Cancer-associated fibroblasts in glioblastoma: from cellular identity to microenvironmental remodeling and therapeutic translation

Cancer-associated fibroblasts (CAFs) are major regulators of extracellular matrix remodeling, immune suppression, and therapeutic resistance in many solid tumors, but their identity and significance in glioblastoma (GBM) have remained controversial because conventional fibroblasts are sparse within the neural parenchyma. The recognition of resident fibroblast populations at CNS border and perivascular compartments, together with recent single-cell and spatial studies, has reframed this question. This review focuses primarily on adult GBM, IDH-wildtype under WHO CNS5, and examines the identity, cellular origins, heterogeneity, biological functions, and therapeutic relevance of CAFs and CAF-like stromal populations. Current evidence supports heterogeneous stromal states that may arise through expansion or activation of resident CNS fibroblasts and through acquisition of fibroblast-like programs by pericytic, mesenchymal, glial, or other local stromal populations. However, substantial marker overlap with pericytes, reactive astrocytes, and mesenchymal-like malignant cells, together with dissociation- and culture-related biases, makes CAF identity strongly dependent on lineage and methodology. Functionally, current evidence suggests that CAFs and CAF-like stromal populations establish complex interaction networks with glioma stem cells, myeloid cells, T cells, and vascular endothelial cells. Functional studies support roles for these interactions in GSC maintenance, ECM remodeling and tumor invasion, immunosuppressive myeloid reprogramming and T-cell dysfunction, vascular niche remodeling, and ferroptosis resistance, whereas evidence for broader metabolic adaptation remains more limited. Preclinical studies support several therapeutic concepts, including FAP-directed cellular therapy, radionuclide therapy, and targeted drug-delivery strategies, whereas current clinical evidence is largely limited to FAP-targeted molecular imaging. Therapeutic translation remains constrained by CNS delivery, target specificity, stromal heterogeneity, and potential toxicity. We argue that the central challenge is no longer simply whether fibroblast-like stromal populations exist in GBM, but how bona fide CAF lineages and pathogenic stromal programs can be distinguished from transcriptional mimicry. Future studies integrating lineage tracing, single-cell and spatial multi-omics, and functional perturbation will be essential for developing precise, biologically selective stroma-directed therapies.