been used, but that persistence/​progression of atypical hyperplasia was less common (RR = 0.39; 95% CI: 0.21, 0.70) among those treated with a progestin (Reed et al., 2009). More generally, to

914

914

PART IV:  Cancers by Tissue of Origin

Table 47–​1.  Name and Description of Estrogen and Progestin Treatment Regimens as used in this Book Chapter Regimen Sequential estrogen plus progestin (SEP) Monthly short-​cycle Monthly long-​cycle Three-​month cycle Continuous combined estrogen plus progestin (CEP)

# Days Estrogen per Month

# Days Progestin per Month

≥ 25 ≥ 25 ≥ 25 ≥ 25

< 10 10–​24 10–​24 per 2–​3 months ≥ 25

counteract the proliferative effects of exogenous estrogen on the endometrium, progestins are added to E-​only hormone therapy (Stefanick, 2005). Over time, several different regimens of EP have been used, and the terminology in the published literature is somewhat inconsistent. Please see Table 47–​1 for the definitions of treatment regimens and their acronyms used here.

EP and Endometrial Hyperplasia

A systematic review and meta-​analysis of randomized studies found little evidence that sequential estrogen plus progestin (SEP) increased the risk for hyperplasia within the relatively short follow-​up times (generally ≤ 3  years) evaluated in these trials (Furness et  al., 2012; Roberts et al., 2014). In three trials of SEP including low-​dose estrogen (and various progestins) and in five trials with SEP including moderate-​dose estrogen (and various progestins), there were no cases of endometrial hyperplasia found in either the treatment or placebo group. In contrast, one trial that had a moderate-​dose estrogen (0.625 mg conjugated equine estrogens) and 10 mg medoxyprogesterone acetate 12  days per cycle reported more hyperplasia in the SEP group (3.3%) than the placebo group (0%) after 2 years of follow-​up. However, two studies comparing high-​dose estrogen SEP and placebo found no difference in the occurrence of hyperplasia after 2 years of therapy. Similarly, for continuous combined estrogen plus progestin (CEP), nine trials using various doses and types of CEP have reported no differences in hyperplasia occurrence between the treatment and placebo groups after 1–​3  years of follow-​up (Furness et  al., 2012; Roberts et al., 2014).

EP and Endometrial Cancer

Because there is increasing evidence that endometrial cancer risk differs by the regimen of EP, the following discussion will focus on risks for specified regimens of EP only; general results that were reported for any type and dosage of EP use will not be discussed here. With respect to observational studies, a meta-​analysis reported that monthly short-​cycle SEP therapy was associated with an excess risk of endometrial cancer (meta-​analysis RR = 1.76; 95% CI: 1.51, 2.05) (Brinton and Felix, 2014). Longer durations of monthly short-​cycle SEP therapy for 5 or more years (Doherty et  al., 2007; Pike et  al., 1997) or 10 or more years (Razavi et al., 2010) are associated with a 3-​to 6-​fold elevation in risk. Results for shorter durations of use do not show a consistent increase in risk (Doherty et al., 2007; Pike et al., 1997; Razavi et al., 2010). These results suggest that fewer than 10  days per month of added progestin is insufficient to counteract the proliferative effects of estrogen, especially for longer durations of use. A meta-​analysis reported that monthly long-​cycle SEP therapy was not associated with an altered risk (meta-​analysis RR  =  1.07; 95% CI: 0.92, 1.24) (Brinton and Felix, 2014). Six or more years of monthly long-​cycle SEP therapy was associated with a doubling of risk in one study (Doherty et al., 2007) and more than 10 years with a 40% risk increase in another study (Jaakkola et al., 2011), but two other studies found no increased risk with longer durations of use (Pike et al., 1997; Razavi et al., 2010).

Most investigations on progestin type and endometrial cancer risk did not differentiate between monthly short-​cycle and monthly long-​ cycle SEP regimens; therefore, both regimen types are combined in the following. It is unclear if sequential administration of progesterone-​derived progestin, specifically medroxyprogesterone acetate (MPA), alters endometrial cancer risk. One study found a small increase in risk (OR  =  1.12; 95% CI:  1.05, 1.20) (Weiderpass et  al., 1999a). Conversely, another study found a reduction within the first 5  years of use (OR = 0.64; 95% CI: 0.43, 0.95), and then no association after more than 5 years of use (OR = 1.42; 95% CI: 0.63, 3.19) (Jaakkola et al., 2011). One study found a slight increase in risk when testosterone-​derived progestin, such as levonorgestrel and norethisterone, was sequentially added to the regimen (OR  =  1.09; 95% CI:  1.02, 1.17) (Weiderpass et  al., 1999a). Two other studies found no association (Beral et  al., 2005; Jaakkola et al., 2011), even for 5 or more years of use (Jaakkola et al., 2011). Research on endometrial cancer risk and 3-​month SEP use is limited, but two record linkage studies from Finland, with similar results, shed some light on this association (Jaakkola et al., 2009, 2011). The most recent one, where the long interval for progestins is indicated as 2–​3 months, reported no more than a modest association within the first 5 years of use (OR = 1.40; 95% CI: 0.82, 2.38). However, women who used 3-​month cycle SEP for 5–​10 years and more than 10 years experienced a greater risk of developing endometrial cancer (OR = 1.63; 95% CI: 1.12, 2.38 and OR = 2.95; 95% CI: 2.40, 3.62, respectively) (Jaakkola et al., 2011). These results suggest that progestins given every 2–​3 months may not be sufficient to reduce the estrogen effects on the endometrium with longer durations (≥ 5 years) of use, but more evidence is needed to confirm these limited results. There have been two randomized trials that have assessed endometrial cancer outcomes with continuous combined estrogen plus progestin (CEP) (Chlebowski et al., 2016; Hulley et al., 2002). The results from the Women’s Health Initiative and the Heart and Estrogen/​ Progestin Replacement Study (HERS) both support a reduction in risk with daily 0.625 mg conjugated equine estrogens (CEE) plus 2.5 mg MPA with an HR = 0.59 (95% CI: 0.40, 0.88) (Chlebowski et al., 2016) and an HR = 0.25 (95% CI: 0.05, 1.18) (Hulley et al., 2002), respectively. In agreement with the randomized trials, a meta-​analysis of non-​ randomized studies reported that ever use of CEP was associated with an overall reduction in risk of endometrial cancer (RR  =  0.78; 95% CI: 0.72, 0.86) (Brinton and Felix, 2014). Among those observational studies that evaluated at least 25 days per month progestin in the CEP regimen, three studies found no association with durations of use of < 5  years (Pike et  al., 1997; Razavi et  al., 2010; Weiderpass et  al., 1999a), whereas two studies reported approximately a 50% reduction in risk with < 5 years of use (Jaakkola et al., 2011; Phipps et al., 2011). The results for long durations of CEP therapy of > 10 years are mixed, with some studies showing a 20%–​60% reduction in risk (Jaakkola et al., 2011; Phipps et al., 2011), no association (Trabert et al., 2013), or a 2-​fold elevation in risk (Razavi et al., 2010). Inconsistencies in results could be due to a lack of statistical power because of a low prevalence of CEP use, particularly for longer durations, in some studies. It is unclear if the type of progestin used in CEP regimens changes the risk of endometrial cancer. Two studies have shown that CEP with progesterone-​derived progestins may reduce risk by about half (Beral et al., 2005; Jaakkola et al., 2011), whereas one study found no association (Weiderpass et al., 1999a). Additionally, two studies found a reduction in risk in women using CEP with testosterone-​derived progestin (Jaakkola et  al., 2011; Weiderpass et  al., 1999a), but another reported no association (Beral et al., 2005).

Selected Estrogen Receptor Modulators (SERMs) and Endometrial Cancer

SERMs are a group of antagonists/​agonists for the estrogen receptor that are tissue specific. One of these, tamoxifen, has been used in the prevention and treatment of breast cancer and is usually given for 5–​ 10 years. Although tamoxifen therapy for less than 2 years does not

 915

Endometrial Cancer appear to elevate endometrial cancer risk, longer durations are consistently associated with a 2-​to 5-​fold elevation in risk (Bergman et al., 2000; Bernstein et al., 1999; Cook et al., 1995; DeMichele et al., 2008; Ribeiro and Swindell, 1992; Sasco et al., 1996). Meta-​analyses report an elevated risk for endometrial cancer in prevention trials with 20 mg/​day (meta-​analysis RR  =  2.16; 95% CI:  1.33, 3.50; Braithwaite et al., 2003; meta-​analysis HR = 2.4; 95% CI: 1.5, 4.0; Cuzick et al., 2003)  and in treatment trials with 20–​ 40 mg/​ day (meta-​ analysis RR = 3.28; 95% CI: 2.09, 5.14; Braithwaite et al., 2003; meta-​analysis HR = 3.4; 95% CI: 1.8, 6.4; Cuzick et al., 2003). It has been proposed that tamoxifen users (both for breast cancer prevention and treatment) can use the levonorgestrel-​releasing intrauterine system (LNG-​IUS) to counteract the estrogenic effects on the endometrium without systemic effects that could compromise the efficacy of tamoxifen with respect to breast cancer reduction (Gizzo et al., 2014; Wan and Holland, 2011). However, only a handful of trials and observational studies have been published regarding LNG-​IUS use in tamoxifen users or E-​only users, and the effect on endometrial cancer risk remains uncertain (Fu and Zhuang, 2014; Gizzo et al., 2014; Wan and Holland, 2011). Another SERM, raloxifene, is used to prevent and treat osteoporosis in postmenopausal women (Jordan, 1995). In contrast to tamoxifen, raloxifene does not appear to stimulate proliferation of the endometrium (Boss et al., 1997), and an observational study (DeMichele et al., 2008) and several randomized studies do not show an elevated endometrial cancer risk with raloxifene versus placebo (Vogel et al., 2006). When raloxifene is compared directly to tamoxifen, uterine cancer risk is approximately 40% lower in the raloxifene users (Vogel et al., 2006). Less is known about the influence of other SERMs, such as bazedoxifene, lasofoxifene, and ospemifene, on the risk of endometrial cancer (Mirkin and Pickar, 2015).

Hormonal Contraception and Endometrial Cancer

Over the past 45 years, hormonal contraception has included sequential and combination (i.e., EP) OCs, EP contraceptive patches, EP vaginal rings, progestin-​only pills, the long-​acting injectable or implanted progestins (e.g., medroxyprogesterone), and progestin-​releasing intrauterine devices (IUDs). While a fair amount of study has been devoted to OC use in relation to endometrial cancer risk, this is not true for the other hormone-​containing contraceptives. Evidence is limited, but progestin-​only pills or injected/​implanted progestins may substantially reduce endometrial cancer risk. Several studies reported no increase in uterine cancer incidence (Liang et al., 1983) or up to a 40%–​80% reduction (Maxwell et al., 2006; WHO Collaborative Study of Neoplasia and Steroid Contraceptives, 1991) in risk with depot medroxyprogesterone acetate (DMPA) injections. Contraceptive implants (progestin-​only) are associated with a reduction in endometrial thickness (Mascarenhas et al., 1998), but long-​term information regarding the impact of use of these implants on the risk for endometrial cancer is lacking. There was also a suggested reduction (OR = 0.6; 95% CI: 0.2, 1.4) in risk with the use of progestin-​only pills (Weiderpass et al., 1999b). OCs were available initially as sequential preparations (E-​ only followed by a short course of EP) or as concurrent EP preparations. Starting in the mid-​1970s, there were reports of endometrial abnormalities (e.g., Lyon and Frisch, 1976) and an increased risk of endometrial cancer (Henderson et al., 1983; Weiss and Sayvetz, 1980) among women who used particular sequential preparations (Oracon) that contained a relatively potent estrogen (ethinyl estradiol) followed by a weak progestogen (dimethisterone). Sequential preparations were removed from the consumer market in the United States and Canada in 1976. In contrast, almost all individual studies, as well as a large pooled analysis of 27,276 endometrial cancer cases and 115,743 control women, found that combination OC users have a 30%–​50% lower risk of endometrial cancer than non-​users (Collaborative Group on Epidemiological Studies on Endometrial, 2015; Cote et al., 2015; Hannaford et al., 2007). The reduced risk appears to be evident within 2–​5 years of initiation of use (Collaborative Group on Epidemiological Studies on Endometrial, 2015; Mueck et al., 2010), and each additional 5 years of use further decreases risk by about 25% (Collaborative Group on Epidemiological Studies on Endometrial, 2015). Although

915

the reduction in risk may be greater with more recent use, those who had 30 or more years since last use still had a substantial reduction in risk (Collaborative Group on Epidemiological Studies on Endometrial, 2015), which was also noted among parous women who had only used OCs before their first birth (Cook et al., 2014). Some studies report that the reduction in risk may be greatest with OCs in which progestogen effects predominate (Hulka et al., 1982) or that contain higher dosages of progestins (Rosenblatt et al., 1991), but others have found a reduced risk irrespective of dose (Maxwell et al., 2006). Contemporary OCs have lower dosages or different types of estrogens/​progestins (Wright and Johnson, 2008) than those used in the majority of studies evaluated; it is uncertain if such substantial reductions in risk will continue with these contemporary formulations. The reduction in risk with increasing duration of OC use appears to be fairly consistent across other risk factors such as age (< 60  years, ≥ 60  years), BMI, parity, hormone therapy (all types, none), menopausal status, smoking status, age at menarche, and alcohol intake (Collaborative Group on Epidemiological Studies on Endometrial, 2015).

