Mendelian Randomization and Bayesian Colocalization Analysis Implicate Glycoprotein VI as a Potential Drug Target for Cardioembolic Stroke in South Asian Populations

Abstract

Background Circulating plasma proteins are clinically useful biomarkers for stroke risk. We examined the causal links between plasma proteins and stroke risk in individuals of South Asian ancestry. Methods and Results We applied proteome‐wide Mendelian randomization and colocalization approaches to understand causality of 2922 plasma proteins on stroke risk in individuals of South Asian ancestry. We obtained genetic instruments (proxies) for plasma proteins from the UK Biobank (N=920). Genome‐wide association studies summary data for strokes (N≤11 312) were sourced from GIGASTROKE consortium. Our primary approach involved the Wald ratio or inverse‐variance‐weighted methods, with statistical significance set at false discovery rate <0.1. Additionally, a Bayesian colocalization approach assessed shared causal variants among proteome, transcriptome, and stroke phenotypes to minimize bias from linkage disequilibrium. We found evidence of a potential causal effect of plasma GP6 (glycoprotein VI) levels on cardioembolic stroke (odds ratio [OR] Wald ratio =2.53 [95% CI, 1.59–4.03]; P =9.2×10 −5 , false discovery rate=0.059). Generalized Mendelian randomization accounting for correlated single nucleotide polymorphisms (SNPs), with the P value threshold at P <5×10 −8 and clumped at r 2 =0.3, showed consistent direction of effect of GP6 on cardioembolic stroke (OR generalized inverse‐variance‐weighted =2.21 [95% CI, 1.46–3.33]; P =1.6×10 −4 ). Colocalization analysis indicated that plasma GP6 levels colocalize with cardioembolic stroke (posterior probability=91.4%). Multitrait colocalization combining transcriptome, proteome, and cardioembolic stroke showed moderate to strong evidence that these 2 traits colocalize with GP6 expression in the coronary artery and brain tissues (multitrait posterior probability>50%). The potential causal effect of GP6 on cardioembolic stroke was not significant in European populations (OR inverse‐variance‐weighted =1.08 [95% CI, 0.93–1.26]; P =0.29). Conclusions Our joint Mendelian randomization and colocalization analyses suggest that genetically predicted GP6 is potentially causally associated with cardioembolic stroke risk in individuals of South Asian ancestry. As genetic data on individuals of South Asian ancestry increase, future Mendelian randomization studies with larger sample size for plasma GP6 levels should be implemented to further validate our findings. Additionally, clinical studies will be necessary to verify GP6 as a therapeutic target for cardioembolic stroke in South Asians.