Abstract
BACKGROUND: Previous studies have assessed the association between antidiabetic drugs and stroke risk, but the results are inconsistent. Mendelian randomization (MR) was used to assess effects of antidiabetic drugs on stroke risk. METHODS: We selected blood glucose-lowering variants in genes encoding antidiabetic drugs targets from genome-wide association studies (GWAS). A two-sample MR and Colocalization analyses were applied to examine associations between antidiabetic drugs and the risk of stroke. For antidiabetic agents that had effect on stroke risk, an independent blood glucose GWAS summary data was used for further verification. RESULTS: Genetic proxies for sulfonylureas targets were associated with reduced risk of any stroke (OR=0.062, 95% CI 0.013-0.295, P=4.65×10(－4)) and any ischemic stroke (OR=0.055, 95% CI 0.010-0.289, P=6.25×10(－4)), but not with intracranial hemorrhage. Colocalization supported shared casual variants for blood glucose with any stroke and any ischemic stroke within the encoding genes for sulfonylureas targets (KCNJ11 and ABCC8) (posterior probability>0.7). Furthermore, genetic variants in the targets of insulin/insulin analogues, glucagon-like peptide-1 analogues, thiazolidinediones, and metformin were not associated with the risk of any stroke, any ischemic stroke and intracranial hemorrhage. The association was consistent in the analysis of sulfonylureas with stroke risk using an independent blood glucose GWAS summary data. CONCLUSIONS: Our findings showed that genetic proxies for sulfonylureas targets by lowering blood glucose were associated with a lower risk of any stroke and any ischemic stroke. The study might be of great significance to guide the selection of glucose-lowering drugs in individuals at high risk of stroke.