OTHER RISK FACTORS Events of Reproductive Life In a meta-​analysis, older versus younger ages at menarche (as reported in each study) were associated with a 32% reduction in endometrial cancer risk, with a 4% reduction for each additional 2-​year delay in menarche (Gong et  al., 2015). Although various age categories are used to define age at menopause, almost all studies that investigated this issue find that women who experience a natural menopause at a relatively late age (i.e., > 50  years or > 55  years) are at a modestly greater risk of endometrial cancer than are other women (Baanders-​ van Halewyn et  al., 1996; Dossus et  al., 2010; Ewertz et  al., 1988; Petridou et al., 2002; Zucchetto et al., 2009). Two studies that determined total number of menstrual cycles or menstrual years between menarche and menopause, accounting for pregnancies, lactation, and OC use, reported about a 50% elevation in risk for those with a total number of cycles above the median (411.5) versus below the median (Wang et al., 2015) and over a 2-​fold excess risk for ≥ 37 versus < 33 menstrual years (Zucchetto et al., 2009). It has been consistently reported that parous women, relative to nulliparous women, have a decreased risk of endometrial cancer, with each birth associated with a 14% reduction in risk (Wu et al., 2015). This parity-​related reduction in risk appears to be unmodified by smoking status (Newcomer et al., 2001), but may be greater among women diagnosed at younger ages (< 50 or < 55 years) than among women diagnosed at older ages (Hachisuga et al., 1998; Parazzini et al., 1998). Although incomplete pregnancies (spontaneous or therapeutic abortions) may enable the removal of cells that have undergone pre-​malignant or malignant transformation, no consistent reduction in endometrial cancer risk has been found for women with incomplete pregnancies (e.g., Brinton et al., 1992; Dossus et al., 2010; Kalandidi et al., 1996; McPherson et al., 1996; Parazzini et al., 1998; Parslov et al., 2000; Pocobelli et al., 2011). Similarly, no consistent alteration in endometrial cancer risk has been found in women who have undergone tubal sterilization (Castellsague et al., 1996; Lacey et al., 2000). After accounting for parity, older age at first birth and an older age at last birth have both, if anything, been associated with a reduction in endometrial cancer risk. However, among the studies that accounted jointly for these variables, an older age at first birth (≥ 30  years or ≥ 35 years) was associated with a 16%–​60% reduction in risk relative to a younger age (≤ 20 years or ≤ 25 years) in all (Bevier et al., 2011; Brinton et al., 2007; Dossus et al., 2010; Karageorgi et al., 2010; Kvale et al., 1988; Lambe et al., 1999; Pocobelli et al., 2011) but one study (Pfeiffer et al., 2009). Evidence is mixed for an older age at last birth (≥ 35  years or ≥ 40  years) relative to a younger age (< 25  years or ≤ 25 years) in studies that account jointly for the factors indicated in the preceding: some report a reduction in risk (Bevier et al., 2011; Kvale

916

916

PART IV:  Cancers by Tissue of Origin

et al., 1988; Lambe et al., 1999; Pfeiffer et al., 2009), while others do not (Brinton et al., 2007; Dossus et al., 2010; Pocobelli et al., 2011). An international study (Rosenblatt and Thomas, 1995), a Japanese study (Okamura et  al., 2006), and a study from Mexico (Salazar-​ Martinez et  al., 1999)  observed that lactation, especially of longer durations, is associated with a lower risk of endometrial cancer. In contrast, studies from the United States, Europe, and Japan have not observed such an association (Brinton et al., 1992; Hirose et al., 1996; Newcomb and Trentham-​Dietz, 2000; Zucchetto et al., 2009), even for longer durations of lactation (Newcomb and Trentham-​Dietz, 2000). The latter study also investigated the use of lactation-​suppressant hormones that usually are composed of estrogens; a slight elevation in risk was noted in women using these hormones two or more times versus non-​users (Newcomb and Trentham-​Dietz, 2000). After accounting for parity, impaired fertility measured in various ways (e.g., 3 or more years of unsuccessfully attempting pregnancy, seeking medical advice for infertility, or self-​reported infertility) has been associated with a 20% increase in endometrial cancer in a pooled study of 14 observational studies (Yang et al., 2015). A meta-​analysis of fertility treatment and the development of endometrial/​uterine cancer reported null findings, but heterogeneity among studies was high (Saso et al., 2015). The type of infertility may be important; in a cohort of women with infertility, a 9-​fold elevation in risk (95% CI: 5.0, 16.0) was noted among women whose infertility was characterized by normal estrogen production but progesterone deficiency (Modan et al., 1998).

Anthropometric Characteristics and Physical Activity Obesity is a well-​established risk factor for endometrial cancer (WCRF/​ AICR, 2013). Obesity leads to a net increase in the amount of endogenous estrogens, both through increased conversion of androstenedione to estrogen (Edman and MacDonald, 1978; MacDonald et al., 1978; MacDonald and Siiteri, 1974) and decreased circulating levels of sex hormone-​binding globulin (Davidson et al., 1981; Kaye et al., 1991; Nyholm et al., 1993). Excess adiposity may also lead to increases in insulin and bioactive insulin-​like growth factor-​1 concentrations in the endometrium, substances that can stimulate cell proliferation and suppresses apoptosis (Calle and Kaaks, 2004; Kaaks et al., 2002). Based on the results of a number of meta-​analyses and summary studies, overweight (BMI: 25–​29 kg/​m2) and obese (BMI: ≥ 30 kg/​m2) women are approximately 2 to 4 times more likely to develop endometrial cancer than those with a lower BMI (Aune et al., 2015; Calle and Kaaks, 2004; Jenabi and Poorolajal, 2015; Zhang et al., 2014). An estimated 45%–​60% of the endometrial cancer cases in the United States and the European Union are attributable to a BMI over 25 kg/​m2 (Calle and Kaaks, 2004). Meta-​ analyses of prospective observational studies suggest that other anthropometric factors, such as greater abdominal adiposity (measured by waist to hip ratio), higher BMI in young adulthood, greater adult weight gain, and greater adult attained height, are associated with an increase in endometrial cancer risk (Aune et  al., 2015; WCRF/​AICR, 2013). In the few studies that adjusted for current BMI, these associations generally were reduced (Aune et al., 2015). The available evidence suggests that physical activity is associated with a reduced risk of endometrial cancer (WCRF/​AICR, 2013), with higher levels of physical activity resulting in an estimated 20%–​40% reduction in risk (Cust et al., 2007; Friedenreich et al., 2007; Moore et al., 2010; Schmid et al., 2015; Voskuil et al., 2007). It is hypothesized that physical activity diminishes the risk for cancer indirectly by decreasing adiposity and directly by strengthening immune function, as well as by decreasing inflammation, sex hormone concentrations, and insulin sensitivity (McTiernan, 2008). Only a small number of studies have evaluated the association between endometrial cancer and physical activity stratified by weight or BMI. A meta-​analysis of these studies has shown that high compared to low physical activity decreases endometrial cancer risk by 31% in overweight and obese women; no relationship was observed among non-​ obese women (Schmid et al., 2015).

Intrauterine Contraceptive Devices Ever-​use of intrauterine devices (IUDs) is associated with a reduction in endometrial cancer risk: a meta-​analysis reported a 60% reduction in risk (Beining et al., 2008), although more modest reductions of 40% were reported in another meta-​analysis (Curtis et al., 2007) and 20% in a pooled analysis (Felix et al., 2015a). Two of these summary studies found a stronger reduction in risk with longer durations of use, and a persistent reduction in risk for many years after use ended (Beining et  al., 2008; Felix et  al., 2015a). There is some evidence to suggest that the reduction in risk may be present for use of inert IUDs (e.g., made of stainless steel) and not present for use of copper IUDs (Felix et al., 2015a). While the newer progestin-​releasing IUDs have a clear hormonal impact, biologically it is still unclear why other IUDs reduce endometrial cancer risk. Theories include a reduction in abnormal, pre-​ cancerous cells through the IUD-​induced inflammatory process (Moyer and Mishell, 1971), more complete shedding of the endometrial lining (Guillebaud et al., 1976), and a possible reduction in hormone receptors (de Castro et al., 1986; Guleria et al., 2004; Punnonen et al., 1984).

Abnormal Glucose Tolerance, Metabolic Syndrome, and Diabetes Mellitus There is consistent evidence from meta-​analyses that women with diabetes mellitus are about twice as likely to develop endometrial cancer compared to those without it (Friberg et al., 2007; Liao et al., 2014; Zhang et al., 2013). Obesity increases both the risk of endometrial cancer and the risk of diabetes, making it possible that diabetes is merely a surrogate for obesity. However, an increased risk has remained after adjustment for body mass and/​or physical activity (Friberg et al., 2007). Evidence is also mounting that pre-​diabetic conditions increase risk. A meta-​analysis of studies that evaluated insulin resistance in relation to endometrial cancer risk found higher mean fasting insulin levels, C-​peptide levels, and homeostatic model assessment-​insulin resistance (HOMA-​IR) levels in cases compared to women without endometrial cancer (Hernandez et al., 2015). A pooled analysis of cohort studies also found a 56% increase in risk with a one standard deviation increase in a metabolic risk score (including BMI, blood pressure, and glucose, cholesterol, and triglyceride levels) (Stocks et al., 2015). Additionally, another meta-​analysis reported a 40% elevation in risk for women with metabolic syndrome relative to those without, with higher BMI being the most predictive single factor (Esposito et al., 2014). High levels of circulating insulin (hyperinsulinemia) indirectly increase the proliferation of endometrial cells by decreasing sex hormone–​binding globulin concentrations and leading to an increase in bioavailable estrogens (Calle and Kaaks, 2004). Further, high glucose levels are also linked to cell proliferation in endometrial cancer cell lines through multiple complex signaling pathways (Han et al., 2015). Metformin, a first-​line treatment to control glycemic levels in type 2 diabetics (Davidson and Peters, 1997), conceivably could reduce the development of endometrial cancer by inhibiting cell proliferation and activating cell cycle arrest and apoptosis in vitro (Febbraro et al., 2014). However, a recent case-​control study did not find an endometrial cancer risk reduction in women with diabetes who used metformin versus not (Becker et al., 2013), and similarly a prospective study reported no association after adjustment for BMI (Luo et al., 2014).

Hypertension Although a relationship between hypertension and endometrial cancer has long been hypothesized, results from epidemiologic studies have been inconsistent. Few studies have accounted for age, weight/​BMI, or the metabolic syndrome/​diabetes—​factors associated with both endometrial cancer incidence and hypertension. Nonetheless, one study that did adjust for BMI reported a 30% higher risk in hypertensives than non-​hypertensives among non-​diabetics, and a 2-​fold higher risk associated with hypertension in diabetics (Lucenteforte et al., 2007).

 917

Endometrial Cancer

917

Diet and Alcohol

Environmental Chemical Exposure

Generally, the results from studies assessing individual-​level data for dietary intake of various foods and nutrients and endometrial cancer occurrence are mixed. There is limited evidence that dietary pattern, specifically a Mediterranean diet, that is generally characterized by high intake of vegetables, fruits/​nuts, cereals, legumes, and fish, and low intake of dairy products and meat, as well as a high monounsaturated to saturated fatty acid ratio, and often moderate alcohol intake, is associated with a reduced risk for endometrial cancer (WCRF/​ AICR, 2013). A  more recent pooled analysis of three Italian case-​ control studies found that a Mediterranean diet was associated with a reduction in risk when a woman’s diet contains 6–​9 of the preceding components relative to 0–​3 components (OR = 0.43; 95% CI: 0.34–​ 0.56) (Filomeno et al., 2015), but no association was found in a meta-​ analysis (meta-​analysis RR = 0.72; 95% CI: 0.4, 1.31) (Schwingshackl and Hoffmann, 2015) or in a prospective analysis of postmenopausal women in the Women’s Health Initiative (RR = 0.98; 95% CI: 0.82, 1.17) (George et  al., 2015). Mixed results are also reported for the individual components of the Mediterranean diet. There is some evidence that other dietary components, such as high intake of dietary acrylamide and a high glycemic index, may increase endometrial cancer risk. Additionally, greater intakes of green tea, dietary vitamin A  (retinol and beta-​carotene), soy products, and dietary folate, as well as dietary and/​or supplement intakes of vitamin C, vitamin E, and multivitamins, may decrease the risk for endometrial cancer. However, the World Cancer Research Fund (WCRF) found that the evidence is too mixed and limited for firm conclusions (WCRF/​AICR, 2013). The WCRF deems there to be “probable” evidence of a 15% increase in risk with a 50-​ unit increase in glycemic load, which reflects the average quantity and quality of carbohydrates consumed (WCRF/​AICR, 2013), a conclusion that is supported by the results of three meta-​analyses (Galeone et  al., 2013; Gnagnarella et  al., 2008; Nagle et al., 2013). The WCRF also finds that there is probable evidence that higher coffee consumption, both caffeinated and decaffeinated, decreases risk with an 8% decrease in risk per cup consumed per day. This conclusion is also supported by two meta-​analyses (Je and Giovannucci, 2012; Yu et al., 2011), although no association was found in another meta-​analysis (Bravi et al., 2009) and a prospective study (Hashibe et al., 2015). A meta-​analysis has observed that there is not an association between alcohol consumption and risk (Sun et al., 2011), which is supported by the results of a subsequent prospective European study (Fedirko et al., 2013). However, several other more recent studies report a reduction in risk, particularly among light drinkers (Friedenreich et al., 2013; Je et al., 2014).

No strong or consistent associations were noted in observational studies that have assessed lipid-​adjusted serum or adipose levels of various organochlorines (Hardell et  al., 2004; Sturgeon et  al., 1998; Weiderpass et al., 2000) with endometrial cancer risk.

Smoking Most studies of endometrial cancer in postmenopausal women that have ascertained cigarette smoking behavior note a reduced risk among smokers, summarized in a meta-​analysis as a 20%–​30% reduction in risk for ever versus never smokers (Zhou et  al., 2008). The risk reduction was greater for current smokers (30%–​40%) than for former smokers (9%–​12%) (Felix et al., 2014; Zhou et al., 2008), but among the former smokers, risk was still substantially reduced with shorter times since cessation (within the preceding 5  years [Felix et al., 2014] or 10 years [Zhou et al., 2008]). There are several possible reasons for these findings. Smokers have, on average, a lower BMI than non-​smokers and therefore lower estrogen levels (Edman and MacDonald, 1978; MacDonald et  al., 1978; MacDonald and Siiteri, 1974), but almost all studies adjusted for weight/​BMI. There may be differential metabolism of estrogens among smokers that favors the 2-​ hydroxylation pathway, producing a metabolite of relatively low estrogenic activity (Key et al., 1996; Michnovicz et al., 1986), and smokers may have higher levels of circulating progesterone that diminish estrogenic effect (Friedman et al., 1987).

Familial and Genetic Predisposition HNPCC, or Lynch syndrome, is an autosomally dominant inheritable condition that accounts for 3%–​5% of all endometrial cancer (Meyer et al., 2009), with a cumulative lifetime risk of approximately 40%–​ 60% in women with the syndrome (Gayther and Pharoah, 2010). It is caused by highly penetrant germline mutations in one of four genes encoding for DNA mismatch repair enzymes (MLH1, MSH2, MSH6, and PMS2), which leads to microsatellite instability (MSI) (Lynch et al., 2014). Mutations in these genes are commonly associated with colorectal cancer, but women from some Lynch syndrome families may be at equal or higher risk for endometrial than colorectal cancer depending on the genetic defect (Gayther and Pharoah, 2010). More generally, several studies have evaluated the risk of endometrial cancer associated with a positive family history of endometrial cancer. Although results have varied across the studies, a systematic review and meta-​analysis estimated an 82% elevation in risk associated with a first-​degree family history of endometrial cancer (Win et al., 2015) that did not differ based on age (< 55 and ≥ 55 years) or menopausal status (Win et al., 2015). Three of these studies excluded families with Lynch syndrome from their analyses, and all observed an increase in risk with a family history (Bharati et al., 2014; Cook et al., 2013; Lorenzo Bermejo et al., 2004). At the present time it is unknown if familial aggregation is due to shared genes or a yet-​undetermined endometrial cancer-​ specific genetic syndrome, or rather to shared environment and behaviors among female family members. Apart from inherited defects in DNA mismatch repair, the genetic basis of endometrial cancer susceptibility is less developed than in other cancers. Candidate gene studies have been relatively small with inconsistent results, and the larger genome-​ wide association studies (GWAS) have been used in an attempt to identify a genetic basis for altered endometrial cancer susceptibility. The GWAS (Long et al., 2012; De Vivo et al., 2014; Cheng et al., 2016; Spurdle et al., 2011) have been summarized by a meta-​analysis (Chen et al., 2016) in which three previously reported loci, 6q22.31 (rs2797160), 13q22.1 (rs9600103), and 17q12 (rs11651052), and one new loci, 6p22.3 (rs1740828), reached genome-​ wide significance with no evidence of significant study heterogeneity (Chen et al., 2016). However, it is estimated that these loci account for only 4.4% of familial endometrial cancer, and the GWAS analyses did not identify candidate genes that showed promising results in other studies, such as sex hormone metabolic genes (e.g., cytochrome P-​450 isozymes) (De Vivo et  al., 2014). Another single nucleotide polymorphism (SNP), rs7679673, near TET2 and previously reported to be associated with prostate cancer risk, was associated with a reduction in endometrial cancer risk (OR = 0.87 [per copy of the C allele]; 95% CI: 0.81, 0.93) in a pleiotropic analysis of SNPs related to several cancers (Setiawan et al., 2014).

