173 A Reproducible Extracellular Matrix Program Defines Mesenchymal Glioblastoma and Identifies Candidate Microenvironmental Vulnerabilities

Abstract Glioblastoma (GBM) comprises molecularly distinct subtypes, with the mesenchymal class associated with invasion, therapeutic resistance, and poor prognosis. Extracellular matrix (ECM) remodelling is a defining feature of this phenotype, yet reproducible ECM drivers across independent cohorts remain incompletely characterised. We performed an integrated bioinformatic analysis across multiple publicly available datasets (GSE16011, GSE13041, GSE4290, and TCGA-GBM with IDH-status annotation). Differential expression was assessed using non-parametric testing with Benjamini–Hochberg correction. Cross-cohort reproducibility was evaluated using Fisher’s meta-analysis, and survival associations were analysed using Cox proportional hazards modelling. A conserved ECM transcriptional program was identified across datasets and was enriched in mesenchymal and IDH-wildtype tumours. Canonical mesenchymal-associated ECM genes, including CD44, FN1, and TNC, demonstrated concordant upregulation across cohorts. Integrated tiering defined a high-confidence core ECM signature reproducibly characterising the mesenchymal phenotype. Several components of this signature were associated with hazard ratios >1, linking elevated ECM expression to adverse survival. These findings reinforce the ECM as a biologically coherent determinant of mesenchymal GBM and support prioritisation of extracellular components as candidate microenvironmental vulnerabilities. Given their extracellular localisation and subtype specificity, these targets represent rational candidates for local therapeutic modulation and functional evaluation in biomimetic 3D systems.