Integrative Science and Therapeutics Glioblastoma treatment: Where to now?

Glioblastoma (GBM) remains an incurable disease with a poor overall survival. Despite extensive research into clinical trials, temozolomide remains the only therapeutic agent to improve patient survival in the past 50 years. This is despite only providing a modest increase of 2.5 months to median survival. Resistance to traditional therapies has become a hallmark of GBM, owing to its complex and undetermined molecular landscape. Studies now suggest that GBM is a disease of genetic subtypes and require tailored approaches to therapeutic care. Further strategies for GBM treatment involve targeting tumour associated neovascularisation. While early attempts to attenuate the tumour vascularisation with anti-VEGF has not been successful, studies are now looking towards other angiogenic factors and novel mechanisms of neovascularisation that have yet to be explored. A shift towards understanding the molecular and biological mechanisms of GBM pathogenesis represents a promising new strategy for treatment. Here we highlight some of the major developments to genetic profiling and anti-neovascularisation therapy.


Introduction
Glioblastoma (GBM) is the most common glioma and one of the most debilitating human cancers [1,2]. Although relatively uncommon, GBM is associated with a disproportionate morbidity and mortality in the population with a median survival of 12-15 months owing to inevitable tumour recurrence [3]. Clinically, most patients present with de novo primary GBM (~90%), with few patients progressing from a lower grade glioma to secondary GBM [4,5]. Histopathological examination of primary and secondary GBM are largely indistinguishable, though secondary GBM is typically diagnosed at a younger age and associates with a more favourable prognosis. The distinction between the clinical presentations is primarily due to distinct molecular signatures that are thought to govern tumourigenesis of each subtype [6]. Despite these clinical and molecular differences, all patients are treated with the same aggressive standard of care consisting of maximal resection, concurrent chemoradiation and adjuvant temozolomide-based chemotherapy for newly diagnosed GBM [7]. O 6 -methylguanine-DNA methyltransferase (MGMT) status is routinely performed to assess a patient's response to temozolomide, with minimal to no benefit derived in patients who lack this methylation [8]. Although younger patients and secondary GBM cases respond markedly better to the standard of care, treatment response is largely dependent on the genetic landscape of the tumour [8,9].

Reassessing approaches to Glioblastoma therapy research
With over a century of GBM research conducted, there have been few advances in GBM treatment. Recent improvements in surgical techniques and neuroimaging modalities have improved tumour care and treatment decision making but have provided minimal impact on patient survival [10]. Even the chemotherapy agent temozolomide, which is commonly regarded as the most significant breakthrough of the past 50 years, has provided a minor median survival improvement of 2.5 months with an optimal standard of care [3]. In real-world clinical situations, most patients do not receive the complete standard of care due to poor prognostic factors and concern of cytotoxicity of chemotherapy, particularly in elderly patients [11][12][13]. Alternate FDA approved treatments are available for newly diagnosed GBM including the use of nitrosoureas, though the use of these agents remains controversial without an established standard of care. The lack of advancement for GBM therapy has led to extensive clinical trials to determine novel therapeutic approaches [14]. The results from these trials have not been encouraging, with temozolomide the only therapeutic agent showing clinical efficacy [14,15]. The low rate of discovery from clinical trials can be largely attributed to the complex biology of GBM, making it highly refractive to standard non-specific treatments [16]. There is a need to shift away from chemotherapy based clinical trials that do not address the underlying etiology of the disease and are commonly associated with high toxicity to the patients [7,17]. Recent discoveries of prognostic factors for patients, demonstrate the importance of GBM pathophysiology in treatment response [6,18]. The success of pursuing such approaches has been demonstrated with therapies targeting HER2-amplified breast cancers, Chronic myeloid leukaemia (CML) harbouring the BCR-ABL translocation, BRAF mutant melanoma among other cancer-specific tumour promotors [19]. By targeting pathways that promote GBM progression there is the potential to provide meaningful clinical responses without increasing the burden to the patient's quality of life.

