DNA Methylation and Ischemic Stroke Risk: An Epigenome-Wide Association Study

Background Ischemic stroke (IS) risk heritability is partly explained by genetics. Other heritable factors, such as epigenetics, could explain an unknown proportion of the IS risk. The objective of this study is to evaluate DNA methylation association with IS using epigenome-wide association studies (EWAS). Methods We performed a two-stage EWAS comprising 1,156 subjects. Differentially methylated positions (DMPs) and differentially methylated regions (DMRs) were assessed using the Infinium 450K and EPIC BeadChip in the discovery cohort (252 IS and 43 controls). Significant DMPs were replicated in an independent cohort (618 IS and 243 controls). Stroke subtype associations were also evaluated. Differentially methylated cell-type (DMCT) was analyzed in the replicated CpG sites using EpiDISH. We additionally performed pathway enrichment analysis and causality analysis with Mendelian randomization for the replicated CpG sites. Results A total of 957 CpG sites were epigenome-wide-significant ( p <= 10 (-7) ) in the discovery cohort, being CpG sites in the top signals (logFC = 0.058, p = 2.35 x 10 (-22) ; logFC = 0.035, p = 3.22 x 10 (-22) , respectively). ZFHX3 and MAP3K1 were among the most significant DMRs. In addition, 697 CpG sites were replicated considering Bonferroni-corrected p -values ( p < 5.22 x 10 (-5) ). All the replicated DMPs were associated with risk of cardioembolic, atherothrombotic, and undetermined stroke. The DMCT analysis demonstrated that the significant associations were driven by natural killer cells. The pathway enrichment analysis showed overrepresentation of genes belonging to certain pathways including oxidative stress. ZFHX3 and MAP3K1 methylation was causally associated with specific stroke-subtype risk. Conclusion Specific DNA methylation pattern is causally associated with IS risk. These results could be useful for specifically predicting stroke occurrence and could potentially be evaluated as therapeutic targets.