The onset of angiogenesis is a key event in the malignant progression of gliomas and marks the transition from anaplastic gliomas to glioblastomas, coinciding with a drastically shortened survival [17]. Various cellular processes and biological mechanisms involved in the vascularization of brain tumors have been implicated [18], including sprouting from existing vessels, co-option of the host vasculature by invading cells, vascular mimicry, vessel intussusception, recruitment of bone marrow-derived endothelial precursors, contribution from M2-polarized macrophages, and transdifferentiation of cancer stemlike cells into endothelial cells [17].

Among these, several studies strongly suggest that vessel formation by endothelial cell proliferation and sprouting from the host capillaries is a main mechanism of angiogenesis in malignant gliomas [18]. Histopathologically, GBMs exhibit microvascular proliferations with endothelial hyperplasia, forming nonluminated structures referred to as glomeruloid bodies or vascular tufts, due to their resemblance with glomeruli in the kidneys. The proliferation index of tumor endothelial cells has been estimated to be 22–29% in GBMs, higher than in lower-grade gliomas, whereas proliferation of endothelial cells in normal brain tissue is largely absent [19]. Furthermore, VEGF expression is particularly upregulated in palisading tumor cells around necrotic regions [20]. In addition, the VEGF receptor is abundantly expressed in these endothelial cells but hardly detectable in the normal brain vasculature. These findings also explain the striking presence of endothelial cell proliferations surrounding necrotic areas.

Gliomas are hallmarked by their infiltrative growth into the surrounding brain parenchyma. During invasion, glioma cells migrate along vessels of the existing host vasculature – a process referred to as co-option [21]. Holash et al. reported that co-option facilitates vascularization of experimental gliomas in early stages. Subsequently, vessels in the tumor center regressed, accompanied by upregulation of angiopoietin-2 (Ang-2) . Simultaneously, however, newly formed capillary sprouts at the tumor periphery also expressed Ang-2. Their data suggested coordinated roles for angiopoietins. Upregulation of Ang-2 in the absence of VEGF induced endothelial cell death, whereas it mediated angiogenesis in the presence of VEGF. Apart from this experimental study, however, limited data are available regarding vessel co-option, and it is not documented that similar processes take place in human tumors.

Animal studies also suggest that GBM cells with a stemlike phenotype had the ability to integrate into the vessel wall and transdifferentiate into endothelial cells [22]. Others, however, have reported that endothelial cells carrying mutations found in the GBM cells occur at a very low rate mostly outside the endothelial lining of the vessel wall [22]. Thus, the experimental findings regarding this phenomenon are conflicting, and no clinical data have established a role for cancer stem cell transdifferentiation in human glioma angiogenesis.

Several studies implicate bone marrow-derived cells in brain tumor angiogenesis. In particular, macrophages may acquire an M1 tumor-inhibitory phenotype or an M2 tumor-promoting phenotype. M2 macrophages exert their effects in vivo by releasing cytokines that promote tumor cell growth and angiogenesis. In experimental studies, integration of bone marrow-derived cells into the vessel wall seems to occur at a low rate and has not been demonstrated in human gliomas [18].

Intussusception refers to invagination of the endothelial vascular wall, leading to extravascular tunnels with tissue traversing the vascular lumen. Although this phenomenon has been described in experimental tumors, no studies have demonstrated a role for vascular intussusception in human gliomas.

It has been well established through histopathological studies that immune cell tumor infiltration is prevalent in glioma patients. Whereas a high degree of lymphocytic infiltration in the perivascular space was reported to be associated with up to 4 months longer survival in one study [27], others found that lymphocytes infiltrating the tumor correlated with a poor prognosis [28]. A predictive value of tumor-infiltrating lymphocytes is not yet established, consistent with the presence of functionally distinct classes of lymphocytes. Furthermore, early studies also reported that glioma patients displayed reduced peripheral cellular and humoral immunity [29]. Since then, numerous experimental and clinical studies have shown that glioma cells interact extensively with the immune system. These interactions involve both the innate and adaptive components of the immune system, as well as local and peripheral immune cells. Despite the presence of the blood-brain barrier and the absence of lymphatic vessels, peripheral immune cells gain access to the brain parenchyma and the tumor bed via multiple routes. The BBB is typically disrupted in malignant gliomas due to reduced pericyte coverage, gaps between the endothelial cells and basement membrane, and defects in the brain tumor vasculature [30, 31], allowing peripheral immune cells to enter the CNS. Moreover interstitial fluid drains to the perivascular space, enabling tumor antigens in the brain parenchyma to reach antigen-presenting cells around the meninges and in the subarachnoid space [32]. It has also been demonstrated that CSF communicates with deep cervical lymphatic drainage and that antigens in the ventricles may induce antibody-producing cells in cervical lymph nodes [33]. T lymphocytes are the main cell type involved in adaptive antitumor immunity. These cells belong to different subclasses with different roles in the host-tumor interplay and can be distinguished by their expression of cell surface markers. CD8+ T cells are cytotoxic and become activated in the presence of antigen-presenting cells and CD4+ helper T cells. Several studies show that the glioma-infiltrating effector cells correlate positively with tumor grade as well as survival in glioblastomas [34, 35]. However, it has also been shown that the glioma microenvironment is immunosuppressive due to tumor-derived factors and regulatory cells, Tregs, that impair the function of these effector cells [36]. Tregs are CD4+ T cells that express the transcription factor Foxp3, and numerous studies have consistently reported their presence in the glioma microenvironment at higher, although varying rates compared to normal brain [37–39]. However, data regarding a correlation between Treg recruitment and survival in glioblastoma patients are conflicting [38, 39].

