Insilico Medicine is progressing into a Phase III human clinical trial to test an AI-identified drug targeting idiopathic pulmonary fibrosis (IPF). This advancement provides an empirical test case for the field of computational drug discovery, advancing AI drugs through early safety evaluation and into late-stage efficacy validation.
IPF causes severe scarring of lung tissue, destroying the ability to breathe. Median patient survival typically reaches 2 to 4 years after diagnosis. When administered orally, the drug lentsertib, identified by the AI, inhibits a kinase that interacts with TRAF2 and NCK, addressing the underlying disease mechanism.
The randomized trial evaluated 71 patients across 22 clinical sites in China and divided participants into a placebo group and an active treatment group. The researchers administered doses of 30 mg or 60 mg per day over a 12-week observation period.
Patients assigned to the 60 mg once-daily regimen showed an increase in mean forced vital capacity of +98.4 mL, in sharp contrast to the 20.3 mL decrease in vital capacity recorded in the placebo group. The safety profile remained manageable and adverse events reflected expected baseline rates in all study arms. U.S. Food and Drug Administration (FDA) regulators granted the asset “orphan drug designation” in February 2023.
Multi-omics algorithmic target prioritization
This development relies entirely on Pharma.AI, a proprietary computational pipeline powered by Insilico Medicine. Workflows are divided into separate engines that handle specific biological and chemical engineering tasks.
PandaOmics performs an initial target detection phase. The system ingests vast biological datasets and processes genomics, clinical trial results, academic literature, and patent intelligence to build comprehensive biological network models. The algorithm applies causal inference mechanisms to identify emerging disease associations hidden within the data architecture.
PandaOmics has isolated TNIK as a key biological target for IPF intervention. This computational system bypassed the receptor tyrosine kinase pathway that is targeted by existing antifibrotic drugs.
The software mapped TNIK as a central node controlling fibrosis and inflammation through Wnt, TGF-β, Hippo/YAP-TAZ, JNK, and NF-κB signaling channels. The target selection process incorporated a framework of aging hallmarks to score biological targets based on their involvement in multiple aging mechanisms, chronic inflammation, and extracellular matrix remodeling.
Dr. Feng Ren, co-CEO and chief scientific officer of Insilico Medicine, said: “IPF is one of the clearest clinical examples of an age-related disease where fibrosis, chronic inflammation, extracellular matrix remodeling, and cellular senescence intersect.
“Lentosertib was not discovered by simply screening more compounds starting from traditional targets. It was born from a biology-first, aging-informed AI workflow that links TNIK to the mechanisms of fibrotic and inflammatory diseases and then uses generative chemistry to create drug candidates with the properties needed for clinical development.”
Performing generative molecular engineering
Following target selection, the Chemistry42 engine performs the generative molecule design. This system differs from traditional high-throughput screening methods. Chemistry42 does not search existing compound libraries. Instead, the system applies generative tensor reinforcement learning to construct molecules that physically align with target protein pockets. This algorithmic engineering process balances structural suitability against desired pharmacological properties.
During the computational generation phase, exactly 79 physical molecules were synthesized to undergo testing. The engineering team selected the 55th iteration to proceed to preclinical testing. This target generation protocol reduced the timeline from project initiation to nomination of preclinical candidates to 18 months.
The basic architecture is derived from the company’s GENTRL methodology, published in Nature Biotechnology in 2019. This platform establishes a reproducible system to control molecule production, avoiding the capital-intensive trial-and-error process that defines standard pharmaceutical chemistry.
Verification of biological effects through proteomics analysis
Clinical evaluation integrates complex proteomic analyzes to validate algorithm-predicted biological interactions. Insilico Medicine will deploy an internal proteomic aging clock framework within the IPF trial to gather exploratory geriatric information.
Chronological age proteomic clocks, including ProtAge, OrganAgechrono, ipfP3GPT, and PAOPAC, track changes in biological age predicted by interventions. Researchers applied age-related trajectories from UK Biobank as an external comparison dataset to contextualize treatment-responsive proteins against broader population data.
Mortality risk-related proteomic clocks such as PAC and OrganAgemortality provide analytical streams that are parallel and orthogonal to standard clinical endpoints. Clinical teams perform SenMayo and CellAge signature analyzes to assess aging and senescence-related secretory phenotype biology within cellular models.
A peer-reviewed study published in the journal Aging and Disaster confirmed that pharmacological TNIK inhibition causes cytoplasmic activity and observable reductions in extracellular matrix remodeling indicators.
Documenting the calculation pipeline
Moving Rentosertib into the clinical pipeline provides a documented, peer-reviewed data trail that is essential for validating AI capabilities in the life sciences. Nature Biotechnology has published a complete journey from discovery to clinic. This publication details algorithmic TNIK target prioritization, production chemistry results, preclinical efficacy data, and human phase I pharmacokinetics.
The Journal of Medicinal Chemistry published structural biology validation detailing the discovery of a novel TNIK inhibitor chemotype and providing structural support via a TNIK kinase domain co-crystal structure. Nature Medicine documented Phase IIa safety and lung function data and provided empirical validation of computational predictions.
Dr. Alex Zhavoronkov, Founder and CEO of Insilico Medicine, commented: “Rentosertib is a defining program for Insilico because it represents the full scope of our mission to use AI to not only move faster, but also to create new biology, new chemistry, and new therapeutic opportunities.”
“This program began with the hypothesis that aging biology could help identify powerful targets for major diseases. We have now progressed through target discovery, molecular design, preclinical validation, Phase I safety, randomized Phase IIa clinical data, and Phase III development. For the AI drug discovery field, this is no longer just a story of speed, it is a story of clinical translation.”
The implementation of AI in biopharmaceuticals requires verifiable data on human outcomes. Phase III trials will provide a final test of the clinical efficacy of the generative algorithm.
Reference: Accelerating Human-Claude Science with NVIDIA BioNeMo
Want to learn more about AI and big data from industry leaders? Check out the AI & Big Data Expos in Amsterdam, California, and London. This comprehensive event is part of TechEx and co-located with other major technology events such as Cyber Security & Cloud Expo. Click here for more information.
AI News is brought to you by TechForge Media. Learn about other upcoming enterprise technology events and webinars.
