An experimental fibrosis drug shifted six protein-based aging clocks toward younger scores. The convergence is unusual and worth attention, but biomarkers are not the same as longer life or better healthspan.
What would count as evidence that a drug slows human aging? A longer life would be persuasive, but waiting decades is not a workable strategy for early clinical trials. Fewer diseases, preserved function and more years in good health would also matter, yet those outcomes still take time.
Researchers therefore look for earlier signals. In a new Nature Biotechnology study, one such signal appeared repeatedly: six different proteomic aging clocks all estimated a lower biological age after patients received rentosertib, an experimental drug designed with artificial intelligence.
That result is interesting. It is not proof of age reversal.
A drug developed for pulmonary fibrosis
Rentosertib, previously known as INS018_055, is a small-molecule inhibitor of TNIK, a signaling protein implicated in fibrosis and several biological processes associated with aging. Insilico Medicine used AI systems to help identify the target and design the molecule.
The drug was not tested as a general anti-aging treatment. It was tested in idiopathic pulmonary fibrosis, or IPF, a serious disease in which scar tissue progressively stiffens the lungs and makes breathing harder. IPF is strongly associated with older age and overlaps with processes studied in geroscience, including cellular senescence, altered immune signaling and disrupted tissue repair.
The underlying Phase 2a trial was randomized, double-blind and placebo-controlled. Phase 2a studies are relatively early human trials designed mainly to look for signs that a drug is biologically active, explore dosing and continue assessing safety. They are not usually large or long enough to prove that a treatment improves survival.
Seventy-one adults with IPF entered the main 12-week trial at sites in China. For the proteomic substudy, the researchers analyzed usable serum samples from 42 participants, whose mean age was 67.1. Participants received placebo or one of three rentosertib regimens: 30 milligrams once daily, 30 milligrams twice daily or 60 milligrams once daily.
What six proteomic clocks measured
Proteomics is the large-scale study of proteins. Proteins carry out much of the body’s day-to-day work, and their concentrations can change with disease, inflammation, metabolism, treatment and age.
A proteomic aging clock is a statistical model trained to recognize patterns of blood proteins associated with chronological age, mortality risk or other aging-related outcomes. Feed a person’s protein measurements into the model and it produces an estimate often described as biological age.
Biological age is not a second birthday hidden in the blood. It is a model-based score. A person can receive different estimates from different clocks because each model uses different proteins, training populations and outcomes. A younger score may be encouraging, but it does not automatically mean tissues have been rejuvenated or that future disease risk has fallen.
The researchers measured 2,841 serum proteins at baseline and at weeks 2, 4 and 12. They then applied six published clocks: ProtAge, two OrganAge variants, PAC, ipfP3GPT and PAOPAC.
All six clocks moved in the same direction
Across the treated groups, all six models estimated lower biological age relative to baseline, while the placebo group showed little change or slight increases. Of the many comparisons across clocks, doses and time points, 21 met the study’s statistical threshold after correction for multiple testing.
The clearest convergence appeared at week 4. Eleven of 18 comparisons at that time point showed significantly lower changes in predicted biological age than placebo. The 30 milligram twice-daily regimen produced the broadest signal, with nine significant comparisons overall. At week 4, five of the six clocks pointed in the same direction for that regimen.
The pattern was not identical across dose groups. The 60 milligram once-daily group had the largest improvement in forced vital capacity, a standard measure of lung function, in the original trial. Yet the 30 milligram twice-daily group produced the most consistent aging-clock response. Changes in lung function explained little of the variation in clock scores, with a median R-squared of 0.06 across the six clocks.
Researchers also found treatment-related changes in 326 proteins. Some involved fibrosis and extracellular matrix remodeling. Others mapped to metabolism, stress resistance, growth-factor signaling and cellular senescence. In the twice-daily group, drug-related protein changes were modestly opposed to age-associated patterns observed in more than 55,000 UK Biobank participants.
Together, those findings make a simple clock malfunction less likely. They raise the plausible possibility that rentosertib affected biology shared by pulmonary fibrosis and aging, and perhaps some processes beyond the lung disease itself.
Why younger scores are not proof of slower aging
The central limitation is that the clocks measured proteins in people with an active age-related disease. Treating pulmonary fibrosis can change inflammation, tissue remodeling and circulating proteins. Many of those same proteins also contribute to aging-clock predictions. The clocks may therefore be registering disease improvement, a direct drug effect, broader aging biology or a mixture of all three.
Even agreement among six clocks cannot fully solve that problem. The models are different, but they draw from overlapping biological signals and the same serum samples. One protein, LTBP2, is involved in fibrosis and appeared as an important contributor in all six clocks.
The study was also small and short. Only 42 participants were included in the proteomic analysis, with nine to 11 people in each group at individual time points. The strongest clock signal peaked at week 4 and then plateaued by week 12. The statistical threshold was less stringent than the conventional 0.05 cutoff for many analyses, and some pathway analyses used an exploratory false-discovery threshold of 0.25.
Most importantly, the study did not show that participants lived longer, avoided other diseases, remained independent or experienced a lasting improvement in healthspan. Those are clinical aging outcomes. A biomarker moving in a favorable direction is association and measurement, not yet evidence of a meaningful geroprotective effect.
The paper itself acknowledges that the modest sample, short duration, computational emphasis and lack of complementary omics measurements prevented a clean separation of anti-fibrotic and anti-aging effects. Several authors are employees of Insilico Medicine, and the company’s founder and chief executive is an author. Independent replication will matter.
A useful blueprint for future longevity trials
The larger contribution may be methodological. Aging is not currently a conventional disease indication, which makes direct drug trials difficult to design and regulate. Disease trials already enroll people, assign treatments, collect safety data and measure clinical outcomes. Adding carefully selected aging biomarkers could reveal potential geroprotective signals much earlier in development.
The right sequence is important. First, include aging measures prospectively as exploratory endpoints. Next, replicate the signal in other diseases or in older people without severe fibrosis. Then connect biomarker changes to functional outcomes, disease incidence or validated surrogate endpoints. Multiple kinds of clocks, tissue measurements and longer follow-up would make the case stronger.
That approach does not turn every disease drug into a longevity therapy. It creates a disciplined way to decide which candidates deserve more rigorous testing.
Why it matters
Six independent proteomic clocks responding in the same direction inside a controlled human trial is an unusual signal. It suggests that conventional drug studies can be designed to look for aging-related effects without pretending that a biomarker is already a clinical benefit.
Bottom line
Rentosertib changed blood-protein patterns in ways that six aging clocks interpreted as younger. That is a promising research finding, not evidence that the drug reversed human aging, extended life or improved long-term healthspan. The next test is whether the signal can be reproduced, separated from pulmonary fibrosis biology and tied to outcomes people actually feel.
Evidence level: Promising
Story type: Human Phase 2a biomarker analysis
Primary source: Integration of proteomic aging clocks in a phase 2a clinical trial supports simultaneous geroprotective assessment, Nature Biotechnology, 2026.
Image: AI-generated conceptual illustration of proteomic age measurement and drug intervention.