Do Sex Hormones Slow Epigenetic Aging? The Best Test Says No

Six epigenetic clocks failed to find a consistent hormone-aging signal

Higher estradiol, DHEAS, or testosterone-related measures did not consistently predict slower epigenetic aging in older adults. That is the central result of a 2025 Berlin Aging Study II analysis, the most comprehensive direct test currently available.

The study examined 1,404 community-dwelling older adults, with a mean age of 68.7 years, across 7.3 years of follow-up. Researchers tested estradiol, DHEAS, and the Free Androgen Index against six DNA-methylation clocks: Horvath, Hannum, 7-CpG, PhenoAge, GrimAge, and DunedinPACE. These clocks do not all measure the same thing. Some estimate deviation from chronological age, while GrimAge and PhenoAge were trained against health-related outcomes, and DunedinPACE estimates the current pace of physiological aging.

If circulating sex hormones had a meaningful, broadly detectable relationship with epigenetic aging, this design gave researchers several chances to find it. Instead, the results varied by clock, sex, and time point. No association survived correction for multiple comparisons. The authors concluded that sex hormones play, at most, a minor role in epigenetic biological aging in this population.

That does not prove hormones have no effect on health, function, bone density, sexual well-being, or symptoms of hormone deficiency. It does mean that "optimize your hormones to slow your epigenetic clock" remains a hypothesis, not an established clinical result.

What BASE-II actually found

Several uncorrected associations reached the conventional nominal threshold of P<0.05. None held up after the researchers accounted for the many comparisons they made across six clocks, three hormone measures, sex-specific analyses, and time points.

• In men at baseline, higher DHEAS was associated with lower Horvath DNA-methylation age acceleration.
• In women at baseline, higher estradiol was associated with a lower DunedinPACE.
• In women at follow-up, higher estradiol was associated with higher GrimAge acceleration.
• Across both sexes longitudinally, a larger decline in DHEAS was associated with an increase in DunedinPACE.

The estradiol findings show why nominal results deserve restraint. In the same population of women, higher estradiol correlated with a slower pace-of-aging measure at baseline but with greater GrimAge acceleration at follow-up. Those directions conflict. A simple story in which more estrogen reliably slows epigenetic aging cannot explain both findings.

With enough statistical tests, a few P values below 0.05 will appear by chance even when no real relationship exists. Multiple-comparison correction addresses that problem. BASE-II's nominal hormone-clock associations did not survive it, so they should not guide treatment decisions.

DHEAS is interesting, but the study does not establish cause

DHEAS deserves attention because its decline with age is well documented, and it produced the most biologically intuitive nominal findings in BASE-II. Men with higher DHEAS had lower Horvath age acceleration at baseline, and people whose DHEAS changed less unfavorably over time also showed more favorable changes in DunedinPACE.

But an observational association cannot tell us what caused what. Better metabolic health, lower inflammation, less visceral fat, and greater physical fitness could support both higher DHEAS and slower biological aging. DHEAS may track those conditions without driving the clock itself. The longitudinal result does not settle that question; two measures can change together because they share an upstream cause.

This distinction matters because a blood test is not a mechanism. A low DHEAS result may identify someone with illness, frailty, or other risk factors. It does not show that raising DHEAS with a supplement will reverse the process that produced the result.

Epigenetic clocks can respond to interventions

The null hormone pattern is informative partly because DNA-methylation aging measures are not frozen readouts. In the DIRECT PLUS randomized trial, researchers assessed several clocks and DunedinPACE in 256 people with abdominal obesity or dyslipidemia before and after an 18-month dietary intervention. The trial reported attenuation of DNA-methylation-assessed biological aging with higher polyphenol intake.

Similarly, the MACRO weight-loss trial followed 144 adults with obesity through a 12-month low-carbohydrate versus low-fat dietary intervention. It evaluated DunedinPACE, PCPhenoAge acceleration, and PCGrimAge acceleration alongside cardiometabolic biomarkers. The study reported improvement in epigenetic age acceleration during weight-loss dietary intervention, although the supplied results do not provide a treatment-effect estimate or confidence interval for a specific clock outcome.

These trials do not prove that every clock change translates into longer life or fewer diseases. Epigenetic clocks are biomarkers, not patient outcomes. Still, they show that the measures can move in response to modifiable exposures. BASE-II's lack of a repeatable sex-hormone signal therefore cannot be dismissed as a failure of the clocks to detect any change at all.

Sex differences do not prove a hormone prescription

Men and women differ, on average, in biological-aging trajectories. A 2024 study of older adults examined sex differences in biological aging and their associations with clinical measures. Such differences make sex steroids a reasonable scientific question, but they do not identify estradiol, DHEAS, or testosterone as the cause.

Sex is a bundle of biology and lived exposure: chromosomes, reproductive history, body composition, medication use, disease patterns, smoking, alcohol, social conditions, and more. Within-person variation in a single hormone cannot be assumed to account for population-level differences between men and women.

The broader literature also resists a single-hormone explanation. A systematic review of factors associated with epigenetic clock acceleration found substantial heterogeneity across clocks and cohorts. Smoking, obesity, alcohol intake, physical inactivity, and socioeconomic adversity have stronger and more recurrent support than sex hormones. Clock-specific results are common, which is a reason to avoid treating one favorable clock result as a general biological-aging verdict.

Important limits of the study

BASE-II was larger and more thorough than most studies in this niche, but it was still observational. It cannot separate direct hormone effects from shared causes. It also enrolled community-dwelling adults who had reached older age, so severely frail or seriously ill people may be underrepresented. That could narrow hormone variation and obscure relationships at the most extreme levels of depletion.

Hormone therapy creates another practical complication. Treatment use can correlate with symptoms, access to care, reproductive history, and baseline health. Those factors can confound comparisons between people with different hormone levels. The available findings do not establish whether hormone therapy changes epigenetic clocks.

There is also no randomized trial evidence with epigenetic clocks as primary endpoints for testosterone replacement, DHEA supplementation, or estrogen therapy. Large testosterone trials may answer other safety and clinical questions, but they do not answer this specific one without pre-specified DNA-methylation outcomes. Likewise, no Mendelian-randomization study in the available evidence tests whether genetically predicted sex-hormone levels cause changes in epigenetic aging.

Other apparently plausible aging links can also disappear under careful testing. For example, a longitudinal study of 831 trauma-exposed middle-aged women found no association between PTSD and six epigenetic-aging measures, despite PTSD's established links with mortality and age-related disease. Epigenetic clocks capture meaningful risk patterns, but they do not automatically register every condition associated with poorer health.

What would make the claim actionable?

A decisive test would randomly assign healthy older adults to DHEA, testosterone, estrogen-based treatment where clinically appropriate, or placebo; pre-register epigenetic age acceleration as a primary endpoint; measure several clocks; and follow participants long enough to distinguish a durable effect from ordinary assay variation. A trial of at least 500 participants would be a meaningful starting point. It would also need clinical outcomes and adverse-event monitoring, because a better biomarker score would not outweigh treatment harms.

Until those data arrive, use hormone treatment for evidence-based indications with a qualified clinician, not as an epigenetic-age intervention. If your goal is to improve health factors that repeatedly associate with slower epigenetic aging, the current evidence points more credibly toward body weight when relevant, diet quality, physical activity, smoking cessation, and lower alcohol exposure. Hormones may yet earn a place on that list. BASE-II does not put them there.