Computational modelling of CAR T-cell therapy: from cellular kinetics to patient-level predictions

Chimeric Antigen Receptor (CAR) T-cell therapy is characterised by the heterogeneous cellular kinetic profile seen across patients. Unlike traditional chemotherapy, which displays predictable dose-exposure relationships resulting from well-understood pharmacokinetic processes, CAR T-cell dynamics rely on complex biologic factors that condition treatment response. Computational approaches hold potential to explore the intricate cellular dynamics arising from CAR T therapy, yet their ability to improve cancer treatment remains elusive. Here we present a comprehensive framework through which to understand, construct, and classify CAR T-cell kinetics models. Current approaches often rely on adapted empirical pharmacokinetic methods that overlook dynamics emerging from cellular interactions, or intricate theoretical multi-population models with limited clinical applicability. Our review shows that the utility of a model does not depend on the complexity of its design but on the strategic selection of its biological constituents, implementation of suitable mathematical tools, and the availability of biological measures from which to fit the model.