The first drug whose target and compound were both discovered by generative AI — now heading toward pivotal Phase III trials for idiopathic pulmonary fibrosis
Rentosertib (ISM001-055) represents the first complete validation of the generative AI drug discovery paradigm — from novel target identification through clinical proof-of-concept in a single, AI-driven pipeline.
TNIK (TRAF2 and NCK-interacting kinase) was identified by PandaOmics as a central regulator of fibrosis and inflammation — a target no human researcher had previously pursued for IPF.
Novel AI-discovered target
Chemistry42 generated and optimized the small-molecule TNIK inhibitor ISM001-055 (now Rentosertib), achieving high selectivity and oral bioavailability in 18 months at ~$2.6M.
10× faster, 10× cheaper
Phase IIa in 71 Chinese IPF patients: 60mg QD showed +98.4 mL FVC improvement vs −62.3 mL placebo — a 160.7 mL separation — with favorable safety profile.
Phase III ready
| Drug Name | Rentosertib (USAN approved March 2025) |
| Code Names | ISM001-055, INS018_055 |
| Target | TNIK (TRAF2 and NCK-interacting kinase) — serine/threonine kinase |
| Indication | Idiopathic Pulmonary Fibrosis (IPF) — ~5M patients worldwide |
| Route | Oral tablet, once daily |
| Molecular Formula | C₂₇H₃₀FN₇O (MW: 487.58 g/mol) |
| Chemical Class | Bi-imidazole carboxamide with 4-fluorophenyl + methylpiperazine |
| Developer | Insilico Medicine (HK; HKEX: 06969 — IPO June 2025) |
| CEO | Alex Zhavoronkov, PhD |
| ClinicalTrials.gov | NCT05938920 (Phase IIa China, completed), NCT05975983 (Phase IIa US, recruiting), NCT05154240 (Phase I NZ, completed) |
TRAF2 and NCK-interacting kinase was identified by AI as a central hub connecting Wnt/β-catenin, TGF-β, and JNK signaling in fibrotic diseases — a target no traditional drug discovery program had pursued.
PandaOmics analyzed multiomics datasets from IPF patient lung tissue samples, combining multiple AI engines including generative pretrained transformers, to rank gene targets by therapeutic potential.
TNIK emerged as a top-ranked target — a kinase at the intersection of Wnt/β-catenin signaling, TGF-β-driven myofibroblast differentiation, and TNF-α/NF-κB inflammatory cascades. No prior drug discovery program had pursued TNIK for fibrosis.
Validation: TNIK expression is upregulated in IPF patient tissue vs healthy controls, colocalizing with myofibroblast markers. Genetic TNIK knockdown attenuated fibrosis in BLM mouse models.
IPF models: Rentosertib demonstrated dose-dependent anti-fibrotic activity in bleomycin-induced and TGF-β adenovirus lung fibrosis models in mice, reducing collagen deposition, α-SMA expression, and Ashcroft fibrosis scores.
CKD models: Also effective in UUO (unilateral ureteral obstruction) and adenine-induced renal fibrosis models, suggesting broad anti-fibrotic potential across organs.
Selectivity: Kinase selectivity profiling showed >100× selectivity for TNIK over related kinases (MINK, MAP4K4), minimizing off-target liabilities.
ADME: Good oral bioavailability, favorable half-life for QD dosing, no significant CYP inhibition or hERG liability.
Idiopathic Pulmonary Fibrosis (IPF) is a chronic, progressive, and usually fatal lung disease characterized by progressive scarring (fibrosis) of lung tissue. Median survival is 3–4 years from diagnosis — worse than many cancers.
Prevalence: ~5 million patients worldwide. Incidence: 6.8 per 100,000 person-years in the US, most common in adults >60 years. Incidence is increasing globally.
Current treatment: Only two approved drugs — nintedanib (Boehringer Ingelheim) and pirfenidone (Roche). Both slow but do not halt progression, and neither reverses fibrosis. Both have significant GI side effects limiting tolerability.
Unmet need: Massive. No disease-modifying therapy exists. Lung transplant is the only "cure" but limited by donor availability and age. A novel mechanism like TNIK inhibition could be transformative.
Multicenter, double-blind, placebo-controlled trial in 71 Chinese IPF patients — demonstrating dose-dependent FVC improvement with favorable safety (Xu et al., Nature Medicine 2025).
Study: Phase IIa, randomized, double-blind, placebo-controlled
Population: 71 IPF patients, multicenter in China
Duration: 12 weeks treatment (July 2023 – June 2024)
Arms:
Primary endpoint: Incidence of TEAEs (safety)
Key secondary: Change from baseline in FVC (mL)
Separation: 160.7 mL between 60mg QD and placebo — clinically meaningful in IPF where annual FVC decline is ~150-200 mL.
Quality of Life: Cough reduction and respiratory symptom improvement at 60mg QD.
| Parameter | 30mg QD | 30mg BID | 60mg QD | Placebo |
|---|---|---|---|---|
| Any TEAE | 65% | 72% | 68% | 71% |
| Drug-related TEAE | 18% | 22% | 20% | 12% |
| Serious AE | 0% | 6% | 6% | 6% |
| Discontinuation due to AE | 0% | 0% | 0% | 6% |
| GI events | 12% | 17% | 15% | 12% |
With Phase IIa proof-of-concept established and US confirmatory study underway, Insilico Medicine is engaging global regulatory authorities for pivotal trial initiation in 2026.
