Abstract A72: Patterns of cancer risk according to weight in early, middle, and older adult life

Abstract

Abstract Background: Body mass index (BMI) assessed in middle and older adulthood is positively associated with risk of multiple cancer types. Whether BMI in early adulthood is similarly associated with the risk of these cancers has not been established. We therefore examined BMI in early, middle, and older adulthood in relation to site-specific cancer risk among 160,867 men and 97,853 postmenopausal women enrolled in the prospective NIH-AARP Diet and Health Study. We additionally examined the joint association of BMI at age 18 and adulthood weight change (from ages 18 to 50) to determine the independent association of each with cancer risk. Methods: Participants completed an initial self-administered questionnaire (sent in 1995–1996) inquiring about current weight and height, demographics, and other health behaviors and a second questionnaire (1996–1997) inquiring about weight at ages 18, 35, and 50. We used Cox regression to estimate hazard ratios (HR) for the associations of BMI at different ages, as well as the joint association between BMI at age 18 and adulthood weight change, with site-specific cancer risk, adjusting for potential confounders. Cancer outcomes were restricted to those previously identified from prospective cohort studies as having consistent positive associations with obesity. Results: During the follow-up period from 1996 through 2006 (median=10.1 years), 11,268 women and 27,707 men were diagnosed with a first primary cancer. Among women, risk of pancreatic and ovarian cancers and multiple myeloma were more strongly associated with BMI at age 18 than BMI later in life (e.g., HRs for pancreatic cancer=1.21, 1.08, 1.03, and 1.02 for BMI at ages 18, 35, 50, and at enrollment, respectively, per 5 kg/m2). This finding was largely confirmed in the analysis of joint effects of BMI at age 18 (<25 vs ≥25 kg/m2) and adulthood weight change (<4 vs ≥4 kg/m2): among women who gained <4 kg/m2 between age 18 and 50, higher BMI at age 18 was positively associated with risk of pancreatic cancer (HR=1.46, 95% CI: 0.93–2.29), and greater adulthood weight gain did not further increase this risk. In contrast, we found stronger positive associations between BMI at older versus younger ages and risk of endometrial and breast cancers and leukemia (e.g. endometrial HRs=1.24, 1.46, 1.50, and 1.55 for BMI at ages 18, 35, 50, and at enrollment). In line with these results, adulthood weight gain was positively associated with risk of breast cancer and leukemia in women irrespective of BMI at age 18, while BMI at age 18 was not positively associated with risk of these cancers after controlling for weight gain. For endometrial cancer, however, we observed independent positive associations for BMI at age 18 (≥25 vs <25 kg/m2) and adulthood weight gain (≥4 vs <4 kg/m2) (joint HR=3.01, 95% CI: 2.09–4.32). In men, associations for BMI at ages 18, 35, 50, and at enrollment were similar by cancer site (e.g., kidney cancer HRs=1.25, 1.36, 1.39, 1.32, respectively). Also in men, we found that BMI at age 18 and adulthood weight gain were independently positively associated with risk for most of the cancer sites examined, including NHL and kidney, thyroid, liver, gastric cardia cancers. Conclusion: There may be sensitive periods in the life course during which body weight has a particularly important role in cancer development; these influences may be site- and gender-specific. For many obesity-related cancers, higher BMI at age 18 and greater weight gain by age 50 had independent positive associations with risk. Citation Information: Cancer Prev Res 2011;4(10 Suppl):A72.