AI-Designed Rentosertib Raises Eyebrows as It Appears To Make Patients Biologically Younger in Small Trial

An AI-designed drug appeared to make patients biologically 'younger' across six different ageing clocks in a small trial of idiopathic pulmonary fibrosis, according to research published in Nature Biotechnology on 7 September 2026.

Rentosertib, an investigational TNIK inhibitor developed by Insilico Medicine, showed its clearest signal after four weeks at 30 mg twice daily in a secondary analysis of a Phase 2a study that was primarily designed to assess safety and tolerability.

The analysis covered 42 patients, with researchers measuring 2,841 proteins and running six ageing-clock models. All six detected a shift towards lower predicted biological age in the treated groups. Insilico reported a reduction of roughly three to four years at week four, with one clock showing a reduction of up to six years.

Those figures are modelled estimates of biological age, not evidence that patients became younger chronologically or gained additional years of life.

The finding is notable because clocks built using different methods and training objectives converged on the same direction of change inside a randomised drug trial. Of 54 treatment-versus-placebo comparisons, 21 reached Q < 0.10.

Eleven of 18 week-four comparisons were significant, while the 30 mg twice-daily regimen produced nine significant results, the strongest pattern among the three dosing groups.

The Clocks Cannot Separate Ageing From Lung Disease

The biggest limitation is biological rather than statistical. Proteins associated with ageing can also change when fibrosis is treated, which makes it difficult for a clock trained to recognise age-related patterns to distinguish the two effects.

The parent Phase 2a trial enrolled 71 people with IPF across 21 locations in China and was designed primarily to assess safety and tolerability over 12 weeks.

Forty-three participants consented to the proteomic analysis, with one excluded because of a missing end-of-trial sample. The remaining 42 had a mean age of 67.1 years and were all Asian.

LTBP2 highlights the problem. It was the only high-impact feature shared by all six clocks and the largest contributor to the BioAge shift.

The protein is associated with fibrosis and transforming growth factor-β signalling, providing a disease-related pathway that could influence the apparent age reduction.

The authors say proteomic clocks alone cannot fully disentangle ageing from disease-specific effects in an IPF population. They point to validation in healthy or non-IPF populations as the route to determining whether the signal reflects genuine modulation of ageing.

The Strongest Clock Signal Came From a Different Dose

In the original Phase 2a trial, the 60 mg once-daily group recorded the strongest mean improvement in forced vital capacity at 12 weeks, with FVC rising by 98.4 ml compared with a 20.3 ml decline for placebo.

The strongest cross-clock biological-age signal instead came from 30 mg twice daily. Changes in FVC also had only a weak relationship with changes in predicted biological age, with a median R² of 0.06.

The clock signal was strongest at week four and less pronounced by week 12, although many of the underlying protein changes persisted. The later plateau therefore does not necessarily show that the drug's broader proteomic effects disappeared.

Phase 3 Is Testing the Lungs, Not Biological Age

Rentosertib has since entered a Phase 3 IPF programme designed to enrol 320 patients at 47 sites in China over 52 weeks. Its primary endpoint is the annual rate of decline in FVC, rather than a biological-age measure. ClinicalTrials.gov listed the study as not yet recruiting in its 7 July 2026 update.

The trial is designed to establish whether Rentosertib can affect the course of IPF, not whether it slows human ageing. Rentosertib remains investigational and is not approved for any use.

The Nature Biotechnology paper was led by Insilico Medicine founder and chief executive Alex Zhavoronkov, who is the first author, alongside other Insilico researchers and external academic collaborators.

The six clocks agree on the direction of the biological-age signal. What they cannot yet establish is whether Rentosertib changed the ageing process itself or changed the molecular signature of a disease being treated.