← Back to Home

AI-designed pulmonary fibrosis drug enters Phase III, moving efficacy testing to one year of follow-up

Rentosertib has entered a Phase III trial that will test early lung function signals in 320 patients over 52 weeks of treatment. In the same month, an AI drug discovery company spun off from ByteDance reportedly raised substantial funding. Capital and clinical development are advancing in parallel, but each addresses a different question of value.

By SURL BioNews

For patients with idiopathic pulmonary fibrosis, the significance of a new drug lies in whether it can slow the progressive loss of lung function. Rentosertib, a drug candidate discovered and designed with the involvement of generative AI, is moving toward longer testing in humans: on September 10, 2026, Insilico Medicine announced that the first patient had been dosed in the Phase III GENESIS-IPF-3 trial at Peking Union Medical College Hospital.

According to the protocol released by the company, this randomized, double-blind, placebo-controlled trial plans to enroll 320 patients at 47 centers in China for 52 weeks of treatment. The primary endpoint is the annual rate of decline in forced vital capacity (FVC), tracking how the volume of air patients can forcibly exhale after a deep breath changes over time. Compared with short-term fluctuations in measurements, a year of follow-up can better test whether the drug consistently slows the deterioration of lung function.

Rentosertib, also known as ISM001-055, is a small-molecule inhibitor targeting the TNIK kinase. Its development process uses AI to identify a therapeutic target associated with fibrosis and then design molecules that act on that target. Experiments and human trials are ultimately still needed to determine whether this biological hypothesis can translate into a therapeutic effect. For a disease in which lung tissue gradually becomes scarred and function continues to decline, blocking fibrosis signaling is a concrete strategy that remains to be validated.

One piece of evidence supporting further development is a Phase IIa trial published in Nature Medicine in 2025. Seventy-one patients received 12 weeks of treatment. In the group receiving 60 milligrams once daily, mean FVC increased by 98.4 milliliters from baseline, while the placebo group saw a decline of 20.3 milliliters. However, the study primarily assessed safety and tolerability, with lung function as a secondary endpoint. The small sample and short follow-up mean these results are not yet sufficient to establish long-term efficacy.

Safety also requires more complete answers. The Phase IIa study recorded treatment discontinuations related to hepatotoxicity, which occurred more frequently in patients also taking the antifibrotic drug nintedanib. The study authors noted that the effects of the individual drugs and their combined use still need to be clarified. Phase III therefore has two tasks: to confirm whether the lung function signal persists and whether longer-term treatment carries acceptable risks. According to the September announcement, rentosertib remains an investigational drug and has not been approved by any regulatory authority.

Beyond clinical validation, funding continues to flow into AI drug discovery. On September 16, Reuters, citing people familiar with the matter, reported that Anew Labs, a company spun off from ByteDance, had raised $290 million at a valuation of $1.5 billion, in a round led by Sequoia China, IDG Capital and Hillhouse Venture Capital. These transaction details came from anonymous sources. ByteDance and the investors involved did not immediately respond to requests for comment at the time, and the fundraising itself cannot serve as evidence of a drug candidate’s efficacy.

An industry analysis by 36Kr focused on the gap in validating the value of AI drug discovery: models and early-stage research and development can iterate rapidly, while evidence from humans takes years to accumulate. Rentosertib’s entry into Phase III provides a more clearly defined setting in which to examine this question. Even if a single drug candidate ultimately succeeds, results from more independent programs will still be needed to determine whether AI broadly improves the success rate of new drug development. The most concrete progress for now is that a therapeutic hypothesis is beginning to undergo testing on a larger scale and over a longer period.

References

  1. eu.36kr.com
  2. Insilico Medicine
  3. Reuters via MarketScreener