Computer-guided dosing of multi-specific T-cell engagers

Patient-specific tumor heterogeneity limits the efficacy of conventional T-cell immunotherapies. We report MATCH, a modular split-antibody T-cell engager platform, and COMPASS, a computational framework that predicts individualized multi-specific therapy compositions. COMPASS integrates T-cell counts, tumor burden, and antigen expression (CD19, CD20, CD38) to identify dose combinations that maximize tumor cell killing while preserving T-cell viability. COMPASS accurately predicted effective multi-specific formulations for chronic lymphocytic leukemia patient PBMC samples and successfully predicted compositions to treat a human lymphoma xenograft murine model that led to significantly improved survival. This lab-in-the-loop experimental-computational approach provides a generalizable blueprint for precision T-cell immunotherapy and offers a new paradigm in patient-specific cancer treatment. A closed-loop experimental-computational framework for individualized T-cell immunotherapy design and precision dosing.