Prospective Study of Serum Vitamin D and Cancer Mortality in the United States

D. M. Freedman, A. C. Looker, S.-C. Chang, B. I. Graubard
2007 Journal of the National Cancer Institute  
Vitamin D has been hypothesized to reduce cancer mortality through its effects on incidence and/or survival. Epidemiologic studies of the association of 25-hydroxyvitamin D [25(OH)D] and the risk of cancer, however, have been largely limited to incident cancers at a few sites . Methods A total of 16 818 participants in the Third National Health and Nutrition Examination Survey who were 17 years or older at enrollment were followed from 1988 -1994 through 2000. Levels of serum 25(OH)D were
more » ... ed at baseline by radioimmunoassay. Cox proportional hazards regression models were used to examine the relationship between serum 25(OH)D levels and total cancer mortality (in the entire population or according to race/ethnicity, sex, age, and retinol status) and mortality from specific cancers. Because serum was collected in the south in cooler months and the north in warmer months, we examined associations by collection season. All statistical tests were two-sided. Results We identified 536 cancer deaths in 146 578 person-years. Total cancer mortality was unrelated to baseline vitamin D status in the entire population, men, women, non-Hispanic whites, non-Hispanic blacks, Mexican Americans, and in persons younger than 70 or 70 years or older. We found no interaction between vitamin D and season or vitamin D and serum retinol. Colorectal cancer mortality was inversely related to serum 25(OH)D level, with levels 80 nmol/L or higher associated with a 72% risk reduction (95% confidence interval = 32% to 89%) compared with lower than 50 nmol/L, P trend = .02. Conclusions Our results do not support an association between 25(OH)D and total cancer mortality, although there was an inverse relationship between 25(OH)D levels and colorectal cancer mortality. JNCI | Articles 1595 * All P values for differences by serum 25(OH)D level were significant at P ≤ .001, except for education ( P value = .01) and alcohol ( P = .004), using chi-square test for categorical variables and linear regression analysis for continuous variables. BMI = body mass index. † Weighted estimate. ‡ Mean value. * RRs were adjusted for age, sex, and smoking history (pack-years) using Cox proportional hazard regression. P trend was based on 25(OH)D as a continuous variable. Categories were selected to reflect alternative cut points of 25(OH)D insufficiency and the full distribution of 25(OH)D values (21,22). * RRs were adjusted for age, sex, race/ethnicity, and smoking history (pack-years) using Cox proportional hazard regression. P trend was based on 25(OH)D as a continuous variable. Categories were selected to reflect alternative cut points of 25(OH)D insufficiency and the full distribution of 25(OH)D values (21, 22) .
doi:10.1093/jnci/djm204 pmid:17971526 fatcat:hiymgddlk5eqvagj3nbkvjpbv4