Aging clocks face a test of their own

# Aging clocks face a test of their own OCTOBER 16, 2026 A claim that a diet or supplement "reversed biological age" may rest on a single test reading, and a new analysis suggests that is often not enough to trust. The study re-examined only three published single-clock results, so it does not show that most such claims fail. It shows that those three did not hold up across a wider panel of clocks. A field becomes more capable when it improves the way it judges its results. Re-examining clock-based claims is part of that work, particularly when small changes are asked to carry large meanings. The work was published in Aging Cell in 2026 by Daniel Borrus, Albert Higgins-Chen and colleagues at Yale. It is a re-analysis of existing data, not a new trial. The team took longitudinal DNA methylation datasets, in which the same people were tested before and after an event or intervention, and ran 12 epigenetic clocks on each. The text available to us does not give participant numbers for the datasets, so we do not report them. These are studies in people, and the results show patterns in test readings. They do not show effects on lifespan. Epigenetic clocks estimate biological age from chemical marks on DNA, read from a blood sample. The clocks differ. Early ones, such as Hannum and Horvath, were trained to predict calendar age. Later ones, such as PhenoAge and GrimAge, were trained to predict mortality risk. The Dunedin measures estimate pace of aging. The team also tested more reliable versions of six clocks, called PC clocks, built to reduce measurement noise. Noise is a real issue. According to the authors' earlier work, re-testing the same person with standard clocks can shift the result by 0.9 to 2.4 years. PC clocks cut that to 0.3 to 0.8 years. A change of a year or two on one test could therefore be chance. The researchers first revisited three published datasets that had reported a significant drop in biological age on one clock. Two were an 8-week diet, exercise and supplement program and a 2-year folic acid and vitamin B12 trial, both reported on the Horvath multi-tissue clock. They calculated 11 more clocks on the same data. None changed significantly in the same direction as the original finding. The study says this raises questions about whether such single-clock results are real or statistical artifacts. Bioengineer.org, which covered the paper, adds that in one dataset ten of the eleven extra clocks trended the opposite way. That detail comes from the coverage. To learn what a trustworthy signal looks like, the team used events known to raise mortality risk: intensive surgery, radiation and chemotherapy, and SARS-CoV-2 infection. In all three datasets, at least 4 of the 12 clocks rose significantly. PCPhenoAge, PCGrimAge and DunedinPACE rose in all three. The older chronological clocks (Hannum, Horvath multi-tissue and Horvath skin-and-blood) showed no significant increase in any. The skin-and-blood clock showed a significant decrease after radiotherapy and chemotherapy. From this, the authors derived three criteria for more confidence. Per Bioengineer.org's summary, the reliable PC clocks should respond, the mortality-trained or pace-of-aging clocks should respond, and multiple clocks should move in the same direction. The authors also suggest averaging all 12 clocks per person as a quick filter for whether the clocks agree. They stress it is not a new biomarker. Some interventions did pass. The study identified three datasets that met the criteria. According to the Bioengineer.org coverage, these included a two-year Mediterranean diet trial with a standard arm and a green Mediterranean arm, and a three-week alcohol abstinence program in chronic drinkers. The coverage reports significant decreases in at least five clocks, and the 12-clock average fell significantly in every case. In plain terms, diet and cutting alcohol produced shifts that showed up across several clocks, which is the pattern the authors consider convincing. This does not mean anyone should change their diet, start or stop a supplement, or stop drinking on their own. Abrupt alcohol cessation in heavy drinkers can be medically dangerous, so talk with your clinician first. There is a further caution. Blood tests reflect the mix of immune cells as well as cell aging. The coverage reports that after adjusting for this, surgery and chemotherapy effects shrank by about three-quarters, while the COVID-19 effect barely changed and the Mediterranean diet arms kept most of theirs. The authors say researchers should report such adjustments. For anyone weighing a biological-age test or an age-reversal advertisement, the study does not offer consumer guidance, but its logic suggests some questions of our own. Does the result come from one clock or many? Do reliable and mortality-based clocks agree? Was the change larger than the known retest noise? These are our extrapolation, not the authors' advice. A clock change also does not by itself mean added years of life. The main limitation is that the criteria were derived from a handful of datasets, and the study is a re-analysis rather than a new trial. Watch next for trials that report several clocks, including reliable ones, and that publish the results that did not reach significance. Agreement across reliable measures would make a claim more credible. Connecting those measures to outcomes would make it more useful. Sources: Aging cell (2026), Bioengineer.org (Yale University / Aging Cell) Aging cell (2026): https://doi.org/10.1111/acel.70756 Bioengineer.org (Yale University / Aging Cell): https://bioengineer.org/epigenetic-clocks-need-a-reality-check-before-anti-aging-claims/ ---