Particle.news

Download on the App Store

AI-Informed Pipeline Accelerates Bi-Specific CAR-T Design and Clears Tumors in Mice

It promises to streamline preclinical CAR-T development for heterogeneous tumors through rapid computational screening.

Image

Overview

  • St. Jude researchers published a Molecular Therapy paper on August 13 detailing an AI-informed pipeline that screens roughly 1,000 tandem CAR constructs in days to predict surface expression and function.
  • The optimized bi-specific CAR targeting IL13Rα2 and B7-H3 restored T cell surface expression and fully eradicated tumors in four of five mice bearing heterogeneous tumors.
  • The algorithm ranks designs using structural and biophysical fitness scores to identify and refine constructs that overcome misfolding and trafficking failures.
  • Laboratory tests showed the computationally optimized versions outperformed non-optimized tandem CARs across multiple antigen pairs, demonstrating the method’s generalizability.
  • Next steps include extended in vivo safety assessments and progression toward clinical trials, with potential combinations alongside other engineered immunotherapies.