Epigenomic prediction of cardiovascular disease risk and interactions with traditional risk metrics

Epigenome-wide association studies for cardiometabolic risk factors have discovered multiple loci associated with incident cardiovascular disease (CVD). However, few studies have sought to directly optimize a predictor of CVD risk. Furthermore, it is challenging to train multivariate models across multiple studies in the presence of study- or batch effects. Here, we analyzed existing DNA methylation data collected using the Illumina HumanMethylation450 microarray to create a predictor of CVD risk across three cohorts: Women’s Health Initiative, Framingham Heart Study Offspring Cohort, and Lothian Birth Cohorts. We trained Cox proportional hazards-based elastic net regressions for incident CVD separately in each cohort, and used a recently-introduced cross-study learning approach to integrate these individual predictions into an ensemble predictor. The methylation-based risk score (MRS) predicted CVD time-to-event in a held-out fraction of the Framingham dataset (HR per SD = 1.28, p = 2e-3) and predicted myocardial infarction status in the independent REGICOR dataset (OR per SD = 2.14, p = 9e-7). These associations remained after adjustment for traditional cardiovascular risk factors and were similar to those from elastic net models trained on a directly merged dataset. Additionally, we investigated interactions between the MRS and both genetic and biochemical CVD risk, showing preliminary evidence of an enhanced predictive power in those with less traditional risk factor elevation. This investigation provides proof-of-concept for a genome-wide, CVD-specific epigenomic risk score and suggests that the DNA methylation data may enable the discovery of high-risk individuals that would be missed by alternative risk metrics.