DISEASE PREVENTION It is highly likely that the sharp decline in the use of E-​only postmenopausal hormone therapy among American women has been responsible for the decline in the occurrence of endometrial cancer in the United States since the mid-​1970s. While OC use cannot be recommended for general prevention, studies are underway to determine if use of OCs can reduce endometrial cancer occurrence in high-​risk women, such as those with Lynch syndrome (Lu et al., 2013). And finally, a decrease in the prevalence of overweight and obesity in women would also reduce the occurrence of endometrial cancer (in addition to having many other health benefits).

918

918

PART IV:  Cancers by Tissue of Origin

FUTURE RESEARCH A clear understanding of etiologic heterogeneity—​whether by the type I/​II classification, by other molecular signatures, or by genetic syndromes—​has promise but is yet to be realized in endometrial cancer. Similarly, the etiologic effects of closely related risk factors such as BMI, physical activity, the metabolic syndrome/​diabetes, and hypertension, will continue to be a research challenge. The incidence of endometrial cancer associated with hormone therapy among postmenopausal women needs to be an active area of research to the extent that the means of providing hormone therapy changes over time. This research will include continued evaluation of E-​only supplemented with more locally targeted progestins, such as progestin-​releasing IUDs. Presumably, the progestin-​releasing IUDs may also reduce risk among obese women, because these devices are used to treat early stage endometrial cancer in obese women. Long-​term evaluations of SERMs—​such as raloxifene, which may provide some of the benefits of E-​only therapy without inducing endometrial proliferation—​are also needed. The goal will be to identify hormones and combinations of hormones that minimize any increase in the occurrence of endometrial cancer while not increasing the risk of other long-​term deleterious health outcomes, such as breast cancer (Rossouw et al., 2002). References American Joint Commission on Cancer (AJCC). 2010. Cancer Staging Manual, 7th ed. Edge S, Byrd D, Compton C, Fritz A, Greene F, and Trotti A (Eds.). New York: Springer. Allen NE, Key TJ, Dossus L, et al. 2008. Endogenous sex hormones and endometrial cancer risk in women in the European Prospective Investigation into Cancer and Nutrition (EPIC). Endocr Relat Cancer, 15(2), 485–​497. PMCID: PMC2396334. Amankwah EK, Friedenreich CM, Magliocco AM, et al. 2013a. Anthropometric measures and the risk of endometrial cancer, overall and by tumor microsatellite status and histological subtype. Am J Epidemiol, 177(12), 1378–​ 1387. PMCID: PMC3732018. Amankwah EK, Friedenreich CM, Magliocco AM, et al. 2013b. Hormonal and reproductive risk factors for sporadic microsatellite stable and unstable endometrial tumors. Cancer Epidemiol Biomarkers Prev, 22(7), 1325–​ 1331. PMCID: PMC4419269. Arab M, and Noghabaei G. 2014. Comparison of age-​standard incidence rate trends of gynecologic and breast cancer in Iran and other countries. Iran J Public Health, 43(10), 1372–​1379. PMCID: PMC4441890. Aune D, Navarro Rosenblatt DA, Chan DS, et al. 2015. Anthropometric factors and endometrial cancer risk: a systematic review and dose-​response meta-​ analysis of prospective studies. Ann Oncol, 26(8), 1635–​ 1648. PMID: 25791635. Austin DF, and Roe KM. 1982. The decreasing incidence of endometrial cancer:  public health implications. Am J Public Health, 72(1), 65–​68. PMCID: PMC1649756. Baanders-​van Halewyn EA, Blankenstein MA, Thijssen JH, de Ridder CM, and de Waard F. 1996. A comparative study of risk factors for hyperplasia and cancer of the endometrium. Eur J Cancer Prev, 5(2), 105–​112. PMID: 8736077. Barry JA, Azizia MM, and Hardiman PJ. 2014. Risk of endometrial, ovarian and breast cancer in women with polycystic ovary syndrome: a systematic review and meta-analysis. Hum Reprod Update, 20(5), 748–758. PMID: 24688118. Basil JB, Goodfellow PJ, Rader JS, Mutch DG, and Herzog TJ. 2000. Clinical significance of microsatellite instability in endometrial carcinoma. Cancer, 89(8), 1758–​1764. PMID: 11042571. Becker C, Jick SS, Meier CR, and Bodmer M. 2013. Metformin and the risk of endometrial cancer:  a case-​control analysis. Gynecol Oncol, 129(3), 565–​569. PMID: 23523618. Beining RM, Dennis LK, Smith EM, and Dokras A. 2008. Meta-​analysis of intrauterine device use and risk of endometrial cancer. Ann Epidemiol, 18(6), 492–​499. PMID: 18261926. Beral V, Bull D, Reeves G, et  al. 2005. Endometrial cancer and hormone-​ replacement therapy in the Million Women Study. Lancet, 365(9470), 1543–​1551. PMID: 15866308. Bergman L, Beelen ML, Gallee MP, et  al. 2000. Risk and prognosis of endometrial cancer after tamoxifen for breast cancer. Comprehensive Cancer Centres’ ALERT Group, Assessment of Liver and Endometrial Cancer Risk Following Tamoxifen. Lancet, 356(9233), 881–​ 887. PMID: 11036892.

Bernstein L, Deapen D, Cerhan JR, et al. 1999. Tamoxifen therapy for breast cancer and endometrial cancer risk. J Natl Cancer Inst, 91(19), 1654–​ 1662. PMID: 10511593. Bevier M, Sundquist J, and Hemminki K. 2011. Does the time interval between first and last birth influence the risk of endometrial and ovarian cancer? Eur J Cancer, 47(4), 586–​591. PMID: 21055917. Bharati R, Jenkins MA, Lindor NM, et al. 2014. Does risk of endometrial cancer for women without a germline mutation in a DNA mismatch repair gene depend on family history of endometrial cancer or colorectal cancer? Gynecol Oncol, 133(2), 287–​292. PMCID: PMC4016812. Bokhman JV. 1983. Two pathogenetic types of endometrial carcinoma. Gynecol Oncol, 15(1), 10–​17. PMID: 6822361. Boks DE, Trujillo AP, Voogd AC, et al. 2002. Survival analysis of endometrial carcinoma associated with hereditary nonpolyposis colorectal cancer. Int J Cancer, 102(2), 198–​200. PMID: 12385019. Boss SM, Huster WJ, Neild JA, et al. 1997. Effects of raloxifene hydrochloride on the endometrium of postmenopausal women. Am J Obstet Gynecol, 177(6), 1458–​1464. PMID: 9423751. Braithwaite RS, Chlebowski RT, Lau J, et  al. 2003. Meta-​analysis of vascular and neoplastic events associated with tamoxifen. J Gen Intern Med, 18(11), 937–​947. PMCID: PMC1494944. Bravi F, Scotti L, Bosetti C, et al. 2009. Coffee drinking and endometrial cancer risk: a metaanalysis of observational studies. Am J Obstet Gynecol, 200(2), 130–​135. PMID: 19110217. Brinton LA, Berman ML, Mortel R et al. 1992. Reproductive, menstrual, and medical risk factors for endometerial cancer: results from a case-​control study. Am J Obstet Gynecol, 167, 1317–​1325. PMID: 1442985. Brinton LA, and Felix AS. 2014. Menopausal hormone therapy and risk of endometrial cancer. J Steroid Biochem Mol Biol, 142, 83–​ 89. PMCID: PMC3775978. Brinton LA, Hoover RN, and The Endometrial Cancer Collaborative Group. 1993. Estrogen replacement therapy and endometrial cancer risk: unresolved issues. Obstet Gynecol, 81, 265–​271. Brinton LA, Sakoda LC, Lissowska J, et  al. 2007. Reproductive risk factors for endometrial cancer among Polish women. Br J Cancer, 96(9), 1450–​ 1456. PMCID: PMC2360184. Brinton LA, Trabert B, Anderson GL, et  al. 2016. Serum estrogens and estrogen metabolites and endometrial cancer risk among postmenopausal women. Cancer Epidemiol Biomarkers Prev, 25(7), 1081–​1089. PMCID: PMC4930692. Calle EE, and Kaaks R. 2004. Overweight, obesity and cancer: epidemiological evidence and proposed mechanisms. Nat Rev Cancer, 4(8), 579–​591. PMID: 15286738. Castellsague X, Thompson WD, and Dubrow R. 1996. Tubal sterilization and the risk of endometrial cancer. Int J Cancer, 65(5), 607–​612. PMID: 8598311. Chapman JA, DiSaia PJ, Osann K, et al. 1996. Estrogen replacement in surgical stage I and II endometrial cancer survivors. Am J Obstet Gynecol, 175(5), 1195–​1200. PMID: 8942487. Chen MM, O’Mara TA, Thompson DJ, et  al. 2016. GWAS meta-​analysis of 16 852 women identifies new susceptibility locus for endometrial cancer. Hum Mol Genet. E-​pub ahead of print. PMID: 27008869. Chlebowski RT, Anderson GL, Sarto GE, et  al. 2016. Continuous combined estrogen plus progestin and endometrial cancer:  The Women’s Health Initiative Randomized Trial. J Natl Cancer Inst, 108(3). PMID: 26668177. Chu J, Schweid AI, and Weiss NS. 1982. Survival among women with endometrial cancer -​a comparison of estrogen users and nonusers. Am J Obstet Gynecol, 143(5), 569–​573. PMID: 7091228. Collaborative Group on Epidemiological Studies on Endometrial Cancer. 2015. Endometrial cancer and oral contraceptives:  an individual participant meta-​analysis of 27 276 women with endometrial cancer from 36 epidemiological studies. Lancet Oncol, 16(9), 1061–​1070. PMID: 26254030. Collins J, Donner A, Allen LH, and Adams O. 1980. Estrogen use and survival in endometrial cancer. Lancet, 2(8201), 961–​ 964. PMID: WOS:A1980KN32900015. Colombo N, Preti E, Landoni F, et al. 2011. Endometrial cancer: ESMO clinical practice guidelines for diagnosis, treatment and follow-​up. Ann Oncol, 22(Suppl 6), vi35–​vi39. PMID: 21908501. Cook LS, Dong Y, Round P, et al. 2014. Hormone contraception before the first birth and endometrial cancer risk. Cancer Epidemiol Biomarkers Prev, 23(2), 356–​361. PMID: 24326627. Cook LS, Nelson HE, Stidley CA, et  al. 2013. Endometrial cancer and a family history of cancer. Gynecol Oncol, 130(2), 334–​ 339. PMCID: PMC4052607. Cook LS, Weiss NS, Schwartz SM, et  al. 1995. Population-​based study of tamoxifen therapy and subsequent ovarian, endometrial, and breast cancers. J Natl Cancer Inst, 87(18), 1359–​1364. PMID: 7658496.

 91

Endometrial Cancer Cote ML, Alhajj T, Ruterbusch JJ, et al. 2015. Risk factors for endometrial cancer in black and white women: a pooled analysis from the Epidemiology of Endometrial Cancer Consortium (E2C2). Cancer Causes Control, 26(2), 287–​296. PMCID: PMC4528374. Curtis KM, Marchbanks PA, and Peterson HB. 2007. Neoplasia with use of intrauterine devices. Contraception, 75(6 Suppl), S60–​S69. PMID: 17531619. Cushing KL, Weiss NS, Voigt LF, McKnight B, and Beresford SAA. 1998. Risk of endometrial cancer in relation to use of low-​dose, unopposed estrogens. Obstet Gynecol, 91(1), 35–​39. PMID: WOS:000071237300007. Cust AE, Armstrong BK, Friedenreich CM, Slimani N, and Bauman A. 2007. Physical activity and endometrial cancer risk: a review of the current evidence, biologic mechanisms and the quality of physical activity assessment methods. Cancer Causes Control, 18(3), 243–​258. PMID: 17206535. Cuzick J, Powles T, Veronesi U, et  al. 2003. Overview of the main outcomes in breast-​cancer prevention trials. Lancet, 361(9354), 296–​300. PMID: 12559863. Davidson BJ, Gambone JC, Lagasse LD, et al. 1981. Free estradiol in postmenopausal women with and without endometrial cancer. J Clin Endocrinol Metab, 52(3), 404–​408. Davidson MB, and Peters AL. 1997. An overview of metformin in the treatment of type 2 diabetes mellitus. Am J Med, 102(1), 99–​110. PMID: 9209206. de Castro A, Gonzalez-​Gancedo P, Contreras F, and Lapena G. 1986. The effect of copper ions in vivo on specific hormonal endometrial receptors. Adv Contracept, 2(4), 399–​404. PMID: 3551524. De Vivo I, Prescott J, Setiawan VW, et  al. 2014. Genome-​wide association study of endometrial cancer in E2C2. Hum Genet, 133(2), 211–​224. PMCID: PMC3898362. Deligdisch L, and Holinka CF. 1987. Endometrial carcinoma:  two diseases? Cancer Detect Prev, 10(3–​4), 237–​246. PMID: 3568022. DeMichele A, Troxel AB, Berlin JA, et al. 2008. Impact of raloxifene or tamoxifen use on endometrial cancer risk:  a population-​based case-​control study. J Clin Oncol, 26(25), 4151–​4159. PMCID: PMC2654370. Dicker RC, Scally MJ, Greenspan JR, et al. 1982. Hysterectomy among women of reproductive age:  trends in the United States, 1970–​1978. JAMA, 248(3), 323–​327. PMID: 7087126. Doherty JA, Cushing-​Haugen KL, Saltzman BS, et  al. 2007. Long-​term use of postmenopausal estrogen and progestin hormone therapies and the risk of endometrial cancer. Am J Obstet Gynecol, 197(2), 139–​ 137. PMID: WOS:000248718900005. Dossus L, Allen N, Kaaks R, et al. 2010. Reproductive risk factors and endometrial cancer: the European Prospective Investigation into Cancer and Nutrition. Int J Cancer, 127(2), 442–​451. PMID: 19924816. Dossus L, Lukanova A, Rinaldi S et al. 2013. Hormonal, metabolic, and inflammatory profiles and endometrial cancer risk within the EPIC cohort:  a factor analysis. Am J Epidemiol, 177(8), 787–​799. PMID: 23492765. Duggan BD, Felix JC, Muderspach LI, et al. 1994. Microsatellite instability in sporadic endometrial carcinoma. J Natl Cancer Inst, 86(16), 1216–​1221. PMID: 8040889. Edman CD, and MacDonald PC. 1978. Effect of obesity on conversion of plasma androstenedione to estrone in ovulatory and anovulator young women. Am J Obstet Gynecol, 130(4), 456–​461. PMID: 629290. Elwood JM, and Boyes DA. 1980. Clinical and pathological features and survival of endometrial cancer-​patients in relation to prior use of estrogens. Gynecol Oncol, 10(2), 173–​187. PMID: WOS:A1980KN01800007. Esposito K, Chiodini P, Capuano A, et  al. 2014. Metabolic syndrome and endometrial cancer:  a meta-​ analysis. Endocrine, 45(1), 28–​ 36. PMID: 23640372. Ewertz M, Schou G, and Boice JD, Jr. 1988. The joint effect of risk factors on endometrial cancer. Eur J Cancer Clin Oncol, 24(2), 189–​194. PMID: 3356205. Farhi DC, Nosanchuk J, and Silverberg SG. 1986. Endometrial adenocarcinoma in women under 25 years of age. Obstet Gynecol, 68(6), 741–​745. PMID: 3785784. Febbraro T, Lengyel E, and Romero IL. 2014. Old drug, new trick: repurposing metformin for gynecologic cancers? Gynecol Oncol, 135(3), 614–​621. PMCID: PMC4268303. Fedirko V, Jenab M, Rinaldi S, et al. 2013. Alcohol drinking and endometrial cancer risk in the European Prospective Investigation into Cancer and Nutrition (EPIC) study. Ann Epidemiol, 23(2), 93–​ 98. PMID: 23273691. Felix AS, Gaudet MM, La Vecchia C, et  al. 2015a. Intrauterine devices and endometrial cancer risk:  a pooled analysis of the Epidemiology of Endometrial Cancer Consortium. Int J Cancer, 136(5), E410–​ E422. PMCID: PMC4267918. Felix AS, Scott McMeekin D, Mutch D, et  al. 2015b. Associations between etiologic factors and mortality after endometrial cancer diagnosis:  the

