📌 Key Takeaways
- Yale researchers published TranslAGE in Nature Medicine (August 21, 2026),
harmonizing 51 human intervention studies and 3,128 samples
to compare 16 epigenetic clocks and 94 DNA methylation biomarkers. - DunedinPACE showed the largest effect size (mean −0.0891, p = 0.0012);
PCGrimAge showed the strongest statistical significance (mean −0.07843, p = 0.0003). - Pharmacological interventions — especially anti-TNF therapy and metformin —
produced consistently larger clock changes than lifestyle or supplements,
though most pharmacological studies enrolled disease populations, not healthy adults. - A change in epigenetic clock score does not prove rejuvenation.
These clocks are best understood as tracking tools, not outcome endpoints.
One of the most persistent questions in longevity science has been deceptively simple:
Which aging clock should we actually use?
On August 21, 2026, a team led by Raghav Sehgal and Albert Higgins-Chen at Yale University
published a landmark answer in Nature Medicine.
By constructing TranslAGE — a harmonized database of 51 human intervention studies
and 3,128 samples — they systematically compared 16 epigenetic clocks
and 94 DNA methylation biomarkers across four categories of intervention.
The findings point clearly to two clocks: DunedinPACE for effect size,
and PCGrimAge for statistical robustness.
They also reveal a hierarchy of interventions
that challenges some popular assumptions in the anti-aging space.
Epigenetic clock: A biomarker that estimates biological age
from DNA methylation patterns at CpG sites.
Examples include DunedinPACE, PCGrimAge, and PhenoAge.
TranslAGE: A harmonized database built by Yale,
integrating 51 human intervention studies
to compare how different clocks respond to different interventions.
MCID (Minimal Clinically Important Difference):
The threshold of change considered clinically meaningful.
For epigenetic clocks, this threshold has not yet been established.
What Are Epigenetic Clocks — and Why Is Measuring Aging So Hard?
Chronological age tells you how many years you have lived.
Biological age attempts to measure how your body has actually aged.
Epigenetic clocks estimate the latter
by analyzing methylation patterns across thousands of CpG sites in the genome.
These patterns shift predictably with age —
but they also respond to disease, diet, stress, and medical intervention.
That responsiveness is precisely what makes them useful for intervention research,
and precisely what makes them difficult to interpret.
First-generation clocks (Horvath 2013, Hannum 2013):
Trained to predict chronological age. High accuracy, low sensitivity to intervention.
Second-generation clocks (PhenoAge, GrimAge, DunedinPACE):
Trained on health outcomes — mortality risk, disease burden, pace of aging.
More sensitive to intervention, more clinically relevant.
TranslAGE processed all 16 clocks through a unified pipeline,
correcting for chronological age,
to evaluate which clocks respond most reliably to human interventions.
What Is TranslAGE — and Why Does the Methodology Matter?
Before TranslAGE, the literature was fragmented.
Study A might show that intervention X changed clock Y —
but Study B used a different clock, different preprocessing, and different populations.
Cross-study comparison was essentially impossible.
TranslAGE solved this by pulling data from public repositories (GEO, EMBL)
and private sources (TruDiagnostic),
then running every sample through the same computational pipeline.
The analytical approach used paired t-tests for pre/post comparisons
and generated effect size heatmaps across intervention-by-biomarker combinations.
Chronological age was regressed out to isolate biological aging signals.
DunedinPACE vs. PCGrimAge — Two Clocks, Two Strengths
The study did not crown a single winner.
Instead, it identified two clocks that are best understood as complementary.
Significant decrease in 16 interventions;
increase in only 1.
Highly sensitive to lifestyle interventions.
Responds even in healthy populations.
Highly sensitive to pharmacological interventions.
Stronger response in disease populations.
Closely linked to mortality prediction.
DunedinPACE measures the pace of aging —
how fast biological aging is occurring at the moment of measurement.
It is particularly sensitive to lifestyle changes
and responds in healthy individuals, making it useful for broad intervention tracking.
PCGrimAge is more tightly linked to mortality risk
and responds more strongly to pharmacological interventions,
especially in populations with existing disease.
The paper recommends that future clinical trials prioritize both clocks.
Neither clock is a definitive endpoint.
The paper recommends focusing on these clocks in future trials —
but that is not the same as saying a lower score proves rejuvenation.
For epigenetic clocks to function as validated surrogate endpoints,
their changes must first be linked to long-term health outcomes
(longevity, disease prevention) through extended follow-up studies.
That evidence does not yet exist at scale.
Pharmacological Interventions Led — But Context Is Everything
Across intervention categories, pharmacological interventions
showed significantly larger effect sizes than all other categories
(vs. lifestyle: p = 0.0063).
Anti-TNF therapy and metformin were the standout performers,
showing consistent, directionally aligned changes across multiple studies.
Anti-TNF (arthritis, IBD) and metformin (diabetes)
showed consistent responses across multiple studies.
Mediterranean diet also showed directional consistency.
but high consistency across studies.
Mediterranean diet showed strong reproducibility.
DunedinPACE was particularly sensitive here.
