Longitudinal changes in epigenetic clocks predict survival in the InCHIANTI cohort

Abstract

Importance: Cross-sectional assessment of epigenetic clocks provides information on the pace of aging. Whether longitudinal acceleration or deceleration of epigenetic clocks over time provides additional mortality prediction is unknown. Objective: To test the independent associations of baseline levels and longitudinal changes in epigenetic clocks with mortality Design: Longitudinal study Setting: InCHIANTI, a population-based study of community dwelling individuals in Tuscany, Italy. Participants: 699 InCHIANTI study participants aged 21-95 years at baseline with longitudinal measurements of DNA methylation. Exposure: Baseline levels and longitudinal changes in seven epigenetic clocks, including two first-generation clocks developed using chronological age for reference (Hannum Clock, Horvath Clock), three second-generation clocks developed using time-to-death for references (DNAmPhenoAge, DNAmGrimAge, DNAmGrimAge Version 2), and two third-generation clocks developed using longitudinal rate of change of multiple phenotypes for reference (DunedinPOAm_38, DunedinPACE). Main Outcomes and Measures: Mortality was the primary outcome. Cox regression was used to estimate independent associations of baseline and longitudinal changes in epigenetic clocks with mortality. Results: Adjusting for age, sex, study sites, and epigenetic clock at the baseline, longitudinal changes of the following epigenetic clocks were associated with mortality: Hannum clock (aHR = 1.14, 95% CI:[1.03, 1.26]), DNAmPhenoAge (aHR = 1.23, 95% CI: [1.10,1.37]), DNAmGrimAge (aHR = 1.13, 95% CI: [1.02,1.26]), DNAmGrimAge Version 2 (aHR = 1.18, 95% CI:[1.06,1.31]), and DunedinPOAm_38 (aHR = 1.15, 95%CI: [1.01,1.30]). Conclusions and Relevance: Our findings confirm that epigenetic clocks capture a dimension of health that is predictive of mortality and add the notion that time changes of epigenetic age reflect changes in health that additionally and independently contribute to mortality prediction. Future studies should test whether interventions that slow down the rate of epigenetic aging are associated with longer healthspan and lifespan.