Insilico's AI-Designed Rentosertib Shows 98.4mL FVC Improvement in Phase IIa, Published in Nature Medicine

Title

Insilico's AI-Designed Rentosertib Shows 98.4mL FVC Improvement in Phase IIa, Published in Nature Medicine

Content

Insilico Medicine, a clinical-stage generative AI biotech headquartered in Cambridge, Massachusetts (HKEX: 3696), reported the first randomized clinical proof-of-concept for an AI-discovered target and AI-designed compound: Rentosertib (formerly ISM001-055), an oral small-molecule TNIK inhibitor for idiopathic pulmonary fibrosis. The 60 mg once-daily arm showed a mean FVC change of +98.4 mL at 12 weeks versus -20.3 mL on placebo, with patients not on concomitant antifibrotic therapy reaching +187.8 mL. The paper's authors framed the milestone directly: prior AI-discovered drugs had experienced phase 2 failure rates similar to non-AI-discovered candidates, and none had progressed through phase 3 trials. Rentosertib is the first to deliver a positive randomized efficacy signal in IPF, where current standard-of-care (nintedanib and pirfenidone) only slows progression. IPF affects 10-60 per 100,000 people in the USA with median survival 2-4 years after diagnosis. Rentosertib targets TNIK, identified de novo via Insilico's PandaOmics biology engine, with the chemical structure generated through Chemistry42. Preclinical nomination took approximately 18 months and Phase 0/1 completion under 30 months from target discovery, a dramatic compression of the typical 10-15 year, $2-3 billion drug development arc. The pipeline evolved from PandaOmics (target identification) to Chemistry42 (compound design) to Medicine42 with inClinico (trial outcome prediction). Phase 0/1 results were published in Nature Biotechnology in 2024; Phase IIa in Nature Medicine in 2025. Insilico initiated a 320-patient Phase III trial (CTR20262475, NCT07687459) on July 7, 2026, led by Professor Zuojun Xu of Peking Union Medical College Hospital. On the trial architecture: 128 patients were screened and 71 randomized across four arms (placebo n=17, 30 mg QD n=18, 30 mg BID n=18, 60 mg QD n=18) over 12 weeks. Primary safety was met with similar TEAE rates (70-83%). Pharmacokinetics showed favorable exposure for 60 mg QD (mean 3,450 h·ng/mL at week 12). Liver toxicity was the main safety signal: 7 of 12 discontinuations related to liver dysfunction. Operational results were dose-dependent. Placebo declined (-20.3 mL), 30 mg QD showed similar decline (-27.0 mL), 30 mg BID turned nominally positive (+19.7 mL), and 60 mg QD reached statistical and clinical significance (+98.4 mL, 95% CI 10.9 to 185.9). Patients on 60 mg QD without concomitant antifibrotic therapy showed the strongest effect (+187.8 mL, 95% CI 68.6 to 306.9). Going forward, Insilico is positioned to deliver the first Phase 3 readout for an AI-discovered drug. The pattern is replicable across age-related fibrotic and inflammatory indications. Open questions: whether the FVC signal scales in larger trials and whether liver toxicity is manageable.