A lung drug designed by artificial intelligence just showed something scientists did not expect: signs that it might turn back the biological clock inside the body.
Quick Take
- Rentosertib, a drug for lung scarring, was first tested in a phase 2a human trial across 21 sites in China.
- A new reanalysis of blood samples from that trial found six separate aging-clock models all pointed the same direction.
- Insilico Medicine, the drug’s maker, reports a peak effect of roughly 3 to 4 years of reversed biological age.
- The lung trial itself already showed a real breathing benefit before anyone looked at aging markers.
A Lung Drug Built By Machines, Not Chemists
Rentosertib started life with a code name, ISM001-055, and a narrow job: treat idiopathic pulmonary fibrosis, a disease that scars the lungs and slowly steals a patient’s breath. Generative artificial intelligence helped find the drug’s target, a protein called TNIK, then helped design the molecule itself. That combination made it one of the first AI-originated drugs to reach human testing at this scale.
Researchers ran a phase 2a trial testing rentosertib against a placebo. Doctors watched patients closely, tracking safety and how well their lungs worked. The study included 21 sites across China, ran from July 2023 to June 2024, and used a double-blind, randomized design, the gold standard for drug testing.
The results looked promising on their own terms. Patients on the highest dose saw their lung function, measured by forced vital capacity, improve by about 98.4 milliliters compared to placebo in the 71-patient study. That number matters because lung capacity naturally declines in fibrosis patients, so any improvement counts as a win in a disease with few good treatments.
Blood Samples Reveal A Second Story
After the lung results came in, Insilico went back to the same trial’s blood samples. Researchers pulled 12 weeks of proteomic data, essentially a snapshot of thousands of blood proteins, from 42 of the IPF patients. They ran that data through six different aging-clock models, tools built by outside groups including Harvard and Oxford researchers to estimate a person’s biological age from their blood.
Every single one of the six models pointed the same way. Insilico says all six “consistently indicated a reduction in predicted biological age” among patients taking rentosertib. That kind of agreement across independently built models is unusual, since aging clocks often disagree with each other when applied to the same data.
The company reports the strongest effect showed up at week four in patients taking 30 milligrams twice daily. At that point, the average reversal was about 3 to 4 years, with one specific clock model showing up to 6 years. Nature Biotechnology published the analysis in September 2026, and outside coverage confirmed the same six-clock structure independently.
Nature Biotechnology | Insilico’s AI-Driven IPF Candidate Rentosertib Shows Potential for Biological Age Reversal, as Assessed by Six Proteomic Aging Clocks https://t.co/qjHiwL9r63
First-in-Class Evaluation of an AI-Driven Candidate: This is the first de-novo drug designed with…
— Insilico Medicine (@InSilicoMeds) September 7, 2026
Why Aging Clocks Are Still A Young Science
Proteomic aging clocks are a fairly new tool in medicine. They work by comparing a patient’s blood proteins to patterns seen in large population studies, then estimating how “old” that blood profile looks compared to the person’s actual birth date. Large studies involving tens of thousands of people have shown these clocks can predict future disease and death risk with real accuracy.
Even so, medical researchers who study these clocks describe them as promising tools, not finished, universally accepted measures of anti-aging success. A shift in a blood-based clock score is not the same claim as proving a person’s body has literally aged backward. The rentosertib finding fits that pattern: a strong signal in a small group, worth serious follow-up, not yet a settled verdict.
The sample size here was 42 patients, small by clinical trial standards, and the aging analysis was not the original goal of the study. Researchers designed the trial to test lung function and safety, then applied the aging clocks afterward. That does not undercut the result, but it does mean bigger, purpose-built studies would carry more weight going forward.
What Comes Next For The Drug And The Science
Insilico has already filed for expanded testing of rentosertib, including an inhaled version now cleared to begin new trials. The company built its reputation on using AI to speed up drug discovery, and rentosertib stands as its highest-profile human proof of that approach working in real patients, not just in a computer model.
For patients with pulmonary fibrosis, the lung-function results alone justify continued study. For the broader public curious about slowing aging itself, the honest takeaway is that six independent tools spotted the same pattern in the same small group of people, and researchers now have a clear reason to test it again, bigger and longer, before anyone calls it proven.
Sources:
insilico.com, unite.ai, thenextweb.com, nytimes.com
