Epigenetic age acceleration and mortality risk prediction in US adults

Abstract

Epigenetic clocks have emerged as novel measures of biological aging and potential predictors of mortality. We examined all-cause, cardiovascular, and cancer mortality prediction by epigenetic age acceleration (EAA) estimated using different epigenetic clocks. Among 2105 participants to the 1999–2002 National Health and Nutrition Examination Survey aged ≥ 50 years old and followed for mortality through 2019, we calculated EAAs from the residuals of nine epigenetic clocks regressed on chronological age. We assessed the association of EAAs and pace of aging with mortality adjusting for covariates. During 17.5 years of median follow-up, 998 deaths occurred, including 272 from cardiovascular disease and 209 from cancer. Overall mortality was most significantly predicted by Grim EAA (P < 0.0001) followed by Hannum (P = 0.005), Pheno (P = 0.004), Horvath (P = 0.03), and Vidal-Bralo (P = 0.04) EAAs. Grim EAA predicted cardiovascular mortality (P < 0.0001), whereas Hannum (P = 0.006), Horvath (P = 0.009), and Grim (P = 0.01) EAAs predicted cancer mortality. Overall mortality prediction differed by race/ethnicity between non-Hispanic White and White participants for Horvath (P_interaction = 0.048), Hannum (P_interaction = 0.01), and Grim (P_interaction = 0.04) EAAs. Hannum prediction of cancer mortality also differed between the two races/ethnicities (P_interaction = 0.007). Despite being predictive in non-Hispanic White participants, Horvath (P = 0.75), Hannum (P = 0.84), and Grim (P = 0.10) EAAs failed to predict overall mortality in Hispanic participants, and Hannum EAA was not associated with cancer mortality in Hispanic participants (P = 0.18). In a US representative sample, Horvath, Hannum, SkinBlood, Pheno, Vidal-Bralo, and Grim EAAs as well as pace of aging predict mortality. Howbeit, Horvath, Hannum, and Grim EAAs were less predictive in Hispanic participants.