Mesothelioma, a rare and aggressive form of cancer typically associated with asbestos exposure, has long been a graveyard for experimental oncology drugs. Characterized by its development in the protective linings of the chest (pleura) or abdomen (peritoneum), the disease leaves patients with limited therapeutic options once standard-of-care treatments fail. However, a breakthrough from Cambridge, Massachusetts-based Insilico Medicine is offering a new glimmer of hope.
Leveraging the power of generative artificial intelligence, Insilico has developed ISM6331, a novel oral TEAD (TEA domain family member) inhibitor. Preliminary clinical results presented at the ESMO Congress in Madrid on October 25, 2026, suggest that this AI-designed molecule could redefine the treatment landscape for patients with refractory mesothelioma.
Main Facts: The Power of Generative Design
The core of this development lies in the intersection of structural biology and machine learning. ISM6331 was synthesized using Insilico’s proprietary Chemistry42 generative chemistry platform. Unlike traditional small-molecule discovery, which relies on high-throughput screening of physical libraries, Insilico’s AI model navigated the chemical space to design a potent, pan-TEAD inhibitor that blocks palmitoylation—a crucial lipid modification required for TEAD proteins to function and drive tumor growth.
TEAD proteins work in tandem with the Hippo pathway’s co-activators, YAP and TAZ. In many cancers, including mesothelioma, the Hippo pathway becomes dysregulated—often due to mutations in the NF2 gene—allowing TEAD to over-activate gene programs that promote unchecked cell proliferation. By targeting all four members of the TEAD family, ISM6331 addresses the compensatory mechanisms that often allow cancer cells to bypass single-target therapies.
The ongoing Phase 1 dose-escalation study, presented by Dr. Melina Marmarelis of the University of Pennsylvania’s Abramson Cancer Center, revealed that among 22 evaluable patients, 18.2% achieved a partial response. Perhaps more impressively, 68.2% of patients experienced disease control, meaning their tumors either shrank or remained stable at the first assessment point.
Chronology: From AI Concept to Clinical Reality
The journey of ISM6331 reflects the accelerating timeline of AI-accelerated drug discovery:

- Pre-2024: Insilico Medicine utilizes its Chemistry42 platform to identify the TEAD binding site and design a molecule capable of deep inhibition of all four TEAD isoforms.
- 2024-2025: The molecule undergoes rigorous preclinical validation, demonstrating strong efficacy in NF2-mutant tumor models and favorable pharmacokinetic profiles.
- Late 2025: Following the publication of positive results for competitor molecules like Vivace Therapeutics’ VT3989, the industry gains confidence in the TEAD-inhibitor class as a viable therapeutic strategy.
- August 14, 2026: Data cutoff for the Phase 1 trial of ISM6331.
- October 25, 2026: Initial clinical results are unveiled at the ESMO Congress in Madrid, highlighting the drug’s potential in heavily pre-treated patients.
- Future Outlook (2027): Insilico plans to initiate a randomized Phase 2a dose-optimization trial to further validate efficacy and establish the optimal therapeutic window.
Supporting Data: Efficacy and Tolerability
The clinical data provided at ESMO provides a nuanced look at the drug’s performance. The trial enrolled patients who had exhausted standard therapies, with participants having received a median of two prior lines of treatment. In some cases, responders included individuals who had undergone as many as five or six previous treatment cycles.
Efficacy Metrics
- Overall Response Rate (ORR): 18.2% (4/22 patients).
- Disease Control Rate (DCR): 68.2%.
- Durability: Three of the four partial responses were observed at the first scan (week 8), and three remained ongoing as of the August data cutoff, with one patient achieving a response lasting nearly a full year.
- Dose Response: At the 130 mg daily dose, two of five evaluable patients achieved a partial response, suggesting that higher concentrations may correlate with better outcomes.
Safety Profile
Safety remains the primary hurdle for TEAD inhibitors, as the pathway is essential for normal tissue homeostasis. As noted by Halle Zhang, Insilico’s vice president of clinical development for oncology, "Proteinuria is the on-target effect described for TEAD inhibitors as a class."
Clinical investigators reported:
- Proteinuria: Four cases observed across 36 patients, all of which were Grade 1 or 2 and reversible.
- Grade 3 Adverse Events: Three patients experienced Grade 3 events at the 130 mg dose, including asymptomatic increases in pancreatic enzymes, fatigue, and drug-related rash.
- Toxicity: No dose-limiting toxicities (DLTs) were reported in the 15–130 mg range, allowing for continued dose escalation.
Official Responses: Insights from the Development Team
Halle Zhang, representing the development team, emphasized the importance of the "pan-TEAD" approach. When asked why the drug targets all four family members, she explained that these proteins are known to compensate for one another. If a drug were to target only a single isoform, the cancer cells could potentially switch to another, rendering the treatment ineffective.
Regarding patient selection, Insilico is currently taking a broad approach. While the trial enrolled patients regardless of their specific Hippo pathway mutation status, the company is actively analyzing tumor biology to identify biomarkers that could predict superior responses. "We are evaluating tumor biology to understand which patients may benefit most," Zhang stated.
The company’s approach to dose selection is equally methodical. Instead of relying solely on the maximum tolerated dose, Insilico intends to determine the optimal dose based on a "totality of data," incorporating pharmacokinetics (PK), pharmacodynamics (PD), safety signals, and overall benefit-risk assessments.

Implications: A New Era for Oncology
The success of ISM6331 places it in a competitive, yet promising, field of experimental medicine. The most prominent rival, Vivace Therapeutics’ VT3989, set a high bar when it reported a 26% response rate in its own Phase 1/2 trial. With Vivace already progressing into a randomized Phase 3 sTEADfast trial, the race to provide a standard-of-care alternative for mesothelioma is clearly heating up.
The AI Advantage
The success of these trials serves as a major vindication for the generative AI model in drug discovery. By reducing the time required for lead optimization and enabling the design of molecules that hit "undruggable" targets, platforms like Chemistry42 are fundamentally shifting the economic and scientific landscape of pharmaceutical R&D.
Addressing Unmet Needs
For patients with mesothelioma, the implications are profound. Because mesothelioma often remains asymptomatic until it reaches an advanced stage, many patients are ineligible for surgery and have limited responses to standard chemotherapy or immunotherapy. A targeted, orally bioavailable drug that can shrink tumors in patients who have failed multiple prior lines of therapy represents a significant advancement in life expectancy and quality of life.
The Path Forward
As Insilico moves toward its 2027 Phase 2a trials, the industry will be watching closely. The key challenges will be twofold: maintaining the favorable safety profile as doses are optimized and definitively proving that the AI-designed molecule offers a survival benefit over existing, albeit limited, therapies.
If the current results hold, ISM6331 could become a cornerstone of future mesothelioma treatment, further cementing the role of generative AI as an essential tool in the modern oncologist’s arsenal. The transition from "in silico" design to clinical "in vivo" success is no longer a theoretical exercise; it is a reality that is currently reshaping the future of cancer care.
