Correlates of serum alpha- and gamma-tocopherol in the Women's Health Initiative

PURPOSE: There is increasing evidence that vitamin E (primarily alpha- and gamma -tocopherol) may reduce the risk of cardiovascular disease and some c ancers, therefore it is important to understand factors that influence bloo d levels. METHODS: The correlates of serum alpha- and gamma -tocopherol were investig ated among participants in the Women's Health Initiative (WHI), a 40-site d isease prevention trial. Subjects were 1047 postmenopausal women aged 50-79 years, who provided fasting blood specimens and detailed information on di et, supplement use, and other factors at entry to the study (1994-96). RESULTS: Total serum cholesterol and triglycerides were highly correlated w ith serum alpha- and gamma -tocopherol concentrations and were controlled f or in all analyses along with age, ethnicity and body mass index (BMI). Alp ha and gamma -tocopherol were strongly negatively correlated (partial r = - 0.69). The strongest predictor of serum tocopherols was average daily intak e of vitamin E from supplements (partial r = 0.60 for alpha, r -0.54 for ga mma). Other factors associated with increased alpha- and/or decreased gamma -tocopherol concentrations were serum retinol and carotenoids, supplementa l vitamin C, alpha -tocopherol intake from food, dietary fiber, and Hispani c ethnicity. Factors associated with lower alpha- and/or higher gamma -toco pherol concentrations included gamma -tocopherol intake from food, total fa t intake, and BMI. Age, income, hormone use, and geographic location were " spuriously" associated with serum tocopherol levels through their associati on with supplement use, i.e., there was no such association among the subse t of women not taking supplements. CONCLUSIONS: Vitamin E intake from supplements and BMI are the major indepe ndent predictors of serum tocopherol levels in women, whereas dietary facto rs only play a small role. Ann Epidemiol 2001;11:136-144. (C) 2001 Elsevier Science Inc. All rights reserved.