Delineating cardioembolic stroke from other stroke subtypes using genetic risk factors for atrial fibrillation

Atrial fibrillation (AF) affects ~34 million individuals worldwide and is associated with a five­fold increased risk of ischemic stroke. Strokes caused by AF are classified as cardioembolic, and secondary prevention of cardioembolic (CE) stroke focuses on identifying AF as a potential cause and preventatively treating with anticoagulants. However, presence of AF can remain occult even after extensive workup, and whether AF in the context of stroke carries the same stroke risk as pre­existing AF is unknown. We thus sought to understand whether AF genetic risk is an important and potentially useful determinant of CE stroke risk. We first assessed whether stroke patients with AF have a genetic predisposition to AF and then evaluated the extent to which genetic risk factors for AF capture the heritability of CE stroke.

We performed a genome­-wide association of AF and CE stroke in 13,390 ischemic stroke cases and 28,026 referents. Associated loci (p<5E-8) were consistent across the phenotypes, as expected. Using AF­-associated SNPs (p<4.4E-4) from an independent meta­-analysis, we observed strong correlation between previously­-reported AF genetic risk, AF in the presence of stroke, and CE stroke (Pearson’s r=0.77 and 0.76, respectively). Heritability estimates revealed AF and CE stroke to be similarly heritable (20.0% and 19.5%, respectively), suggesting that CE stroke does not have a substantial heritable component beyond the primary AF risk factor.

We next derived an AF polygenic risk score (PRS) using 934 previously­implicated AF SNPs. The PRS, adjusted for clinical AF risk factors, was associated with CE stroke (odds ratio (OR) per standard deviation (SD)=1.40, p=1.45x10­48) and explained ~20% of the heritable component of CE stroke risk. The AF PRS also associated with stroke of undetermined cause (p=0.004), but none of the other primary stroke subtypes (all p>0.1).

Despite the fact that CE stroke often affects older adults with multiple clinical comorbidities, increasing AF risk due to non­-genetic factors, our results suggest that CE stroke patients have enrichment for AF genetic risk. Future work in AF and CE stroke may reveal how PRSs based on AF­-associated SNPs could be used either to predict individuals at highest risk of CE stroke to be more closely monitored, or help distinguish between clinical etiopathologic subtypes of stroke.