Six proteomic clocks detect a younger biological age signal in an IPF trial – but separating geroprotection from treatment remains key.
Six proteomic clocks, six different research pedigrees, one direction of travel: that is the headline finding of a new study in Nature Biotechnology, and it is arresting enough on its own terms. The drug in question is rentosertib, an AI-designed inhibitor of TNIK (TRAF2- and NCK-interacting kinase) being developed for idiopathic pulmonary fibrosis; the researchers tracked 2,841 proteins in longitudinal serum samples from the 42 participants of its Phase 2a trial, then set half a dozen independently built aging clocks loose on the data to see what they made of it [1].
What makes rentosertib worth watching is less the readout than the philosophy behind it. Insilico Medicine‘s AI platform flagged TNIK as a target sitting at the intersection of fibrosis and six of the recognized hallmarks of aging – rather an efficient piece of biological real estate – and the company built that dual relevance into the drug from the start, shaping target selection, preclinical work, indication choice and trial design around aging biology rather than bolting geroscience on after the fact. This new analysis is the clinical test of that premise: can an ordinary disease trial, run for entirely conventional regulatory reasons, also double as evidence for whether a drug modifies aging itself?
Longevity.Technology: There is something compelling about the way this study was constructed; six proteomic clocks, built differently and trained to look at aging from different angles, all ended up pointing broadly in the same direction. Convergence is not proof, but in a field where clocks have a habit of disagreeing with one another (sometimes quite vociferously), it is certainly more interesting than a single flattering readout. The authors are appropriately careful about the central ambiguity – whether rentosertib is modifying aging biology itself or simply making the proteome of people with IPF look younger as their disease improves – and that distinction matters enormously. Yet perhaps the more consequential idea here is not the headline-grabbing suggestion of biological age reversal, but the architecture underneath it: build aging biology into target selection, indication choice and clinical development from the outset, then use conventional disease trials to ask a second question about geroprotection at the same time. If that dual-purpose model works, it could offer geroscience a pragmatic route into the clinic without waiting for aging itself to become a neatly approvable indication – regulators do, after all, tend to prefer diseases they can put in a box. But the next experiment may matter almost as much as this one: reproduce the signal in a population without IPF, where improvement in pulmonary fibrosis can no longer muddy the waters, and the case for genuine geroprotection becomes considerably harder to wave away.
Six clocks, one direction
The six clocks were not variations on precisely the same algorithm. They included ProtAge, ipfP3GPT, PAOPAC, PAC and chronological and mortality variants of OrganAge, spanning classical machine learning and deep learning models trained against different outcomes. Despite those differences, each detected treatment-associated age reduction; the most consistent response occurred at week four with the 30 mg twice-daily regimen, where five of the six clocks showed negative effect sizes [1].
That agreement caught the attention of Nobel laureate Michael Levitt, who said: “What convinces me is not the size of the effect but the agreement [2].” It is an important distinction. Proteomic clocks infer aspects of biological aging from circulating proteins; they do not establish that a person has literally become several years younger. It’s worth noting here that while the clocks were independently built, they are not independently interested; four of the five clock-development teams behind these six models – including Insilico’s own ipfP3GPT – are co-authors on the paper itself. Only PAC, developed independently at the University of Connecticut, sits fully outside the author list.
The dose response adds another wrinkle. In the original trial, the largest improvement in forced vital capacity (FVC) occurred with 60 mg once daily, whereas the most consistent clock response was seen with 30 mg twice daily. Across the six clocks, changes in FVC explained little of the variance in predicted biological age, with a median R² of 0.06. The authors therefore argue against a simple disease-improvement explanation – while acknowledging that FVC is an incomplete proxy for overall disease status [1].
Following the proteins
The proteome offered clues beyond the clocks. Proteins uniquely affected by twice-daily dosing were enriched in pentose phosphate, glutathione and cholesterol metabolism, while the 60 mg once-daily regimen produced stronger signatures involving Wnt signaling and immune activity more closely associated with fibrosis. Treated groups also showed circulating protein patterns consistent with a senomorphic effect, including changes across established senescence signatures [1].
Interesting, yes. Definitive, no. As the authors put it, “proteomic clocks cannot answer” on their own whether the apparent age reduction represents aging modulation or a consequence of anti-fibrotic activity [1]. The study is small, short and computationally heavy; complementary omics, direct measurements of senescent-cell burden and biomarkers tied prospectively to functional outcomes will be needed to make the biological interpretation sturdier.
A dual-purpose route
The broader proposition is to make such measurements routine exploratory passengers in trials of drugs for age-related disease, rather than waiting until an approved medicine attracts geroscience interest years later. The authors propose a staged route: collect aging and senescence biomarkers prospectively, replicate any signal in populations where IPF improvement cannot drive it, then pursue qualified biomarkers or composite clinical endpoints capable of supporting geroprotective claims [1].
Insilico founder Alex Zhavoronkov sees a much larger prize, describing “the social and economic opportunity for drugs reversing biological age” as “vast” [2]. He is not a man given to modest arithmetic: three added years of life expectancy, scaled across the global population, would by his own reckoning amount to something like 25 billion additional life years – more, he noted, than humanity has lost across every war it has ever fought [2]. It is a big claim to hang on 42 patients and a median R² of 0.06 – but big claims are what get drugs out of Phase 2a and into Phase 3, which is precisely where rentosertib now sits, having entered Phase 3 development for IPF in China.
When the clocks leave the lung
For now, the clocks have raised a question rather than settled one. Testing rentosertib beyond severe pulmonary fibrosis should reveal whether its younger proteomic signature travels with the drug – or whether it belongs, at least partly, to the recovering lung. Either answer would be useful; geroscience advances by learning what its biomarkers are actually measuring.
Photographs courtesy of Alex Zhavoronkov
[1] https://www.nature.com/articles/s41587-026-03286-y
[2] https://www.prnewswire.com/news-releases/nature-biotechnology–insilicos-ai-driven-ipf-candidate-rentosertib-shows-potential-for-biological-age-reversal-as-assessed-by-six-proteomic-aging-clocks-302871277.html
