Abstract
The complexity of epigenetic changes that accompany aging has been distilled into a number of molecular timepieces-termed epigenetic clocks-that characterize the pace of biological aging to differing degrees. Here, we develop and validate a DNA methylation-based Physiological health Age (PhysAge) score, comprised of eight DNA methylation surrogates to represent multi-system physiology and developed from commonly measured clinical biomarkers: CRP, peak flow, pulse pressure, HDL-cholesterol, Hba1c, waist-to-height ratio (WHR), cystatin C, and dehydroepianrosterone sulphate (DHEAS). We use data from the population-representative US Health and Retirement Study (HRS), split into a training (n = 1589) and test sample (n = 1588) and corroborate findings in two independent cohorts: The Irish Longitudinal Study of Aging (TILDA; n = 488) and the Northern Ireland Cohort for the Longitudinal Study of Ageing (NICOLA; n = 1830). PhysAge and the predominant second-generation epigenetic clocks, PhenoAge, GrimAge2, and DunedinPACE, were tested for their prediction of mortality and multiple age-related clinical measures (i.e., grip strength, gait speed, cognitive function, disability, frailty). PhysAge was comparable to extant clocks in predicting health measures and was indistinguishable from GrimAge2 in predicting mortality, despite not being trained on mortality. Moreover, the eight individual surrogates comprising PhysAge predicted health outcomes better than the measured values in many instances. The established clinical relevance of the biomarkers from which surrogates were derived opens up new opportunities for cross-study and cross-country comparisons of population health. Findings suggest that the DNA methylation PhysAge can be leveraged as a single biomarker to represent multiple physiological systems and offers utility in the context of clinical monitoring.