Aging measures and cancer in the Health and Retirement Study (HRS).

Abstract

Cancer survivors may have higher biological age (BA) than cancer-free persons (controls). In HRS, we examined the associations of BA with cancer prevalence and mortality. BA was estimated by the Klemera and Doubal method (KDM-BA), phenotypic age (PhenoAge), and subjective age (SA) among 946 cancer survivors and 4555 controls; and by epigenetic clocks (Horvath, Hannum, Levine, GrimAge, Zhang Score (ZS), and methylation-based pace of aging (mPOA)) among 582 cancer survivors and 2805 controls. Age acceleration is estimated as residuals regressed on chronological age. There are significant multivariable associations with cancer prevalence for Hannum, GrimAge, and SA, and ZS (logistic regression), and with mortality for PhenoAge, Hannum, Levine, GrimAge, and ZS in cancer survivors, and for KDM-BA, PhenoAge, and ZS in controls (Cox regression). The strongest association in cancer survivors is for GrimAge (HR per 1 SD = 1.80, p < 0.001). PhenoAge and first- and second-generation epigenetic clocks hold promise for predicting mortality in cancer survivors.