Epigenetic Clocks (Horvath, GrimAge, DunedinPACE)

What they are

Epigenetic clocks estimate biological age (or the rate at which someone is aging) from DNA-methylation levels at a chosen set of CpG sites, using a trained regression model. Major variants:

• Horvath 2013 (multi-tissue): original pan-tissue clock; ~353 CpGs.
• Hannum 2013 (blood): trained on whole-blood methylation.
• PhenoAge (Levine 2018): trained against a composite of clinical biomarkers; better mortality prediction.
• GrimAge (Lu 2019): trained against plasma-protein surrogates and smoking-pack-years; strongest mortality prediction of the “1st-gen” clocks.
• DunedinPACE (Belsky 2022): estimates the rate of aging from a single time-point, calibrated against longitudinal phenotypic data from the Dunedin cohort.

Why they matter

Epigenetic clocks are the most-used biological-age readout in geroscience research today. They predict all-cause mortality independently of chronological age and respond (modestly) to known interventions: caloric restriction, exercise, and possibly rapamycin.

Limitations

• Different clocks measure different things — results from different clocks are not interchangeable.
• Consumer tests vary widely in quality and interpretation.
• A single measurement is noisy; longitudinal trends are more informative.
• “Reducing biological age by X years” from a short intervention is often over-interpreted by direct-to-consumer marketing.

What it’s useful for

• Research endpoints for intervention trials.
• Population-level epidemiology of biological aging.
• Personal use is reasonable for tracking trends over years, with appropriate scepticism about absolute numbers from single tests.

Related entries

See also: Epigenetic alterations, Caloric restriction.