919

NRG Oncology/​Gynecologic Oncology Group 210 trial. Gynecol Oncol, 139(1), 70–​76. PMCID: PMC4587355. Felix AS, Yang HP, Gierach GL, Park Y, and Brinton LA. 2014. Cigarette smoking and endometrial carcinoma risk: the role of effect modification and tumor heterogeneity. Cancer Causes Control, 25(4), 479–​489. PMCID: PMC4151521. Ferlay J, Soerjomataram I, Ervik M, et  al. 2013. GLOBOCAN 2012 v1.0, Cancer incidence and mortality worldwide:  IARC CancerBase No. 11 [Internet]. Available from: http://​globocan.iarc.fr. Accessed July 20, 2016. Filomeno M, Bosetti C, Bidoli E, et al. 2015. Mediterranean diet and risk of endometrial cancer: a pooled analysis of three Italian case-​control studies. Br J Cancer, 112(11), 1816–​1821. PMCID: PMC4647248. Food and Drug Administration. 1976. Estrogens and endometrial cancer. FDA Drug Bulletin, 6, 18–​20. Friberg E, Mantzoros CS, and Wolk A. 2007. Diabetes and risk of endometrial cancer: a population-​based prospective cohort study. Cancer Epidemiol Biomarkers Prev, 16(2), 276–​280. PMID: 17301260. Friedenreich C, Cust A, Lahmann PH, et al. 2007. Physical activity and risk of endometrial cancer:  the European prospective investigation into cancer and nutrition. Int J Cancer, 121(2), 347–​355. PMID: 17357139. Friedenreich CM, Speidel TP, Neilson HK, et  al. 2013. Case-​control study of lifetime alcohol consumption and endometrial cancer risk. Cancer Causes Control, 24(11), 1995–​2003. PMCID: PMC3824213. Friedman AJ, Ravnikar VA, and Barbieri RL. 1987. Serum steroid hormone profiles in postmenopausal smokers and nonsmokers. Fertil Steril, 47(3), 398–​401. PMID: 2951278. Fu Y, and Zhuang Z. 2014. Long-​ term effects of levonorgestrel-​ releasing intrauterine system on tamoxifen-​treated breast cancer patients: a meta-​ analysis. Int J Clin Exp Pathol, 7(10), 6419–​6429. PMCID: PMC4230119. Furness S, Roberts H, Marjoribanks J, and Lethaby A. 2012. Hormone therapy in postmenopausal women and risk of endometrial hyperplasia. Cochrane Database Syst Rev, 8(8), CD000402. PMID: 22895916. Galeone C, Augustin LS, Filomeno M, et al. 2013. Dietary glycemic index, glycemic load, and the risk of endometrial cancer: a case-​control study and meta-​analysis. Eur J Cancer Prev, 22(1), 38–​45. PMID: 22584216. Gayther SA, and Pharoah PD. 2010. The inherited genetics of ovarian and endometrial cancer. Curr Opin Genet Dev, 20(3), 231–​ 238. PMCID: PMC3073171. George SM, Ballard R, Shikany JM, Crane TE, and Neuhouser ML. 2015. A prospective analysis of diet quality and endometrial cancer among 84,415 postmenopausal women in the Women’s Health Initiative. Ann Epidemiol, 25(10), 788–​793. PMID: 26260777. Gizzo S, Di Gangi S, Bertocco A, et al. 2014. Levonorgestrel intrauterine system in adjuvant tamoxifen treatment: balance of breast risks and endometrial benefits-​-​systematic review of literature. Reprod Sci, 21(4), 423–​431. PMID: 24060633. Gnagnarella P, Gandini S, La Vecchia C, and Maisonneuve P. 2008. Glycemic index, glycemic load, and cancer risk: a meta-​analysis. Am J Clin Nutr, 87(6), 1793–​1801. PMID: 18541570. Gong TT, Wang YL, and Ma XX. 2015. Age at menarche and endometrial cancer risk: a dose-​response meta-​analysis of prospective studies. Sci Rep, 5, 14051. PMCID: PMC4566123. Grady D, and Ernster VL. 1996. Endometrial cancer. In: Schottenfeld D, and Fraumeni JF (Eds.), Cancer Epidemiology and Prevention, 2nd ed., pp. 1058–​1089 New York: Oxford University Press. Grady D, Gebretsadik T, Kerlikowske K, Ernster V, and Petitti D. 1995. Hormone replacement therapy and endometrial cancer risk:  a meta-​ analysis. Obstet Gynecol, 85(2), 304–​313. PMID: 7824251. Gray LA, Christopherson WM, and Hoover RN. 1977. Estrogens and endometrial carcinoma. Obstet Gynecol, 49(4), 385–​ 389. PMID: WOS:A1977DB65400001. Green PK, Weiss NS, McKnight B, Voigt LF, and Beresford SA. 1996. Risk of endometrial cancer following cessation of menopausal hormone use (Washington, United States). Cancer Causes Control, 7(6), 575–​ 580. PMID: 8932917. Gruber JS, and Luciani CT. 1986. Physicians changing postmenopausal sex hormone prescribing regimens. Prog Clin Biol Res, 216, 325–​ 335. PMID: 3725795. Guillebaud J, Bonnar J, Morehead J, and Matthews A. 1976. Menstrual blood-​loss with intrauterine devices. Lancet, 1(7956), 387–​390. PMID: 55650. Guleria K, Agarwal N, Mishra K, Gulati R, and Mehendiratta A. 2004. Evaluation of endometrial steroid receptors and cell mitotic activity in women using copper intrauterine device: can Cu-​T prevent endometrial cancer? J Obstet Gynaecol Res, 30(3), 181–​187. PMID: 15210039. Gusberg SB, and Hall RE. 1961. Precursors of corpus cancer. III. The appearance of cancer of the endometrium in estrogenically conditioned patients. Obstet Gynecol, 17, 397–​412. PMID: 13709910.

920

920

PART IV:  Cancers by Tissue of Origin

Hachisuga T, Fukuda K, Hirakawa T, and Kawarabayashi T. 2001. The effect of nulliparity on survival in endometrial cancer at different ages. Gynecol Oncol, 82(1), 122–​126. PMID: 11426973. Hachisuga T, Kaetsu A, Sugimori H, et  al. 1998. Risk factors for endometrial cancer in Japanese women. Int J Gynecol Cancer, 8(4), 292–​297. PMID: WOS:000075928200006. Hall IJ, Moorman PG, Millikan RC, and Newman B. 2005. Comparative analysis of breast cancer risk factors among African-​American women and white women. Am J Epidemiol, 161(1), 40–​51. PMID: 15615914. Han J, Zhang L, Guo H, et  al. 2015. Glucose promotes cell proliferation, glucose uptake and invasion in endometrial cancer cells via AMPK/​ mTOR/​ S6 and MAPK signaling. Gynecol Oncol, 138(3), 668–​ 675. PMCID: PMC4672629. Hanahan D, and Weinberg RA. 2011. Hallmarks of cancer: the next generation. Cell, 144(5), 646–​674. PMID: 21376230. Hannaford PC, Selvaraj S, Elliott AM, et  al. 2007. Cancer risk among users of oral contraceptives:  cohort data from the Royal College of General Practitioner’s oral contraception study. BMJ, 335(7621), 651. PMCID: PMC1995533. Hardell L, van Bavel B, Lindstrom G, et al. 2004. Adipose tissue concentrations of p,p’-​DDE and the risk for endometrial cancer. Gynecol Oncol, 95(3), 706–​711. PMID: 15581986. Hashibe M, Galeone C, Buys SS, et al. 2015. Coffee, tea, caffeine intake, and the risk of cancer in the PLCO cohort. Br J Cancer, 113(5), 809–​816. PMCID: PMC4559834. Henderson BE, Casagrande JT, Pike MC, et  al. 1983. The epidemiology of endometrial cancer in young women. Br J Cancer, 47(6), 749–​ 756. PMCID: PMC2011352. Hernandez AV, Pasupuleti V, Benites-​Zapata VA, et al. 2015. Insulin resistance and endometrial cancer risk: a systematic review and meta-​analysis. Eur J Cancer, 51(18), 2747–​2758. PMID: 26597445. Herrington LJ, and Weiss NS. 1993. Postmenopausal unopposed estrogens:  characteristics of use in relation to the risk of endometrial carcinoma. Ann Epidemiol, 3, 308–​318. Hirose K, Tajima K, Hamajima N, et al. 1996. Subsite (cervix/​endometrium)-​ specific risk and protective factors in uterus cancer. Jpn J Cancer Res, 87(9), 1001–​1009. PMID: 8878465. Howe HL. 1984. Age-​specific hysterectomy and oophorectomy prevalence rates and the risks for cancer of the reproductive system. Am J Public Health, 74(6), 560–​563. PMCID: PMC1651658. Hulka BS, Chambless LE, Kaufman DG, Fowler WC, Jr., and Greenberg BG. 1982. Protection against endometrial carcinoma by combination-​product oral contraceptives. JAMA, 247(4), 475–​477. PMID: 7033575. Hulley S, Furberg C, Barrett-​Connor E, et al. 2002. Noncardiovascular disease outcomes during 6.8 years of hormone therapy: Heart and Estrogen/​progestin Replacement Study follow-​up (HERS II). JAMA, 288(1), 58–​66. Jaakkola S, Lyytinen H, Pukkala E, and Ylikorkala O. 2009. Endometrial cancer in postmenopausal women using estradiol-​progestin therapy. Obstet Gynecol, 114(6), 1197–​1204. PMID: 19935019. Jaakkola S, Lyytinen HK, Dyba T, Ylikorkala O, and Pukkala E. 2011. Endometrial cancer associated with various forms of postmenopausal hormone therapy: a case control study. Int J Cancer, 128(7), 1644–​1651. PMID: 21280035. Jamison PM, Noone AM, Ries LA, Lee NC, and Edwards BK. 2013. Trends in endometrial cancer incidence by race and histology with a correction for the prevalence of hysterectomy, SEER 1992 to 2008. Cancer Epidemiol Biomarkers Prev, 22(2), 233–​241. PMID: 23239812. Je Y, DeVivo I, and Giovannucci E. 2014. Long-​term alcohol intake and risk of endometrial cancer in the Nurses’ Health Study, 1980–​2010. Br J Cancer, 111(1), 186–​194. PMCID: PMC4090729. Je Y, and Giovannucci E. 2012. Coffee consumption and risk of endometrial cancer:  findings from a large up-​to-​date meta-​analysis. Int J Cancer, 131(7), 1700–​1710. PMID: 22190017. Jenabi E, and Poorolajal J. 2015. The effect of body mass index on endometrial cancer:  a meta-​ analysis. Public Health, 129(7), 872–​ 880. PMID: 26026348. Jewett PI, Gangnon RE, Trentham-​Dietz A, and Sprague BL. 2014. Trends of postmenopausal estrogen plus progestin prevalence in the United States between 1970 and 2010. Obstet Gynecol, 124(4), 727–​ 733. PMCID: PMC4172523. Jordan VC. 1995. Tamoxifen:  toxicities and drug resistance during the treatment and prevention of breast cancer. Annu Rev Pharmacol Toxicol, 35, 195–​211. PMID: 7598491. Kaaks R, and Lukanova A. 2002. Effects of weight control and physical activity in cancer prevention: role of endogenous hormone metabolism. Ann N Y Acad Sci, 963, 268–​281. PMID: 12095952.

