382 Precision and Individualized Medicine: Quantification of Brain Cancer Cell Invasion in Confined Microfluidic Spaces Predicts Prognosis

INTRODUCTION: Clinical prognosis of patients with glioblastoma (GBM) is calculated with clinical scores, extent of resection, molecular markers, and tumor subtypes. However, these are often hampered by patient comorbidities, drug toxicity, and other confounding factors. We designed a confined 3D microfluidic technology based on the brain cytoarchitecture by mimicking the diameters of the perivascular spaces and white matter tracts utilized by GBM to invade. METHODS: Patient-derived GBM cell lines (n = 13) were obtained from patients undergoing surgery. After cell culture, the GBM cells’ abilities to migrate and proliferate with MAqCI were evaluated. MAqCI score was calculated via logistic regression. Patients were categorized into low- and high-PFS. Predicted scores were correlated with the clinical status of patients. Cells were treated with gold standard and experimental drugs, and scores were calculated to predict response to treatment. RESULTS: MAqCI Predicted PFS with 83% accuracy, 84.6% sensitivity, and 75% specificity and allowed for the classification of short- and long-PFS (*p = 0.035). Survival based on MAqCI scores separated by short-term (n = 5) and long-term (n = 8) PFS, showed a significant difference between groups (**p = 0.002). After treatment with gold-standard therapy, responder patients were identified based on changes in the MAqCI score. RNAseq of highly motile cells isolated from MAqCI allowed for the identification of druggable targets. An FDA-approved antidepressant against the adrenomedullin gene was selected for MAqCI testing. Changes in MAqCI score after treatment allowed for the identification of responder patients. CONCLUSIONS: Cell invasion properties are directly correlated with patient prognosis. MAqCI could be utilized clinically as a companion assay for predicting tumor recurrence and selecting drugs for individualized GBM treatment.