Generative AI-Designed TEAD Inhibitor Shows 40% Response Rate in Heavily Pretreated Mesothelioma

A first-in-human Phase 1 trial of ISM6331, an oral pan-TEAD inhibitor designed with the Chemistry42 generative artificial intelligence platform, has delivered an objective response rate of 40.0 percent at the 130 mg once-daily dose among response-evaluable participants with heavily pretreated mesothelioma. The data, presented as a Rapid Oral at the European Society for Medical Oncology Congress 2026 in Madrid by Melina E. Marmarelis, MD, of the Abramson Cancer Center at the University of Pennsylvania, represent one of the clearest clinical signals yet reported for a drug candidate originating from generative chemistry. Insilico Medicine, the clinical-stage company behind the molecule and listed on the Hong Kong Stock Exchange under code 3696, announced the findings as its investigational therapy advances toward later-stage testing in a cancer with notoriously limited options for patients whose disease has already progressed through multiple lines of treatment.

The trial, registered as NCT06566079, enrolled 36 participants across dose levels ranging from 15 mg to 130 mg, with a data cutoff of August 14, 2026. The study population was heavily pretreated: the median number of prior lines of therapy was two, with a range extending from one to twelve, underscoring how advanced these patients’ cancers were before receiving the experimental compound. Dose escalation remains ongoing as the company works toward completing Phase 1 dose selection. Within this framework, the mesothelioma cohort produced the headline results: across the 30 mg to 130 mg dose range, the objective response rate was 18.2 percent, corresponding to four of 22 response-evaluable participants, while the disease control rate reached 68.2 percent.

The responses were not confined to a single dose level. Partial responses were documented at 30 mg, 50 mg, and 130 mg, suggesting that antitumor activity emerges across a span of exposures rather than clustering at the maximum tested dose. At the top dose of 130 mg once daily, two of five response-evaluable participants achieved objective responses, with target-lesion reductions of 58 percent and 36 percent. Three of the four responders showed evidence of response at their first Week-8 scan, indicating that the drug’s effects became visible early in the treatment course. Notably, partial responses were observed in patients who had already received five and six prior lines of therapy, a level of pretreatment where responses to any new agent are uncommon.

Durability data, though still preliminary, added weight to the signal. Three of the four partial responses were ongoing at the data cutoff, including one that had lasted approximately one year. For a population of mesothelioma patients who had exhausted standard options, the persistence of tumor shrinkage over such an interval is a meaningful observation, even within the small numbers inherent to an early-phase dose-escalation study. The company and its investigators framed these results as supporting continued investigation rather than as definitive evidence of clinical benefit, which is the appropriate reading of Phase 1 data at this stage.

Activity was not limited to mesothelioma. The trial recorded target-lesion shrinkage in heavily pretreated non-small cell lung cancer and head and neck squamous cell carcinoma, in patients who had received seven and eight prior lines of therapy respectively. Stable disease beyond four months was also observed in epithelioid hemangioendothelioma and colorectal cancer. These early signals across multiple tumor types are consistent with the drug’s mechanism, which targets a pathway implicated in a broad range of cancers, though the company has not yet reported response rates for these other indications and the observations remain exploratory at this stage of development.

The safety profile reported to date has been a central part of the announcement. No dose-limiting toxicity was reported from 15 mg through 130 mg, and no treatment-related adverse events led to discontinuation. Treatment-related adverse events were mostly low grade, at Grade 1 or 2. Grade 3 treatment-related adverse events occurred in 8.3 percent of participants and included asymptomatic increases in amylase and lipase, fatigue, and drug dermatitis. Treatment-related proteinuria, designated an adverse event of special interest, occurred in 11.1 percent of participants and was reported as low-grade and reversible. Taken together, these findings supported the company’s characterization of a manageable safety profile across the dose range explored so far.

