AI drug discovery has spent years promising faster ways to find medicines. Rentosertib is now raising a more unexpected possibility.
Insilico Medicine said patients who received the AI-assisted experimental drug in an earlier pulmonary-fibrosis trial showed changes in biological-age measurements associated with younger profiles. The finding does not establish that rentosertib slows human aging, and improvements in the underlying lung disease could partly explain the signal.
What makes the result more intriguing is its history: the connection between the drug’s target and aging biology appears to predate the latest clinical analysis, stretching back to the AI-assisted work that helped identify the target in the first place.
AI helped choose the target and build the drug
Insilico used PandaOmics to identify TNIK, a protein involved in cell signaling and fibrosis, as a drug target. Chemistry42 then generated and refined compounds against TNIK until researchers selected the molecule that became the experimental treatment.
About 18 months after the search began, the team had a preclinical candidate ready to advance.
AI handled the early search and molecule generation. Human testing would decide whether the treatment could continue through the next stages of AI drug development.
Rentosertib advances into late-stage human testing
Phase 3 trial was announced in July, with plans to enroll 320 people across 47 sites in China and follow treatment for 52 weeks.
Earlier human results were strong enough for the program to advance. Late-stage drug trials give AI in healthcare a clinical benchmark that software demos cannot provide.
Nobel laureate Michael Levitt said the consistency of the aging findings caught his attention, though improvements in lung disease could be influencing the readings. “The experiment in healthy volunteers is the one I want to see next,” he said.
Researchers have therefore cautioned that the current evidence does not show that rentosertib slows aging in humans.
What eWeek found: Aging was part of the AI target work years earlier
eWeek compared Insilico’s 2024 discovery paper with the new aging study and found that the longevity angle predates the latest clinical analysis. Earlier research had linked TNIK to aging biology years before human data produced the new aging signal.
Researchers therefore were not adding an aging rationale after seeing the latest results. Part of the connection was present during AI-assisted target selection, before the drug entered clinical testing.
Pharma teams evaluating AI companies for drug discovery should ask what happens to secondary findings uncovered during target selection. Keeping those relationships traceable could let researchers revisit them years later when clinical evidence points in an unexpected direction.
Platform buyers should also ask whether vendors retain enough evidence to reconstruct why an AI system ranked a target highly in the first place. Rentosertib suggests early biological connections can remain relevant long after a molecule leaves discovery, potentially guiding researchers toward another therapeutic use.
Rentosertib still has a long way to go before anyone can call it an anti-aging treatment. A dedicated study in healthy volunteers would help separate a direct effect on aging biology from changes caused by treating pulmonary fibrosis.
For now, the more defensible finding is also the more interesting one: an AI-assisted drug program built around lung disease has resurfaced an aging connection that was hiding in the target biology years before the human data arrived.
More AI news: Apple Vision Pro entered a Duke Health operating room to keep arthroscopic video, CT scans, and planning tools directly in the surgeon’s field of view.
