Exploring a targeted epigenetic clock based on mortality-associated CpGs as a potential biomarker for frailty

Age-related DNA methylation changes are promising biomarkers to track the individual aging process. Particularly second-generation epigenetic clocks capture aspects of biological age more accurately, but the requirement of genome wide profiles hampers implementation into practice. We therefore aimed to develop a simplified, targeted approach based on individual age- and mortality-associated CpG sites measurable by digital PCR.

We selected three CpG sites strongly associated with all-cause mortality in the Lothian Birth Cohorts and with chronological age in multiple publicly available repositories to establish the targeted age- and mortality-associated epigenetic clock (TaM clock). For comparison, we applied a previously published three-CpG signature selected solely for correlation with chronological age (TaC clock). These signatures were initially benchmarked using DNA methylation profiles from 20 frail and 20 non-frail participants of the ActiFE cohort. In fact, in this subset the TaM clock revealed significant association between the delta age and the frailty status based on a 32-item frailty index. Furthermore, the TaM clock outperformed the TaC clock at capturing a significant increase in epigenetic age in Down syndrome, Werner syndrome and HIV. We subsequently developed digital PCR assays to analyse 446 samples from the ActiFE cohort. One TaM clock site (cg20595453) showed significant association with both mortality and frailty. However, predictions generated by either targeted clock were not significantly associated with mortality and there was no clear relationship with frailty or age-related clinical parameters in this larger cohort.

Our targeted signatures were not sensitive enough to reliably predict frailty or mortality in a relatively healthy study population, which seems to be a general challenge for epigenetic clocks. It may be necessary to include additional disease-associated target sites to support frailty analysis in personalized medicine.