Can anything you do actually move an epigenetic clock?

TL;DR

• One randomised trial has moved one clock - CALERIE, 220 adults, two years of roughly 25% caloric restriction, slowing DunedinPACE by about 3%.
• It did not move the others - PhenoAge and GrimAge showed no significant change, because outcome-trained clocks are loaded with lifetime exposure history.
• The 15% mortality figure is separate - it comes from observational cohort work, not from CALERIE, and combining the two overstates both.
• Replicate noise rivals the effect size - for some clocks the spread between technical replicates is comparable to what an intervention is expected to produce.
• "Did my biological age go down" is under-specified - two providers using different clocks can hand you different answers from the same blood draw.
• No trial has shown that moving a clock moves an outcome - the case for exercise, sleep and metabolic health rests on decades of hard endpoint data instead.
• Peptides sold for this, including epithalon, have not been shown to lower a validated clock - if you want to explore one, the physician-discussion section below explains how to measure it honestly, starting from a baseline.

Educational content, not medical advice.

---

If you're holding a biological-age result and deciding whether to spend money on changing it, you're asking a sharper question than most of the content on this topic answers. This article assumes you already know roughly what a methylation clock is - if not, epigenetic clocks and what a biological age test actually measures cover the definitions properly, and this piece won't repeat them.

The question behind the question

There are two versions of "can I lower my epigenetic age," and they have different answers.

Version one: can something I do change the number the lab sends back? Almost certainly yes, and that's a low bar - the number moves for several reasons, only some of which are biological.

Version two: can something I do change the underlying rate at which I am ageing, in a way that translates into more healthy years? That is the question you actually care about, and it is much harder to answer, because it requires a randomised trial, a validated instrument, and ideally long-term outcome follow-up that mostly doesn't exist yet.

Keeping these separate is the entire discipline of this topic. A clock is a measurement instrument, and moving an instrument's output is not the same as moving the thing it was built to estimate.

Measurement noise, first

Before any intervention discussion, the boring point that deserves top billing.

Methylation arrays have real technical variability. Run the same DNA sample twice and the clock estimates will not be identical. For some clocks, the spread between technical replicates is comparable in magnitude to the effect sizes people are hoping to detect from an intervention. This is not a fringe complaint - it's an active methodological concern in the field, and it's the reason serious work reports reliability statistics alongside results.

The practical consequences:

• A single before-and-after pair, on a consumer test, is not evidence. It's one draw from a distribution.
• "My biological age dropped four years in six months" is a claim that could be entirely explained by measurement variability plus regression to the mean, without any biology changing at all.
• Trends across several repeated measurements are far more informative than any two points, and this is why serious testing programmes re-measure on a schedule rather than reacting to one result.

Some providers reduce this by averaging across multiple technical replicates or using principal-component versions of the clocks, which are meaningfully more reliable. Whether yours does is a reasonable thing to ask before you buy a second test to compare against the first.

What the randomised data actually shows

The cleanest evidence available comes from CALERIE, and it is worth stating precisely because it is routinely overstated.

CALERIE randomised 220 adults without obesity to either roughly 25% caloric restriction or an ad-libitum control condition for two years. A published analysis in Nature Aging in 2023 examined the effect on epigenetic measures. The findings:

• The intervention slowed the pace of ageing as measured by DunedinPACE, by roughly 3%.
• It did not produce significant changes in biological-age estimates from first- and second-generation clocks, including PhenoAge and GrimAge.

That is the state of the randomised evidence. One trial, one instrument moved, a small effect, and the other instruments unmoved.

A figure often attached to this deserves careful handling. Independent observational cohort work associates a roughly 3% slower DunedinPACE with approximately 15% lower mortality risk. That association is real and it comes from separate observational data - it is not a finding of the CALERIE trial, and CALERIE did not demonstrate a mortality benefit. Combining the two into "caloric restriction cuts mortality risk 15%" is exactly the move this article exists to warn against. An observed association in one population plus a demonstrated instrument change in another does not equal a demonstrated intervention effect on outcomes.

Why the clocks don't move together

This is the article's spine, and CALERIE is its clearest demonstration. The three families of clock were trained on different things, so they answer different questions, and there is no reason to expect them to respond in unison.

Read that table against the CALERIE result and it stops being surprising. A two-year intervention would not be expected to substantially move a mortality-trained clock heavily loaded with lifetime exposure history. It would plausibly move a rate-of-change measure. That is precisely what happened.

The implication for your report is direct: asking "did my biological age go down" is under-specified. Down on which instrument, measured how, and does that instrument have any business responding to what you did? Two providers using different clocks can hand you genuinely different answers from the same blood draw without either being wrong. The same structural point applies to the newer organ-specific ageing clocks, which fragment the question further.

