Arthur Schatzkin

Individuals who smoke generally have a lower body mass index (BMI) than nonsmokers. The relative roles of energy expenditure and energy intake in maintaining the lower BMI, however, remain controversial. We tested the hypothesis that current smokers have higher total energy expenditure than never smokers in 308 adults aged 40-69 years old of which 47 were current smokers. Energy expenditure was measured by doubly labeled water during a two week period in which the subjects lived at home and performed their normal activities. Smoking status was determined by questionnaire. There were no significant differences in mean BMI (mean ± SD) between smokers and never smokers for either males (27.8+5.1 kg/m² vs. 27.5+4.0 kg/m²) or females (26.5+5.3 kg/m² vs. 28.1+6.6 kg/m²), although the difference in females was of similar magnitude to previous reports. Similarly, total energy expenditure of male smokers (3069+764 kcal/d) was not significantly different from that of never smokers (2854+468 kcal/d), and that of female smokers (2266+387 kcal/d) was not different from that of never smokers (2330+415 kcal/d). These findings did not change after adjustment for age, fat-free mass and self-reported physical activity. Using doubly labeled water, we found no evidence of increased energy expenditure among smokers, however, it should be noted that BMI differences in this cohort also did not differ by smoking status.

Identifying diet-disease relationships in nutritional cohort studies is plagued by the measurement error in self-reported intakes.The authors propose using biomarkers known to be correlated with dietary intake, so as to strengthen analyses of diet-disease hypotheses. The authors consider combining self-reported intakes and biomarker levels using principal components, Howe's method, or a joint statistical test of effects in a bivariate model. They compared the statistical power of these methods with that of conventional univariate analyses of self-reported intake or of biomarker level. They used computer simulation of different disease risk models, with input parameters based on data from the literature on the relationship between lutein intake and age-related macular degeneration.The results showed that if the dietary effect on disease was fully mediated through the biomarker level, then the univariate analysis of the biomarker was the most powerful approach. However, combination methods, particularly principal components and Howe's method, were not greatly inferior in this situation, and were as good as, or better than, univariate biomarker analysis if mediation was only partial or non-existent. In some circumstances sample size requirements were reduced to 20-50% of those required for conventional analyses of self-reported intake.The authors conclude that (i) including biomarker data in addition to the usual dietary data in a cohort could greatly strengthen the investigation of diet-disease relationships, and (ii) when the extent of mediation through the biomarker is unknown, use of principal components or Howe's method appears a good strategy.

Observational epidemiologic studies of nutrition and cancer have faced formidable methodologic obstacles, including dietary measurement error and confounding. We consider whether Mendelian randomization can help surmount these obstacles. The Mendelian randomization strategy, building on both the accuracy of genotyping and the random assortment of alleles at meiosis, involves searching for an association between a nutritional exposure–mimicking gene variant (a type of “instrumental variable”) and cancer outcome. Necessary assumptions are that the gene is independent of cancer, given the exposure, and also independent of potential confounders. An allelic variant can serve as a proxy for diet and other nutritional factors through its effects on either metabolic processes or consumption behavior. Such a genetic proxy is measured with little error and usually is not confounded by nongenetic characteristics. Examples of potentially informative genes include LCT (lactase), ALDH2 (aldehyde dehydrogenase), and HFE (hemochromatosis), proxies, respectively, for dairy product intake, alcoholic beverage drinking, and serum iron levels. We show that use of these and other genes in Mendelian randomization studies of nutrition and cancer may be more complicated than previously recognized and discuss factors that can invalidate the instrumental variable assumptions or cloud the interpretation of these studies. Sample size requirements for Mendelian randomization studies of nutrition and cancer are shown to be potentially daunting; strong genetic proxies for exposure are necessary to make such studies feasible. We conclude that Mendelian randomization is not universally applicable, but, under the right conditions, can complement evidence for causal associations from conventional epidemiologic studies.

Although many epidemiologists use the National Death Index (NDI) as the “gold standard” for ascertainment of US mortality, high search costs per year and per subject for large cohorts warrant consideration of less costly alternatives. In this study, for 1995–2001 deaths, the authors compared matches of a random sample of 11,968 National Institutes of Health (NIH)-AARP Diet and Health Study subjects to the Social Security Administration's Death Master File (DMF) and commercial list updates (CLU) with matches of those subjects to the NDI. They examined how varying the lower limits of estimated DMF match probabilities (m scores of 0.60, 0.20, and 0.05) altered the benefits and costs of mortality ascertainment. Observed DMF/CLU ascertainment of NDI-identified decedents increased from 89.8% to 95.1% as m decreased from 0.60 (stringent) to 0.20 (less stringent) and increased further to 96.4% as m decreased to 0.05 (least stringent). At these same cutpoints, the false-match probability increased from 0.4% of the sample to 0.6% and then 2.3%. Limiting NDI cause-of-death searches to subjects found in DMF searches using less stringent match criteria, further supplemented by CLU vital status updates, improves vital status assessment while increasing substantially the cost-effectiveness of ascertaining mortality in large prospective cohort studies.

