Novel insights into vascularization patterns and angiogenic factors in glioblastoma subclasses

Abstract

Glioblastoma (GBM) is a highly vascularized and aggressive type of primary brain tumor in adults with dismal survival. Molecular subtypes of GBM have been identified that are related to clinical outcome and response to therapy. Although the mesenchymal type has been ascribed higher angiogenic activity, extensive characterization of the vascular component in GBM subtypes has not been performed. Therefore, we aimed to investigate the differential vascular status and angiogenic signaling levels in molecular subtypes. GBM tissue samples representing proneural IDH1 mutant, classical-like and mesenchymal-like subtypes were analyzed by morphometry for the number of vessels, vessel size and vessel maturity. Also the expression levels of factors from multiple angiogenic signaling pathways were determined. We found that necrotic and hypoxic areas were relatively larger in mesenchymal-like tumors and these tumors also had larger vessels. However, the number of vessels, basement membrane deposition and pericyte coverage did not vary between the subtypes. Regarding signaling patterns the majority of factors were expressed at similar levels in the subtypes, and only ANGPT2, MMP2, TIMP1, VEGFA and MMP9/TIMP2 were higher expressed in GBMs of the classical-like subtype. In conclusion, although morphological differences were observed between the subtypes, the angiogenic signaling status of GBM subtypes seemed to be rather similar. These results challenge the concept of mesenchymal GBMs being more angiogenic than other subclasses.

Keywords: Angiogenesis; Glioblastoma; Subclasses; Subtypes.

Fig. 1

The rate of necrosis is associated with total tumor volume and MES-like tumors have a relatively higher percentage of hypoxic tumor area. Pre-operative radiographic scans of patients were analyzed for the volume of necrosis and total tumor, in which the total tumor area (right image) and the central necrosis (middle image) were delineated (a). The volume of necrosis was found to gradually increase with total tumor volume (SR = 0.824, P < 0.01) (b), but the percentage of necrosis only showed a trend for a lower level of necrosis in PN IDH1 ^mut GBMs (c). A representative micrograph of CAIX staining is shown (d), where ‘N’ indicates necrotic tissue, and the arrowheads indicate staining for CAIX. A rim surrounding the necrosis is depicted that was positive for CAIX expression. Quantification of these positive CAIX sites and comparing them against total vital tumor area indicated that the percentage of hypoxic tissue is higher in MES-like GBMs compared to CLAS-like GBMs (e). The percentage of hypoxic tumor area was found to be associated with the proportion of necrosis measured on the MRIs (SR = 0.4838, P < 0.05) (f). Horizontal lines represent median score of the groups; scale bar 250 µm; *P < 0.05

Fig. 2

Endothelial marker expression is increased in MES-like GBMs that on average have larger vessels. The expression patterns of endothelial marker CD34 (a) and neo-endothelial marker ENG (d) is lowest in PN IDH1 ^mut tumors and highest in MES tumors at mRNA level (b, e). When this is evaluated at protein level, the differences between the subclasses do not maintain for CD34 expression (c), but a trend can still be observed for elevated expression of ENG in MES-like tumors (f). Morphometric assessments of vessels, including Chalkley grid score (a), average vessel area (c), and vessel perimeter (d) showed that MES-like GBMs have larger vessels than PN IDH1 ^mut GBMs, but the number of vessels per mm² (b) did not differ. Horizontal lines represent median values; scale bar 50 µm; *P < 0.05, **P < 0.01, ***P < 0.001

Fig. 3

The vascular maturation status of GBM subclasses is rather similar. The mRNA expression of basement membrane marker ColIV (a) is highest in CLAS and MES GBMs (b). At protein level the expression of ColIV still shows a trend for higher expression in MES-like GBMs when corrected for total tissue area (c), but this trend disappears when the layer thickness of ColIV is determined through division of the positive pixels by the total vascular perimeter (d). For α-SMA (e) an increased mRNA expression level was also observed for MES tumors (f), but this remained a significant difference at protein level when we corrected for total tissue area (g). The correction for the total vascular perimeter eventually reduced the differences in α-SMA expression to a trend for elevated expression in MES-like tumors (h). Horizontal lines represent median values; scale bar 50 µm; **P < 0.01, ***P < 0.001

Fig. 4

Ample angiogenic signaling factors are differentially expressed in GBM subclasses, preferably affecting the CLAS-like GBMs. Of the 31 angiogenic signaling factors that were analyzed, 4 were found to be differentially expressed between the subclasses as was evaluated by qRT-PCR. The pro-angiogenic factors ANGPT2 (a) and VEGFA (b) were found to be highly expressed in CLAS-like GBMs. VEGFA expression was also high in MES-like tumors, but with a bigger variation in expression level and therefore not significantly different. MMP2 (c) was upregulated in CLAS-like GBMs, but this was accompanied by upregulated expression of its endogenous inhibitor TIMP1 (e). MMP9 (d) and its inhibitor TIMP2 (f) were not differentially expressed in the subclasses. Dividing the expression of the tissue proteases by the expression levels of their endogenous inhibitors revealed that there was no differential expression of MMP2/TIMP1 (g), but MMP9/TIMP2 was actually upregulated in CLAS-like GBMs (h). In the Venn diagram the differential expression of angiogenic factors can be appreciated, with the majority of markers in the middle shared region, thereby reflecting the similarity in signaling patterns of the molecular subclasses (i). Individual mRNA ratios per tumor are displayed and horizontal lines represent median expression values; *P < 0.05