The interventions with the most credible signal

"Most credible" here is a comparative statement in a field where the absolute standard is low.

Caloric restriction has the only randomised evidence of the kind described above - a single trial, a single instrument, a ~3% effect on pace of ageing, no significant movement on the outcome-trained clocks. It is the strongest thing on this list, and it is still one trial.

Exercise, sleep, smoking cessation and metabolic health have extensive observational associations with slower clock measures and, far more importantly, decades of hard outcome data on morbidity and mortality that does not depend on epigenetic clocks at all. This is the most useful reframe available: the interventions with the best evidence for actually ageing well are well-established on their own merits. Whether they also move a clock is close to irrelevant to the decision. You would do them anyway.

Chronic inflammation management sits in a similar place - mechanistically coherent, associated with clock measures observationally, and independently supported for health outcomes. The inflammaging discussion covers what's established and what isn't.

Note what's happening across that list: the strength of the case tracks the non-clock evidence, not the clock evidence. That's the tell.

The interventions sold for this that have none

Being direct, because the marketing here is aggressive and the buyers are motivated.

The overwhelming majority of supplements and protocols marketed on the promise of lowering biological age have no randomised evidence of moving a validated clock. Not weak evidence - none. What's typically offered instead:

• Single-arm results. A group takes something, clocks are measured before and after, the average drops. With no control arm this cannot distinguish an effect from measurement noise, regression to the mean, seasonal variation, or the other things people change simultaneously when they start caring about their health.
• Customer-reported clock changes. Self-selected, unverified, and drawn from exactly the population most motivated to report a good result. This is testimonial, not data.
• Mechanistic extrapolation. A compound affects a pathway plausibly connected to ageing, therefore it lowers biological age. Where preclinical work is cited, note the species - an effect in mice is a reason to run a human trial, not a substitute for one.
• Association presented as effect. People with higher levels of X have slower clocks, therefore taking X slows your clock. The observational-to-causal leap, made without comment.

None of this is disqualifying evidence that these things don't work. It is an accurate description of the fact that nobody has established that they do. If you are being asked to pay for a clock change, the reasonable question is: which randomised trial, in humans, using which validated clock, showed what effect size?

What a lower number would still not tell you

Suppose you did everything right - a good provider, replicate-averaged measurement, a real change on a real instrument. Some things that result still would not establish:

• That you'll live longer or better. The link from clock movement to outcomes rests on observational associations. No trial has yet shown that changing a clock changes an endpoint that matters.
• Which part of you changed. Methylation clocks pool signal across tissues sampled from blood. A whole-body number doesn't localise anything.
• That it was the intervention. Unless you changed one thing in isolation - and nobody does - attribution is guesswork.
• That it will hold. Two-year effects are two-year effects.
• That the instrument is the territory. A clock is a model fitted to data. It is a useful summary, not a direct readout of ageing itself.

How to use a result sensibly

A biological-age result is genuinely useful - as a baseline and a trend line, interpreted alongside things with established clinical meaning.

Know which clock you bought, and what it was trained on. This determines what your result can and can't tell you, and whether it could plausibly respond to anything you do in the next two years.

Ask about reliability. Replicate averaging or principal-component clock versions materially reduce the noise problem. If a provider can't tell you what they do, that itself is information.

Re-test on a sensible interval, not reactively. Annual is defensible. Quarterly mostly buys noise.

Don't let it displace validated markers. Standard bloodwork, blood pressure, metabolic panel, body composition and functional capacity all have decades of clinical meaning behind them. The clock is one layer on top of that, not a replacement for it.

Interpret it with a clinician who will read it alongside the rest, and who is willing to tell you when the honest answer is that the number doesn't change what you should do.

The bottom line - and where to start

One randomised trial has moved one clock, by about 3%, and left the others unmoved - and the mortality figure attached to that number comes from separate observational data, not from the trial. Measurement noise on these arrays is large enough to explain a lot of what people report as personal success. The interventions with the strongest case are the ones that were already well-supported for reasons having nothing to do with methylation. Almost everything sold specifically to move the number has no randomised evidence that it does.

That is not a reason to skip testing. It's a reason to test properly, understand which instrument you're reading, and spend your money on interpretation and trend rather than on a number you're hoping to buy down.

Epithalon and related peptides: options to discuss with a physician

Some readers arrive here because a peptide has been suggested to them as a way to lower biological age. The same standard this article applies to everything else applies here: none of the compounds below has been shown in a randomised human trial to reverse or lower epigenetic age. If you still want to explore one, Vivre Labs offers access, and we recommend starting with baseline bloodwork.