Beyond tumor shrinkage, the trial generated pharmacodynamic and pharmacokinetic evidence that the drug is engaging its intended biological target. Sustained reductions in circulating CCN2, a biomarker downstream of TEAD activity, support the conclusion that the TEAD pathway is being engaged in patients. The compound’s terminal half-life of approximately 18 to 32 hours supports continuous once-daily dosing, a practical advantage for an oral therapy intended for long-term administration. These measurements matter in early-phase oncology because they connect the molecular design to biological activity in humans, bridging the gap between preclinical optimization and clinical response.

ISM6331 itself is an investigational oral, non-covalent pan-TEAD1-4 inhibitor, meaning it blocks all four TEAD family members rather than a subset. The molecule was designed to block TEAD-dependent transcription downstream of the Hippo pathway, a signaling network whose dysregulation drives tumor growth in multiple cancers. According to Insilico, the compound was maximally multiparameter-optimized using the Chemistry42 generative AI platform, balancing continuous once-daily pharmacokinetics with patient CCN2 pharmacodynamic engagement. The program holds dual U.S. FDA designations for mesothelioma: Orphan Drug Designation granted in June 2024 and Fast Track Designation granted in July 2026, the latter marking Insilico’s first Fast Track program.

The company’s leadership emphasized both the scientific difficulty of the target and the role of artificial intelligence in reaching this point. Alex Zhavoronkov, PhD, Founder and CEO of Insilico Medicine, said that TEAD had sat for years in the industry’s too hard category, and that the company had now reported first-in-human data for a pan-TEAD inhibitor designed using generative AI with a 40 percent ORR at 130 mg in heavily pretreated mesothelioma, deep lesion reductions of 58 percent and 36 percent, disease control in more than two-thirds of evaluable participants, and responses by week eight, including after five and six prior lines. He noted that no dose-limiting toxicities had been seen so far and that falling blood CCN2 indicated the drug was hitting TEAD in patients, describing the result as generative chemistry becoming a clinical signal. He also positioned the Hippo-TEAD axis at the intersection of cancer and aging biology for the company, describing a multi-indication opportunity spanning Hippo-driven cancers and, potentially, age-related disease, consistent with Insilico’s dual-purpose discovery strategy. Halle Zhang, PhD, Vice President of Clinical Development for Oncology at Insilico, said the 40 percent ORR at 130 mg and manageable safety profile in previously treated mesothelioma support continued investigation, and that as dose selection is completed the company plans to advance to Phase 2a randomized dose optimization.

The next steps are clearly defined. Insilico plans to complete Phase 1 dose selection and, subject to emerging data, initiate a Phase 2a randomized dose-optimization study in mesothelioma. Such a study would be the first randomized evaluation of the compound and would be needed to establish whether the response signals seen in small, single-arm cohorts translate into meaningful clinical outcomes. For now, the reported findings remain those of an ongoing Phase 1 trial in a limited number of heavily pretreated patients, with dose escalation still underway. Yet the combination of objective responses across multiple dose levels, early and durable shrinkage in a refractory population, early activity signals in other tumor types, a safety profile without dose-limiting toxicities to date, and biomarker evidence of target engagement gives the program a coherent evidentiary foundation as it moves toward the randomized phase. For a field watching whether AI-designed molecules can survive the transition from computational design to clinical activity, ISM6331’s trajectory through ESMO 2026 offers a concrete, if early, data point.

Subject of Research: First-in-human Phase 1 results of the generative AI-designed pan-TEAD inhibitor ISM6331 in heavily pretreated mesothelioma and advanced solid tumors

Article Title: Insilico Medicine reports 40% ORR at 130 mg for Pan-TEAD inhibitor ISM6331 designed using generative AI in heavily pretreated mesothelioma at ESMO 2026

Article References: Insilico Medicine reports 40% ORR at 130 mg for Pan-TEAD inhibitor ISM6331 designed using generative AI in heavily pretreated mesothelioma at ESMO 2026. Original publication (Bibliographic details unavailable.)

Image Credits: AI Generated

DOI: Not provided

Keywords: ISM6331, Insilico Medicine, pan-TEAD inhibitor, mesothelioma, generative AI, Phase 1 trial, ESMO 2026, Hippo pathway, objective response rate, drug discovery, oncology, CCN2 biomarker