ABSTRACT
BACKGROUND: The impact of selenium on coronary artery disease (CAD) and type 2 diabetes (T2D) remains unclear with inconsistent results from observational studies and randomized controlled trials. We used Mendelian randomization to obtain unconfounded estimates of the effect of selenium on CAD, T2D, lipids and glycemic traits. METHODS: We applied genetic variants strongly (P < 5 × 10-8) associated with blood and toenail selenium to publicly available summary statistics from large consortia genome-wide association studies of CAD (76,014 cases and 264,785 non-cases), T2D (74,124 cases and 824,006 controls), lipids and glycemic traits. Variant specific Wald estimates were combined using inverse variance weighting, with several sensitivity analyses. RESULTS: Genetically predicted selenium was associated with higher T2D (OR 1.27, 95% CI 1.07-1.50, P = 0.006). There was little evidence of an association with CAD. Genetically predicted selenium was associated with lower low-density lipoprotein (LDL) cholesterol, lower high-density lipoprotein (HDL) cholesterol, higher fasting insulin and higher homeostasis model assessment of insulin resistance. These results were not robust to all sensitivity analyses. No associations with triglycerides, fasting glucose or homeostasis model assessment of ß-cell function were evident. CONCLUSIONS: Our study suggests selenium may increase the risk of T2D, possibly through insulin resistance rather than pancreatic beta cell function, but may reduce lipids. We found little evidence of an association with CAD, although an inverse association cannot be definitively excluded. The effect of selenium on these outcomes warrants further investigation.