• Epithalon (epitalon). A short synthetic peptide from the Khavinson research programme in Russia. The telomerase and epigenetic claims made for it rest mostly on small Russian studies and on cell and animal work, and no trial has shown that it lowers a validated epigenetic clock in humans. It is investigational and not FDA-approved. An FDA advisory committee (PCAC) recommended epitalon for the 503A bulks list in July 2026; that vote is advisory only, and it has not been added to the list. The epithalon page sets out its evidence and regulatory status.
• The cellular energy and longevity protocol. This combines SS-31, MOTS-c and epithalon. SS-31 (elamipretide) is FDA-approved only for Barth syndrome, so any other use is off-label. MOTS-c has no completed interventional human trial and is investigational. Neither the combination nor any of its components has been shown to change epigenetic age. Details are on the cellular energy and longevity protocol page.
• NAD+. A coenzyme involved in cellular energy metabolism whose levels fall with age in some tissues. Small human studies show that supplementation can raise NAD+ levels in blood, which is a change in a level, not evidence that epigenetic age or any ageing outcome has changed. See the NAD+ page.

The honest way to look. If you and a physician decide to try any of these, measure before and after rather than relying on how you feel or on one test result. That means a baseline biomarker panel first and, if you want a clock reading as well, the same clock from the same provider, ideally a replicate-averaged or principal-component version, repeated after a sensible interval. Given the measurement noise described above, a single repeat that moves is not evidence that a compound worked, and the physician should read it alongside validated clinical markers.

Explore it with verified supply. Every Vivre Labs lot is independently tested and the certificate of analysis is published, so you know exactly what is in the vial.

Frequently asked questions

Is biological age reversible? The honest answer is that one randomised trial (CALERIE, 220 adults, two years of ~25% caloric restriction) slowed the pace of ageing measured by DunedinPACE by roughly 3%, while producing no significant change on PhenoAge or GrimAge. So a measurement can move. Whether that corresponds to a change in your actual ageing trajectory, and whether it translates to more healthy years, has not been demonstrated by any trial to date.

Why did two biological age tests give me different results? Most likely because they used different clocks, which were trained on different targets - chronological age, mortality risk, or rate of change - and are genuinely measuring different things. On top of that, methylation arrays carry technical variability that for some clocks is comparable in size to the effects people are trying to detect. Disagreement between tests is expected, not a fault in one of them.

Do supplements lower epigenetic age? There is no randomised evidence that supplements marketed for this move a validated epigenetic clock. What's usually presented instead is single-arm before-and-after data with no control group, customer-reported results, or extrapolation from mechanism or from animal work. Those cannot distinguish a real effect from measurement noise and regression to the mean.

If a clock moved, does that mean I'll live longer? Not established. The link between clock measures and mortality comes from observational cohorts - for instance, work associating a roughly 3% slower DunedinPACE with around 15% lower mortality risk. That is an association observed across a population, not a demonstrated consequence of changing your own number. No trial has yet shown that moving a clock changes an outcome.

Related reading

• Epigenetic clocks: how biological age is measured · What a biological age test actually measures · Organ-specific ageing clocks · Does rapamycin slow aging in humans? · Inflammaging · CGMs for non-diabetics · Are GLP-1s anti-aging drugs? · Klotho: the longevity protein
• Part of Testing & Biological Age

---

Educational content, not medical advice. No doses, schedules or dosing instructions appear. Compounds and protocols named in the physician-discussion section are identified with their evidence tier and regulatory status, and none is presented as able to lower epigenetic age. Effect sizes cited are from the primary sources listed below; observational associations are identified as such and are not evidence of an intervention effect. Epigenetic test results should be interpreted by a qualified clinician alongside validated clinical markers. Claims for the Malaysian market are subject to MAB requirements.

References

• CALERIE randomised controlled trial: 220 adults without obesity randomised to approximately 25% caloric restriction or ad-libitum control for 2 years; epigenetic analysis published in Nature Aging, 2023 - intervention slowed pace of ageing as measured by DunedinPACE by roughly 3%, with no significant change in biological-age estimates from first- and second-generation clocks including PhenoAge and GrimAge. · Independent observational cohort work associating approximately 3% slower DunedinPACE with roughly 15% lower mortality risk - an association from separate observational data, not an outcome demonstrated by the CALERIE trial. · Clock training targets: Horvath and Hannum trained on chronological age; PhenoAge and GrimAge trained on mortality and clinical biomarkers; DunedinPACE trained on longitudinal rate of change in organ-system biomarkers. · Technical replicate variability in DNA methylation arrays, and improved reliability of principal-component and replicate-averaged clock implementations. · Absence of randomised trial evidence, as of October 2026, that marketed supplement or peptide interventions, including epithalon, move a validated epigenetic clock. · FDA Pharmacy Compounding Advisory Committee, July 2026: advisory recommendation of epitalon and MOTS-c for the 503A bulks list; not added to the list as of October 2026. · Elamipretide (SS-31): FDA approval September 2025 for Barth syndrome only.