Genomic alterations define glioblastoma
Early genome-wide profiling of GBM demonstrated a remarkable genomic heterogeneity within the tumour suggesting the existence of molecular subclasses that may clinically impact treatment [20,21]. The Cancer Genome Atlas (TCGA) group set out to extensively characterise the genomic landscape of GBM and identify the major cancer-causing genomic alterations [22,23]. The study identified major alterations to the Receptor Tyrosine Kinase (RTK)/RAS/PI3K pathway, in addition to p53 and RB mutations. EGFR activating mutations or amplification was the most common alteration found to be expressed in 57.4% of GBM cases and has gained much interest as a primary driver of tumour proliferation and survival. Furthermore 50% of GBM tumours with a EGFR amplification, harbor a EGFR variant (EGFRvIII) with in frame deletion of exons2-7 resulting in constitutive activation and enhanced RAS/PI3K signaling [24][25][26]. Mutations in PI3K (25.1%) and deletions/ mutations in PTEN (41%) were also commonly found and reported to be mutually exclusive with 59.4% of GBM presenting with one or the other [22]. These genomic alterations reaffirm a strong association between RTK/RAS/PI3K pathways and tumourigenesis.
The identification of IDH mutations in GBM provided differentiation between what had until then been only identified as histopathological primary and secondary GBM [6,27]. IDH1 mutations were identified in over 80% of grade II and III gliomas and were conserved during transformation to secondary GBM [5]. In contrast IDH mutations in primary GBM are rare occurring in under 5% of cases, most associated with younger age and genetic profiles more similar to secondary GBM. IDH mutations are thought to be an early initiator of tumourigenesis and progression to secondary GBM requires further genomic alterations [4]. The majority of secondary GBM cases have IDH1 and TP53 mutations, whereas primary GBM is most commonly associated with EGFR amplification and loss of PTEN function. This molecular characterization has been further expanded to four GBM subtypes Proneural, Neural, Classical and Mesenchymal each with its own specific differentiation lineage and prognostic outcome [6,21]. While these findings present a unique opportunity for individualised subtype-specific therapy, recent studies have reported a proneural-mesenchymal shift following irradiation, contributing to radioresistance [28][29][30]. The inherent plasticity in GBM discerns the need for individualised treatment and highlights some of the limitations in current clinical trial developments. Through identification of treatment-induced genetic alterations patients may receive adaptive and specific tailored-therapy with improved clinical outcomes [18].
The identification of multiple genetic signature pathways to GBM tumourigenesis underscore the complexity of the disease and obstacles to treatment. While current diagnosis and treatment is standard regardless of molecular subtype, successful development of new therapeutic targets will need to account for the intrinsic cellular differences regulating GBM behaviour.

