Lancet Healthy Longev. 2026 Oct 07. pii: S2666-7568(25)00092-3. [Epub ahead of print] 100773
BACKGROUND: Frailty is an age-related condition characterised by multisystem physiological decline, which increases vulnerability to adverse outcomes. Biomarkers of ageing might identify individuals at risk and enable early interventions. This systematic review and meta-analysis aimed to examine cross-sectional and longitudinal associations between DNA methylation-based biological age metrics (eg, DNA methylation age, epigenetic-age acceleration [EAA], and age deviation) and frailty.
METHODS: In a systematic search of six databases (Embase, Cochrane Central Register of Controlled Trials, PubMed, Ovid, Scopus, and Web of Science) from Jan 1, 2011, to June 6, 2025, we identified population-based cohort studies reporting associations between DNA methylation age, EAA, or age deviation and frailty from general or disease-specific populations with a control group. Risk of bias was assessed using an adapted Newcastle-Ottawa Scale. Random-effects meta-analyses with Hartung-Knapp adjustments were performed on standardised β coefficients and SEs. Publication bias, influence, and sensitivity analyses were conducted.
FINDINGS: From 34 437 records screened, 24 studies met the inclusion criteria (17 cross-sectional studies, one longitudinal study, and six studies that were both cross-sectional and longitudinal), encompassing 28 325 participants (14 757 [52·1%] female; median of mean age 65·2 years [IQR 62·2-69·4]). DNA methylation age and age deviation showed no association with frailty. In cross-sectional meta-analyses, higher Hannum EAA (nine studies; n=11 162; standardised β coefficient 0·06 [95% CI 0·02-0·09], I2=71·4%), PhenoAge EAA (eight studies; n=10 371; 0·07 [0·03-0·11], I2=81·7%), GrimAge EAA (eight studies; n=10 371; 0·11 [0·06-0·15], I2=90·5%), and pace of ageing (five studies; n=7895; 0·10 [0·01-0·19], I2=91·0%) were significantly associated with higher frailty. In longitudinal meta-analyses, higher GrimAge EAA (five studies; n=6143; 0·02 [0·00-0·05], I2=46·0%, p=0·0481) was significantly associated with increases in frailty, whereas PhenoAge EAA and pace of ageing were not significantly associated with frailty.
INTERPRETATION: Higher GrimAge EAA is consistently associated with higher frailty. Future research should focus on developing and validating DNA methylation clocks that integrate molecular surrogates of health risk and are specifically trained to predict frailty in large, harmonised, longitudinal cohorts, enabling their translation into clinical practice.
FUNDING: National University of Singapore.