Kaaks R, Lukanova A, and Kurzer MS. 2002. Obesity, endogenous hormones, and endometrial cancer risk:  a synthetic review. Cancer Epidemiol Biomarkers Prev, 11(12), 1531–​1543. PMID: 12496040. Kalandidi A, Tzonou A, Lipworth L et al. 1996. A case-​control study of endometrial cancer in relation to reproductive, somatometric, and life-​style variables. Oncology, 53(5), 354–​359. PMID: 8784467. Karageorgi S, Hankinson SE, Kraft P, and De Vivo I. 2010. Reproductive factors and postmenopausal hormone use in relation to endometrial cancer risk in the Nurses’ Health Study cohort 1976–​2004. Int J Cancer, 126(1), 208–​216. PMCID: PMC2784268. Kaye SA, Folsom AR, Soler JT, Prineas RJ, and Potter JD. 1991. Association of body mass and fat distribution with sex hormone concentrations in postmenopausal women. Int J Epidemiol, 20, 151–​156. Kelsey JL, LiVolsi VA, Holford TR et al. 1982. A case-​control study of cancer of the endometrium. Am J Epidemiol, 116(2), 333–​342. PMID: 7114042. Kennedy DL, Baum C, and Forbes MB. 1985. Noncontraceptive estrogens and progestins: use patterns over time. Obstet Gynecol, 65(3), 441–​446. PMID: WOS:A1985ACQ8200030. Key TJ, Pike MC, Brown JB, et al. 1996. Cigarette smoking and urinary oestrogen excretion in premenopausal and post-​menopausal women. Br J Cancer, 74(8), 1313–​1316. PMCID: PMC2075917. King RJB, and Whitehead MI. 1983. Estrogen and progestin effects on epithelium and stroma from pre-​and postmenopausal endometria: application to clinical studies of the climacteric syndrome. In: Jasonni VM (Ed.), Steroids and Endometrial Cancer. New York: Raven Press. Koepsell TD, Weiss NS, Thompson DJ, and Martin DP. 1980. Prevalence of prior hysterectomy in the Seattle-​Tacoma area. Am J Public Health, 70(1), 40–​47. PMCID: PMC1619330. Kurman RJ, Kaminski PF, and Norris HJ. 1985. The behavior of endometrial hyperplasia: a long-​term study of “untreated” hyperplasia in 170 patients. Cancer, 56(2), 403–​412. PMID: 4005805. Kvale G, Heuch I, and Ursin G. 1988. Reporductive factors and risk of cancer of the uterine corpus: a prospective study. Cancer Res, 48, 6217–​6221. Lacey JV, Jr., Brinton LA, Lubin JH, et al. 2005. Endometrial carcinoma risks among menopausal estrogen plus progestin and unopposed estrogen users in a cohort of postmenopausal women. Cancer Epidemiol Biomarkers Prev, 14(7), 1724–​1731. PMID: 16030108. Lacey JV, Jr., Brinton LA, Mortel R, et  al. 2000. Tubal sterilization and risk of cancer of the endometrium. Gynecol Oncol, 79(3), 482–​484. PMID: 11104624. Lambe M, Wuu J, Weiderpass E, and Hsieh CC. 1999. Childbearing at older age and endometrial cancer risk (Sweden). Cancer Causes Control, 10(1), 43–​49. PMID: WOS:000079727100006. Lee JY, Kim EY, Jung KW, et  al. 2014. Trends in gynecologic cancer mortality in East Asian regions. J Gynecol Oncol, 25(3), 174–​ 182. PMCID: PMC4102735. Liang AP, Levenson AG, Layde PM, et al. 1983. Risk of breast, uterine corpus, and ovarian cancer in women receiving medroxyprogesterone injections. JAMA, 249(21), 2909–​2912. PMID: 6842804. Liao C, Zhang D, Mungo C, Tompkins DA, and Zeidan AM. 2014. Is diabetes mellitus associated with increased incidence and disease-​specific mortality in endometrial cancer? A  systematic review and meta-​analysis of cohort studies. Gynecol Oncol, 135(1), 163–​171. PMCID: PMC4404750. Liedtke S, Schmidt ME, Vrieling A, et al. 2012. Postmenopausal sex hormones in relation to body fat distribution. Obesity (Silver Spring), 20(5), 1088–​ 1095. PMID: 22240723. Lindemann K, Cvancarova M, and Eskild A. 2015. Body mass index, diabetes and survival after diagnosis of endometrial cancer: a report from the HUNT-​Survey. Gynecol Oncol, 139(3), 476–​480. PMID: 26434365. Long B, Liu FW, and Bristow RE. 2013. Disparities in uterine cancer epidemiology, treatment, and survival among African Americans in the United States. Gynecol Oncol, 130(3), 652–​659. PMCID: PMC4074587. Long J, Zheng W, Xiang YB, et  al. 2012. Genome-​wide association study identifies a possible susceptibility locus for endometrial cancer. Cancer Epidemiol Biomarkers Prev, 21(6), 980–​987. PMCID: PMC3372671. Lorenzo Bermejo J, Buchner FL, and Hemminki K. 2004. Familial risk of endometrial cancer after exclusion of families that fulfilled Amsterdam, Japanese or Bethesda criteria for HNPCC. Ann Oncol, 15(4), 598–​604. PMID: 15033666. Lorenzoni C, Vilajeliu A, Carrilho C, et al. 2015. Trends in cancer incidence in Maputo, Mozambique, 1991–​ 2008. PLoS One, 10(6), e0130469. PMCID: PMC4481529. Lu KH, Loose DS, Yates MS, et al. 2013. Prospective multicenter randomized intermediate biomarker study of oral contraceptive versus depo-​provera for prevention of endometrial cancer in women with Lynch syndrome. Cancer Prev Res (Phila), 6(8), 774–​781. PMCID: PMC3737517.

 921

Endometrial Cancer Lucenteforte E, Bosetti C, Talamini R, et al. 2007. Diabetes and endometrial cancer: effect modification by body weight, physical activity and hypertension. Br J Cancer, 97(7), 995–​998. PMCID: PMC2360421. Luhn P, Dallal CM, Weiss JM, et  al. 2013. Circulating adipokine levels and endometrial cancer risk in the prostate, lung, colorectal, and ovarian cancer screening trial. Cancer Epidemiol Biomarkers Prev, 22(7), 1304–​ 1312. PMCID: PMC3819202. Lukanova A, Lundin E, Micheli A, et al. 2004. Circulating levels of sex steroid hormones and risk of endometrial cancer in postmenopausal women. Int J Cancer, 108(3), 425–​432. PMID: 14648710. Luo J, Beresford S, Chen C et al. 2014. Association between diabetes, diabetes treatment and risk of developing endometrial cancer. Br J Cancer, 111(7), 1432–​1439. PMCID: PMC4183842. Lynch HT, Drescher K, Knezetic J, and Lanspa S. 2014. Genetics, biomarkers, hereditary cancer syndrome diagnosis, heterogeneity and treatment:  a review. Curr Treat Options Oncol, 15(3), 429–​ 442. PMID: 24827900. Lyon FA, and Frisch MJ. 1976. Endometrial abnormalities occurring in young women on long-​ term sequential oral contraceptives. Obstet Gynecol, 47(6), 639–​643. Lyon JL, and Gardner JW. 1977. The rising frequency of hysterectomy:  its effect on uterine cancer rates. Am J Epidemiol, 105(5), 439–​443. PMID: 860706. MacDonald ND, Salvesen HB, Ryan A, et  al. 2000. Frequency and prognostic impact of microsatellite instability in a large population-​based study of endometrial carcinomas. Cancer Res, 60(6), 1750–​ 1752. PMID: 10749149. MacDonald PC, Edman CD, Hemsell DL, Porter JC, and Siiteri PK. 1978. Effect of obesity on conversion of plasma androstenedione to estrone in postmenopausal women with and without endometrial cancer. Am J Obstet Gynecol, 130(4), 448–​455. MacDonald PC, and Siiteri PK. 1974. The relationship between the extraglandular production of estrone and the occurrence of endometrial neoplasia. Gynecol Oncol, 2(2–​3), 259–​263. Mascarenhas L, van Beek A, Bennink HC, and Newton J. 1998. A  2-​year comparative study of endometrial histology and cervical cytology of contraceptive implant users in Birmingham, UK. Hum Reprod, 13(11), 3057–​3060. PMID: 9853856. Maxwell GL, Schildkraut JM, Calingaert B, et al. 2006. Progestin and estrogen potency of combination oral contraceptives and endometrial cancer risk. Gynecol Oncol, 103(2), 535–​540. PMID: 16740300. McPherson CP, Sellers TA, Potter JD, Bostick RM, and Folsom AR. 1996. Reproductive factors and risk of endometrial cancer. The Iowa Women’s Health Study. Am J Epidemiol, 143(12), 1195–​1202. PMID: 8651218. McTiernan A. 2008. Mechanisms linking physical activity with cancer. Nat Rev Cancer, 8(3), 205–​211. PMID: 18235448. Meyer LA, Broaddus RR, and Lu KH. 2009. Endometrial cancer and Lynch syndrome: clinical and pathologic considerations. Cancer Control, 16(1), 14–​22. PMCID: PMC3693757. Michnovicz JJ, Hershcopf RJ, Naganuma H, Bradlow HL, and Fishman J. 1986. Increased 2-​hydroxylation of estradiol as a possible mechanism for the anti-​estrogenic effect of cigarette smoking. N Engl J Med, 315(21), 1305–​1309. Mirkin S, and Pickar JH. 2015. Selective estrogen receptor modulators (SERMs):  a review of clinical data. Maturitas, 80(1), 52–​ 57. PMID: 25466304. Mittal KR, and Barwick KW. 1993. Endometrial adenocarcinoma involving adenomyosis without true myometrial invasion is characterized by frequent preceding estrogen therapy, low histologic grades, and excellent prognosis. Gynecol Oncol, 49(2), 197–​201. PMID: 8504988. Modan B, Ron E, Lerner-​Geva L et al. 1998. Cancer incidence in a cohort of infertile women. Am J Epidemiol, 147(11), 1038–​1042. PMID: 9620047. Moore SC, Gierach GL, Schatzkin A, and Matthews CE. 2010. Physical activity, sedentary behaviours, and the prevention of endometrial cancer. Br J Cancer, 103(7), 933–​938. PMCID: PMC2965881. Morch LS, Kjaer SK, Keiding N, Lokkegaard E, and Lidegaard O. 2016. The influence of hormone therapies on type I  and II endometrial cancer:  a nationwide cohort study. Int J Cancer, 138(6), 1506–​ 1515. PMID: 26421912. Moyer DL, and Mishell DR, Jr. 1971. Reactions of human endometrium to the intrauterine foreign body. II. Long term effects on the endometrial histology and cytology. Am J Obstet Gynecol, 111(1), 66–​80. PMID: 5096359. Mueck AO, Seeger H, and Rabe T. 2010. Hormonal contraception and risk of endometrial cancer:  a systematic review. Endocr Relat Cancer, 17(4), R263–​R271. PMID: 20870686.

921

Nagle CM, Olsen CM, Ibiebele TI, et  al. 2013. Glycemic index, glycemic load and endometrial cancer risk:  results from the Australian National Endometrial Cancer study and an updated systematic review and meta-​ analysis. Eur J Nutr, 52(2), 705–​715. PMID: 22648201. Newcomb PA, and Trentham-​Dietz A. 2000. Breast feeding practices in relation to endometrial cancer risk, USA. Cancer Causes Control, 11(7), 663–​ 667. PMID: 10977111. Newcomer LM, Newcomb PA, Trentham-​ Dietz A, and Storer BE. 2001. Hormonal risk factors for endometrial cancer:  modification by cigarette smoking (United States). Cancer Causes Control, 12(9), 829–​835. PMID: 11714111. Ni J, Zhu T, Zhao L, et al. 2015. Metabolic syndrome is an independent prognostic factor for endometrial adenocarcinoma. Clin Transl Oncol, 17(10), 835–​839. PMID: 26260911. Nyholm HC, Nielsen AL, Lyndrup J, et al. 1993. Plasma oestrogens in postmenopausal women with endometrial cancer. Br J Obstet Gynaecol, 100(12), 1115–​1119. PMID: 8297845. Okamura C, Tsubono Y, Ito K et al. 2006. Lactation and risk of endometrial cancer in Japan: a case-​control study. Tohoku J Exp Med, 208(2), 109–​ 115. PMID: 16434833. Orgeas CC, Hall P, Wedren S, Dickman PW, and Czene K. 2009. The influence of menopausal hormone therapy on tumour characteristics and survival in endometrial cancer patients. Eur J Cancer, 45(17), 3064–​3073. PMCID: PMC2783257. Ottolina J, Ferrandina G, Gadducci A, et al. 2015. Is the endometrial evaluation routinely required in patients with adult granulosa cell tumors of the ovary? Gynecol Oncol, 136(2), 230–​234. PMID: 25527364. Paganini-​Hill A, Ross RK, and Henderson BE. 1989. Endometrial cancer and patterns of use of oestrogen replacement therapy:  a cohort study. Br J Cancer, 59(3), 445–​447. Parazzini F, Negri E, La Vecchia C, et al. 1998. Role of reproductive factors on the risk of endometrial cancer. Int J Cancer, 76(6), 784–​786. PMID: 9626340. Parslov M, Lidegaard O, Klintorp S, et  al. 2000. Risk factors among young women with endometrial cancer:  a Danish case-​ control study. Am J Obstet Gynecol, 182(1), 23–​29. PMID: WOS:000084987100005. Persson I, Adami HO, Bergkvist L, et al. 1989. Risk of endometrial cancer after treatment with oestrogens alone or in conjunction with progestogens: results of a prospective study. BMJ, 298(6667), 147–​151. PMCID: PMC1835464. Petitti DB, Perlman JA, and Sidney S. 1987. Noncontraceptive estrogens and mortality:  long-​term follow-​up of women in the Walnut Creek Study. Obstet Gynecol, 70(3 Pt 1), 289–​293. PMID: 3627576. Petridou E, Koukoulomatis P, Dessypris N, et  al. 2002. Why is endometrial cancer less common in Greece than in other European Union countries? Eur J Cancer Prev, 11(5), 427–​432. Pettersson B, Adami HO, Lindgren A, and Hesselius I. 1985. Endometrial polyps and hyperplasia as risk-​factors for endometrial carcinoma:  a case-​ control study of curettage specimens. Acta Obstet Gynecol Scan, 64(8), 653–​659. PMID: WOS:A1985A352900007. Pfeiffer RM, Mitani A, Landgren O, et al. 2009. Timing of births and endometrial cancer risk in Swedish women. Cancer Causes Control, 20(8), 1441–​1449. PMCID: PMC2843814. Phipps AI, Doherty JA, Voigt LF et  al. 2011. Long-​term use of continuous-​ combined estrogen-​progestin hormone therapy and risk of endometrial cancer. Cancer Causes Control, 22(12), 1639–​1646. PMCID: PMC3206981. Pike MC, Peters RK, Cozen W, et  al. 1997. Estrogen-​progestin replacement therapy and endometrial cancer. J Natl Cancer Inst, 89(15), 1110–​1116. PMID: 9262248. Pitot HC. 1986. Fundamentals of Oncology. New York: Marcel Dekker. Pocobelli G, Doherty JA, Voigt LF, et al. 2011. Pregnancy history and risk of endometrial cancer. Epidemiology, 22(5), 638–​645. PMCID: PMC3152311. Punnonen R, Pettersson K, and Vanharanta R. 1984. Androgen, estrogen and progestin cytosol receptor concentrations in the normal human endometrium. Effects of intrauterine device. Gynecol Obstet Invest, 17(2), 73–​77. PMID: 6323283. Razavi P, Pike MC, Horn-​Ross PL, et  al. 2010. Long-​term postmenopausal hormone therapy and endometrial cancer. Cancer Epidemiol Biomarkers Prev, 19(2), 475–​483. PMCID: PMC2821215. Reed SD, Voigt LF, Newton KM, et  al. 2009. Progestin therapy of complex endometrial hyperplasia with and without atypia. Obstet Gynecol, 113(3), 655–​662. PMCID: PMC2680496. Ribeiro G, and Swindell R. 1992. The Christie Hospital adjuvant tamoxifen trial. J Natl Cancer Inst Monogr, 11(11), 121–​125. PMID: 1627418. Robboy SJ, and Bradley R. 1979. Changing trends and prognostic features in endometrial cancer associated with exogenous estrogen therapy. Obstet Gynecol, 54(3), 269–​277. PMID: 471366.

