Genomic heterogeneity drives distinct infiltration patterns in glioblastoma.

Intertumoral heterogeneity in glioblastoma-driven by both genomic and transcriptomic variation-complicates our understanding of how different tumor cell populations contribute to disease progression. Infiltrating tumor cells, which invade surrounding brain tissue and evade surgical resection, are thought to play a central role in recurrence. To address this, we aimed to characterize the gene expression profiles and cellular states of infiltrative tumor cells in glioblastoma. We performed high-plex spatial transcriptomics using the CosMx Spatial Molecular Imager (NanoString) on tumor tissue from eight glioblastoma patients. Formalin-fixed paraffin-embedded samples were selected to capture both the tumor core and invasive margin. A targeted panel of 1,000 genes enabled spatially resolved gene expression profiling at single-cell resolution, allowing precise identification and localization of malignant and non-malignant cell states. We show that malignant cells can be distinguished from non-malignant populations by using patient-specific clustering. Based on this annotation, we identified several known malignant states-including AC-, OPC-, NPC-, and MES-like cells-as well as a recently characterized glial-progenitor (GPC)-like state. This population co-expressed genes associated with both astrocytic and oligodendrocyte progenitor lineages and was found to be more proliferative than the traditional AC-like state. The GPC-like state was most enriched in the classical glioblastoma subtype and was strongly associated with EGFR amplification or mutation. Spatial analyses investigating malignant differences between tumor and infiltrated tissue showed heterogeneous infiltration patterns across patients. In the most extreme case, the dominant GPC-like population in the tumor core gave way to increased proportions of AC-like cells in infiltrated regions. Our study highlights diverging infiltration patterns across glioblastoma tumors, with indications of a GPC-like to AC-like transition occurring in classical-subtyped tumors. This shift is associated with a decrease in cell proliferation and may have implications for clinical treatment.