For stroke, a recent 90-SNP GRS derived from the MEGASTROKE GWAS meta-analysis4 showed that genetic and lifestyle factors are independently associated with incident stroke24, and that even among individuals with high GRS, lifestyle factors had a large impact on risk, implying that risk could be reduced in those with high genetic predisposition for stroke. However, in contrast to GRSs for other cardiovascular diseases like coronary artery disease (CAD)21–23, the predictive power of previous GRS for stroke has been limited25–27, likely because of limited genetic data for stroke and the well-known heterogeneity of the stroke phenotype4,7. Recent analytical advances have enabled more powerful GRS construction, such as those leveraging multiple sets of GWAS summary statistics21,28, potentially allowing for power and heterogeneity limitations to be overcome. Specifically, for CAD, an approach where multiple GRSs are combined into one meta-score (metaGRS) was found to improve risk prediction over any one of the individual CAD GRS21. Such an approach may be widened to provide substantively improved genomic prediction of stroke.
