DNA methylation patterns predict primary myocardial infarction

Introduction Previous work has linked epigenetic modifications such as DNA methylation (DNAm) to atherosclerotic cardiovascular disease (CVD), while advances in epigenetic editing might soon enable targeting DNAm as a therapeutic target. Therefore, we aimed to identify differentially methylated sites associated with the primary onset of MI, to elucidate putative pathomechanisms mediated by the epigenome. Methods Blood DNA methylation (DNAm) at 766,135 CpG sites was assessed in a subset of study participants of the Gutenberg Health Study (GHS, n=1,560), a prospective, population-based cohort. The MyoVasc study, a prospective cohort on heart failure (HF) and major CVD outcomes, was used for validation (n=2,042). Fatal or non-fatal primary MI was defined as a composite endpoint combining incident MI or cardiac death in individuals without history of MI. To identify CpG sites predictive of fatal or non-fatal primary MI, an epigenome-wide association study (EWAS) adjusted for age, sex and blood cell composition with 5% false discovery rate (FDR) correction was performed. A polyepigenetic primary MI risk score was trained in the derivation cohort using ridge regression, and tested in the validation cohort. CpG sites located in gene bodies or promotor regions were tested for their association with respective gene expression using linear regression adjusted for age, sex and cell-type abundance. Putatively causal genes were identified using Two-sample Mendelian Randomization (MR), integrating expression quantitative trait loci (eQTL) and MI or coronary artery disease (CAD) summary statistics from the eQTL and CARDIoGRAMplusC4D consortia, respectively. Results In the derivation cohort (GHS), 42.3% were women (mean age 57.2±10.1 years), with 369 events during 3.9 years median follow-up (IQR: 3.5); corresponding values in the validation cohort were 30.7%, and [67.4±9.9 years], n=124 events and 2 years (IQR: 1.7). In EWAS, 1,115 CpG sites were associated with incident MI. Genes annotated to these sites were enriched for KEGG pathways such as dilated cardiomyopathy, TNF signalling and MAPK signalling. The polyepigenetic risk score derived from these CpG sites predicted primary MI (HR: 1.43, 95%-CI:1.18–1.74, p = 0.0003) and all-cause death (p < 0.0001) in the validation cohort, but were not or only weakly predictive for stroke (p = 0.82) or recurrent MI (p = 0.053). Of 579 genes with available mRNA expression proximal to MI-related CpG sites, 81 were significantly associated at 5% FDR. Of these, eight were putatively causal for MI or CAD, including genes with roles in innate immunity, cardiomyopathy and cardiometabolic traits. Conclusion This study has identified differential DNA methylation in both established and novel loci associated with atherothrombosis. A subset of identified genes was both putatively causal for MI or its precursor CAD, and likely regulated through DNAm, representing potential targets for therapeutic epigenetic editing.