Senolytics: two randomised trials, and no confirmed clinical benefit yet

The senescence theory of ageing is easy to state and hard to test. Cells that have stopped dividing but refuse to die accumulate with age, secrete inflammatory signals, and — in mice — clearing them delays multiple age-related conditions. Drugs that clear them are called senolytics, and two of them have now been through randomised trials in people with results published in major journals.

Neither trial produced the result the theory predicts. That is worth reading in detail, because senolytic supplements are already sold on the strength of the mouse literature.

The bone trial

A phase 2 randomised controlled trial published in Nature Medicine tested the senolytic combination dasatinib plus quercetin, given intermittently, in 60 postmenopausal women [s1].

The primary endpoint was the percentage change at 20 weeks in C-terminal telopeptide of type 1 collagen, or CTx — the standard marker of bone resorption. It did not differ between groups. The median change was −4.1% on dasatinib plus quercetin, with an interquartile range of −13.2 to 2.6, against −7.7% in the control group, interquartile range −20.1 to 14.3, at P = 0.611 [s1]. The control group's bone resorption fell more, though the intervals overlap so completely that neither direction means anything.

The secondary endpoint moved and then stopped moving. The bone formation marker P1NP rose significantly relative to control at 2 weeks (+16%, P = 0.020) and at 4 weeks (+16%, P = 0.024), but by 20 weeks it was not different from control (−9%, P = 0.149) [s1]. No serious adverse events occurred [s1].

The interesting result is exploratory and should be read as a hypothesis rather than a finding. In the women with the highest senescent cell burden — the top tertile for T cell p16 messenger RNA — dasatinib plus quercetin both raised P1NP (+34%, P = 0.035) and lowered CTx (−11%, P = 0.049) at 2 weeks, and increased radius bone mineral density by 2.7% at 20 weeks (P = 0.004) [s1]. The authors' own conclusion is that intermittent treatment did not reduce bone resorption in the overall group, and that further studies are needed to test whether underlying senescent cell burden dictates the response [s1].

That is a coherent explanation for a null result. It is also the explanation that would be offered whether or not it were true, which is why it needs a trial designed around it rather than a subgroup found inside one.

The eye trial

The second randomised senolytic result comes from a different drug and a different disease. UBX1325, also called foselutoclax, inhibits the anti-apoptotic protein BCL-xL and was given as a single intravitreal injection of 10 μg to people with diabetic macular oedema who had responded poorly to anti-VEGF treatment [s2].

Sixty-five participants were randomly assigned, 32 to UBX1325 and 33 to sham, and followed for up to 48 weeks [s2]. The trial's stated primary objective was safety, and on that measure there was little to separate the groups: four grade 3 or higher treatment-emergent adverse events in the sham group, three of them serious, against five serious events of grade 3 or higher on UBX1325 [s2]. No adverse events led to discontinuation [s2].

Efficacy was a secondary objective. The difference between UBX1325 and sham in mean change in best corrected visual acuity to week 48 was 5.6 ETDRS letters, with a 95% confidence interval of −1.5 to 12.7 [s2]. Five and a half letters would matter to a patient. An interval running from slightly worse to substantially better does not establish that the drug delivered it. The investigators describe the findings as trends suggestive of potential efficacy requiring larger trials [s2].

The biomarker problem underneath both

A 2026 study in Aging Cell asked a prior question: can DNA methylation even detect senescent cell clearance? Existing epigenetic clocks have shown inconsistent associations with cellular senescence and no reduction after senolytic treatment, so the authors built clocks specifically on the senescence signal [s3].

They found that senescence, age and mortality risk intersect at only a small fraction of the methylome — 9,363 CpG sites out of 396,333 analysed, or 2.4% [s3]. They then trained three predictors on those sites, one each for in vitro senescence, age and mortality. All three stayed the same or accelerated after senolytic treatment, in vivo and in vitro alike [s3]. The authors conclude that this calls into question whether cellular senescence can be captured by DNA methylation at all, and challenges the assumption that ageing biomarkers should fall after geroscience interventions [s3].

That last point deserves emphasis. Trials of ageing interventions increasingly report epigenetic clock movement as evidence of effect. If the clocks do not respond to a treatment that demonstrably kills senescent cells, then a clock reading is not a general-purpose measure of whether an intervention worked.

Where this leaves things

Two randomised trials, both small, both in specific diseases rather than in healthy ageing. One missed its primary endpoint; the other met a safety objective and produced an efficacy signal too imprecise to interpret. Neither is a failure of the senescence hypothesis, and neither is support for taking a senolytic.

Several authors of the bone trial hold patents on senolytic drugs through their institution, a conflict the paper discloses and which was reviewed by its conflict-of-interest board [s1].

Dasatinib is a prescription cancer drug. Nothing in this article is medical advice, and no dose or regimen described here should be read as guidance.

Sources

1. [s1] Effects of intermittent senolytic therapy on bone metabolism in postmenopausal women: a phase 2 randomized controlled trial. Nature Medicine, published online July 2, 2024. https://doi.org/10.1038/s41591-024-03096-2
2. [s2] Safety and Efficacy of Senolytic UBX1325 in Diabetic Macular Edema. NEJM Evidence, published online April 22, 2025. https://doi.org/10.1056/EVIDoa2400009
3. [s3] DNA Methylation Signatures of Cellular Senescence Are Not Reversed by Senolytic Treatment. Aging Cell, published online February 2026. https://doi.org/10.1111/acel.70430