e15027 Background: Lung cancer is the leading cause of death worldwide. Metastatic lung cancer patients often relapse or grow refractory to chemotherapy, radiotherapy, and checkpoint inhibitors. Adoptive T cell therapy (ACT) has been garnering more attention due to longitudinal complete responses. Tumor resections from patients harbor tumor-infiltrating lymphocytes (TILs) which can be cultured ex vivo and assessed for tumor reactivity before infusion. This generally follows a patient lymphodepletion regimen which allows the transferred T cells an optimal environment to proliferate and survive in the patient. TIL ACT can produce complete responses – rarely ever observed using traditional onco-therapy – in metastatic melanoma patients. However, while TILs specific to the neo-antigens expressed by tumor cells can be expanded ex vivo, this observed specificity is low in the clinic. Researchers have attempted to solve this by priming ex vivo expanded T cells with antigen-presenting cells previously pulsed with peptides representative of the neo-antigen repertoire of the matched tumor, but have rarely observed complete responses, likely due to current biomarkers for T cell activity poorly predicting anti-tumor cytotoxicity. To date there has been no acceptable potency assay for manufactured TILs, a requirement by the FDA for approval to use them in the clinic. Thus, the need to assess potency of ex vivo engineered T-cells against matched tumor cells is evident. Methods: We have developed a novel diagnostic immune-oncology (IO) pipeline, which uses a membrane-microfluidic platform to culture patient-derived tumor micro-organospheres (MOs) in extracellular matrix droplets. MOs can be rapidly established following patient tumor sample acquisition through biopsies or resection, and preserve the stromal cell populations in the original tumor microenvironment, as characterized by both flow cytometry and single-cell RNA-seq. An automated imaging assay was further established to robustly quantify the amount of immune-induced apoptosis of tumor cells in the MOs by patient-matched TILs, which is highly specific and yields minimal background. Results: We find that this method is not only amenable to high-throughput microscopy, but the larger surface-area-to-volume ratio of micro-organospheres also allows greater TIL infiltration and interaction with tumor cells. The resulting highly-sensitive assay requires far fewer input immune and tumor cells to achieve robust, clinical grade sensitivity response, making it the first clinically feasible assay for testing personalized TIL potency. Conclusions: The MO IO technology is currently being used for assessing clinical efficacy of manufactured TIL products for an upcoming ACT trial for non-small cell lung cancer patients. This technology also provides a companion to TIL ACT, CAR T therapy, and other immunotherapies, for which the ability to predict clinical potency is generally lacking.