Insilico Medicine Advances AI-Designed Drug to Phase III Trials

Idiopathic pulmonary fibrosis (IPF) is a devastating condition characterized by severe lung tissue scarring, typically leaving patients with a survival expectancy of only two to four years post-diagnosis. Rentosertib, an oral inhibitor of the TRAF2- and NCK-interacting kinase (TNIK), aims to disrupt these disease mechanisms. In a randomized trial involving 71 patients across 22 Chinese clinical sites, those receiving a 60 mg daily dose showed a mean forced vital capacity gain of 98.4 mL, while the placebo group suffered a 20.3 mL loss.

The Computational Pipeline

The development of rentosertib relied on Pharma.AI, a platform that segments biological and chemical engineering into distinct engines. The PandaOmics system utilized causal inference to identify TNIK as a central node in fibrosis, bypassing traditional receptor tyrosine kinase pathways. Following target selection, the Chemistry42 engine employed Generative Tensorial Reinforcement Learning to design molecules that fit the target protein pocket. This approach condensed the timeline from project initiation to preclinical candidate nomination to just 18 months, with the team synthesizing 79 molecules before selecting the 55th iteration for testing.

Beyond basic efficacy, the clinical program incorporates internal proteomic aging-clock frameworks—such as ProtAge and OrganAge—to capture geroscience readouts. Peer-reviewed studies in journals including Nature Medicine and the Journal of Medicinal Chemistry have documented the discovery-to-clinic progression. Alex Zhavoronkov, CEO of Insilico Medicine, noted that the program’s trajectory from target discovery through Phase IIa data validates the transition of AI from a speed-oriented tool to a clinical translation engine. The U.S. Food and Drug Administration granted the asset Orphan Drug Designation in February 2023.