Biotechnology · global
AI-designed antifibrotic drug enters Phase 3 lineup, but the real answer will have to wait for lung-function curves
Rentosertib’s Phase 3 trial has moved from company announcements to clinical registry records; this time, the AI drug-development narrative no longer stops at how the molecule was found, but at whether it can slow the decline in lung function among IPF patients over 52 weeks.
Idiopathic pulmonary fibrosis (IPF) is a slow but cruel lung disease, in which scarred tissue gradually replaces the alveolar spaces that exchange gases. For AI drug development, this kind of disease is also especially demanding: models can help propose targets and molecules, but what they ultimately face is whether patients’ lung function truly declines more slowly.
Insilico Medicine recently announced that its oral small-molecule drug rentosertib has entered Phase 3 clinical testing. According to clinical trial registry records, the study identifiers include NCT07687459 and the China registration number CTR20262475. It is designed as a randomized, double-blind, placebo-controlled, parallel-group trial, with an expected enrollment of 320 IPF patients and a 52-week treatment period. The company announcement said the trial has been initiated; the registry records, however, show the study status as not yet recruiting, with an estimated start date of August 30, 2026. This indicates that “initiated” may refer to regulatory and trial-preparation procedures having begun, rather than patients already having been enrolled.
Rentosertib, also known as INS018_055 or ISM001-055, is an oral inhibitor targeting TNIK. Insilico says the drug candidate was discovered by its AI biology and generative chemistry platforms to target pathological pathways associated with fibrosis. The key point in this claim is not the two letters “AI” themselves, but whether it can translate a target hypothesis derived from data and models into human efficacy that is reproducible, quantifiable, and acceptable to regulators.
The primary endpoint of the Phase 3 trial is the annualized rate of decline in forced vital capacity (FVC) over 52 weeks. FVC is a commonly used measure in IPF clinical trials, reflecting the amount of air a patient can forcefully exhale in one breath; it is not the same as the entirety of a patient’s day-to-day experience, but it is an important yardstick for assessing disease progression and drug effect. Secondary endpoints listed in the registry records also include time to disease progression, fibrosis burden seen on high-resolution computed tomography, changes in DLCO, and quality-of-life questionnaires.
These design features make the test for rentosertib clear: it must not only withstand a longer treatment course in terms of safety, but also produce effects sufficient to explain a clinical difference beyond existing treatment options. Insilico has said the Phase 3 trial is built on published Phase 2a human data, but the currently public summary is still insufficient on its own to judge the magnitude of efficacy, population differences, and long-term risks. For general readers with a higher level of education, the more cautious interpretation is this: it is an important clinical milestone, not proof that efficacy has been established.
Background Context
In recent years, AI drug development has moved from “finding new molecules” into a more difficult stretch: how to leave verifiable evidence across clinical endpoints, manufacturing quality, patient heterogeneity, and regulatory requirements. Rentosertib’s Phase 3 study makes that path concrete. If the trial recruits smoothly and is completed, it will provide a rare case to examine whether an antifibrotic drug discovered with AI assistance can truly change the trajectory of declining lung function in a disease such as IPF, where progression is slow and endpoints are stringent.