What happened
Researchers from the Perelman School of Medicine at the University of Pennsylvania and Penn's Abramson Cancer Center developed a human-in-the-loop AI framework to identify viable target antigens for CAR T cell therapy. Published in Cell, the framework combines large language models with human expertise, nominating glycoprotein non-metastatic melanoma protein B (GPNMB) as a top candidate. Preclinical testing demonstrated GPNMB-targeted CAR T cells effectively killed tumours in mouse models across multiple cancer types, including melanoma, leukaemia, and colorectal cancer. This AI-driven process reduced target discovery time from months or years to weeks.
Why it matters
Accelerating CAR T cell target discovery significantly reduces the timeline and cost for developing new immunotherapies, directly impacting pharmaceutical R&D teams and clinical trial sponsors. This modular, disease-agnostic AI framework, leveraging public datasets, democratises target identification, making advanced research accessible beyond major institutions. The ability to rapidly validate targets like GPNMB across multiple cancer types expands the potential application of CAR T therapies beyond blood cancers, where current FDA-approved treatments are limited.
