AI Platform Aims to Revolutionize Cancer Immunotherapy

A new AI platform developed by researchers at the Technical University of Denmark and the Scripps Research Institute has dramatically accelerated the design of specialized protein molecules called “minibinders” to enhance cancer immunotherapy. Normally, T cells recognize cancer cells by detecting protein fragments (peptides) presented on their surface by molecules known as pMHCs, but harnessing this natural mechanism for treatment has been slow and challenging due to the variability of T-cell receptors.

This AI system rapidly generates minibinders, custom proteins that precisely guide a patient’s T cells to recognize and kill cancer cells by targeting these pMHC molecules. Remarkably, the platform can design effective minibinders in just 4 to 6 weeks, a process that traditionally takes years. Laboratory tests showed that minibinders designed against a well-known cancer target, NY-ESO-1, created engineered immune cells called IMPAC-T cells that efficiently eliminated cancer cells. The platform also successfully created minibinders targeting unique cancer markers from individual patients, demonstrating its ability to tailor treatments to specific tumor profiles.

A vital innovation of this platform is the integrated “virtual safety check,” where AI screens minibinders for potential harmful cross-reactions with healthy cells before lab experiments, significantly reducing risks and improving treatment safety. This virtual filtering enhances precision by reducing side effects typically associated with immunotherapies.

The clinical pathway envisioned is similar to current CAR-T therapies: patients undergo a standard blood draw, their immune cells are isolated and genetically modified with AI-designed minibinders, then reinfused to seek and destroy cancer cells with high specificity. Researchers project this approach could enter human clinical trials within about five years, thus significantly shortening the timeline from discovery to patient treatment.

Source.