Abstract Body: Background DNA methylation reflects environmental and lifestyle exposures and may contribute to peripheral artery disease (PAD), but causal CpG markers remain poorly characterized.

Methods In the VA Million Veteran Program, we conducted an epigenome-wide association study (EWAS) of incident PAD. Following, we performed de novo CpG-GWAS of PAD-associated CpGs and performed Mendelian randomization (MR) using PAD summary-level data among 31,307 PAD cases and 211,753 controls to strengthen causal inference. Network MR assessed 11 cardiometabolic mediators. We constructed methylation risk scores (methRS) for incident and prevalent PAD using LASSO-penalized logistic regression (10-fold cross-validation; 1-SE λ) with a 70/30 (train/test) split. Performance was compared with a clinical model including age, sex, ancestry, smoking status, and cardiometabolic comorbidities.

Results After excluding participants without methylation data, failing quality control, or with prevalent PAD (n=5,594), EWAS tested 757,266 CpGs among 3,433 participants with incident PAD and 30,719 without PAD. After BACON adjustment and Bonferroni correction, 420 CpGs were significantly associated with PAD; of these, 79.3% showed hypomethylation associated with increased PAD risk. MR supported 30 CpGs at nominal significance, of which 8 remained after false discovery rate correction. Adiposity (visceral adiposity and body mass index) mediated up to 33.1% (95% CI 9.5–56.6%) of the cg02738868–PAD association. Blood lipids mediated up to 27.4% (95% CI 14.9–39.9%) of the association between LDLR CpGs (cg19751789, cg19305903) and PAD. LASSO selected 65 CpGs for incident-PAD methRS and 276 CpGs for prevalent-PAD methRS. In the test set, methRS achieved AUC 0.705 (95% CI 0.690–0.719) for incident PAD and 0.797 (95% CI 0.787–0.808) for prevalent PAD. Adding methRS to the clinical model improved AUC to 0.741 (95% CI 0.727–0.755) and 0.847 (95% CI 0.837–0.857), respectively.

Conclusion Integrating EWAS with MR prioritized 30 putatively causal DNA methylation markers for PAD and highlighted major cardiometabolic mechanisms. MethRS provided non-redundant prediction beyond clinical factors, supporting identification of individuals at high PAD risk.