Abstract
DNA methylation is an ideal biomarker for many applications, because it has both the stability for prognostic use and the plasticity to be altered by environmental variables including diet. Methylation risk scores (MRS) can be constructed to take advantage of epigenetic information in predicting disease incidence, and have notably been used in `biological age` measurements that robustly predict mortality risk. Thus, we hypothesized that DNA methylation could be used in a similar way to construct a specific predictor of cardiovascular disease (CVD) risk. We performed an analysis of methylation data, previously collected using the Illumina Infinium HumanMethylation450k platform, on samples from the longitudinal Framingham Heart Study Offspring Cohort. After preprocessing, methylation M-values were used to conduct a probe-wise epigenome-wide association scan employing Cox models to predict incident cardiovascular events. Significant loci from this step were taken forward into a combined regression whose fit defined the MRS as a weighted sum of methylation levels at these loci. Pending datasets will allow the validation of this MRS in an independent cohort, including assessments of the predictive performance of the model, both independently and with respect to established CVD predictors including the Framingham Risk Score. Additionally, component loci will be examined for biological significance in relation to known diet-responsive loci across the genome. The resulting MRS has potential for use in both clinical risk prediction and evaluation of the effectiveness of personalized dietary patterns.
Keywords
epigenetics, methylation, cardiovascular, CVD