Consistent Inverse Associations of Total, "Bioavailable", Free, and "Non-Bioavailable" Vitamin D with Incidence of Diabetes among Older Adults with Lower Baseline HbA_1c (6%) Levels

Background: Serum 25-hydroxyvitamin (25(OH)D) levels are inversely associated with risk of diabetes. The "free hormone hypothesis" suggests potential effects to be mainly related to concentrations of "bioavailable" and free rather than total 25(OH)D. We assessed associations of serum concentrations of vitamin D-binding protein (VDBP), as well as total "bioavailable", complementary "non-bioavailable", and free 25(OH)D, with the risk of developing diabetes among non-diabetic older adults in a large population-based cohort study in Germany. Methods: We included 4841 non-diabetic older adults aged 50-75 years at the baseline exam from the ESTHER cohort conducted in Saarland, Germany, in 2001-2002. Concentrations of "bioavailable" and free 25(OH)D were derived from serum concentrations of VDBP, total 25(OH)D, and albumin. Incidence of diabetes was ascertained during up to 14 years of follow-up. Associations were quantified by multivariable Cox proportional hazards regression models with comprehensive confounder adjustment. Results: During a median follow-up of 10.6 years, 837 non-diabetic participants developed diabetes. We observed similar inverse associations with developing diabetes for VDBP (hazard ratio (HR) for lowest versus highest quintile: 1.37, 95% confidence interval (CI): 1.09, 1.72), total 25(OH)D (HR: 1.31, 95% CI: 1.03, 1.66), and "non-bioavailable" 25(OH)D (HR: 1.30, 95% CI: 1.02, 1.65). Associations were smaller and statistically insignificant for "bioavailable" and free 25(OH)D. However, associations of total "non-bioavailable", "bioavailable", and free 25(OH)D with incidence of diabetes were much stronger among, and essentially restricted to, participants with lower baseline HbA(1c) (<= 6%) levels. Conclusions: This large prospective cohort study of older Caucasian adults, in agreement with results from randomized trials and Mendelian randomization studies, supports a protective effect of vitamin D against development of diabetes. The "free hormone theory" may not be relevant in this context. However, our results underline the importance of adequate vitamin D status among those who have not yet shown any sign of impaired glucose tolerance.