Abstract BACKGROUND The emergence of personalized therapy, which targets oncogenic drivers, has provided a novel adjunct to the standard-of-care therapy for glioblastoma, the most common central nervous system malignancy. Among the less frequently observed oncogenic drivers in glioblastoma are mutations in the proto-oncogene B-Raf. Although treatment options for BRAF-mutated malignancies have been extensively investigated, the implications of BRAF and MEK inhibition in gliomas remain ambiguous. In this study, we present a model system designed to predict cellular temporal trajectories, employing human neocortical slice models and patient-derived tumor cells, coupled with spatially resolved and single-cell transcriptomic analyses for comprehensive computational modeling. MATERIAL AND METHODS Non-neoplastic cortical tissue was procured from a BRAF(V600E) mutant glioblastoma patient during surgical intervention. The tissue was sectioned and inoculated with autologous glioblastoma cells to establish an ex-vivo personalized brain slice model. To facilitate adequate tumor growth, slices were cultured for seven days with clinically approved BRAF/MEK inhibitors and respective controls. For high-dimensional analysis, the tumor specimen and slices in all treatment conditions were profiled using live cell imaging, spatially resolved transcriptomics (Visium 10X) and single nucleus RNA-sequencing (10X). Advanced computational modeling techniques, encompassing graph neural networks and single-cell deconvolution, were employed to scrutinize the spatio-temporal evolution of the tumor and its microenvironment. RESULTS Live cell imaging showed varied treatment responses in tumor growth during culture. Immunofluorescence and immunohistochemistry analyses (using NeuN, GFAP, and Iba-1 cell markers) indicated no toxic effects from drug combinations on the tumor microenvironment. Single nucleus profiling revealed a tumoral trajectory towards induction of mesenchymal-like transcriptional programs under BRAF/MEK inhibition along with up-regulation of alternative receptor tyrosine kinases such as HER2/3 and EGFR. Spatial projection of single cell trajectories using cytospace confirmed the escape mechanism of BRAF/MEK inhibition due to acquisition of anti-inflammatory myeloid cells in close proximity to the tumor niches and enhancement of mesenchymal transcriptional programs. In line with the experimental set-up, we observed a rapid progression of the tumor und BRAF/MEK inhibition in the patient. CONCLUSION Our reported model system, employing high-dimensional readouts and advanced computational modeling, showcases a promising avenue to address uncertainties in personalized medicine, emphasizing that conventional oncological drivers may necessitate alternative treatment strategies in central nervous system malignancies.