Insilico Medicine, a company that uses artificial intelligence to design drugs, has released the first results in people for its experimental daily pill for mesothelioma, a hard-to-treat cancer. On October 26, 2026, the company said tumors shrank in 2 of 5 mesothelioma patients who could be assessed at the highest dose tested so far, 130 mg once a day1. The drug, called ISM6331, was designed with help from Insilico’s generative AI software, Chemistry421.
The results were presented at the European Society for Medical Oncology (ESMO) Congress 2026 in Madrid, a major cancer meeting, by Melina E. Marmarelis, MD, of the Abramson Cancer Center at the University of Pennsylvania1. The talk was scheduled for October 252. The numbers below come from the company’s announcement; no full paper has been published.
What the company reported
The study is a Phase 1 trial, the first stage of testing in people, which mainly checks safety and what dose to use3. By the data cutoff of August 14, 2026, 36 people with advanced cancers had been enrolled at doses ranging from 15 mg to 130 mg a day, and higher doses were still being tested1. Participants had received a median of 2 earlier lines of treatment, ranging from 1 to 121.
The main numbers for mesothelioma, as reported by Insilico1:
- At 130 mg a day: tumors shrank by a set amount (a “response”) in 2 of the 5 patients who could be assessed, or 40%. The measured tumors shrank by 58% and 36%.
- At doses from 30 mg to 130 mg: tumors shrank in 4 of 22 mesothelioma patients who could be assessed, or 18.2%. The four responses were seen at 30, 50 and 130 mg.
- Disease held in check: among those same 22 patients, the company reported a “disease control rate” of 68.2%, a measure that counts tumors that shrank or stayed stable.
- Timing and length: three of the four responses showed up at the first scan, at week 8. Two came in patients who had already had five and six earlier treatments. Three of the four were still ongoing at the cutoff, one lasting about a year.
The company also described signs in other cancers: measured tumors shrank in a patient with non-small cell lung cancer and one with head and neck cancer, who had already had seven and eight earlier treatments, and the disease stayed stable for more than four months in a patient with a rare blood-vessel tumor (epithelioid hemangioendothelioma) and one with colorectal cancer1.
Side effects
Insilico said no dose caused side effects serious enough to count as “dose-limiting” from 15 mg through 130 mg, and nobody stopped the drug because of side effects1. Most side effects linked to the drug were mild or moderate. Severe (grade 3) ones occurred in 8.3% of participants and included raised levels of the pancreas enzymes amylase and lipase without symptoms, tiredness and a drug skin rash1. Protein in the urine, a side effect the researchers were watching closely, occurred in 11.1%, and the company said it was mild and went away1.
The company said blood levels of a protein called CCN2 fell and stayed lower, which it reads as a sign the drug is hitting its target, and that the drug lasts about 18 to 32 hours in the body, enough for once-daily dosing1.
What the drug is meant to do
ISM6331 is meant to block a family of proteins called TEAD, which help switch on genes involved in cell growth, survival and tissue repair3. They sit at the end of a signaling system called the Hippo pathway, which normally tells cells when to stop dividing3. In mesothelioma, gene changes such as the loss of a gene called NF2 can disrupt this system and let growth signals run unchecked4. The drug is designed to block all four TEAD proteins, which is why it is called a “pan-TEAD” inhibitor2.
Insilico says its software generated the drug’s new chemical backbone using the shape of the target protein and then scored candidate molecules for potency, selectivity and drug-like properties2. As one patient-information site pointed out, using AI does not mean a drug skips normal testing; it still has to pass through lab studies, animal research and clinical trials like any other medicine3.
What these results do not show yet
The groups are very small. The 40% figure rests on 5 patients, so one more or one fewer response would move it a lot. Everyone in the study got the drug; there was no comparison group on another treatment or a dummy pill. The results come from a conference talk described in a company press release, not a peer-reviewed paper.
The study was not built to show whether the drug helps people live longer. Its main goals are safety measures and finding the dose for later studies; tumor response and how long it lasts are secondary goals5. Testing of higher doses was still under way at the cutoff1, and the full trial plans to enroll about 100 adults3.
Patients with mesothelioma in the trial must already have had platinum chemotherapy and an immune checkpoint drug, a type of immunotherapy, while patients with other cancers must show signs of a disrupted Hippo pathway4. So the results apply to people with advanced disease who have run through standard options, not to newly diagnosed patients.
Who is behind it
Insilico Medicine is based in Cambridge, Massachusetts, and has been listed on the Hong Kong Stock Exchange since December 30, 20251. ISM6331 is part of the company’s wholly owned drug pipeline6. The trial runs at sites in China and the United States, and the first patient was dosed in January 20254. The sources do not describe any financial ties of the presenting doctor to the company.
Alex Zhavoronkov, Insilico’s founder and CEO, framed the results as a win for AI-designed chemistry1:
No dose-limiting toxicities so far, and falling blood CCN2 tells us we are hitting TEAD in patients. That is generative chemistry becoming a clinical signal.
He also said the Hippo-TEAD system “sits at the intersection of cancer and aging biology” and could one day matter for age-related disease and “potentially” aging itself1. An earlier company release was more cautious, saying that within cancer, its longevity strategy “does not mean treating aging itself” but going after mechanisms behind serious age-related cancers2. No data on aging were reported.
How regulators have treated it
The U.S. Food and Drug Administration (FDA) gave the drug orphan drug status for mesothelioma in June 2024, a label that supports work on rare diseases2. It cleared the start of human testing in August 20244. In July 2026, the FDA granted Fast Track status for adults whose pleural mesothelioma (in the lining around the lungs) cannot be removed by surgery and has worsened after an anti-PD-1 immune drug and platinum chemotherapy2. Fast Track brings more frequent meetings with the FDA and may open faster review routes if certain criteria are met later5. It was the first Fast Track status for any Insilico program2. None of these steps is an approval.
What to watch next
Insilico says it plans to finish choosing a dose and then, depending on the data, start a Phase 2a study in mesothelioma, a mid-stage trial that would randomly assign patients to different doses to find the best one1. Key questions for that next step: whether the response rate holds up in larger groups, how long responses last, whether side effects change at the chosen dose, and whether a peer-reviewed paper follows with the full data.
