BIOM-72. INTEGRATIVE SINGLE-NUCLEI SPATIAL TRANSCRIPTOMIC PROFILING OF PRIMARY AND RECURRENT HUMAN IDH-WILDTYPE GLIOBLASTOMA REVEALS CELL-SPECIFIC EGFRVIII EXPRESSION AND ANXA1 SIGNALING AXIS

Abstract The World Health Organization provides clear criteria for diagnosing IDH1-wildtype glioblastoma (GBM), relying on traditional approaches to define pathoetiological features, genomic mutations, copy number variation, and aberrant expression of prognostic genes. While these methods guide diagnosis, they may lack the resolution to fully capture tumor heterogeneity, therapeutic response, or outcomes. To address this, we applied leading-edge technologies to 14 primary and recurrent tumors to add granularity to current diagnostic metrics and clinical classification. Aiming to better inform tumor subtyping and improve patient care, this integrative approach enabled us to identify malignant gene programs and localize prognostically relevant features at both single-cell and spatial levels. We analyzed 14 GBM tumors from 6 patients using spatial in situ sequencing (ISS) and single-nuclei RNA-sequencing. A custom ISS gene panel enabled spatial mapping of tumor-associated cell types and state markers, while matched snRNA-seq from adjacent tumor tissue provided transcriptome-wide profiles. One patient contributed 5 serially collected recurrent tumors, enabling longitudinal analysis of tumor evolution. Profiling these tumors using snRNA-seq identified canonical copy number alterations (+7/–10) and ligand-receptor interaction analysis showed an evocative ANXA1 signaling axis progress across recurrent tumors, which suggests possible therapeutic opportunities. Finally, the dataset enabled subcellular and spatial localization of the prognostic GBM biomarker, EGFRvIII, being enriched in oligodendrocyte progenitor (OPC)-like tumor cells, which provides exciting evidence of these cancer cell subtypes as the putative cell of origin for GBM. By extending beyond conventional diagnostic benchmarks, our molecular profiling reveals new insight into GBM biology that may support more refined tumor subtyping and prognostic assessment.