Evaluation of stroke as a potential moderator of polygenic risk in Alzheimer's disease among non-Hispanic white and Caribbean Hispanic participants.

Abstract

Recent studies have evaluated Alzheimer's disease (AD) associated with cerebrovascular and cardiovascular risk factors. Stroke history mediates association between late-onset AD and cardiovascular disease risk factors (Tosto et al., 2016). Our research investigated stroke as a potential moderator between polygenic risk score (PRS) and Alzheimer's disease. Participants were enrolled in the multi-ethnic Washington Heights-Inwood Columbia Aging Project (WHICAP). 653 non-Hispanic white and 2285 Caribbean Hispanic participants were included based on GWAS data and stroke status (self-reported history of stroke or MRI-visualized stroke). A concordance analysis assessed stroke group agreement. PRS from GWAS statistics was calculated for both populations. Following calculation of moderators by stroke status on population-specific PRS, binomial logistic regression models were conducted. Model outcome was Alzheimer's disease, which incorporated four diagnoses: Pure Alzheimer's disease, Probable Alzheimer's disease with Stroke, Alzheimer's disease with Parkinsonism, and Atypical Alzheimer's disease. Another set of models was conducted for the Caribbean Hispanic population towards a general outcome of Alzheimer's disease alone. Concordance analyses among all 4987 WHICAP participants showed Cohen's kappa of -0.033 and p of 0.001* (p* < 0.05), indicating no concordance between self-reported and MRI-visualized stroke groups; this is consistent with observation by Reitz et al., 2009. Within the Caribbean Hispanic population, MRI-visualized stroke models for Alzheimer's disease (four types) showed moderator significance (p of 0.002*, 0.007*, and 0.001* for Models 1, 2, 3 respectively) and MRI-visualized stroke models towards generalized Alzheimer's disease showed moderator significance (p of 0.152, 0.041*, and 0.020* for Models 1, 2, 3 respectively). Stroke is likely to be an important moderator of PRS and Alzheimer's disease risk in MRI-visualized stroke models, while a self-reported history of stroke was not related.