Given that the metaGRS is composed of GRSs for stroke and stroke risk factors, we conducted several complementary analyses to assess the association of the metaGRS with these risk factors, and whether the metaGRS was associated with IS risk independently of these risk factors. As expected, the IS metaGRS was positively and significantly associated with all seven risk factors (Supplementary Table 2). Adjusting for these risk factors as well as BP-lowering and/or lipid-lowering medication status only modestly attenuated the association of the metaGRS with incident IS (Supplementary Fig. 8), indicating that the information contained in the metaGRS was only partially explained by these factors. On the other hand, adjusting for the metaGRS modestly but consistently attenuated the association of each risk factor itself with IS risk (Supplementary Fig. 7). There was no evidence for statistical interaction of the metaGRS effects on incident IS with medication status at assessment (Wald test in logistic regression, P = 0.23 and P = 0.82 for interaction of the metaGRS with BP medication and cholesterol-lowering medication, respectively).
