DNA methylation clocks: what the number means

A DNA methylation clock uses methylation measurements at selected CpG sites. At these positions, cytosine can carry a methyl group on carbon 5, forming 5-methylcytosine. The statistical model converts the measured proportions into a predicted age or another trained outcome.

Two clocks can give different results from the same blood sample because their training targets differ. Some estimate calendar age; others were calibrated against health-related traits, mortality or rates of physiological change.

Clocks trained on age and health outcomes

Early epigenetic clocks were trained to estimate chronological age. GrimAge uses mortality-related predictors, while DunedinPACE estimates the pace of ageing-related physiological change. A change of one predicted year and a change in pace are different quantities.

In August 2026, researchers recalculated 16 clocks across 51 human intervention studies. Thirteen interventions showed reductions across the clock panel after multiple-testing correction, but responses differed between interventions and clock families.

The study measures biomarker responsiveness. Whether a treatment-induced clock change predicts a corresponding improvement in survival or health is still untested for most interventions.

Measurement error and blood-cell composition

Methylation arrays have technical measurement error. A blood sample also contains varying proportions of immune-cell types. Illness or treatment may change those proportions, shifting a whole-blood clock even if methylation within each cell type remains stable.

An October 2026 Aging Cell study applied additional clocks to intervention datasets originally reported using one clock. None of the newly tested clocks reached statistical significance. The authors recommended technically reliable principal-component clocks, agreement between models and at least one mortality- or pace-trained measure.

Testing a clock change in a clinical trial

Trials should specify the clock analysis in advance, randomise participants when feasible and report all clocks measured. They should also follow outcomes patients experience, such as mobility, organ function and disease progression.

A reduced methylation age means that a specified model gave a lower reading in a particular sample. The size of any corresponding gain in healthy life remains unknown.