Single-cell profiling identifies a pro-tumoral VCAN positive macrophage subset and defines a prognostic signature in glioblastoma.

Glioblastoma (GBM), the most aggressive primary brain tumor, develops within a tumor microenvironment (TME) dominated by tumor-associated macrophages (TAMs) that critically influence disease progression. Through single-cell RNA sequencing (scRNA-seq) and bioinformatic analysis, we delineated macrophage heterogeneity within the GBM TME and identified a distinct VCAN⁺ macrophage subpopulation. Pseudotime trajectory analysis revealed these cells at the terminal stage of macrophage differentiation, where they exhibit enhanced granulocyte migration and chemotaxis pathways and exhibit a pro-tumorigenic phenotype that diverges from classical M1/M2 polarization. These VCAN⁺ macrophages displayed a distinct polarization state driven by tumor necrosis factor-α (TNF-α), contributing to both a pro-inflammatory and an immunosuppressive TME. CellChat analysis demonstrated their pivotal role in intercellular communication-predominantly mediating crosstalk with GBM tumor cells, endothelial cells, and CD8⁺ T cells via SPP1 signaling: SPP1 binding to CD44 on tumor cells enhances their invasiveness, while its interaction with CD47 on CD8⁺ T cells inhibits anti-tumor immunity. Transcriptional regulatory network analysis identified that CDX2 and MXI1 serve as key transcription factors modulating VCAN⁺ macrophage function and maintaining their specific polarization and homeostasis. Through machine learning, we identified seven hub genes-C1QA, C1QC, C3, CCL4, CD44, SERPINE1, and TREM2 (all highly expressed in VCAN⁺ macrophages and involved in their polarization or intercellular communication)-and constructed an effective GBM prognostic model (AUC = 0.83 in validation cohort), underscoring their roles in immune regulation, extracellular matrix remodeling, and VCAN⁺ macrophage-mediated tumor progression. Further studies with larger cohorts and functional validation will clarify this subpopulation's therapeutic potential and spatial distribution patterns.