In a prospective multinational cohort of 2,594 heart failure patients across 40 countries, fifteen DNA methylation (DNAm)-based scores and epigenetic clocks independently predicted mortality after adjustment for age, sex, ancestry, NT-proBNP, and the established MAGGIC risk score. GrimAge acceleration carried a hazard ratio of 1.36 per standard deviation for all-cause death; DNAm frailty and mortality scores reached HRs of 1.44 and 1.48, respectively. Diet-related methylation signatures (Alternative Healthy Eating Index, Mediterranean Diet Score) were protective, with HRs of 0.81 and 0.86. Critically, patients accumulating four or five extreme-high scores faced more than double the mortality risk (HR 2.27), and epigenetic signals were most discriminating among lower-clinical-risk patients (MAGGIC <17).
This work is notable for its scale — few epigenetic studies recruit from 40 countries — and for demonstrating additive prognostic value atop already-strong biomarkers like NT-proBNP. The diet-methylation signal is particularly intriguing: it suggests cumulative dietary patterns leave a measurable molecular imprint that modifies prognosis independently of disease severity. However, important caveats apply. Whole-blood DNAm proxies tissue-averaged signals and may not reflect cardiac-specific biology. Causality cannot be established from observational Cox models. The cohort skews male (66%) and middle-aged, limiting generalizability. Whether intervening on modifiable methylation signatures — via diet, exercise, or pharmacology — would alter outcomes remains untested. As a preprint posted to medRxiv and not yet peer-reviewed, methodology and effect estimates may shift upon formal scrutiny. Still, this represents a meaningful step toward biologically personalized heart failure risk stratification.
