BackgroundThe causality between plasma branched-chain amino acids (BCAAs) levels and stroke remains uncertain and the stratified research on the association between BCAAs levels and subtypes of stroke is not well studied. Therefore, the association of genetically proxied circulating BCAA levels with the risks of stroke and its subtypes was explored by Mendelian randomization (MR) in this study. MethodsSummary-level data derived from the published genome-wide association studies (GWAS) were employed for analyses. Data for plasma BCAA levels (n = 16,596) were obtained from a meta-analysis of GWAS. The MEGASTROKE consortium provided data for ischemic stroke (n = 440,328) and its subtypes and data for hemorrhagic stroke were available from 2 meta-analyses of GWAS of European-ancestry groups (intracerebral hemorrhage, n = 3,026; subarachnoid hemorrhage, n = 77,074). The inverse variance weighted (IVW) method was selected as the primary MR analysis. Supplementary analysis used included the weighted median, MR-Egger regression, Cochran's Q statistic, MR Pleiotropy Residual Sum and Outlier global test, and leave-one-out analysis method. ResultsAccording to IVW analysis, 1-SD increment in genetically determined circulating isoleucine was associated with increased risks of cardioembolic stroke (CES) (OR: 1.56, 95% CI: 1.21-2.20, P = 0.0007), but not with risks of other stroke subtypes. We could not discover any proof that leucine and valine levels could increase risk of any stroke subtype. All heterogeneity tests produced stable findings, and there was no concrete evidence to indicate the perturbation of horizontal multiplicity. ConclusionIncreasing plasma isoleucine level had a causal effect on the risk of CES but not on the risk of other stroke subtypes. Further research is needed to identify the mechanisms of the causal associations between BCAAs and stroke subtypes.