Gender-affirming hormone therapy does not simply change appearance or the endocrine balance: its effects reach as far as the way cells regulate DNA without altering its letters. One open question was whether that impact extends to the so-called epigenetic clocks, the estimators of biological age built from methylation marks, and whether it behaves the same way in trans women and trans men.
A team with US and Australian authors followed twenty-six transgender people —thirteen women and thirteen men, aged 18 to 73— for twelve months while they started hormone treatment in Melbourne, Australia. They analysed their blood DNA before the first dose, at six months and at one year, and published the results in The Yale Journal of Biology and Medicine.
What an epigenetic clock measures
DNA methylation is a chemical tag that attaches to the genome and regulates which genes are switched on. From hundreds of thousands of those marks, the algorithms known as epigenetic clocks calculate a biological age that can run ahead of or behind calendar years. The study used three of the most cited —Horvath, Hannum and PhenoAge—, a methylation-based estimate of telomere length (DNAmTL) and DunedinPACE, an indicator of the pace at which the body ages each year.
The starting picture: mismatched clocks
Before the first dose, the Horvath and Hannum clocks showed a biological age ahead of chronological age, more markedly so in trans men. PhenoAge and DunedinPACE, by contrast, pointed to a slower pace of ageing in both groups. The authors note that this contradiction might reflect the effect of minority stress in an otherwise healthy cohort: accumulated biological wear is not always mirrored equally by every indicator.
How the markers responded to treatment
The central result calls for caution: over the twelve months studied, hormone therapy did not significantly alter the three classic clocks. The differences appeared in the dynamic indicators. In trans women there was a modest rise in DunedinPACE —a slightly faster pace of ageing— and a minimal change in estimated telomere length. In trans men, DunedinPACE stayed stable or fell slightly, while the telomere-length indicator dropped significantly.
The authors themselves frame the finding in its biological context: estradiol acts as a telomerase activator, and testosterone influences blood cell proliferation and telomere attrition. Different treatments can push these markers in opposite directions without that implying harm in itself.
A lot of variation between individuals
The most uncomfortable detail for any general conclusion is the spread: individual trajectories were so different that the authors prefer to speak of an individualised response. Two people on the same treatment, of the same age and with a similar starting point did not move the same way during the year of follow-up.
The limits, in the authors’ own words
Twenty-six participants and twelve months are small numbers, and the cohort’s hormone regimens were heterogeneous. The paper itself describes its results as preliminary and hypothesis-generating, and calls for 24- to 36-month follow-ups to learn whether these changes stabilise, progress or reverse after the initiation phase of treatment.
Why it matters
Gender-affirming hormone therapy is a long-term treatment, in many cases lifelong, and research into its long-term effects remains scarce. Work like this does not answer whether hormones make people age more or less: it shows that the biology of ageing can be measured in this context and that it responds unevenly depending on the type of treatment and on the individual. That map, still blurry, is the starting point for larger studies.
Source: Tracking Epigenetic Biomarkers of Health and Aging During the Initial Year of Gender-Affirming Hormone Therapy. Mozhui K, Henriksen BA, Bell LA, Bretherton I, Cheung AS, Novakovic B. The Yale Journal of Biology and Medicine 2025;98(2):105-115. DOI: 10.59249/xmgo7523.
Read the publication: The Yale Journal of Biology and Medicine.
Article published on dosier.es. Image generated with artificial intelligence (marta ai → marta.syf.es).
