T-cell bispecific antibodies (TCBs) are a promising class of immunotherapeutic agents that promote tumor cell killing by physical crosslinking of effector T-cells to target expressing cells.
While TCBs are effective in targeting less-immunogenic tumors, they are subject to safety liabilities in normal tissues, which may express low levels of the target. Preclinical assessment of safety risks is crucial but species differences between human and rodent immune responses necessitate the development of advanced human cell-based models for TCB safety profiling. While conventional killing assays are experimentally tractable, they lack physiological organization and cytoarchitecture and often fail to accurately predict efficacy and off-tumor cytotoxicity. Here, we show the robustness of our Intestine-Chip to capture variability in immune cell activation at physiologically relevant TCB concentrations when immune cells are directly applied to the apical epithelium. With increased confidence in the predictive capabilities of the model, we then confirmed expected regional dependence to gastrointestinal (GI) toxicity of the TCB by showing elevated immune cell activation in the large- versus small-intestine. Further, we have demonstrated vascular recruitment and transmigration of circulating immune cells to the intestinal epithelium to more accurately capture the in vivo mechanisms of TCB-mediated toxicity. Together, these data show that our models are suitable for safety profiling of novel engineered immunotherapies and provide clinically relevant results.