AI-designed “armored” CAR T enters first human trial for platinum-resistant ovarian cancer

When ovarian cancer becomes resistant to platinum-based chemotherapy, enabling immune cells to reach tumors and sustain their activity becomes a hurdle that new therapies must overcome. On October 1, Roswell Park Comprehensive Cancer Center and Generate Biomedicines announced that the first patient had been dosed with the experimental cell therapy GB-5267, bringing a concept that combines AI design and immune engineering into its first human trial. No patient safety or efficacy results have been released so far.

GB-5267 is an autologous chimeric antigen receptor T-cell therapy (CAR T): patients’ own T cells are collected, genetically engineered and expanded in culture, then infused back into their bodies. It targets MUC16; the research team says this antigen is highly expressed in many ovarian tumors. Manufacturing after cell collection is handled by Roswell Park’s engineering and cell manufacturing facilities, which comply with good manufacturing practice (GMP).

“Armored” refers to the engineering design that incorporates the cytokine interleukin-18 (IL-18). The microenvironment surrounding solid tumors suppresses immune cells, weakening their ability to enter tumors and function; the team hopes IL-18 will help the modified T cells remain active within that environment. This is the therapy’s design objective, and whether it can be achieved in patients is precisely what the trial will test.

AI’s role in this therapy is to help design and evaluate CAR receptor architectures. Generate says the selection criteria include the T cells’ ability to kill tumor cells bearing MUC16, expand and sustain activity, rather than estimating performance solely from the strength of receptor binding to the antigen. However, the announcement did not provide the scale of the AI training data, complete comparative preclinical data or a quantified advantage over other design methods; the first dosing also cannot yet demonstrate that this design approach delivers clinical benefit.

Trial information listed by the U.S. National Cancer Institute (NCI) provides further details on the treatment arrangements. The Phase 1 study, numbered NCT07489287 and led by Emese Zsiros, enrolls adults with platinum-resistant ovarian, peritoneal or fallopian tube cancer. Participants undergo leukapheresis and receive a cell infusion approximately six to eight weeks later; one group receives intravenous administration only, while another receives both intravenous and intraperitoneal administration. The purpose of the intraperitoneal route is to bring the treatment closer to sites within the abdominal cavity where ovarian cancer commonly spreads.

Another design feature is that chemotherapy for lymphodepletion is not required before infusion. This type of pretreatment is typically used to reduce patients’ existing immune cells and create space for the infused CAR T cells; omitting this step avoids the chemotherapy exposure associated with that pretreatment. However, the side effects of the cell therapy itself and its overall tolerability still require confirmation through human data.

According to the NCI information, the study will gradually increase the dose, primarily assessing safety, tolerability and the maximum tolerated dose, while making preliminary observations of antitumor activity using RECIST 1.1 and iRECIST criteria, with follow-up extending to as long as 15 years. GB-5267 has therefore crossed the threshold into human testing; whether it can control tumors, how long its effects may last and whether the cells can expand sufficiently without lymphodepletion all remain questions for subsequent results to answer.