Insilico Medicine’s generative AI platform didn’t just design a drug for lung disease; it accidentally stumbled onto a potential longevity breakthrough. Rentosertib, originally targeted at idiopathic pulmonary fibrosis (IPF), has shown signs of reversing biological aging markers in a small cohort of patients. The findings, published in Nature Biotechnology, rely on data from a 42-patient trial where researchers analyzed blood samples alongside lung function metrics. What makes this notable isn’t just the result, but the consensus: six independent AI aging clocks, developed by teams at Harvard, Oxford, Beijing, and Insilico, all agreed that treated patients appeared biologically younger than those on placebo.
The magnitude of the shift is striking. By week four, the strongest models predicted a biological age reduction of three to four years, with one clock registering a drop of up to six years. This doesn’t mean patients physically rejuvenated overnight. Rather, their blood protein patterns shifted to resemble those of younger individuals. Crucially, the optimal dose for lowering biological age (30 mg twice daily) differed from the dose that best improved lung function (60 mg once daily). This divergence suggests the anti-aging effect isn’t merely a byproduct of reduced physical strain from better breathing, but a distinct biological mechanism. The company further validated this by comparing patient profiles against 55,000 records in the UK Biobank, finding that rentosertib reversed protein changes typically associated with aging.
Skepticism remains warranted, however. Nobel laureate Michael Levitt praised the agreement among disparate models, noting they share neither features nor training data. Yet experts like cardiologist Eric Topol and Harvard’s Vadim Gladyshev urge caution. The sample size is tiny, and the subjects were sick with IPF, not healthy adults seeking longevity interventions. Gladyshev calls it the first clear evidence that predicted biological age can be reduced, but admits we lack definitive trials in healthy populations. Insilico used two AI systems to identify the TNIK protein target and generate the molecule in just 18 months, a timeline far shorter than traditional pharma cycles. With Eli Lilly now invested and Phase III trials underway for IPF, the real test will be whether this AI-designed molecule can deliver consistent results beyond a small, diseased cohort.
