Multimodal profiling of CAR T cells against glioblastoma using a microengineered 3D tumor-on-a-chip model.

Immunotherapies such as chimeric antigen receptor (CAR) T cells have shown promising outcomes in hematological cancer but face challenges in targeting solid tumors like glioblastoma (GBM). Advancing this therapy for GBM has been hindered by the lack of preclinical tools that accurately model the complex interplay between CAR T cells and tumor cells within the tumor microenvironment (TME) - interactions critical for optimizing CAR constructs and improving efficacy. Physiologically relevant models that closely mimic the solid TME are therefore highly sought after in developing CAR T therapies. Here, we report a microengineered glioblastoma-on-a-chip (GOC) model with a functional vascular network to investigate the efficacy and selectivity of IL-13 mutein CAR T cells (TV-13) against U87 GBM tumor cells expressing high interleukin-13 receptor alpha-2 (IL13Rα2), compared with the ubiquitously expressed IL13Rα1. This biomimetic platform recapitulates the GBM TME and enables dynamic evaluation of CAR T cell responses under locoregional administration, paralleling clinical approaches. Using the organotypic GOC model, we evaluated CAR T cell-mediated inhibition of GBM invasion, monitored real-time dynamic CAR T-U87 interactions, and quantified the release of cytotoxic, proinflammatory, and stimulation-associated cytokines as measures of T cell effector function. CAR T cells induced a density-dependent reduction in U87 migration, accompanied by robust cytokine release, while TV-13 maintained specificity towards IL13Rα2 tumor antigen over IL13Rα1. Additionally, we further demonstrated the efficacy of CAR T cells against patient-derived GBM cells within the GOC model. Collectively, these findings highlight the GOC platform as a powerful preclinical screening tool for cancer immunotherapy optimization.