Abstract Introduction: Despite significant advancements in scientific and diagnostic technologies, clinical trials for cancer often fail due to a lack of efficacy. Access to a platform that accurately predicts patient response is crucial to increase the success rate of clinical studies, and to minimize unnecessary exposure to ineffective and potentially harmful drug testing for patients. Because of the complexity of the disease, it would be beneficial to deploy patient samples for drug testing in a side-by-side comparison during the clinical trial. An accurate prediction of patient response earlier in the clinical study would aid decision making to determine the appropriate duration of treatment for each patient. Methods: We developed a 3D human cell co-culture model using our proprietary MPS technology, the Curio-spheroid Curiochip. Using patient samples isolated and expanded from either ascites fluids or biopsies to recapitulate the clinical phenotypes of advanced gastric cancer, we quantified the migratory properties of the patient cancer cells (i.e., the unique metastatic potential), and measured the level of secretion of VEGF. Somatic mutational analyses using NGS and RNAseq were performed for subtyping and clustering the samples. The patient cancer cells were expanded within 2 weeks post-collection, and used to form a spheroid that was loaded into a hydrogel embedded within the Curio-Spheroid, enabling the patient spheroids to interact through the hydrogel with neighboring (healthy) human endothelial cells. The patient spheroids induced aggressive vasculogenesis in the co-culture model, which was subdued by the anti-VEGFR2 antibody Ramucirumab. The platform was retrospectively validated with the Standard-of-Care (SOC) treatments that the patients received. Results: The patient cell testing took 3-4 weeks from sample collection to imaging. The patient cells preserved their migratory capacity, pro-angiogenic potential and drug responses in the vascularized model. The dose responses to SOC in the platform were consistent with the actual patient response assessments based on RECIST criteria, supporting the potential to use this platform to accurately predict patient clinical responses. Conclusions: The Curio-spheroid 3D co-culture model of vascularized gastric cancer provides a dynamic microenvironment to preserve the phenotypes of the patient cells and cell-cell interactions that are known to influence drug responses. This platform is compatible with high-throughput screening (HTS) for target or drug discovery with a good assay window. The ability to accurately recapitulate patient responses to SOC in the clinic make the Curio-Spheroid platform a promising method to test patient samples and measure drug responses of patients in real time during ongoing clinical trials. Citation Format: Yeongmin Choi, Sumin Kim, Youngsook Song, Yangji Kim, Jin Hawn Jang, Suryong Kim, Jihye Baek, Delaney Donnelly, Bushra Rajiput, Byungjun Lee, Woo Sun Kwon, Tsung-Li Liu, Kyusuk Baek, Sanghee Yoo, Sun Young Rha. Development of an in vitro gastric cancer 3D co-culture platform using a microphysiological system (MPS) to improve outcomes in oncology clinical trials [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6789.