Higher selenium was associated with higher risk of diabetes: Consistent evidence from longitudinal and cross-sectional studies based on nail and serum selenium measures

Although the association between selenium (Se) and diabetes has been well-discussed in recent years, few studies have focused on the effects of long-term natural Se exposure and rarely concerned the effects of different Se biomarkers. To address this question, we carried out a 7-year longitudinal study on older adults aged over 65 and another crosssectional study on middle-aged and older adults aged 40 and above from Chinese soil Se-deplete and Se-optimum areas. Cox proportional hazard models were used to evaluate the associations between nail Se levels and incidence risk of diabetes. Unconditional logistic regression models and analysis of variance models were used to examine the associations between serum Se levels and the prevalence risk of diabetes. The nail and serum Se levels were 0.47 +/- 0.20 mu g/g and 111.09 +/- 55.01 mu g/L for the two study populations, respectively. For both of the independent studies, higher Se levels were observed to be associated with a higher risk of diabetes and prediabetes. Compared with the Second nail Se quartile (Q2), the adjusted hazard ratios (HRs) and 95 % confidence intervals (95 % CIs) of diabetes for Q1, Q3 and Q4 were 1.24(0.70, 2.21), 1.53(0.98, 2.39) and 1.31(0.76, 2.26), respectively, and the adjusted HRs (95 % CIs) of prediabetes were 1.47(0.77, 2.81), 1.38(0.83, 2.30), and 1.97(1.13, 3.44), respectively. Compared with the first serum Se quintile (Q1), the adjusted odds ratios (ORs) and 95 % CIs of diabetes for higher quintiles were 1.12(0.75, 1.66), 1.05(0.71, 1.57), 1.09(0.73, 1.62) and 1.51(1.02, 2.19), and the adjusted ORs (95 % CIs) of prediabetes were 1.27(0.77, 2.09), 1.70(1.05, 2.74), 1.94(1.21, 3.11) and 1.67(1.03, 2.71). Our findings consistently suggest that higher Se status is associated with a higher risk of diabetes in adults.