92

922

PART IV:  Cancers by Tissue of Origin

Roberts H, Hickey M, and Lethaby A. 2014. Hormone therapy in postmenopausal women and risk of endometrial hyperplasia:  a Cochrane review summary. Maturitas, 77(1), 4–​6. PMID: 24182544. Rosenblatt KA, and Thomas DB. 1995. Prolonged lactation and endometrial cancer. WHO Collaborative Study of Neoplasia and Steroid Contraceptives. Int J Epidemiol, 24(3), 499–​503. PMID: 7672888. Rosenblatt KA, Thomas DB, and The WHO Collaborative Study of Neoplasia and Steroid Contraceptives. 1991. Hormonal content of combined oral contraceptives in relation to the reduced risk of endometrial cancer. Int J Cancer, 49, 870–​874. Ross RK, Paganini-​Hill A, Roy S, Chao A, and Henderson BE. 1988. Past and present preferred prescribing practices of hormone replacement therapy among Los Angeles gynecologists:  possible implications for public health. Am J Public Health, 78(5), 516–​519. PMCID: PMC1349329. Rossouw JE, Anderson GL, Prentice RL, et  al. 2002. Risks and benefits of estrogen plus progestin in healthy postmenopausal women:  principal results From the Women’s Health Initiative randomized controlled trial. JAMA, 288(3), 321–​333. PMID: 12117397. Rubin GL, Peterson HB, Lee NC, et al. 1990. Estrogen replacement therapy and the risk of endometrial cancer: remaining controversies. Am J Obstet Gynecol, 162(1), 148–​154. PMID: 2301483. Salazar-​Martinez E, Lazcano-​Ponce EC, Gonzalez Lira-​Lira G, et  al. 1999. Reproductive factors of ovarian and endometrial cancer risk in a high fertility population in Mexico. Cancer Res, 59(15), 3658–​3662. PMID: 10446978. Salvesen HB, Akslen LA, Albrektsen G, and Iversen OE. 1998. Poorer survival of nulliparous women with endometrial carcinoma. Cancer, 82(7), 1328–​ 1333. PMID: 9529025. Sasco AJ, Chaplain G, Amoros E, and Saez S. 1996. Endometrial cancer following breast cancer: Effect of tamoxifen and castration by radiotherapy. Epidemiology, 7(1), 9–​13. PMID: 8664408. Saso S, Louis LS, Doctor F, et al. 2015. Does fertility treatment increase the risk of uterine cancer? A  meta-​analysis. Eur J Obstet Gynecol Reprod Biol, 195, 52–​60. PMID: 26476799. Schairer C, Adami HO, Hoover R, and Persson I. 1997. Cause-​specific mortality in women receiving hormone replacement therapy. Epidemiology, 8(1), 59–​65. PMID: 9116097. Schmid D, Behrens G, Keimling M, et al. 2015. A systematic review and meta-​ analysis of physical activity and endometrial cancer risk. Eur J Epidemiol, 30(5), 397–​412. PMID: 25800123. Schwartzbaum JA, Hulka BS, Fowler WC, Jr., Kaufman DG, and Hoberman D. 1987. The influence of exogenous estrogen use on survival after diagnosis of endometrial cancer. Am J Epidemiol, 126(5), 851–​ 860. PMID: 3661533. Schwingshackl L, and Hoffmann G. 2015. Adherence to Mediterranean diet and risk of cancer: an updated systematic review and meta-​analysis of observational studies. Cancer Med, 4(12), 1933–​1947. PMID: 26471010. Secord AA, Hasselblad V, Von Gruenigen VE, et al. 2016. Body mass index and mortality in endometrial cancer: a systematic review and meta-​analysis. Gynecol Oncol, 140(1), 184–​190. PMID: 26524722. SEER. 2016. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) Research Data (1973–​ 2013), National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released April 2016, based on the November 2015 submission. Setiawan VW, Schumacher F, Prescott J, et  al. 2014. Cross-​ cancer pleiotropic analysis of endometrial cancer:  PAGE and E2C2 consortia. Carcinogenesis, 35(9), 2068–​2073. PMCID: PMC4146418. Setiawan VW, Yang HP, Pike MC, et al. 2013. Type I and II endometrial cancers: have they different risk factors? J Clin Oncol, 31(20), 2607–​2618. PMCID: PMC3699726. Shapiro JA, Weiss NS, Beresford SAA, and Voigt LF. 1998. Menopausal hormone use and endometrial cancer, by tumor grade and invasion. Epidemiology, 9(1), 99–​101. PMID: WOS:000071089500021. Siiteri PK, and MacDonald PC. 1973. The role of extra-​ glandular estrogen in human endocrinology. Handbook of Physiology (Section 7, Endocrinology), Volume II, Female Reproductive System, Part I, page 615. Washington, DC: American Physiology Society. Simon J, Nachtigall L, Ulrich LG, Eugster-​Hausmann M, and Gut R. 2010. Endometrial safety of ultra-​low-​dose estradiol vaginal tablets. Obstet Gynecol, 116(4), 876–​883. PMID: 20859151. Smith DC, Prentice RL, and Bauermeister DE. 1981. Endometrial carcinoma: histopathology, survival, and exogenous estrogens. Gynecol Obstet Invest, 12(4), 169–​179. PMID: 7250780. Spurdle AB, Thompson DJ, Ahmed S, et al. 2011. Genome-​wide association study identifies a common variant associated with risk of endometrial cancer. Nat Genet, 43(5), 451–​454. PMCID: PMC3770523.

Standeven M, Criqui MH, Klauber MR, Gabriel S, and Barrett-​Connor E. 1986. Correlates of change in postmenopausal estrogen use in a population-​based study. Am J Epidemiol, 124(2), 268–​274. PMID: 3728443. Stefanick ML. 2005. Estrogens and progestins:  background and history, trends in use, and guidelines and regimens approved by the US Food and Drug Administration. Am J Med, 118(Suppl 12B), 64–​73. PMID: 16414329. Stocks T, Bjorge T, Ulmer H, et  al. 2015. Metabolic risk score and cancer risk: pooled analysis of seven cohorts. Int J Epidemiol, 44(4), 1353–​1363. PMCID: PMC4588859. Sturgeon SR, Brock JW, Potischman N, et al. 1998. Serum concentrations of organochlorine compounds and endometrial cancer risk (United States). Cancer Causes Control, 9(4), 417–​424. PMID: 9794174. Sun Q, Xu L, Zhou B, et al. 2011. Alcohol consumption and the risk of endometrial cancer:  a meta-​analysis. Asia Pac J Clin Nutr, 20(1), 125–​133. PMID: 21393120. Szekely DR, Weiss NS, and Schweid AI. 1978. Incidence of endometrial carcinoma in King County, Washington: a standardized histologic review. J Natl Cancer Inst, 60(5), 985–​989. PMID: 642039. Thom MH, White PJ, Williams RM, et al. 1979. Prevention and treatment of endometrial disease in climacteric women receiving oestrogen therapy. Lancet, 2(8140), 455–​457. PMID: 89511. Trabert B, Wentzensen N, Yang HP, et  al. 2013. Is estrogen plus progestin menopausal hormone therapy safe with respect to endometrial cancer risk? Int J Cancer, 132(2), 417–​426. PMCID: PMC3427719. Vogel VG, Costantino JP, Wickerham DL, et al. 2006. Effects of tamoxifen vs raloxifene on the risk of developing invasive breast cancer and other disease outcomes: the NSABP Study of Tamoxifen and Raloxifene (STAR) P-​2 trial. JAMA, 295(23), 2727–​2741. PMID: 16754727. Voskuil DW, Monninkhof EM, Elias SG, et  al. 2007. Physical activity and endometrial cancer risk:  a systematic review of current evidence. Cancer Epidemiol Biomarkers Prev, 16(4), 639–​648. PMID: 17416752. Wan YL, and Holland C. 2011. The efficacy of levonorgestrel intrauterine systems for endometrial protection: a systematic review. Climacteric, 14(6), 622–​632. PMID: 22017273. Wang Z, Risch H, Lu L, et al. 2015. Joint effect of genotypic and phenotypic features of reproductive factors on endometrial cancer risk. Sci Rep, 5, 15582. PMCID: PMC4620445. WCRF/​ AICR. 2013. Continuous Update Project Report. Food, Nutrition, Physical Activity, and the Prevention of Endometrial Cancer. Available from http://​www.dietandcancerreport.org. Accessed July 1, 2016. Weiderpass E, Adami HO, Baron JA, et al. 1999a. Risk of endometrial cancer following estrogen replacement with and without progestins. J Natl Cancer Inst, 91(13), 1131–​1137. PMID: WOS:000081416500012. Weiderpass E, Adami HO, Baron JA, et al. 1999b. Use of oral contraceptives and endometrial cancer risk (Sweden). Cancer Causes Control, 10(4), 277–​284. PMID: WOS:000081282400005. Weiderpass E, Adami HO, Baron JA, et al. 2000. Organochlorines and endometrial cancer risk. Cancer Epidemiol Biomarkers Prev, 9(5), 487–​493. PMID: 10815693. Weiderpass E, Antoine J, Bray FI, Oh JK, and Arbyn M. 2014. Trends in corpus uteri cancer mortality in member states of the European Union. Eur J Cancer, 50(9), 1675–​1684. PMID: 24656568. Weiderpass E, Baron JA, Adami HO, et  al. 1999c. Low-​potency oestrogen and risk of endometrial cancer: a case-​control study. Lancet, 353(9167), 1824–​1828. PMID: 10359406. Weiss NS, and Sayvetz TA. 1980. Incidence of Endometrial Cancer in Relation to the Use of Oral-​Contraceptives. N Engl J Med, 302(10), 551–​554. PMID: WOS:A1980JG43900004. Weiss NS, Szekely DR, and Austin DF. 1976. Increasing incidence of endometrial cancer in the United States. N Engl J Med, 294(23), 1259–​1262. PMID: 1264151. Westhoff C, Gentile G, Lee J, Zacur H, and Helbig D. 1996. Predictors of ovarian steroid secretion in reproductive-​age women. Am J Epidemiol, 144(4), 381–​388. PMID: 8712195. WHO. 2000. International Classification of Diseases for Oncology, 3rd ed. Fritz APC, Jack A, Shanmugaratnam K, Sobin L, Parkin DM, Whelan S (Eds.). Geneva: World Health Organization. WHO Collaborative Study of Neoplasia and Steroid Contraceptives. 1991. Depot medroxyprogesterone acetate (DMPA) and risk of endometrial cancer. Int J Cancer, 49, 186–​190. Win AK, Reece JC, and Ryan S. 2015. Family history and risk of endometrial cancer: a systematic review and meta-​analysis. Obstet Gynecol, 125(1), 89–​98. PMID: 25560109.

 923

Endometrial Cancer Wright JD, Herzog TJ, Tsui J, et  al. 2013. Nationwide trends in the performance of inpatient hysterectomy in the United States. Obstet Gynecol, 122(2 Pt 1), 233–​241. PMCID: PMC3913114. Wright KP, and Johnson JV. 2008. Evaluation of extended and continuous use oral contraceptives. Ther Clin Risk Manag, 4(5), 905–​911. PMCID: PMC2621397. Wu QJ, Li YY, Tu C, et  al. 2015. Parity and endometrial cancer risk:  a meta-​analysis of epidemiological studies. Sci Rep, 5, 14243. PMCID: PMC4642705. Wysowski DK, Golden L, and Burke L. 1995. Use of menopausal estrogens and medroxyprogesterone in the United-​States, 1982–​1992. Obstet Gynecol, 85(1), 6–​10. PMID: WOS:A1995PX70600002. Yang HP, Cook LS, Weiderpass E, et  al. 2015. Infertility and incident endometrial cancer risk:  a pooled analysis from the epidemiology of endometrial cancer consortium (E2C2). Br J Cancer, 112(5), 925–​ 933. PMCID: PMC4453954. Yu X, Bao Z, Zou J, and Dong J. 2011. Coffee consumption and risk of cancers: a meta-​analysis of cohort studies. BMC Cancer, 11, 96. PMCID: PMC3066123.

923

Zadnik V, and Krajc M. 2016. Epidemiological trends of hormone-​related cancers in Slovenia. Arh Hig Rada Toksikol, 67(2), 83–​ 92. PMID: 27331295. Zeleniuch-​Jacquotte A, Akhmedkhanov A, Kato I, et al. 2001. Postmenopausal endogenous oestrogens and risk of endometrial cancer: results of a prospective study. Br J Cancer, 84(7), 975–​981. PMCID: PMC2363831. Zhang Y, Liu H, Yang S, et al. 2014. Overweight, obesity and endometrial cancer risk:  results from a systematic review and meta-​analysis. Int J Biol Markers, 29(1), e21–​29. PMID: 24170556. Zhang ZH, Su PY, Hao JH, and Sun YH. 2013. The role of preexisting diabetes mellitus on incidence and mortality of endometrial cancer:  a meta-​ analysis of prospective cohort studies. Int J Gynecol Cancer, 23(2), 294–​303. PMID: 23287960. Zhou B, Yang L, Sun Q, et al. 2008. Cigarette smoking and the risk of endometrial cancer:  a meta-​ analysis. Am J Med, 121(6), 501–​ 508 e503. PMID: 18501231. Zucchetto A, Serraino D, Polesel J, et al. 2009. Hormone-​related factors and gynecological conditions in relation to endometrial cancer risk. Eur J Cancer Prev, 18(4), 316–​321. PMID: 19554665.

924

 925

48 Cervical Cancer ROLANDO HERRERO AND RAUL MURILLO

OVERVIEW Cervical cancer is the fourth most common cancer in women worldwide, with more than 500,000 cases and 250,000 deaths per year. The disease is characterized by marked regional differences, with more than 80% of the cases and deaths occurring in developing countries. The etiology and natural history of the disease are very well studied, with persistent infection with one of 13 human papillomavirus (HPV) types now considered as a necessary cause. The molecular mechanisms have also been elucidated and are mediated mainly by the expression of viral oncogenes that interfere with cellular pathways. The two most common HPV types, namely HPV 16 and HPV 18, are associated with about 70% of all cases around the world. Immune (e.g., HIV infection), hormonal (e.g., high parity), environmental (e.g., smoking), and genetic (e.g., HLA types) cofactors determine the risk of persistence and cancer among women harboring HPV infection. Recent developments for control of this disease include highly effective and safe vaccines against the main causal HPV types, which since 2006 have been introduced in national vaccination programs around the world. In addition, highly sensitive HPV DNA tests are now available, and it has been demonstrated that they can accurately identify women at highest risk who require additional evaluations or treatment. Intensive efforts to improve access of underserved populations to these new interventions are underway, and research continues to improve the vaccines, the testing methods, and the effectiveness of the interventions. There is hope that the massive implementation of these two powerful public health tools will significantly reduce the burden of disease and reduce associated inequalities in the near future.

mortality, but it is generally accepted that screening programs have played an important role in controlling the burden of the disease, although other factors have also probably contributed (e.g., reductions in parity and smoking). For example, in the United States, cervical cancer incidence declined since 1978 at an annual percent change of  –​2.2% (Kurdgelashvili et al., 2013). Given their complexity, high-​ quality cytology-​ based screening programs have been very difficult or impossible to implement and maintain in developing countries, and cervical cancer rates have not declined to the same extent in such countries. Failure to achieve significant coverage, poor quality control of the laboratories, and lack of follow-​up of detected abnormalities are among the main reasons for the failure of cytology-​based screening programs. In many countries, however, particularly in Africa, there are no established programs for cervical cancer control, and the disease generally follows its natural course, presenting as advanced symptomatic cancer, leading to painful and often protracted terminal disease. The recently accumulated knowledge about the role of HPV infection as the cause of cervical cancer and the better understanding of the natural history and the role of HPV cofactors has led to the development, evaluation, and implementation of new methods for primary and secondary prevention. In the last 10 years there have been major changes in the technology available for controlling this disease, and opportunities for implementing control programs have arisen. The development of the highly effective and safe vaccines against HPV is a major recent achievement, and screening based on HPV testing is revolutionizing public health programs around the world.

