Abstract
BACKGROUND: Systemic studies of association of genome-wide DNA methylated sites with cardiovascular disease (CVD) in prospective cohorts are lacking. Our aim was to identify DNA methylation sites associated with the risk of CVD and further investigate their potential predictive value in CVD development for high-risk subjects. METHODS: We performed an epigenome-wide association study (EWAS) to identify CpGs related to CVD development in a Chinese population.We adopted a nested case-control design based on data from China PEACE Million Persons Project. A total of 83 cases who developed CVD events during follow-up and 83 controls who were matched with cases by age, sex, BMI, ethnicity, medications treatment and behavior risk factors were included in the discovery stage. Genome-wide DNA methylation from whole blood was detected using Infinium Human Methylation EPIC Beadchip (850 K). For significant CpGs [FDR(false discovery rate) < 0.005], we further validated in an independent cohort including 38 cases and 38 controls. RESULTS: In discovery set, we identified 8 significant CpGs (FDR < 0.005) associated with the risk of CVD after adjustment for cell components, demographic and cardiac risk factors and the first 5 principal components. Two of these identified CpGs (cg06901278 and cg09306458 in UACA) were replicated in another independent set (p < 0.05). Enrichment analysis in 787 individual genes from 1036 CpGs in discovery set revealed a significant enrichment for anatomical structure homeostasis as well as regulation of vesicle-mediated transport. Receiver operating characteristic (ROC) analysis showed that the model combined 8 CVD-related CpGs with baseline characteristics showed much better predictive effect for CVD occurrence compared with the model with baseline characteristics only [AUC (area under the curve) = 0.967, 95% CI (0.942 - 0.991); AUC = 0.621, 95% CI (0.536 - 0.706); p = 9.716E-15]. CONCLUSIONS: Our study identified the novel CpGs associated with CVD development and revealed their additional predictive power in the risk of CVD for high-risk subjects.