A human multiorgan microphysiological system for modeling immunotherapies tumor clearance and normal tissue immune injury

Despite the clinical success of CTLA-4 blockade, immune checkpoint inhibitors can trigger immune-related adverse events (irAEs) that remain difficult to predict preclinically, in part because conventional in vitro and animal models lack the capacity to simultaneously evaluate ECM-dependent T-cell infiltration, tumor cytotoxicity, and normal tissue immune tolerance. We developed a multiorgan system to quantify how fibrin matrix density, T-cell donor identity, HLA compatibility, and ipilimumab dose collectively shape immune cytotoxic responses across tumor and normal tissues. Imaging experiments compared fibrin concentrations from 2.5-10 mg/mL to assess physical barrier effects on immune infiltration. Brain (SK-N-SH/HMC3), gut (T84) tumor spheroids, or human primary hepatocyte spheroids with or without Kupffer cells were embedded in a defined fibrin hydrogel (6 mg/mL fibrinogen, 4 U/mL thrombin). Tumor models used two T-cell donors, while the liver model was perfused with either matched or mismatched CD8⁺ T-cell donors. Ipilimumab was applied from 0-10 µg/mL. Spheroid density, derived from fluorescence-based viability measurements, was plotted against antibody concentration.T-cell penetration into spheroids was strongly dependent on fibrin density with low-to-moderate matrix (2.5-6 mg/mL) supported deep infiltration, whereas high-density fibrin (8-10 mg/mL) restricted immune access and created an immune-excluded phenotype. CTLA-4 blockade overcame this barrier in part, enabling infiltration even through dense matrices. Dose-response analysis demonstrated robust tumor killing in both brain and gut spheroids, exhibiting a steep viability decline beginning at 1 µg/mL and reached near-complete loss at 10 µg/mL.Response magnitude varied between donors, reproducing clinically observed heterogeneity in immune potency. In contrast, normal hepatocyte spheroids remained intact under matched donor conditions across all doses. However, mismatched T cells produced clear dose-dependent injury that was significantly amplified in the presence of Kupffer cells, revealing a macrophage-dependent mechanism of immune-mediated hepatotoxicity.This multiorgan platform recapitulates fibrin-driven immune exclusion, checkpoint-enhanced infiltration, effective tumor cytotoxicity, and HLA mismatch-dependent normal tissue injury, mirroring key therapeutic and toxic outcomes observed in patients. We evaluated antigen-specific T cells, bispecific antibodies, and antibody-drug conjugates to assess on-target efficacy, off-target toxicity, and bystander effects. These findings support the use of human microphysiological systems as predictive tools for immunotherapy, enabling mechanistic dissection of efficacy-toxicity trade-offs prior to clinical testing and supporting FDA-aligned strategies to reduce non-human primate use in monoclonal antibody safety testing.