Abstract P157: Serum Vitamin D is Not Associated With Risk of Incident Diabetes in African Americans: The Jackson Heart Study

A recent meta-analysis of cohort studies with participants primarily of European ancestry, concluded that the risk of incident diabetes was inversely proportional to serum 25-hydroxy-vitamin D levels (25-OH-D). African Americans have lower levels of vitamin D than European Americans. But, lower levels of vitamin D do not seem to carry the same risk for low bone density or vascular calcification for African Americans. We sought to test the hypothesis that low vitamin D elevates the risk of diabetes in African Americans. During 2001-5, 5301 African American adults in Jackson, MS, were examined and blood samples drawn; mean age was 55±13. Serum 25-OH-D2 and 25-OH-D3 were measured from stored frozen serum; mean total 25-OH-D (25-OH-D2 plus 25-OH-D3) =14.5±6.7ng/ml. A seasonal pattern was evident for 25-OH-D3 but not for 25-OH-D2 levels. A cosinor model adjusted for seasonality of 25-OH-D3; mean annualized concentrations and seasonal amplitude were significantly higher for men and for persons of normal BMI compared to overweight or obese persons. Total 25-OH-D, as the sum of predicted annualized mean 25-OH-D3 and measured 25-OH-D2, was used in subsequent analysis. The analysis evaluated 3363 participants after exclusions for missing serum vitamin D (n=141), prevalent diabetes (n=1152), incomplete ascertainment of diabetes (n=62), or no follow-up (n=725). During a mean follow-up of 7.1 years, there were 584 new cases of diabetes. Using a Cox Proportional Hazards model controlling for age and sex, the risk of incident diabetes was significantly and inversely associated with total serum 25-OH-D; after adding BMI to the model, 25-OH-D was not significantly associated with risk of diabetes (Table 1). Physical activity, smoking, and alcohol use did not predict incident diabetes in any model. In conclusion, we found that in this cohort of African-Americans, with low mean 25-OH-D and a relatively narrow range of 25-OH-D, the risk of incident diabetes was not associated with serum 25-OH-D levels after controlling for sex, age, and BMI.