Moreover, numerous studies have clearly demonstrated that malignant gliomas are heavily infiltrated with microglia [40] that are recruited by glioma-derived chemoattractants such as MCP-1 and CSF-1 [41]. Importantly, microglia exert immunosuppressive effects through multiple mechanisms. In the presence of glioma cells, microglia display impaired MHC class II antigen presentation [42] and produce anti-inflammatory IL-10, which inhibits cytotoxic T cell function [43]. Moreover, tumor-associated microglia express FASL and upregulate the immunosuppressive molecule B7-H1, both capable of inducing T cell apoptosis [44, 45].

Recently, myeloid-derived suppressor cells (MDSC) have been identified both in peripheral blood and tumor tissue of glioma patients [46]. These cells have the ability to suppress T cell function through depletion of amino acids that are critical for T cell function and through production of reactive oxygen species [47]. Notably, an association between tumor-infiltrating granulocytic MDSCs and CD4+ T effector memory function has been reported [48].

Glial cell types including astrocytes and oligodendrocytes represent the most abundant cell types in the CNS. These cells provide structural support and maintain homeostasis in the normal brain but are also present in the brain tumor microenvironment (Fig. 20.2), where they can influence multiple aspects of glioma growth and sensitivity to treatment (Table 20.1).

Reactive astrocytes are abundantly present in the tumor bed among infiltrating glioma cells, and brain tumor growth is accompanied by astrogliosis [53]. Activated astrocytes secrete various factors [54] and have been shown to increase the proliferation of malignant cells in vitro [55]. Moreover, they also secrete factors that may promote the proliferation of glioma cells, including EGF, IGF-1, GDF-15, and TGF-β [55–57]. However, in vitro or in vivo studies directly investigating the effects of astrocytes or oligodendroglia on glioma cell growth have not been published.

Conversely, several studies have clearly demonstrated a proinvasive effect of tumor-associated astrocytes on glioma cells both in vitro [54, 57] and in vivo [58, 59], involving multiple mechanisms: Astrocytes have been shown to secrete an inactive preform of matrix metalloproteinase-2, a proteolytic enzyme linked to cancer cell invasion, which is converted into an active form by glioma cells [54]. Moreover, connective tissue growth factor (CNTF), which has been implicated in cancer metastasis, has been shown to be secreted from reactive astrocytes surrounding infiltrative gliomas and bind to tyrosine kinase receptor type A (TrkA). Furthermore, targeting either CNTF or TrkA both reduced glioma cell infiltration [59]. Of note, it was reported that the gap junction protein connexin 43 (CX43) is overexpressed by astrocytes in the tumor bed [58]. Furthermore, gliomas in CX43 knockout mice had more circumscribed margins suggesting that CX43 promotes glioma cell detachment from the tumor core.

One study also suggests that oligodendrocyte progenitor cells may enhance angiogenesis in gliomas by disrupting the blood-brain barrier, thereby abrogating the effect of perivascular pericytes and promoting vessel sprouting and tubule formation [60].

Gliomas are characterized by chemoresistance, and several studies show that astrocytes can modulate the response of the tumor cell compartment to chemotherapy. Co-cultures of astrocytes and a panel of glioma cell lines showed that the presence of astrocytes increased glioma cell survival after treatment with temozolomide and doxorubicin in several cell lines [61]. However, the gap junction channel inhibitor CBX as well as CX43 siRNA knockdown abolished this protective effect suggesting a major role for gap junction channels between glioma cells and astrocytes in mediating chemoresistance [62].

Malignant gliomas are characterized by dynamic interactions with the immune system, glial cells in the tumor bed as well as recruitment of host vasculature. Since these interactions are critical to tumor progression, they may be attractive targets for glioma therapy. Unlike some other cancer types, however, neither anti-angiogenic nor immune-based therapies have demonstrated any survival benefit in glioma patients, suggesting that the mechanisms that regulate tumor-host interactions in the CNS are in many ways unique to gliomas. Thus, therapeutic progress in this field requires that those mechanisms be explored in further detail.