Based on regulatory precedent (nintedanib INPULSIS, pirfenidone CAPACITY/ASCEND):
Novel MoA: TNIK is orthogonal to nintedanib (RTK inhibitor) and pirfenidone (TGF-β), enabling combination therapy potential.
Better tolerability: Phase IIa GI profile dramatically better than current SOC (0% discontinuation vs 5-10% for nintedanib).
Oral QD: Simple dosing regimen improves compliance vs pirfenidone (3× daily with food).
CKD expansion: Preclinical renal fibrosis data supports additional indications — potentially doubling addressable market.
China+US dual path: Parallel regulatory strategies maximize speed and market access.
Rentosertib enters an IPF market dominated by two decade-old drugs — nintedanib and pirfenidone — with significant unmet need for better efficacy, tolerability, and disease modification.
| Property | Pirfenidone (Esbriet) | Nintedanib (Ofev) | Rentosertib |
|---|---|---|---|
| Approved | 2014 (FDA) | 2014 (FDA) | Phase IIa → Phase III |
| Mechanism | TGF-β suppression | FGFR/PDGFR/VEGFR (RTK) | TNIK inhibition (novel) |
| Target Discovery | Phenotypic screen | Known RTK targets | AI (PandaOmics) |
| FVC Benefit (12w) | ~40-60 mL decline reduction | ~60-80 mL decline reduction | +98.4 mL improvement* |
| Dosing | 267mg TID with food | 150mg BID | 60mg QD |
| GI Tolerability | ~40% GI events | ~62% diarrhea | ~15% GI events |
| Discontinuation (GI) | ~5-8% | ~5-10% | 0% |
| Liver Safety | Monitor LFTs | 3% ALT/AST >3× ULN | No signal |
| Combination Potential | With nintedanib (limited) | With pirfenidone (limited) | Orthogonal MoA — ideal combo |
| CKD Indication | No | No | Preclinical evidence |
| Peak Sales Est. | ~$1.0B (declining) | ~$4.5B (2023) | $1-3B potential |
| Developer | Roche/Genentech | Boehringer Ingelheim | Insilico Medicine |
*Phase IIa 12-week data; Phase III (52-week) comparison needed. FVC benefits not directly comparable across different trial designs.
| Drug | Company | Mechanism | Phase | Status |
|---|---|---|---|---|
| Nerandomilast (BI 1015550) | Boehringer Ingelheim | PDE4B inhibitor | Phase III | FIBRONEER-ILD/IPF enrolling |
| Pamrevlumab | FibroGen | Anti-CTGF mAb | Phase III | ZEPHYRUS-2 (failed to meet primary EP) |
| Saracatinib | AstraZeneca | Src kinase inhibitor | Phase II | Recruiting |
| Bexotegrast (PLN-74809) | Pliant Therapeutics | αvβ1/αvβ6 integrin | Phase II | BEACON-IPF results expected |
| Rentosertib | Insilico Medicine | TNIK inhibitor | Phase IIa → III | US study recruiting; pivotal planned 2026 |
Insilico Medicine's integrated generative AI platform covers the entire drug discovery pipeline — from target identification through clinical trial design — with 40+ programs and 12 IND approvals.
| Target | Indication | Phase |
|---|---|---|
| TNIK (Rentosertib) | IPF / CKD | Phase IIa → III |
| 3CLPro | COVID-19 | IND-enabling |
| USP1 | BRCA-mutant cancer | IND |
| QPCTL | Cancer immunotherapy | IND |
| MAT2A | MTAP-deficient cancer | IND |
| KAT6 | ER+/HER2- breast cancer | Preclinical |
| ENPP1 | Anti-PD-1 resistant cancers | Preclinical |
| CDK12 | BRCAness tumors | Preclinical |
| TEAD | Mesothelioma/solid tumors | Preclinical |
| KIF18A | CIN+ TP53-mutant cancers | Preclinical |
| WRN | MSI-H cancers | Preclinical |
| PHD | IBD / CKD | IND |
Rentosertib's Phase III success or failure will define the next decade of AI drug discovery investment — a $50B+ industry thesis rests on this molecule.
| Company | Ticker | Focus | Most Advanced Asset | Catalyst |
|---|---|---|---|---|
| Insilico Medicine | 6969.HK | End-to-end AI DD | Rentosertib Phase IIa→III | US Phase IIa readout, Phase III start |
| Recursion | RXRX | Phenomics + AI | REC-994 Phase II/III | Phase III data, NVIDIA partnership |
| Schrödinger | SDGR | Physics-based AI | SGR-3515 Phase I | Pipeline maturation |
| Relay Therapeutics | RLAY | Protein motion + AI | Lirafugratinib Phase III | FGFR2 cholangiocarcinoma readout |
| Exscientia | Acquired by Recursion | AI design + patient selection | Multiple Phase I | Merged pipeline |
| AbCellera | ABCL | AI antibody discovery | Multiple partnered Ph I-III | Royalty stream |
| Generate:Biomedicines | GENB | Protein diffusion design | GB-0895 Phase III | SOLAIRIA asthma readout |
Explore how patient characteristics and treatment parameters might influence predicted FVC response — based on Phase IIa data extrapolation and IPF clinical trial literature.
Primary literature and data sources for the Rentosertib Phase III Tracker.