Idiopathic pulmonary fibrosis gives patients a median survival of two to four years after diagnosis. That grim math is why the latest news out of Insilico Medicine matters: the company’s AI-discovered drug rentosertib is heading into Phase III trials, and this time the data backing it up is hard to wave away.
What Rentosertib Actually Does
Rentosertib is an oral drug that inhibits TNIK, a kinase most existing antifibrotic drugs never touched. In a randomized trial <cite index=”1-8,1-9″>across 22 Chinese clinical sites, 71 patients split between placebo and active treatment groups received either 30 mg or 60 mg daily doses over 12 weeks</cite>.
The results were the kind that make a trial worth funding further. <cite index=”1-10″>Patients on the 60 mg once-daily dose gained a mean of 98.4 mL in forced vital capacity, while the placebo group lost 20.3 mL over the same period</cite>. Lung function that improves instead of declining is not something IPF patients see often. Safety held up too, with <cite index=”1-10″>adverse events tracking close to baseline rates across every arm</cite>, and the drug already carries <cite index=”1-10″>FDA Orphan Drug Designation, granted back in February 2023</cite>.
How the AI Actually Found the Target
Here’s the part that separates this from a typical biotech headline. <cite index=”1-13,1-14″>Insilico’s PandaOmics engine mined genomics data, clinical trial results, scientific literature, and patent records to build biological network models</cite>, and it landed on TNIK as the key target, <cite index=”1-16″>a node that regulates fibrosis and inflammation through Wnt, TGF-beta, Hippo/YAP-TAZ, JNK, and NF-kB signaling</cite>. No human researcher pointed it there first.
From Molecule Design to the Clinic
Once TNIK was locked in, the Chemistry42 engine took over. Rather than screening existing compound libraries, it used generative reinforcement learning to design molecules that would physically fit the target protein. <cite index=”1-24″>The system synthesized 79 candidate molecules, and the 55th became the preclinical candidate</cite>, a process that <cite index=”1-24″>cut the timeline from project start to preclinical nomination down to 18 months</cite>.
Why This Trial Matters Beyond Insilico
This isn’t just a pipeline update. Co-CEO Feng Ren put it plainly: rentosertib “came from a biology-first, ageing-informed AI workflow,” not a conventional screen. Founder Alex Zhavoronkov called it a shift from a speed story to a “clinical translation story.” For an industry full of AI drug discovery claims, a Phase III readout offers something rarer: a real answer.
The Bottom Line
Plenty of companies say AI can find drugs faster. Few have carried one from an algorithm’s first guess through Phase IIa efficacy data and into a Phase III trial with numbers this clean. If rentosertib holds up, it becomes a genuine proof point for AI-driven drug discovery, not just another pitch deck slide.
Curious how AI is reshaping healthcare software beyond pharma? Check out our healthcare technology insights for more on where the industry is headed. For the full clinical breakdown, read the original coverage on AI News.
