Antonios Samiotakis: No financial relationships to disclose
Cole Zmurchok: No financial relationships to disclose
Objectives: Bispecific T-cell engagers (TCEs) are a promising therapeutic modality in cancer immunotherapy. Discovery and development of TCEs often involves testing molecules in vitro and in humanized mouse xenograft studies to assess anti-tumor efficacy and obtain an early understanding of pharmacokinetic and pharmacodynamic properties. We observed a range in in vivo tumor efficacy from a set of PSMA x CD3 bispecific TCEs, and developed a simple Quantitative Systems Pharmacology (QSP) model encoding key mechanisms of action to explain these diverse outcomes. The QSP model provides insight into the molecular properties required for efficacy, and can be used to inform molecule selection for in vivo testing and further development.
Methods: We engineered and assessed PSMA x CD3 TCEs with diverse molecular properties in vitro and in vivo, and developed a simple QSP model that describes the formation of CD3-antibody-PSMA trimers. The model can be configured as a single compartment to fit in vitro T cell-dependent cellular cytotoxicity (TDCC) data, or as three compartments (central, peripheral, and tumor) to fit in vivo mouse xenograft tumor efficacy studies. The model accounts for the key mechanisms of action, including binding CD3 on T cells and PSMA expressed on cancer cells, avidity, internalization, distribution, and elimination, and quantifies immune synapse formation through CD3-antibody-PSMA trimers.
Results: Trimer formation, as calculated by the model, better explains TDCC activity than drug concentration for a set of PSMA x CD3 antibodies with diverse molecular properties across varying target expression levels and effector-to-tumor-cell ratios. Sensitivity analysis shows that avidity and CD3 binding affinity are key parameters for the in vivo QSP model. The avidity parameter was calibrated from the TDCC data. When fitting data from in vivo studies, a linear relationship between efficacy and model-predicted trimers was established, explaining most of the differences in observed efficacy. Structural information on the formation of the trimer complex beyond the mechanisms included in the QSP model is likely to provide additional insight into the remaining unexplained differences in the efficacy-trimer relationship.
Conclusions: The simple mechanistic model captures key aspects of bispecific T-cell engagement in vitro and in vivo, demonstrating the value of trimer predictions in understanding the properties of diverse antibodies. Additional structural information on the immune synapse could play an important role in predicting efficacy and a better understanding of the in-vitro-to-in-vivo correlation in T-cell engaging bispecific molecules. Future work will focus on incorporating aspects of immune cell dynamics and cytokine release into the model.
Citations: N/A
Keywords: T cell engagers, in vitro and in vivo efficacy, model-predicted trimers