Education inequalities in cancer incidence have long been noted. It is not clear, however, whether such inequalities persist in the United States, especially for less common malignancies and after adjustment for individual risk factors. Within the NIH–AARP Diet and Health Study, we examined the association between education and the risk of developing cancers in a prospective cohort of 498 455 participants who were 50–71 year old and without cancer at enrollment in 1995/96. During a maximum 8.2 years of follow–up we identified 40 443 cancers in men and 18 367 in women. In age-adjusted models, the least educated men (<high school), compared to those with the most education (post–graduate), had increased risks of developing cancers of the esophagus (RR: 2.64, 95%CI:1.86–3.75), head and neck (1.98, 1.54–2.54), stomach (2.32, 1.68–3.18), colon (1.31, 1.12–1. 53), rectum (1.68, 1.32–2.13), liver (1.90, 1.22–2.95), lung (3.67, 3.25–4.15), pleura (4.01, 1.91–8.42), bladder (1.56,1.33–1.83) and combined smoking–related cancers (2.41, 2.22–2.62). In contrast, lower education level was associated with a decreased risk of melanoma of the skin (0.43, 0.35–0.54) and local prostate cancers (0.79, 0.74–0.85). Women with the least education had increased risks of colon (1.60, 1.24–2.05), lung (2.14, 1.79–2.56), kidney (1.68, 1.12–2.54) and combined smoking–related cancers (1.66, 1.43–1.92) but a lower risk of melanoma of the skin (0.33, 0.22–0.51), endometrial (0.67, 0.51–0.89) and invasive breast cancers (0.72, 0.61–0.84). Adjustment for smoking and other risk factors did not eliminate these associations, except those for cancers of the head and neck, colon, and liver in men and kidney in women. We found a higher risk of malignant disease, particularly smoking– related cancers, among those in the lowest educational attainment category. Only some of the educational gradient is attributable to smoking. The persistence of substantial education inequalities in cancer incidence poses a challenge for etiologic research and public health policy.

Introduction: To prospectively examine the relation of total, vigorous and non-vigorous physical activity to postmenopausal breast cancer risk. Methods: We studied 32,269 women enrolled in the Breast Cancer Detection Demonstration Project Follow-up Study. Usual physical activity (including household, occupational and leisure activities) throughout the previous year was assessed at baseline using a self-administered questionnaire. Postmenopausal breast cancer cases were identified through self-reports, death certificates and linkage to state cancer registries. A Cox proportional hazards regression was used to estimate the relative risk and 95% confidence intervals of postmenopausal breast cancer associated with physical activity. Results: During 269,792 person-years of follow-up from 1987 to 1998, 1506 new incident cases of postmenopausal breast cancer were ascertained. After adjusting for potential risk factors of breast cancer, a weak inverse association between total physical activity and postmenopausal breast cancer was suggested (relative risk comparing extreme quintiles = 0.87; 95% confidence interval = 0.74 to 1.02; p for trend = 0.21). That relation was almost entirely contributed by vigorous activity (relative risk comparing extreme categories = 0.87; 95% confidence interval = 0.74 to 1.02; p for trend = 0.08). The inverse association with vigorous activity was limited to women who were lean (ie, body mass index 2: relative risk = 0.68; 95% confidence interval = 0.54 to 0.85). In contrast, no association with vigorous activity was noted among women who were overweight or obese (ie, body mass index ≥ 25.0 kg/m²: relative risk = 1.18; 95% confidence interval = 0.93 to 1.49; p for interaction = 0.008). Non-vigorous activity showed no relation to breast cancer (relative risk comparing extreme quintiles = 1.02; 95% confidence interval = 0.87 to 1.19; p for trend = 0.86). The physical activity and breast cancer relation was not specific to a certain hormone receptor subtype. Conclusions: In this cohort of postmenopausal women, breast cancer risk reduction appeared to be limited to vigorous forms of activity; it was apparent among normal weight women but not overweight women, and the relation did not vary by hormone receptor status. Our findings suggest that physical activity acts through underlying biological mechanisms that are independent of body weight control.

Introduction: By inhibiting cyclooxygenase-2, nonsteroidal anti-inflammatory drugs (NSAIDs) decrease aromatase activity and might reduce breast cancer risk by suppressing estrogen synthesis. Epidemiologic evidence for a protective role of NSAIDs in breast cancer, however, is equivocal. Methods: We tested NSAID use for its association with breast cancer incidence in the National Institutes of Health–AARP Diet and Health Study, where 127,383 female AARP (formerly known as the American Association of Retired Persons) members with no history of cancer, aged 51 to 72 years, completed a mailed questionnaire (1996 to 1997). We estimated relative risks of breast cancer for NSAID exposures using multivariate Cox proportional hazards regression models. The state cancer registry and mortality index linkage identified 4,501 primary incident breast cancers through 31 December 2003, including 1,439 estrogen receptor (ER)-positive cancers and 280 ER-negative cancers. Results: Proportional hazards models revealed no statistically significant association between overall NSAIDs and total breast cancer. As cyclooxygenase inhibition by aspirin (but not other NSAIDs) is irreversible, we tested associations by NSAID type. Although we observed no significant differences in risk for daily use (versus nonuse) of aspirin (relative risk = 0.93, 95% confidence interval = 0.85 to 1.01) or nonaspirin NSAIDS (relative risk = 0.96, 95% confidence interval = 0.87 to 1.05), risk of ER-positive breast cancer was significantly reduced with daily aspirin use (relative risk = 0.84, 95% confidence interval = 0.71 to 0.98) – a relationship not observed for nonaspirin NSAIDS. Neither aspirin nor nonaspirin NSAIDs were associated with risk of ER-negative breast cancer. Conclusion: Breast cancer risk was not significantly associated with NSAID use, but daily aspirin use was associated with a modest reduction in ER-positive breast cancer. Our results provide support for further evaluating relationships by NSAID type and breast cancer subtype.

Several epidemiologic studies have examined the association between physical activity and pancreatic cancer risk; however, the results of these studies are not consistent.