Abstract 11229: Clinical and Genetic Atrial Fibrillation Risk and Discrimination of Cardioembolic from Non-Cardioembolic Stroke

Introduction: Prior studies demonstrate correlation between atrial fibrillation (AF) polygenic risk score (PRS) and ischemic stroke, especially cardioembolic (CE) stroke, suggesting shared genetic components. In this study, we hypothesized that an AF PRS can discriminate CE from non-CE strokes. Methods: We evaluated AF and stroke risk in 26,145 individuals of European descent from the NINDS Stroke Genetics Network (SiGN) with study-adjudicated TOAST subtypes. AF genetic risk was estimated using two newly derived PRS (LDpred-funct and sBayesR) and two previously validated PRS (pruning and thresholding and LDpred). A clinical risk score (CRS) was derived within separate individuals in UK Biobank utilizing components of the Cohorts for Aging and Genomic Research in Epidemiology (CHARGE-AF) model available in SiGN. We regressed each AF PRS on AF status and separately CE stroke in logistic regression models adjusted for CRS, imputation group, and the first 10 principal components. We calculated discrimination of AF and CE stroke, and compared across models using 2000-iteration bootstrapping and pairwise Z-testing. We also assessed category-free reclassification of CE stroke risk with the addition of PRS to a) CRS, and b) a modified CHA 2 DS 2 -VASc score including all available score components. Results: Each AF PRS was significantly associated with AF and CE stroke after adjustment for the CRS. Addition of any AF PRS significantly improved model discrimination as compared to the CRS alone, and LDpred discriminated both AF and CE better than other PRSs ( Figure ; P <0.005). Adding LDpred PRS to CRS resulted in appropriate reclassification of CE stroke risk compared to the CRS or the modified CHA 2 DS 2 -VASc score alone ( Figure ). Conclusions: Addition of AF polygenic risk to clinical risk factors improves discrimination of CE from non-CE strokes, as well as reclassification of stroke subtype. AF polygenic risk may be a useful biomarker for identifying AF-related strokes.