Insilico Medicine, the company behind rentosertib, one of the first drugs whose target and molecule were both found with generative AI, has published an analysis suggesting the drug shifted blood markers of biological age in a small trial of patients with idiopathic pulmonary fibrosis (IPF). The paper appeared in Nature Biotechnology on 7 September 20261. It applied six separate proteomic aging clocks to serum samples collected during a 12-week phase 2a trial, and all six predicted a lower biological age in treated patients than in those on placebo13.
The authors themselves say the design cannot separate an effect on aging from an effect on the lung disease being treated15. That caveat sits at the center of how to read the result.
What happened
Rentosertib is an oral small molecule that blocks an enzyme called TNIK. Insilico says its AI platform picked TNIK as a target linked to both fibrosis and aging biology, and its generative chemistry software then designed the molecule3. The drug was first tested for safety in a randomized phase 2a trial (NCT05938920) reported in Nature Medicine in June 20252.
That trial randomized 71 IPF patients to one of three rentosertib regimens or placebo for 12 weeks2. During it, researchers collected serum and measured about 2,800 proteins with the Olink platform3. In the new paper, the team took those protein profiles from 42 participants and ran them through six aging clocks: ProtAge, OrganAge (one version trained on chronological age, one on mortality), PAC, ipfP3GPT and PAOPAC1. The clocks were built by different groups, including teams at Harvard, Oxford, Peking University and Insilico itself3.
The evidence so far
The abstract states that "all six clocks consistently predicted lower biological age in treated arms"1. According to an independent summary of the paper, the 42 analyzed patients were Asian with a mean age of 67.1 years, placebo patients showed "minimal changes or slight increases" in predicted age, and the strongest response was seen at week 45. The 30 mg twice-daily arm gave the most consistent signal across clocks5.
Insilico's press release puts the peak effect at about 3 to 4 years lower predicted age at week 4 in that arm, and up to 6 years on one clock3. Those figures come from the company, and the clocks estimate age from protein patterns; they do not measure how long anyone will live.
| Measure | Reported result | Source |
|---|---|---|
| Patients with proteomic data analyzed | 42 | Paper, company |
| Proteins with shifted trajectories, drug arms | 326 | Coverage of paper |
| Proteins with shifted trajectories, placebo | 2 | Coverage of paper |
| Lung function (FVC) change, 60 mg once daily | +98.4 mL | 2025 trial |
| Lung function (FVC) change, placebo | -20.3 mL | 2025 trial |
Beyond the clocks, the team ran pathway analyses. The abstract says these identified "potential anti-aging shifts in senescence and metabolic processes alongside the anti-fibrotic activity of rentosertib"1. News-Medical's summary reports that the drug changed the trajectories of 326 proteins versus two in placebo, and that growth-factor pathways such as PI3K-Akt and RAS signaling were turned down in treated groups5. Insilico also says the dose with the best lung-function result was not the dose with the strongest clock signal, which it reads as a sign that the aging effect is not just a side effect of a healthier lung3.
The underlying 2025 trial was mainly a safety study. Rates of treatment-emergent adverse events were similar across arms (72.2% to 83.3% on drug, 70.6% on placebo), and the events most often leading to stopping treatment were liver toxicity and diarrhea2.
What it does not show yet
The paper's abstract is direct: "proteomic clocks alone cannot deconvolute aging- and disease-specific effects"1. IPF is itself an age-related disease, so a drug that calms fibrosis could move age-linked proteins without slowing aging in the rest of the body.
According to News-Medical's summary, the authors list "the small sample size, short follow-up, the computational nature of much of the analysis, the lack of complementary omics data, and the inability to fully separate anti-fibrotic from anti-aging effects" as reasons that "preclude definitive conclusions about geroprotection"5.
Other gaps worth naming:
- The clock analysis was not the trial's primary goal. The trial's primary endpoint was safety2.
- Twelve weeks is short, and the strongest signal appeared at week 4, not at the end of treatment5.
- All participants had IPF and were treated in China, so the results say nothing yet about healthy people5.
- No clinical aging outcome, such as frailty, disease onset or survival, was measured.
Outside doctors who spoke to Medical News Today were cautious. Dung Trinh, an internist at MemorialCare Medical Group who was not involved, told the outlet6:
"However, this does not prove that the drug reverses human aging. The study was small, involved patients with idiopathic pulmonary fibrosis, and lasted only 12 weeks, so some of the biomarker changes may reflect improvement in the underlying disease."
Zeeshan Khan, chief of geriatric medicine at Hackensack Meridian Jersey Shore University Medical Center, added: "What ultimately matters is whether patients live healthier and more functional lives"6.
Who is behind it
The first author is Alex Zhavoronkov, Insilico's founder and chief executive. The competing-interests statement says seven other authors are Insilico employees and that the company is developing rentosertib for IPF1. Co-authors also include researchers at Brigham and Women's Hospital and Harvard Medical School (among them Vadim Gladyshev's group), the Broad Institute, Massachusetts General Hospital, Stanford, RWTH Aachen, Peking University, Westlake University and Peking Union Medical College Hospital1. Several of the clocks used were developed by these co-authors' own groups3.
Insilico, listed in Hong Kong, used its press release to highlight its finances, including about $106 million in first-half 2026 revenue3. The release also quotes Zhavoronkov on the economic value of drugs that reverse biological age, a claim that goes well past what the paper tests. A more measured comment in the same release came from Nobel laureate Michael Levitt: "This trial cannot yet separate slower aging from a treated lung, and the authors say so plainly"3.
How it fits the field
Geroscience researchers have long argued that drugs for age-related diseases might also slow aging, but standard trials are not built to detect that. The paper frames its approach as a "dual-purpose" trial design that adds aging endpoints to studies of a specific disease1. It also argues that proteomic clocks may be easier to interpret than epigenetic clocks, which it says "often yield inconsistent, hard-to-interpret results"1.
The practical question is validation. No aging clock is yet accepted by regulators as a surrogate endpoint. Insilico says it wants to pursue biomarker qualification through the FDA's existing programs3, but that is a stated plan, not an outcome.
What to watch next
Two days after the paper, Insilico said it had dosed the first patient in GENESIS-IPF-3, a phase 3 trial planned to enroll 320 patients at 47 centers in China, with a 52-week primary endpoint on lung-function decline4. A longer, larger trial gives a chance to see whether clock shifts hold up past week 4 and track with clinical results, if aging biomarkers are collected. The phase 3 release does not say whether they will be.
Levitt's own suggestion is a study in healthy volunteers3. That would be the clearest test of whether the signal reflects aging itself rather than a treated disease.
