Int. J. Cancer: 118, 1004–1009 (2006) ' 2005 Wiley-Liss, Inc.
Occupational physical activity and risk of adenocarcinomas of the esophagus and stomach Cheryl Vigen, Leslie Bernstein and Anna H. Wu* Department of Preventive Medicine, USC/Norris Comprehensive Cancer Center, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA Physical activity may have a role in many cancers, but little is known about its effect on esophageal and gastric adenocarcinoma risk. We investigated occupational physical activity and esophageal and gastric adenocarcinoma risk in a population-based, casecontrol study including 212 esophageal, 264 gastric cardia and 389 distal gastric cancer cases, and 1,330 controls in Los Angeles County. Lifetime occupational histories were obtained during inperson interviews, and total lifetime occupational activity (Total Activity Index) was calculated using US Census job codes classified as sedentary, or moderately or highly physically active. Average Annual Activity Index was a per-year Total Activity Index counterpart. Unconditional logistic regression was used to calculate odds ratios, 95% confidence intervals and trend tests adjusting for gender, race, age, birthplace, education, smoking, body mass index (BMI) and number of years worked. Esophageal adenocarcinoma risk tended to decrease with increasing Total Activity Index (OR 5 0.67, 95% CI 5 0.38,1.19 for highest versus lowest quartile), but neither gastric cardia nor distal gastric cancer was associated with the Total Activity Index. This inverse association held for esophageal adenocarcinoma (OR 5 0.61, 95% CI 5 0.38,0.99 for highest vs. lowest quartile) and modest associations were observed for gastric cardia (OR 5 0.76, 95% CI 5 0.49,1.18) and distal gastric cancer (OR 5 0.77, 95% CI 5 0.52,1.14) when based on Average Annual Activity Index before age 65 years. Analyses stratified by gender, race, age, BMI, education and years worked provided similar results. We found a modest protective effect of Total Activity Index on esophageal adenocarcinoma. Future studies with more complete information on occupational and recreational physical activity are needed to confirm and further investigate the suggested protective effect of physical activity on these tumor types. ' 2005 Wiley-Liss, Inc. Key words: esophageal/gastric adenocarcinomas; physical activity; occupation
Physical activity is associated with reduced risk of several types of cancer, with particularly strong evidence for colon and breast cancer.1–7 Recreational or occupational activities have been used as the exposures of interest in most studies of these relationships. How much of the recent increase in the incidence of esophageal and gastric cardia adenocarcinomas may be due to misclassification, as opposed to a result of lifestyle changes, including increasing body weight and decreasing physical activity, is controversial.8–11 While high body mass index (BMI) has consistently been found to be an important risk factor for esophageal and gastric cardia adenocarcinomas,12–14 virtually no information exists on the relationship between physical activity and risk of these cancers. In our study, we investigated the relationship between occupational physical activity and gastric and esophageal adenocarcinomas using estimated total lifetime occupational physical activity and average occupational activity level prior to age 65 years as the exposure metrics. Material and methods This study population has been previously described.13 Briefly, cases for our study were identified by the Los Angeles County Cancer Surveillance Program and included men and women aged 30–74 years when diagnosed. Esophageal and gastric cardia adenocarcinoma patients were diagnosed between 1992 and 1997, and distal gastric cancer cases were diagnosed between 1992 and Publication of the International Union Against Cancer
1994. Control subjects were individually matched to the interviewed case patients on gender, race, date of birth (65 years) and neighborhood of residence. A reference date for each case patient was defined as the date that was 1 year before the case patient’s date of diagnosis, and this same date was used as the reference date for the case’s matched control(s). The case group included 222 esophageal adenocarcinoma (ADC) patients and 277 gastric cardia and 443 distal gastric cancer patients who were compared to 1,356 control subjects (528 case patients had one control, 382 had 2 or more controls, 37 had no control and 5 were unable to complete the interview). Our study was approved by the Institutional Review Board of the Keck School of Medicine of the University of Southern California, and written informed consent was obtained from each study participant before interview. Trained interviewers, blinded to the study hypotheses, administered in-person interviews with study subjects or a case patient’s next-of-kin (NOK) when the case was unavailable due to death or illness. NOK interviews accounted for 240 (63 esophageal adenocarcinomas, 77 gastric cardia cancers and 100 distal gastric cancers) of the 865 interviews with case patients in this analysis (exclusions are described below). The structured interview included questions on demographic characteristics, lifetime smoking habits and alcohol use, height and weight at ages 20 years and 40 years and at the reference date, as well as lifetime occupational history. Subjects were asked to provide information on each job held