DNA methylation signatures of diet quality and treatment-related cardiotoxicity in childhood cancer survivors

Background: Childhood cancer survivors are at increased risk of cardiometabolic risk factors (CMRF) and cardiovascular disease (CVD) due to cancer treatment exposures. Whether DNA methylation (DNAm) signature of diet quality can remediate these risks remains unclear. Methods: We examined blood DNAm signatures of diet quality and their associations with CMRF/CVD among 2,191 adult survivors of childhood cancer and 283 community controls from the St. Jude Lifetime Cohort. Diet quality was assessed using the Healthy Eating Index (HEI)-2015 and alternate Mediterranean Diet Score (aMED), respectively. Genome-wide DNAm was profiled using the Illumina EPIC BeadChip V1. We identified differentially methylated positions (DMPs), genes (DMGs) and regions (DMRs) for diet quality-associated DNAm signatures in blood, from which DNAm-based scores for a healthy diet (DNAmDiet) were further derived. Multivariable logistic regression examined associations between DNAmDiet and seven outcomes: abnormal glucose metabolism, hypertension, hypertriglyceridemia, hypercholesterolemia, obesity, cardiomyopathy, and myocardial infarction. We also compared diet- and treatment-related methylation effects, focusing on CpGs where healthy diet-associated methylation changes occurred in the opposite direction of treatment-induced alterations, suggesting potential remediation of treatment-related DNAm alterations. Results: We identified 29 DMPs, 113 DMRs, and 53 DMGs significantly associated with HEI-2015, and 9 DMPs, 16 DMRs, and 16 DMGs associated with aMED (P<9×10⁻⁸ for DMPs; FDR<0.05 for DMRs or DMGs). Diet-associated DNAm signatures mapped to key genes including AHRR, CPOX, CLDND1, GPR15, LY9, and MAP4K4, many of which are involved in immune regulation, cellular signaling, nutrient-sensing, suggesting that healthy diet may influence epigenetic regulation of immune and metabolic pathways. A DNAmDiet score based on aMED signatures was associated with lower risk of cardiomyopathy (Odds Ratio [OR]=0.76; 95% CI, 0.65-0.92), hypercholesterolemia (OR=0.65; 95% CI, 0.50-0.84), and hypertriglyceridemia (OR=0.78; 95% CI, 0.63-0.97). Effect sizes for diet-associated DMPs were, on average, twice as large in survivors as in community controls (|betasurvivor|/|betacontrol|=2.02; P<0.001), despite strong correlations between populations (r=0.85; P<0.001). We found that up to 87.5% of treatment-associated DMPs showed opposite associations with HEI-2015 or aMED scores, with up to 8.9% reaching statistical significance. Conclusions: DNAm signatures of healthy diet were associated with lower risks of specific cardiometabolic and cardiovascular outcomes. Importantly, healthy diet-related DNAm changes were in the opposite direction of treatment-induced DNAm alterations, suggesting potential for epigenetic recovery through dietary interventions.