Search

Cancer-associated fibroblasts (CAFs) constitute a key component of the tumor microenvironment. Pro-tumoral cancer-associated fibroblasts were presumed absent in glioblastoma given the lack of brain fibroblasts. Using single-cell RNA sequencing we identified CAFs in patient GBMs. CAFs were identified using a negative selection strategy to filter endothelial, epithelial, immune cells and pericytes and for the positive expression of previously defined CAF markers. Copy number variation (CNV) analysis was performed to distinguish CAFs from malignant cells. Single-cell spatial transcriptomics from 16 GBM patients confirmed the proximity of CAFs to mesenchymal GBM stem cells (GSCs), endothelial cells, and M2-macrophages. Serial trypsinization of glioblastoma specimens yielded cells with CAF morphology and transcriptomic profile. CAFs were chemotactically attracted to GSCs and CAFs induced GSC proliferation. To identify CAF and GSC interaction mediators, we created a resource of inferred crosstalk by mapping expression of receptors to their cognate ligands, identifying PDGF and TGFB as mediators of GSC effects on CAFs, and osteopontin and HGF as mediators of CAF-induced GSC enrichment. Furthermore, CAFs were found to induce M2-macrophage polarization by producing the EDA fibronectin variant which binds macrophage toll-like receptor 4 (TLR4) in a targetable manner. Glioblastoma CAFs were enriched in the subventricular zone which houses the neural stem cells that houses GSCs. Including CAFs in GSC-derived xenografts induced in vivo tumor growth and reduced survival in two different xenograft models. These findings are among the first to identify glioblastoma CAFs and their GSC interactions, making them an intriguing target for Glioblastomas.

Cancer-associated fibroblasts (CAFs) constitute a key component of the tumor microenvironment. While pro-tumoral CAFs have been identified in some cancers, CAFs had been presumed absent in glioblastoma given the lack of brain fibroblasts. We found that serial trypsinization of primary glioblastoma cultures yields cells that morphologically resemble fibroblasts and transcriptomically resemble CAFs as shown by bulk RNA-seq and single-cell RNA-seq. Moreover, Single-cell RNA-seq from patient GBMs showed a mesenchymal lineage for CAFs. We demonstrate that Glioblastoma CAFs are chemotactically attracted to glioblastoma stem cells (GSCs) and CAFs enriched GSCs. To identify CAF/GSC interaction mediators, we created a resource of inferred crosstalk by mapping the expression of receptors to their cognate ligands/agonists. This analysis suggested PDGF-b and TGF-b as mediators of GSC recruitment and proliferation of CAFs, and osteopontin and hepatocyte growth factor (HGF) as mediators of CAF-induced GSC enrichment, hypothesis confirmed by blocking antibodies. Glioblastoma CAFs also induce hypertrophied vessels and M2 macrophage polarization, the latter through unique CAF production of the EDA fibronectin variant which binds macrophage toll-like receptor 4 (TLR4) in a targetable manner. Glioblastoma CAFs were enriched in the subventricular zone which houses the neural stem cells that produce GSCs. Depleting CAFs in GSC-derived xenografts slowed their in vivo growth. These findings are among the first to identify glioblastoma CAFs and reveal their involvement with GSCs, making them an intriguing target.

While cancer-associated fibroblasts (CAFs) and their pro-tumoral effects have been demonstrated in systemic cancers, CAFs had been presumed absent in glioblastoma given the lack of normal fibroblasts in the healthy brain. Here, we show that 5–26% (mean=12%) of cells in human glioblastomas express CAF markers α-SMA or PDGFR-β, morphologically resemble fibroblasts, and transcriptomically resemble by RNA-seq CAFs from other cancers. Glioblastoma CAFs were chemotactically attracted to glioblastoma-initiating stem cells (P=0.02). While glioblastoma CAFs did not affect differentiated glioblastoma cell proliferation (P=0.4), CAFs increased glioblastoma stem cell proliferation (P=0.002) and expression of glioblastoma stem cell-associated genes (P< 0.001). To identify mediators of CAF/glioblastoma stem cell interactions, we created a resource of inferred crosstalk by mapping the expression of receptors to that of their cognate ligands/agonists, using our RNA-seq results from glioblastoma CAFs and stem cells, revealing PDGF-β/PDGFR and osteopontin/CD44 to mediate stem cell recruitment of CAFs and CAF enrichment of stem cells, as confirmed by blocking antibodies (P=0.02–0.03). CAFs also render the glioblastoma microenvironment more pro-tumoral by promoting M2 polarization of tumor-associated macrophages (P=0.01), an effect we found to arise from unique CAF production of the EDA splice variant of fibronectin binding toll-like receptor 4 (TLR4), a known EDA receptor expressed by macrophages (P=0.02). In patient glioblastomas, CAFs were enriched 3-fold in the subventricular zone (SVZ) (P=0.04) which houses the neural stem cells that generate glioblastoma stem cells. SVZs from epilepsy cases or autopsies of glioblastoma-containing brains without ventricular involvement lacked CAFs. Depleting CAFs in xenografts derived from neurosphere-containing glioblastoma stem cells slowed their growth in vivo (P< 0.001). These findings are among the first to identify and profile glioblastoma CAFs. CAF recruitment by glioblastoma stem cells and creation of a pro-tumoral microenvironment in the perivascular niche housing glioblastoma stem cells, particularly in the SVZ, makes them an intriguing therapeutic target.

INTRODUCTION: Glioblastoma carries a high economic burden for patients and caregivers. We investigated the hospital costs of surgery for newly-diagnosed glioblastoma patients based on insurance status. METHODS: Retrospective review of GBM patients undergoing first resection at UCSF and corresponding hospital charges from 2010–2015. RESULTS: Of 227 patients (median age= 62; females= 37.9%), 31 (13.7%) had Medicaid, 94 (41.4%) had Medicare, and 102 (44.9%) had private insurance. Medicaid patients had 30% higher mean overall hospital costs for resecting GBM compared to non-Medicaid insurance patients ($50,285 vs $38,800; p= 0.015). Sub-cost analysis revealed Medicaid patients had higher ICU, OR and imaging costs versus non-Medicaid insured patients ($13,400/$16,470/$2,182 vs $9,700/$14,770/$1,357; p= 0.01/p= 0.03/p< 0.0001). Kaplan-Meier survival analysis showed Medicaid patients had the shortest overall survival (10.7 months, Medicare = 12.8 months, Private insurance = 15.8 months; p=0.02). Tumor diameter at diagnosis was largest for Medicaid (4.7 cm) versus Medicare (4.1 cm) and privately insured patients (4.2 cm; p=0.03). Only 67.74% of Medicaid patients had PCPs versus 91.5% and 86.27% of Medicare and privately insured patients, respectively (p=0.004), at their initial visit to our institution. Medicaid patients had longer overall and ICU lengths of stay (6.9 and 2.6 days) versus Medicare (4.0 and 1.5 days) and privately insured (3.9 and 1.8 days; p < 0.01) cohorts. Moreover, Medicaid patients had similar comorbidity rates as Medicare patients (67.8% vs. 67.18%), but both groups had higher comorbidity rates than privately insured patients (40.4%; p< 0.0001). CONCLUSIONS: Despite higher surgical costs and longer lengths of stay, GBM patients with Medicaid have poorer survival. This may reflect that these patients lacking PCPs and, thus, having more comorbidities and presenting later in the disease course with larger tumors consume more hospital resources such as OR time and confer increased operative risk.