Epigenetic clocks are not yet a validated stand-in for health

On 24 October 2026 the journal Experimental Gerontology published a systematic review from researchers at Henan University of Chinese Medicine in Zhengzhou, China, that asks a plain question. When an intervention makes a DNA-methylation clock look younger, do people's health measures improve along with it? The team gathered human intervention studies that reported a change in at least one such clock and at least one clinical outcome, before and after treatment. They found 26 independent studies: 13 comparative studies and 13 single-arm before-and-after studies.

Putting the clock next to the clinic

For each study the authors set the change in the clock beside the change in clinical measures and classified whether the two moved together. Fifteen studies gave mixed results, six were indeterminate and four were jointly inconclusive. Only one uncontrolled study was uniformly concordant, and that was for intermediate measures rather than outcomes that people feel or that shorten life. No comparative study, and no study of patient-important outcomes, was uniformly concordant.

Eleven studies went further and looked at individuals: whether the people whose clock improved most also improved most clinically, whether responders differed from non-responders, and whether the clock change helped explain the clinical effect. Clear findings appeared only in single studies and were not repeated in independent ones. No combination of intervention, clock and clinical outcome turned up often enough for a meta-analysis, and the authors found no trial-level validation of any clock as a surrogate endpoint. They rated the certainty of the evidence as very low across everything they assessed.

A measurement is not yet a stand-in for health

A surrogate endpoint is a measurement that trials can use instead of the outcome that matters, because moving it has been shown, across many trials, to move that outcome. The review concludes that epigenetic clocks are not there yet: for now they should be treated as exploratory or supporting measures of how the body responds to an intervention, not as validated substitutes for clinical results.

That does not make the clocks meaningless. In observational data they carry real information. DunedinPACE, a blood DNA-methylation measure of the pace of ageing built from the Dunedin Study 1972–1973 birth cohort, was associated with later illness, disability and death in the datasets used to test it. The open question is a different one: whether pushing the number down with a diet, a pill or a programme changes those risks.

What the larger randomised trials have shown

The two randomised trial analyses cited here were secondary or post hoc analyses and reported small effects on some clocks. In CALERIE, 220 adults without obesity were randomly assigned to 25% calorie restriction or their usual diet for two years. Restriction slowed DunedinPACE but did not significantly change the PhenoAge or GrimAge clocks, and the authors described the effects as small. In DO-HEALTH, a post hoc analysis of 777 older adults over three years found that omega-3 alone slowed three clocks, with effects equivalent to 2.9 to 3.8 months.

Neither report establishes that slowing a clock changes disease or mortality outcomes. The CALERIE authors themselves wrote that a conclusive test will need long-term trials with chronic disease and mortality as endpoints.

How to read a biological-age result today

For someone without a diagnosis who has paid for an epigenetic age test, or who sees a supplement or clinic advertise that it turns biological age back by a number of years, the review offers a working rule. A younger clock reading after an intervention is a change in a laboratory measure. It is not evidence that the intervention lowers the risk of disease or lengthens life. The review found no trial-level validation of any clock as a stand-in for such outcomes. Anyone weighing an intervention because of a clock result should discuss it with a doctor.

The review has limits of its own. Thirteen of the 26 studies had no comparison group, and the classification of agreement is the authors' own. The abstract does not list which interventions were included or how the work was funded; the authors declared no competing interests. It was published online ahead of print.

Sources

• Yi M, et al. Concordance between changes in epigenetic clocks and changes in clinical outcomes in geroscience intervention trials: a systematic review. Experimental Gerontology, 24 October 2026. PubMed 42926120

• Belsky DW, et al. DunedinPACE, a DNA methylation biomarker of the pace of aging. eLife, 2022. PubMed 35029144

• Waziry R, et al. Effect of long-term caloric restriction on DNA methylation measures of biological aging in healthy adults from the CALERIE trial. Nature Aging, 2023. PubMed 37118425

• Bischoff-Ferrari HA, et al. Individual and additive effects of vitamin D, omega-3 and exercise on DNA methylation clocks of biological aging in older adults from the DO-HEALTH trial. Nature Aging, 2025. PubMed 39900648