outside the home, including both full-time and part-time jobs (at least half-time), paid and volunteer, lasting at least 1 year. Data collected for each job included the year the job began, number of years worked, job title and specific substances to which the subject may have been exposed. Years of work after the reference date were excluded. In this analysis, each job title was coded according to the Bureau of the Census 1970 Index,15 which included over 400 job titles. Each of these job titles was assigned a physical activity score of 0 for sedentary, 1 for moderate or 2 for high using methods published by Garabrant et al.1 Sedentary jobs were those involving little or no physical activity and included occupations such as secretary and clerical worker. Moderate activity jobs involved light physical activity and included occupations such as most categories of sales workers and most types of teachers. High activity jobs involved extensive physical activity and included occupations such as farm laborer and gardener. For each subject, we calculated total lifetime occupational physical activity (Total Activity Index) as the sum of the years worked weighted by the activity score for each of the jobs. For example, someone who had
Abbreviations: ADC, adenocarcinoma; BMI, body mass index; CI, confidence interval; NOK, next-of-kin. Grant sponsor: California Tobacco Related Research Program; Grant numbers: 3RT-0122, 10RT-0251; Grant sponsor: National Cancer Institute; Grant number: CA59636; Grant sponsor: National Institute of Environmental Heath Sciences; Grant number: P30 ES07048. *Correspondence to: USC/Norris Comprehensive Cancer Center, 1441 Eastlake Avenue, Los Angeles, CA 90089-9175, USA. Fax: 1011-1-323-865-0139. E-mail:
[email protected] Received 29 March 2005; Accepted after revision 22 June 2005 DOI 10.1002/ijc.21419 Published online 8 September 2005 in Wiley InterScience (www.interscience. wiley.com).
ESOPHAGUS/STOMACH CANCER AND PHYSICAL ACTIVITY
worked as a secretary for 5 years, as a teacher for 25 years and as a gardener for 15 years would have a Total Activity Index of 55 [ 5 (5)(0) 1 (25)(1) 1 (15)(2)]. Part-time jobs were counted proportionally for the number of years worked and consequently for the Total Activity Index. When subjects listed 2 jobs covering the same time period, we determined the subject’s usual job based on which job had been reported during another period of the job history, or on reports of the typical number of hours worked per week. The usual job was coded as full-time with the second job coded as half-time. We excluded 58 cases (8 esophageal, 10 cardia and 40 distal gastric adenocarcinoma) and 25 controls for whom height or weight at the reference date was not known, since body mass index is an important risk factor for the cancers of interest and may be associated with physical activity. Twenty subjects [10 self-reporting cases (1 esophageal, 1 gastric cardia and 8 distal gastric), 9 NOK-reporting cases (1 esophageal, 2 gastric cardia and 6 distal gastric) and 1 control] provided limited or no occupational information and these subjects were excluded from our analyses. For 24 subjects (4 esophageal, 3 gastric cardia, and 11 distal gastric cancer patients and 6 controls), the information provided was incomplete and we imputed some value for their years worked when a job title was reported. Specifically, for 15 of the 24 subjects we estimated the missing number of years worked by filling in the gap between other fully reported occupations (these subjects showed no evidence of extended periods of time of unemployment). For the remaining 9 subjects, the missing years were either at the beginning or the end of the job history, and we imputed an expected number of years worked in the following manner. These 9 subjects had complete reports covering at least 50% of the expected number of years worked, and the missing years worked were imputed as the expected number of years worked minus the number of years worked in fully reported jobs, with expected number of years worked calculated according to average years worked by individuals (both cases and controls) of the same gender, age at diagnosis and education level in our study population. Results reported below are based on 212 esophageal, 264 gastric cardia and 389 distal gastric adenocarcinomas, and 1,330 control subjects. Data analysis Odds ratios (OR) and 95% confidence intervals (CI) for the OR were estimated to assess the association between the Total Activity Index and each of the 3 tumor types using unconditional logistic regression. Quartile cut points for Total Activity Index were selected using all control subjects combined. Covariates included in the models were age (in 10-year groupings), race (white, African-American, Latino-American and Asian-American), birthplace (US born or non-US born), education (less than high school, high school, some college and college graduate or higher), smoking status (never, former and current), BMI calculated as weight in kilograms divided by the square of height in meters (less than 25, 25 to 29.9 and 30 and above) and total years worked (quartiles). Analyses were repeated using BMI at age 20 years and BMI at age 40 years as a covariate. Additional analyses were stratified by sex, race (white or nonwhite), age (60 years or younger or greater than 60 years), BMI at reference date (less than 25, 25 or greater), years worked [less than, greater than the median (
