← Back to Home

AI-Designed Cancer Drug Shows Early Signals: Tumors Shrink in Two of Five Patients in Mesothelioma Trial

Insilico Medicine has reported first-in-human trial data for its TEAD inhibitor ISM6331, with an objective response rate of 40% in the 130 mg group, based on just five evaluable patients. The findings offer clues for further research, but durable efficacy remains to be demonstrated.

By SURL BioNews

For patients with mesothelioma who have received multiple lines of treatment, whether a new drug can shrink tumors is an important clue in identifying the next treatment option. Insilico Medicine recently reported antitumor activity in a Phase I trial of ISM6331, an oral drug candidate designed with the involvement of generative AI. The objective response rate in the 130 mg daily group was 40%. However, that figure represents two of five evaluable patients and must be interpreted alongside the sample size.

The company released the findings on October 25, and the results were also presented at the 2026 European Society for Medical Oncology Congress in Madrid. The two patients’ target lesions shrank by 58% and 36%, respectively. Across the combined 30–130 mg dose groups, four of 22 evaluable patients with mesothelioma responded, yielding an objective response rate of 18.2%. The disease control rate at the first tumor assessment was 68.2%, but this included stable disease and cannot be interpreted as significant tumor shrinkage in every case.

S&P Capital IQ’s report on the same event, published through MarketScreener, also noted that three of the four patients with partial responses were still responding at the data cutoff, including one whose response had lasted approximately one year. As of August 14, 2026, trial NCT06566079 had enrolled 36 participants across doses ranging from 15 to 130 mg. Participants had received a median of two prior lines of treatment, with a range of one to 12. This background shows that the study involved previously treated patients and also underscores how widely individual disease courses may vary.

ISM6331 is designed to inhibit TEAD1 through TEAD4, blocking TEAD-dependent gene transcription downstream of the Hippo signaling pathway. According to the company, the Chemistry42 generative AI platform was involved in molecular design and property optimization. The study also observed a decrease in blood CCN2 levels, supporting an effect on the intended pathway in humans. However, such biomarker changes cannot yet substitute for evidence that patients live longer or experience improvements in quality of life, nor can this trial alone establish that AI design is superior to other drug development methods.

On safety, the company said that no dose-limiting toxicities or discontinuations due to treatment-related adverse events had occurred as of the data cutoff. Grade 3 treatment-related adverse events nevertheless occurred in 8.3% of participants, with reported events including asymptomatic elevations in amylase and lipase, fatigue, and drug-induced dermatitis. Treatment-related proteinuria occurred in 11.1%, with all cases low-grade and reversible. These risks will require continued monitoring as more patients receive the drug and treatment duration increases.

Currently available public information comes primarily from the company’s announcement and reporting on the same event, rather than two independent sets of clinical validation. Dose escalation remains ongoing in the Phase I trial, and the response rate in a five-patient group can change substantially as new patients are added. The different dose groups also cannot be compared directly, and the findings are insufficient to establish 130 mg as the optimal dose. The current results provide a reason to continue research but have not established a benefit relative to other therapies.

Insilico Medicine said it will next complete Phase I dose selection, then use subsequent data to decide whether to initiate a randomized Phase IIa dose-optimization trial in mesothelioma. The key next steps are to confirm whether responses can be reproduced in more patients, how long they last, and which dose offers a better balance between efficacy and toxicity.

References

  1. BigGo Finance
  2. Insilico Medicine
  3. S&P Capital IQ via MarketScreener