Low consistency across studies —
some showed clocks moving in opposite directions
for the same supplement.
System-specific changes observed,
but overall effect size smaller than pharmacological.
The finding that pharmacological interventions moved epigenetic clocks more
does not mean drugs are more effective at reversing aging.
Three critical caveats apply:
① Anti-TNF studies enrolled patients with arthritis and IBD —
disease populations where intervention effects are amplified
by regression toward a healthier baseline.
② Metformin is a diabetes drug.
Its use in healthy adults for longevity purposes
is not supported by this data.
③ Clock changes have not been causally linked to longer lifespan
in long-term follow-up studies.
“Pharmacological > supplements” describes a pattern in clock responsiveness —
not a prescription for healthy people to take medication.
Five Limitations the Authors Themselves Acknowledge
disease status, intervention type, and duration.
Pooling them enables comparison but limits causal interpretation.
differed across source studies,
potentially affecting cross-study comparability.
were trained on intermediate biomarkers like CRP,
meaning they may reflect inflammation and metabolic state
rather than aging per se.
how much a clock score must change to be clinically meaningful.
An effect size of 0.2 cannot yet be translated into
“X years of biological age reversed.”
are driven largely by disease cohorts (arthritis, IBD).
Effect sizes in healthy older or younger adults were comparatively smaller.
Implications for Aesthetic Medicine and Longevity Clinics
① Choosing an epigenetic test for clinical use:
For longevity clinics and anti-aging practices considering epigenetic testing,
DunedinPACE and PCGrimAge now have the strongest evidence base
for tracking intervention response over time.
② Skepticism toward supplement-based “age reversal” claims:
Supplement interventions showed low cross-study consistency —
with some studies showing clocks moving in opposite directions
for the same compound.
Claims that a supplement will “reduce your epigenetic age by X years”
are not supported by this body of evidence.
③ Lifestyle interventions remain the most reproducible:
Mediterranean diet showed consistent directional effects across multiple studies.
The effect size is modest, but the reproducibility is high —
which matters more for long-term clinical guidance
than a single dramatic result.
The field has long struggled with too many clocks
and no clear guidance on which to trust.
This study gives a practical answer:
anchor your tracking to DunedinPACE and PCGrimAge.
But use them as tools to monitor intervention response —
not as proof that aging has been reversed.
For clinic operators designing longevity programs,
the takeaway is clear:
if you are incorporating epigenetic testing,
build your protocol around DunedinPACE and PCGrimAge
as pre/post tracking instruments.
The honest framing — “we are tracking your pace of aging,
not proving you are younger” —
is both more accurate and more likely
to build lasting patient trust.
Summary
- Yale’s Sehgal, Higgins-Chen et al. published “Responsiveness of epigenetic aging biomarkers
to longevity interventions in humans” in Nature Medicine on August 21, 2026
(DOI: 10.1038/s41591-026-04562-9).
The TranslAGE database harmonized 51 studies and 3,128 samples
to compare 16 epigenetic clocks and 94 DNA methylation biomarkers. - Second-generation clocks outperformed first-generation clocks in intervention responsiveness.
DunedinPACE led on effect size (mean −0.0891, p = 0.0012);
PCGrimAge led on statistical significance (mean −0.07843, p = 0.0003).
The authors recommend both as priorities for future clinical trials. - Pharmacological interventions (anti-TNF, metformin) showed the largest effect sizes —
but this reflects the disease populations studied,
not a recommendation for healthy adults to use these drugs.
Supplements showed low cross-study consistency.
Mediterranean diet showed high reproducibility. - Key limitations: study heterogeneity, preprocessing inconsistency,
no established MCID, and larger effects in disease vs. healthy populations.
A change in clock score indicates intervention responsiveness —
not proven rejuvenation.
Validation as surrogate endpoints requires long-term outcome data. (NERO editorial note)
integrating 51 human intervention studies and 3,128 samples
to compare 16 epigenetic clocks under a unified analytical pipeline.
Before TranslAGE, cross-study comparison of aging clocks was nearly impossible
due to inconsistent methods and different clock choices.
This database enables, for the first time, a systematic ranking
of which clocks respond most reliably to which interventions.
DunedinPACE showed the largest effect size across interventions,
while PCGrimAge showed the strongest statistical significance.
The authors recommend both as the primary clocks
for future longevity clinical trials.
Neither should be interpreted as a definitive proof of rejuvenation —
they are best used as tracking instruments for intervention response.
are largely explained by the study populations involved —
patients with arthritis, IBD, and diabetes,
where interventions correct disease-driven biological aging signals.
Metformin is a diabetes medication;
anti-TNF therapy is used for autoimmune disease.
This study does not support their use in healthy adults for longevity purposes.
The supplement findings (low consistency) also do not support
specific supplement recommendations.
Source: Sehgal R, Borrus D, Armstrong JF, et al. “Responsiveness of epigenetic aging biomarkers to longevity interventions in humans.” Nature Medicine. 2026;32:3477–3490. doi:10.1038/s41591-026-04562-9 (Full text directly verified by NERO editorial team)
