Epigenetic Clocks of Aging: Can You Really Turn Back Time

The idea is tempting: give a blood or saliva sample, get a number — "your biological age is 47, even though your passport says 52" — and then make it drop. A whole industry of tests and supplements has grown out of this idea. But what is under the hood, and how much can you trust it? Let's break it down using recent peer-reviewed work.

What epigenetic clocks are

With age, methyl groups attach to DNA at specific points — this is called methylation. The pattern of these marks changes predictably, and in 2013 Steve Horvath showed that methylation at several hundred sites can predict a person's chronological age with good accuracy. That is how the first "clocks" appeared.

Later, the algorithms began to be trained not on age, but on health. The DunedinPACE clock (Belsky et al., 2022) measures not age but the pace of aging — like a "speedometer." It was calibrated against real changes in 19 markers across the cardiovascular, metabolic, renal, immune, and other systems in a cohort of 817 people followed from age 26 to 45; the final algorithm uses 173 methylation points. A value of 1.0 is the average pace, 0.85 means aging roughly 15% slower than normal.

How well they predict disease

This is where second-generation clocks shine. In the Framingham study (2,471 people), an accelerated DunedinPACE was linked to a markedly higher risk: cardiovascular disease — HR 1.39, and all-cause mortality — HR 1.65. Test-retest reliability is also high: the ICC coefficient reaches 0.97.

The largest comparison to date appeared in Nature Communications in December 2025: 14 different clocks were tested against 174 future diagnoses in 18,859 people from the Generation Scotland cohort. The conclusion is both important and sober: second- and third-generation clocks (GrimAge, DunedinPACE) clearly outperform the first generation at predicting disease and death — but no single clock came out best for every outcome, and the estimates from different algorithms often diverge from one another.

Can you actually grow younger

This is the key question — and here you have to be careful. The most frequently cited study is Kara Fitzgerald's randomized pilot trial (Aging, 2021): 43 healthy men aged 50–72, 8 weeks of diet, sleep, exercise, and stress work. In the intervention group, epigenetic age measured by the Horvath clock was 3.23 years lower than in the control group (p=0.018).

It sounds like a breakthrough, but the caveats are many. This is a small, short study on a narrow sample. In 2024 a correction was issued for the paper, refining how within-group changes were calculated to account for elapsed time. And, most importantly, lowering the number on the clock does not equal a proven extension of lifespan — it is still an indirect marker, not a confirmed result.

Where the clocks fall short

The core problem is correlation versus cause. The algorithms are trained to spot marks that accompany aging, but they do not distinguish the "drivers" of aging from the "passengers". Different clocks can be built on almost non-overlapping sets of points — and all of them will "work." Vadim Gladyshev's team tried to fix this in 2024 (Nature Aging) by splitting the clock into DamAge (damaging changes, linked to mortality) and AdaptAge (protective ones, linked to longevity), but this is only the beginning of the path.

Critics go further. A review in npj Aging (Kriukov et al., 2025) reminds us: biological age exists only as the output of the algorithm that measures it. In the CALERIE calorie-restriction study, one clock showed slower aging while another showed acceleration from the very same intervention. That is why the authors explicitly warn against using clocks as a reliable "ruler" to test claims of rejuvenation.

Is it worth taking the test right now

For medical decisions — not yet. A review in eBioMedicine (2026) calls such tests promising but premature for routine clinical use: gaps in standardization need to be closed and the mechanisms understood first. As a curious benchmark every couple of years — why not, but be skeptical of the exact number and don't chase tenths of a point. The paradox is that the best thing you can do for these numbers has long been known without any test: sleep, regular movement, decent nutrition, quitting smoking — exactly what the clocks measure.