CLASSIFICATION INTRODUCTION Over the next 10 years, cervical cancer will kill approximately 2.5 million relatively young women of low socioeconomic status around the world (Ferlay et al., 2013), with direct consequences for their families and society at large. The social problem is magnified if we consider also that multiple births are one of the risk factors for the disease and that women in low socioeconomic strata are very often the sole support of their families. It has been estimated that almost 90% of cervical cancer deaths occur in developing countries, where there is no access to screening programs and where the cultural characteristics of the population predispose women to the disease. In developed countries where cervical cancer rates are low, the disease also affects disadvantaged families (Singh et  al., 2011). Notably, according to GLOBOCAN, there are still more than 6000 cervical cancer deaths in the United States and 13,000 in the European Union every year (Ferlay et al., 2013). Great efforts have been made over the years in developed countries to control this disease with the establishment of more or less organized screening programs, generally based on cytologic screening and referral of women with positive results to colposcopy and biopsy, followed by treatment of cervical cancer precursors. This process requires very organized and quality-​controlled laboratories and coordination between multiple components within the system. Most important, it requires multiple visits by the women to different types of medical personnel. There was never a randomized clinical trial to demonstrate the efficacy of cytology screening for the reduction of cervical cancer

Anatomic Distribution The cervix is the lower fibromuscular part of the uterus, divided from the muscular corpus by the internal os, which forms a sphincter that can be damaged and can affect reproductive function. It has a supravaginal and a vaginal portion, and the vaginal mucosa is reflected around it, forming the vaginal fornices (Figure 48–1). The cervix is cylindrical and measures approximately 3 cm long and 2 cm in diameter, but is modified by hormonal and reproductive events. It communicates with the vagina via the external os; the portion between the internal and external os is the horizontally flattened endocervical canal. The underlying stroma is composed of elastic tissue and a small amount of smooth muscle. The endocervix is covered by cylindric glandular epithelium and the ectocervix is covered by non-​keratinized stratified squamous epithelium similar to the vaginal epithelium.

The Cervical Transformation Zone The endometrial cavity and the endocervical canal are covered by tall, mucus-​secreting columnar epithelium derived from the Müllerian duct. This endocervical epithelium comes into contact distally with the vaginal epithelium derived from the urogenital sinus. During fetal life, the columnar epithelium extends into the ectocervix, constituting the congenital ectropium. During late fetal life and adolescence, under the influence of hormonally and bacterially mediated acidification, the columnar epithelium in contact with the vaginal milieu

925

926

926

PART IV:  Cancers by Tissue of Origin with HPV, with a stronger relative detection of HPV 18, which can be found in approximately 30% of the cases (Castellsague et al., 2006). Cervical sarcomas are very rare, accounting for less than 1% of all cancer cases. This histological subtype represents at least six different clinical entities including leiomyosarcomas and liposarcomas, among others.

Falloplan tube

Uterine cavity

Precursor Lesions Endocervical canal

Ectocervix (Portio vaginalis) Vagina

Squamocolumnar juction (Transformation zone)

Figure 48–1.  Aspects of cervical anatomy that are essential to understanding cervical carcinogenesis.

Cervical cancer progresses through a series of epithelial changes mediated by HPV that can be assessed molecularly, microscopically, and macroscopically. The initial step in the carcinogenic process is persistent HPV infection that sometimes manifests itself as a low-​grade intraepithelial lesion. These lesions are considered a manifestation of productive HPV infection and are not necessarily cancer precursors; they usually regress. In a small fraction of women, persistent infection may progress to a high-​grade lesion, which constitutes an abortive infection where the late events in the HPV life cycle are not supported. High-​grade lesions often can progress to cancer and are therefore considered true cancer precursors. However, it is impossible with current technology to accurately distinguish lesions that will regress from those that will progress, and active research is ongoing to investigate potential biomarkers for this purpose.

HUMAN PAPILLOMAVIRUS INFECTION is replaced with squamous epithelium in a process called squamous metaplasia. The boundary between the original columnar and squamous epithelium constitutes the original transformation zone, and the metaplastic area proximal to this line is the transformation zone, where most cervical cancers occur. The elements of instability of this junctional interface have long been considered responsible for the increased risk of cancer at this anatomic site compared to other similar epithelia. Women with an increased area of squamous metaplasia, as described after exposure to diethylstilbestrol in utero, were at increased risk of cervical neoplasia (Robboy et al., 1984). Recently, a population of squamo-​ columnar junction residual embryonic cells with a unique morphology and gene-​expression profile has been described (Herfs et al., 2012). These cells are susceptible to HPV infection (Mirkovic et al., 2015), are able to differentiate, and are vulnerable to neoplastic transformation. Biomarkers associated with them are expressed in cervical lesions but not in normal epithelium or in vulvar, vaginal, or penile lesions, likely explaining the large differences in incidence between cervical cancer and HPV-​related tumors in other anatomic localizations. Furthermore, there is some evidence that these embryonic cells do not regenerate after excision of the transformation zone.

Histopathology The two major histologic types of cervical cancer are squamous cell carcinoma and adenocarcinoma, with the former accounting for approximately 85% of the cancers of known histologic type worldwide (Li et  al., 2011; Parkin et  al., 2006). In recent decades, the relative proportion of these two histologic types has changed, particularly in developed countries with advanced screening programs where a significant reduction of squamous cell carcinoma and a slight increase of adenocarcinoma has been observed (Kurdgelashvili et al., 2013). Squamous cell carcinomas are further subdivided according to their degree of differentiation and morphologic characteristics into keratinizing, non-​keratinizing, papillary, basaloid, warty, verrucous, squamo-​ transitional, and lymphoepithelioma-​like, but few of these variations impact therapy or prognosis beyond stage and grade. Among the adenocarcinomas, the classification includes adenocarcinoma in situ, endocervical adenocarcinoma, mucinous, mucinous gastric type, mucinous intestinal type, mucinous signet-​ring cell type, endometrioid carcinoma, clear cell carcinoma, serous carcinoma, mesonephric carcinoma, and adenocarcinoma admixed with neuroendocrine carcinoma. Adenocarcinoma of the cervix is also associated

More than 200 HPV types have been described that infect the skin and mucosa, and many of them are associated with benign lesions of the skin and genitals, but in this book we will concentrate on mucosal oncogenic HPV types. Persistent infection with approximately 12 oncogenic types of HPV (HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, and 59)  is an established cause of cervical cancer (International Agency for Research on Cancer [IARC], 2007). The epidemiology of infection is described in more detail in Chapter 24 of this volume. Detection of HPV DNA in cervical cells or tissue is the earliest expression of exposure to the virus and is a highly sensitive marker for the presence of cervical cancer precursors. However, cervical HPV infection is usually transient and therefore viral detection lacks specificity for lesion detection. The method for detection of current HPV infection is HPV DNA or RNA detection in cervical exfoliated cells or biopsy. HPV serology, based on detection of antibodies against L1 virus-​like particles, is very useful for monitoring of vaccination, and high serologic titers have been shown to predict natural acquired immunity against specific HPV types (Safaeian et  al., 2010a) (Castellsague PMID 24610876). However, HPV serology is not a sensitive marker of current or past infection and has limited utility to assess the presence of infection or cervical disease, because only about 50%–​70% of exposed women seroconvert (Stanley et al., 2012).

Microscopic Diagnosis of Early HPV Infection As described in the previous edition of this book (Schiffman and Hildesheim, 2006), Papanicolaou proposed a cytology classification to predict risk of invasive cancer, extended later to detect even earlier cancer precursors. In the mid-​1950s, Koss and Durfee described a morphologic abnormality of squamous cells, which they termed “koilocytotic atypia” (Koss, 1989). (Koilos is a Greek word meaning “hollow.”) Synonyms for koilocytotic atypia include “condylomatous atypia” and “warty atypia.” Koilocytes have hyper-​chromatic, enlarged, wrinkled, and sometimes multiple nuclei, surrounded by perinuclear clear zones (halos). Twenty years later, Meisels and Fortin, and Purola and Savia, were the first groups to propose formally that flat cervical lesions demonstrating koilocytotic atypia (often associated with acanthosis, hyperkeratosis, and parakeratosis) were cervical equivalents of exophytic condyloma acuminatum (genital warts), already known by then to be caused by HPV. Electronic microscopic

 927

Cervical Cancer studies showed that koilocytotic cells contained crystalline arrays of assembled HPV virions in their abnormal nuclei. Richart and colleagues (1969) noted the continuity of cervical cancer precursor lesions and introduced the concept of “cervical intraepithelial neoplasia (CIN),” defined as a continuum (CIN 1, CIN 2, and CIN 3)  of increasingly severe pre-​invasive changes. Throughout the 1970s and 1980s, pathologists learned that the cytopathologic effects of HPV classified as koilocytotic atypia were impossible to distinguish from “precancerous” changes called CIN 1. In fact, CIN 1 is now considered a histologic sign of HPV infection, usually transient and linked more closely to viral exposure than cancer risk. This recognition led to the combination of HPV effects and CIN 1 for the purposes of cancer screening in the Bethesda System of cervical cytology (Kurman et al., 1991). In the Bethesda System, which has been updated twice (Solomon et  al., 2002), the category “low-​grade squamous intraepithelial lesion (LSIL)” subsumes older cytologic terms such as CIN 1, mild dysplasia, mild dyskaryosis (British), koilocytotic atypia, condylomatous atypia, warty atypia, and HPV effect. The Bethesda System is used in the United States and many other countries for cytology screening, The CIN scale remains more accepted worldwide when discussing histopathology (i.e., biopsies). Although characteristically flat and nearly invisible without staining, early lesions due to oncogenic HPV infections may often be diagnosed by application of 5% acetic acid (vinegar) or Lugol’s iodine solution (Ferris et  al., 1994). Lugol’s solution stains glycogen, which is often lacking in HPV-​associated lesions. HPV-​induced lesions are usually “acetowhite,” meaning they temporarily turn white after vinegar is applied. Visual inspection with acetic acid (VIA) followed by immediate treatment of visible abnormalities has been recommended in screening for HPV-​induced cervical lesions in low-​resource settings (Blumenthal et al., 2001; Denny et al., 2000; Sankaranarayanan et al., 2003).

Cervical Precancer Introduction

Oncogenic HPV infection, with or without macroscopic/​microscopic signs, is extremely common and usually benign, particularly in young women. On the other hand, cervical precancer is relatively rare, and represents a truly premalignant lesion in the most severe cases (CIN 3 or carcinoma in situ). The clear definition of the presence of precancer is difficult due to heterogeneity of the histologic interpretation. Some lesions classified as precancer certainly represent acute HPV infections of particularly bad microscopic appearance that are destined to regress. Others are true precancer destined to persist with high risk of invasion. Non-​ oncogenic HPV infections are capable of producing lesions diagnosed as precancer, showing that this level of abnormality is not a perfect surrogate for cancer risk (Clifford et al., 2003). For safety and concern over loss to follow-​up, treating precancer is a valid clinical strategy to provide a margin of safety, given that it is not yet possible to know which lesions pose a threat. Better accuracy based on molecular profiling is the goal. As described in the following, a recent proposed classification incorporates p16 immunostaining to define the presence or absence of precancer (Darragh et al., 2012).

Microscopic Diagnosis of Precancer

A morphologic continuum of changes exists at the microscopic level leading from HPV infection to precancer, without a current clear cut-​ point. The gradient from infection to precancer is characterized by increasing nuclear atypia and failure of cellular differentiation in progressively more superficial levels of epithelium. When non-​differentiation extends beyond the basal third of the epithelium to the middle third of the epithelium, the diagnosis changes from CIN 1 to CIN 2 (“moderate dysplasia”). In turn, non-​differentiation reaching the upper third is called CIN 3, which includes carcinoma in situ representing full thickness replacement with undifferentiated, immortalized, atypical cells. CIN 3 is the best standard of precancer for epidemiologic studies (ASCUS-LSIL Triage Study [ALTS] Group, 2003). CIN 2, in contrast, encompasses much of the heterogeneity in the precancer

927

stage, combining the worst appearing HPV infections still destined to regress with incipient precancer in a mixture that is not yet separable. Nonetheless, as a clinical endpoint of importance, US (but not all European) clinicians treat CIN 2 lesions, with rare exceptions in young women among whom fertility is a major concern (Nobbenhuis et al., 1999; Wright, Jr., et al., 2003). The new proposed histologic classification (Darragh et  al., 2012)  includes immunostatining with p16 of CIN2 lesions and other abnormalities and considers those that are p16 positive as true precancer.

Macroscopic Diagnosis of Precancer

When HPV infections progress to precancer, the clonal expansion usually occurs at the proximal (endocervical) edge of the acetowhite HPV lesion. Precancer has a thicker, grayer appearance than early HPV infection (Ferris et al., 1994). As a function of neoplastic angiogenesis, abnormally large and unusual blood vessels supplying the precancer rise to the surface to create coarse mosaic patterns or even evident dead-​end vessels. These changes, though highly predictive when they all occur, are variably present. Incipient precancerous lesions are usually very small and can easily be missed by the colposcopist on examination, especially during biopsy placement. Colposcopically directed biopsies have been the clinical reference standard for defining the severity of anogenital HPV-​related disease, specifically, to distinguish between early infection and precancer requiring treatment. However, the choice of biopsy site and the histopathologic diagnosis of resultant biopsies tend to be variable and subjective (Pretorius et al., 2011; Stoler et al., 2011). Recent studies have demonstrated that systematic collection of more than one biopsy improves the diagnostic performance of colposcopy (Wentzensen et al., 2015). Novel, computer-​assisted diagnosis in colposcopy is attempting to exploit the same and additional tissue characteristics to permit diagnosis, perhaps without an expert clinician (Mehlhorn et al., 2012), but the results of this and other novel approaches have not demonstrated clinical utility, in particular for differentiation of precancer from atypical metaplasia. High-​resolution microendoscopy evaluates lesion severity by real-​time analysis and nuclear features using a probe and is currently under evaluation (Pierce et al., 2012). Thus, the colposcopically guided biopsy tends to both under-​diagnosis of precancer through missed lesions and over-​diagnosis because histologic diagnoses of CIN2–​CIN3 include lesions that would not invade if untreated. Both colposcopic and histologic interpretations have limited accuracy and reproducibility (Pretorius et  al., 2011; Stoler et  al., 2011), and the number of biopsies collected influences the accuracy of the results. CIN2 remains a borderline category that includes high and low grade lesions.

