Insilico Medicine Pushes AI-Discovered IPF Drug Into Phase III Trials

 

AI is now entering the fight against deadly lung disease.

Insilico Medicine is moving its AI-discovered drug candidate for idiopathic pulmonary fibrosis, or IPF, into Phase III clinical trials, giving the AI drug discovery industry one of its most important real-world tests so far.

The drug, called rentosertib, was developed to target IPF, a severe and progressive lung disease that causes scarring of lung tissue and gradually reduces a patient’s ability to breathe. Many IPF patients survive only two to four years after diagnosis, making the search for better treatments urgent.

Rentosertib works by inhibiting TRAF2- and NCK-interacting kinase, or TNIK, a biological target linked to fibrosis and inflammation. The drug is taken orally and was identified through Insilico Medicine’s AI-driven drug discovery platform, Pharma.AI.

In an earlier randomized clinical trial, 71 patients across 22 clinical sites in China were divided into placebo and treatment groups. Patients received either 30 mg or 60 mg of rentosertib once daily over a 12-week period.

The strongest signal came from the 60 mg group, where patients showed an average forced vital capacity gain of 98.4 mL. In contrast, the placebo group recorded a 20.3 mL loss in lung capacity. Safety results were considered manageable, with adverse events generally consistent across the study groups.

The U.S. Food and Drug Administration granted rentosertib Orphan Drug Designation in February 2023, a status given to treatments intended for rare diseases.

How AI Helped Find the Target

The development of rentosertib was driven by Insilico Medicine’s Pharma.AI platform, which combines different AI engines for target discovery and molecular design.

Its PandaOmics system analyzed large biological datasets, including genomics, clinical trial data, academic research, and patent information. From this, the platform identified TNIK as a key biological target in IPF.

Rather than following the same receptor tyrosine kinase pathways used by existing antifibrotic drugs, the AI system pointed to TNIK as a central regulator of fibrosis and inflammation. The target was linked to several important biological pathways, including Wnt, TGF-beta, Hippo/YAP-TAZ, JNK, and NF-kB signaling.

Insilico also applied an aging-biology framework, looking at how chronic inflammation, cellular senescence, and extracellular matrix remodeling contribute to IPF.

Feng Ren, PhD, Co-CEO and Chief Scientific Officer of Insilico Medicine, said IPF is one of the clearest examples of an age-related disease where fibrosis, inflammation, tissue remodeling, and cellular aging intersect.

According to Ren, rentosertib was not discovered by simply screening more compounds against a conventional target. Instead, it came from a biology-first, aging-informed AI workflow that connected TNIK to fibrotic and inflammatory disease mechanisms.

Generative AI Designed the Molecule

After the target was selected, Insilico used its Chemistry42 platform for generative molecular design.

Unlike traditional drug discovery methods that search through existing compound libraries, Chemistry42 generates new molecular structures designed to fit the target protein pocket. The system uses generative reinforcement learning to balance molecular fit, drug-like properties, and clinical development requirements.

Insilico synthesized 79 physical molecules for testing. The company selected the 55th generated molecule for preclinical development. This helped shorten the timeline from project launch to preclinical candidate nomination to about 18 months.

The approach builds on Insilico’s earlier work on generative AI for molecular design, including its GENTRL methodology published in Nature Biotechnology in 2019.

Measuring Biological Impact

Insilico’s clinical program also includes advanced proteomic analysis to study whether the drug affects biological aging signals linked to IPF.

Researchers are using proteomic aging clocks and mortality-risk-related protein models to observe treatment-related changes. These include systems designed to track age-associated proteins, organ-aging markers, and senescence-related biological patterns.

The goal is not only to see whether rentosertib improves lung function, but also to understand whether it influences deeper biological mechanisms connected to fibrosis, inflammation, and aging.

Peer-reviewed research has also shown that TNIK inhibition may produce senomorphic effects, meaning it may help reduce harmful features of aging cells without necessarily killing the cells outright.

Why This Matters for AI Drug Discovery

Rentosertib is important because it offers a documented case of an AI-discovered target and AI-designed molecule moving through the full clinical pipeline.

The program has gone from AI-based target discovery to generative molecule design, preclinical validation, Phase I safety testing, randomized Phase IIa clinical data, and now Phase III development.

Alex Zhavoronkov, PhD, Founder and CEO of Insilico Medicine, said rentosertib represents the full arc of Insilico’s mission: using AI not only to move faster, but to originate new biology, new chemistry, and new therapeutic opportunities.

For the AI drug discovery field, this is no longer just a story about speed. It is now a story about clinical translation.

News AI PH Takeaway

AI in medicine is entering a serious proving ground.

It is one thing for AI to identify a target or generate a promising molecule. It is another thing for that drug to survive human testing and move into Phase III, where safety and efficacy must be proven in a larger patient population.

Rentosertib is not yet an approved medicine. Phase III remains a major test. But its progress gives the biotech and AI industries a concrete example of how artificial intelligence may help reshape the future of drug discovery.

From chatbots to clinical trials, AI is no longer just creating content.

It may soon help create life-saving treatments.