Molecular Characteristics of Cervical Cancer Oncogenic HPV DNA.  The human papillomaviruses exhibit

a circular double-​stranded DNA of about 8 kb and belong to the papillomaviridae family. Within the family, genera (designated with Greek letters) are defined on the basis of quantitative thresholds in nucleotide sequence comparisons and biologically distinguishing features like host species, target tissues, pathogenicity, and genome organization (Chen et  al., 1999). There are about 25 PV genera, and HPVs are part of five (Alpha, Beta, Gamma, Mu, and Nu-​PV). Within the genera there are species (designated with the Greek letter of the genus and a number) and types, designated with numbers. For example, HPV type 16 (together with HPV 31, 33, 35, 52, 58, and 67) belong to the genus Alpha, species Alpha 9. Similarly HPV 18, 39, 45, 59, 68, 70, 85, and 97 belong to species Alpha 7. More than 200 human papillomaviruses have been described to date, and 12 have been classified by IARC as carcinogenic (see further discussion later in this chapter). The species that have been linked to anogenital and oropharyngeal cancer are part of the genus Alpha, while the beta and gamma types are more common in skin and mucosal epithelium. The HPV genome encodes approximately eight genes. According to protein expression during the viral cycle, two functional genome regions have been identified: (1) a coding region containing the early genes (referring to the phase of their expression in

928

928

PART IV:  Cancers by Tissue of Origin

the life cycle), E1, E2, E4, E5, E6, and E7; and (2) a region containing two late genes, the major (L1) and minor (L2) capsid proteins. In addition, the HPV genome has a non-​coding region, termed “long control region (LCR),” involved in control of viral DNA replication and transcription. HPV E6 and E7 genes encode the major transforming proteins. HPV types are further subdivided into variant lineages (1%–​10% variation) and sublineages (0.5–​1.0% variation), currently identified by sequence determination of viral fragments. These variants have been associated with different risk of HPV persistence and cervical neoplasia (Burk et al., 2013).

HPV DNA Integration. In the normal viral life cycle, HPV

genomes exist in circular (episomal) state and are present in basal cells of squamous epithelia. A  common event occurring during progression to cervical cancer is the transition from episomal to the integrated form in the host genome. Integration is observed in approximately 90% of invasive carcinomas (Pett et  al., 2007). Integration is a risk factor for deregulation of viral oncogenes E6 and E7 and represents a form of the virus that is resistant to host mechanisms of viral clearance, enabling infected cells to maintain viral oncogene expression and avoid cell death.

DESCRIPTIVE EPIDEMIOLOGY (WORLDWIDE) Cervical cancer is the fourth leading malignancy among women in the world, after breast and colorectal cancer, with an estimated 530,000 new cases and 270,000 deaths in 2012 (Ferlay et al., 2013). Cervical cancer incidence and mortality rates have been declining in most areas around the world in the last 30 years, at a worldwide rate of about 1.6% per year (Forouzanfar et  al., 2011), as a result of increasing access to health services, reduction in risk factors for the disease (e.g., parity), improvements in treatment, and successful cytology-​based (i.e., Pap smear) screening programs. However, cervical cancer remains the leading female cancer in several developing countries in Africa, Asia, and the Americas. Furthermore, some areas report recent increases in rates, including several Eastern European countries (Arbyn et al., 2011). Despite the declining rates, the number of new cases and deaths has increased constantly at a rate of about 0.5% per year because of population aging and increase in life expectancy among women. For example, in Latin America, Parkin et al. (2008) have estimated that the number of new cases would increase by 75% between 2002 and 2025 if incidence rates remained at 2002 levels, due to population growth and aging alone. A striking characteristic of cervical cancer is its geographic variation, with a generally strong inverse correlation between level of development and incidence and mortality. The disease is also strongly influenced by cultural and religious practices that modify sexual behavior and therefore transmission of HPV. The African continent has the highest estimated age-​standardized incidence rates of cervical cancer, with overall rates in Eastern Africa exceeding 40/​100,000, as recorded, for example, in Malawi, Mozambique, Comoros, Zambia, Zimbabwe, Tanzania, and Swaziland (Ferlay et  al., 2013). At the other extreme are the Northern African countries, where sexual contact is limited for cultural and religious reasons, with rates under 3/​100,000 in Iran, Iraq, Saudi Arabia, Syria, Jordan, Egypt, and the State of Palestine. For comparison, in Western countries where successful screening is prevalent, the rates are around 5–​7/​100,000. In most settings with both high and low rates, the highest incidence and mortality rates are found among the poorest and most marginalized women. For example, in the United States, where the average rates are low and cervical cancer has consistently declined over the last decades, strong disparities still exist between black and white women and by socioeconomic status (Singh et  al., 2011). Cervical cancer affects relatively young women, with a median age at death of 54 years, and the burden of disease among women under 40 years is high compared to other cancers, due both to the large numbers of women in those age groups in developing countries and the fact that cervical cancer rates begin to rise at younger ages than other cancers.

A recent analysis ranked cervical cancer highest among cancers according to a disability-​ adjusted life-​ years (DALYs) metric (Soerjomataram et  al., 2012)  because the disease affects relatively young women. In the same study, when DALYs were stratified by the human development index (HDI), a composite of three dimensions of human development (life expectancy, adult literacy, and standard of living), breast cancer DALYs ranged from a high of 566 age-​adjusted DALYs per 100,000 population in populations with very high HDI to 387 in those with low HDI. In contrast, cervical cancer-​related DALYs ranged from 84 per 100,000 women in areas of very high HDI to 595 per 100,000 in areas of low HDI.

ENVIRONMENTAL RISK FACTORS FOR CERVICAL CANCER HPV Infection HPV infection is considered a necessary but not sufficient cause of cervical cancer (Bosch et al., 2002; Walboomers et al., 1999), as it fulfills all causality criteria, including strength and consistency of the association, time sequence, specificity, and coherence with biologic and epidemiologic evidence. Twelve HPV types are classified as Group 1 carcinogens by the IARC Monographs (HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, and 59); HPV 68 is classified as probably carcinogenic (Group  2a); and HPV 26, 53, 66, 67, 70, 73, and 82 are considered possibly carcinogenic (Group  2b) (IARC, 2007). In a recent meta-​ analysis, additional evidence for the carcinogenicity of HPV 68, 26, 66, 67, 73, and 82 was observed, with the suggestion from the authors to upgrade their classification (Arbyn et  al., 2014a). However, these types are likely to contribute only in a very small fraction of cases. HPV is a very common sexually transmitted infection, usually acquired after initiation of sexual activity. Most HPV infections clear spontaneously after a few months, but those that persist, particularly HPV 16 and 18, may progress to cervical cancer precursors, and ultimately to invasive cervical cancer. Oncogenic types of HPV are identified in nearly all cancers of the cervix, and the relative risk of cervical cancer associated with persistent, ongoing infection with high-​risk types of HPV is higher than the risk of lung cancer associated with smoking. Munoz et al. (2003) pooled data from 11 case-​control studies involving almost 2000 cervical cancer cases and a similar number of controls. The pooled odds ratio for cervical cancer associated with detection of any HPV infection was 158.2 (95% CI:  113.4–​220.6). A  study that evaluated HPV infection in 10,575 histologically confirmed cases of invasive cancer from 38 countries using paraffin-​ embedded samples found that 85% of the cases were positive for HPV DNA (de Sanjose et  al., 2010). Ninety-​one percent of HPV-​positive cases harbored HPV 16, 18, 31, 33, 35, 45, 52, and 58. HPV types 16, 18, and 45 were the three most common types in each histologic form of cervical cancer (squamous cell, adenocarcinoma, and adenosquamous carcinoma). There is little geographic variation in the predominant HPV types associated with cervical cancer. Figure 48–2 presents the percent attributable fraction of each HPV type according to geographic region (Guan et al., 2012). HPV 16 is the most common HPV type in cervical cancers in all regions of the world. HPV 18 is the second most common, also in all regions of the world. The third most common type associated with cancer in all areas is HPV 45, except in Europe, where it is HPV 33, and in Eastern Asia, where it is HPV 58. Recently, an analysis of 682 cases of properly characterized cervical adenocarcinomas from five continents (Pirog et  al., 2014), revealed HPV DNA in 72% of cases of classical adenocarcinoma, with much lower frequencies for other, relatively rare types of adenocarcinoma (HPV DNA-​positive in 27% endometrioid, 25% serous, 20% clear cell, 14% NOS, 8% minimal deviation [adenoma malignum]). HPV types 16, 18, and 45 predominated and accounted for 94% of HPV-​ positive tumors. An interesting analysis investigating time trends in the detection of type-​specific HPV infection in invasive cervical cancer was reported in 2014 by Alemany et  al. They studied approximately

 92

929

Cervical Cancer 100% 90%

HPV 58

80%

HPV 52

70%

HPV 45

60%

HPV 35

50% HPV 33

40%

HPV 31

30% 20%

HPV 18

10%

HPV 16

0% Africa

North S/C Oceania Eastern W/C Asia Europe Asia America America

Figure 48–2.  HPV types associated with cervical cancer by region. Compiled from Guan et al., Int J Cancer 2012;131:2349–​2359.

5000 tumor specimens from patients diagnosed between 1940 and 2007 from 11 countries in Central-​South America, Asia, and Europe. HPV detection was performed with SPF-​10/​DEIA/​LIPA25. HPV 16 and 18 were the two most common types, and their relative contribution was stable over the entire period (Figure 48–3). In contrast, the relative contribution of HPV 16 to adenocarcinomas increased over time. Precursors and invasive cervical cancers associated with HPV 16 and possibly HPV 18 tend to occur at younger ages than those associated with other HPV types (Munoz et al., 2003; Porras et al., 2009), probably as a manifestation of the increased carcinogenicity of those types. In a large study in Portland, Oregon, where women were followed for 18 years (Figure 48–4a and 48–4b) (Schiffman et al., 2011), the high risk of developing CIN 3 among HPV-​positive women, particularly for HPV 16, contrasted with the very low risk over the entire period among HPV-​negative women, indicating the high negative predictive value of HPV testing and possibly useful positive predictive value of HPV partial genotyping.

Socioeconomic Determinants Cervical cancer incidence and mortality have consistently shown to be more common in low-​income populations (Cheng et  al., 2011; Parikh et al., 2003), indicating a major role of socioeconomic status in disease development. Limited access to preventive services linked to unfavorable socioeconomic conditions significantly contributes to the progression of precancerous lesions; however, the entire burden of poverty-​linked cervical cancer cannot be explained solely by screening uptake (Khan et al., 2005). In North America and Europe, HPV infection has been reported to be more frequent among low-​income women (Kavanagh et al., 2013; Shi et al., 2014). In general, sexually transmitted diseases, along with other infections, are more frequent in low socioeconomic groups where segregation, sexual networks, and women’s failure to influence better sexual health practices with their partners play a relevant role (Centers for Disease Control and Prevention [CDC], 2013). Accordingly, HPV prevalence is also higher in less developed than in more developed regions (Bruni et al., 2010); in addition, several well-​recognized cervical cancer cofactors, including smoking and parity, are more frequent among low-​income populations; in contrast, hormonal contraception is more prevalent in high-​income groups. A few population-​based studies from developing countries analyzed the association between poverty and cervical cancer. Data from China showed no association of HPV positivity and education level as an

indicator of socioeconomic status (Dai et al., 2006), and a study from Mexico revealed a higher prevalence among lower income levels only for low-​risk HPV; however, in the latter study, stratified data showed a higher overall HPV prevalence between ages 35 and 54 as well as a higher prevalence of multiple infections across all ages for medium-​ and low-​income women (Lazcano-​Ponce et al., 2001).

Demographic Factors Early studies proposed an association between the progression of HPV infection and immune dysfunction and cell senescence, an observation supported by the incremental risk with advancing age (Mandelblatt, 1993; Thulaseedharan et al., 2013). It has been suggested that cervical cancer appears at younger ages than other cancers as it is related with incident HPV infections in young women, but in fact, the relationship between age and cervical cancer is highly influenced by screening practices (Gustafsson et al., 1997). Recent reports show that women with adequate negative screening history have a low risk of cervical cancer after age 65 (Castanon et al., 2014; Sasieni et al., 2009); nevertheless, older women not covered by screening programs reveal high risk in most countries with or without organized screening (Figure 48–5). On the other hand, it has been proposed that the age pattern of cervical cancer incidence in unscreened populations resembles cancer of the breast and other hormone-​dependent epithelia (Plummer et al., 2012). The exact effect of screening in age-​specific incidence rates is difficult to determine due to the lack of controlled studies. However, distinctive patterns can be observed in relation to the differential development of screening (Figure 48–5). Countries with highly organized and long-​standing programs have shifted the peak of invasive cancer to younger ages (around 30) (Figure 48–5d) (Herbert et al., 1996), while trend lines with the highest incidence for women over 65 years old might represent the low impact of screening given a recent introduction or scant implementation of organized programs (Figure 48–5a) (Jin et al., 2013). Yet, some of the latter countries have experienced a reduction in cervical cancer mortality probably due to improvement in socioeconomic determinants and associated cofactors. Some countries with risk peaks in intermediate points between the two described extremes may suggest either partial progress of screening programs (high coverage of opportunistic screening or medium term organized programs) (Figure 48–5c) (Saraiya et  al., 2013), or initially, organized programs suffered difficulties in implementation during the last decades as indicated by an initial reduction in cervical cancer mortality but flat trends in more

930

930

PART IV:  Cancers by Tissue of Origin GLOBAL

100% 80%

HPV16

60% 40% 20%

HPV18 0% 1940–1959

1960–1969

1970–7199

1980–1989

1990–1999

2000–2007

Time at diagnosis SQUAMOUS CELL CARCINOMA

100% 80%

HPV16

60% 40% 20%

HPV18

0% 1940–1959

1960–1969

1970–1979

1980–1989

1990–1999

2000–2007

Time at diagnosis ADENOCARCINOMA

100% 80%

HPV16* 60% 40% HPV18*

20% 0% 1940–1959

1960–1969

1970–1979

1980–1989

1990–1999

2000–2007

Time at diagnosis

Figure 48–3.  Trends of squamous cell carcinoma and adenocarcinoma. Alemany et al., Int J Cancer 2014;135:88–​95. © 2013 UICC. Used with permission.

recent years (Figure 48–5b) (Valerianova et  al., 2010; World Health Organization [WHO], 2014).

Cofactors The identification of HPV-​persistent infection as a necessary cause of cervical cancer had great impact on the epidemiologic study of

cervical cancer. The initially established risk factors for cervical cancer are now recognized by two major groups: those that favor incidence of HPV infection, and those that favor persistence and progression of HPV infections by acting only in its presence. As described in the previous edition of this book (Schiffman and Hildesheim, 2006), several methodological issues in modern epidemiology of HPV-​related carcinogenesis have been described, such as proper selection of controls

 931

Cervical Cancer

931

Smoking Cumulative incidence (%)

HPV–

Carcinogenic HPV+

HPV16+

20

10

0 0

2

4

6

8

10

12

14

16

22

Time (years) HPV16+ Carcinogenic+ HPV–

Women at risk 249 187

154

125

104

88

71

545

428

360

307

269

217

171

52

3757

3054

2621

2300

1953

1649

1281

429

Figure 48–4a. Women