Exploring neovascularisation as a therapeutic target
GBM tumours are among the most highly vascularised of all solid malignancies and are distinguished from lower grade tumours by necrosis and microvascular hyperplasia [31][32][33]. This histopathological classification is independent of tumour cell morphology and carries an inordinate degree of prognostic power suggesting that they mechanistically linked to tumour progression [34,35]. Tumours require adequate blood supply for growth and survival, therefore neovascularisation presents as promising therapeutic targets [36]. Targeting abnormally activated tumour vasculature has the additional benefit of overcoming many problems associated with chemotherapy such as tumour resistance, high levels of cytotoxicity and lack of efficient distribution [37]. Therefore, there has been much interest into the study of angiogenesis which is thought to be a key mediator of microvascular hyperplasia in all forms of vascular cancers [38].
The success of antiangiogenic therapy for metastatic colorectal cancer, accelerated the FDA approval of Bevacizumab (Avastin®) in 2009 for use in GBM following an uncontrolled phase II GBM clinical trial [39]. Bevacizumab is a monoclonal antibody targeting angiogenesis through inhibition of VEGF ligand. While there are many signalling pathways involved in angiogenesis, VEGF, has been the most extensively studied and has been reported in plasma and tumour samples obtained from GBM patients, where its overexpression correlated with poorer prognosis [40,41]. Despite the early promise that bevacizumab would revolutionise GBM treatment, all clinical trials have failed to improve overall survival for both newly diagnosed and recurrent GBM [14,42]. The reasons for this lack of efficacy remain controversial yet no study has shown a specificity of bevacizumab for tumour-associated vascularisation. More recent clinical trials have investigated the efficacy of VEGF traps, VEGFR kinase inhibitors and monoclonal antibodies for recurrent GBM [43]. These clinical trials have also been disappointing as single or concomitant agents producing no improvement to overall survival. The dismal progress of these antiangiogenic inhibitors strongly suggests that there are alternate pathways to tumour induced-neovascularisation in GBM that requires a more comprehensive understanding of the underlying mechanisms.
FGF-2 is another important contributor to angiogenesis by promoting proliferation and migration of endothelial cells [44]. While its role in early vascular development remains controversial, in vivo GBM studies with dominant-negative FGFR2 or FGFR1 inhibited glioma C6 tumour development in rats and decreased microvessel density [45]. A novel inhibitor 2,5DHPS targeting FGF was also found to strongly inhibit GBM invasion and suppress associated angiogenesis in a C6 orthotopic glioma rat model [46]. These early studies suggest a potential mechanism of FGF-2 in GBM vascularisation.
Among the other well characterized angiogenesis cytokines, TGF-β signalling has been shown to have a key role in microvascular modulation and is amplified in GBM tumours conferring to poorer prognosis [47,48]. Genetic mutations to the TGF-β type I receptor ALK1 and its accessory receptor endoglin cause the vascular condition hereditary hemorrhagic telangiectasia (HTT) in humans that is characterized by arteriovenous malformations in organs [49]. Endothelial cell specific deletion of ALK1 and endoglin in vivo completely recapitulate the vascular abnormalities seen in HHT underscoring the importance of TGF-β signalling in vascular development [50][51][52]. Loss of ID1, downstream of TGF-β/ALK1 signaling in GBM tumour endothelial cells, results in downregulation of several key proangiogenic genes, providing the potential for multiple angiogenic pathways [53]. In the GBM microenvironment endoglin has been identified as a sensitive marker of angiogenic blood vessel formation and associates with poorer patient survival [54,55]. Current studies suggest an important role for TGF-β/ALK1 signalling in tumour angiogenesis and further study of its role in GBM pathobiology is needed to harness its potential [56].
Despite the interest of angiogenic inhibitors over the last decade for the treatment of GBM, few studies have fully examined the contribution of angiogenesis-independent pathways to neovascularisation. Vascular co-option and de novo vasculogenesis have both been reported in GBM [57,58]. Possible molecular links between hypoxic and angiopoietin pathways are suspected to mediate GBM vascular co-option and have been previously described as an initial step to GBM vascularisation [59]. Differentiation of circulating bone marrow-derived cells (BMDCs) has also been identified to contribute to vasculogenesis of GBM. In Id1 mutant mice angiogenic defects were observed to inhibit the growth of PTEN +/tumour xenografts [53]. This phenotype was partially rescued by BMDCs. Other studies, however, have shown only a minor contribution of BMDC to GBM vasculature following VEGF inhibition [60]. It is yet to be determined whether BMDCs represent a novel target for GBM therapy.
A recently identified mechanism of glioma vascularisation involves the formation of perfusable vessel-like networks by tumour cells [61]. These structures are completely devoid of endothelial cells and have been termed vasculogenic mimicry (VM), following their ability to create pseudo de novo vascular channels [62]. Histologically VM structures are confirmed by PAS + CD31/CD34vascular patterns. Further molecular characterization of these tumour cells demonstrate expression of endothelial cell associated genes that recapitulated the embryonic development of vasculogenesis [62,63]. These observations led to the four defining characteristics of VM: 1) patterned vascular channels of aggressive and primary tumours are different from endothelial-derived angiogenic vessels; 2) highly invasive tumour cells but not poorly invasive ones have the intrinsic ability to form patterned vascular channels in absence of endothelium; 3) Tumour cells that generate these patterns are highly plastic and aberrantly express genes associated with embryonic stem cells; and 4) the generation of these patterned vascular channels is a novel pathway to generate microcirculation. While these structures have been identified in GBM patient samples and are associated with poorer prognosis, the mechanisms involved in the formation of these structures remains unclear [64,65].
Even more recently GBM stem cells have been observed to transdifferentiate into an endothelial cell phenotype [66][67][68][69]. While these structures have been found to form separate vascular channels, in contrast to VM these endothelial-like tumour cells can also integrate into existing endothelial cell lined blood vessels forming mosaic blood vessels [67,68]. The biological significance and mechanisms regulating this transdifferentiating behaviour are still unknown but may offer new explanation and opportunities for neovascularisation treatment.
GBM-induced neovascularisation is undoubtedly more complex than early VEGF therapies anticipated. It is possible that multiple signalling pathways are involved within angiogenesis and new strategies for multiple targeting of angiogenic pathways are needed. Additionally, further research is needed to understand the contribution of angiogenesis-independent pathways to neovascularisation. By improving our understanding of basic GBM vascular pathology we may still realize the potential of neovascularisation inhibitors.

Conclusions
Previous approaches to advancing GBM treatment through non-specific treatments have been unsuccessful providing marginal improvements to patient outcome in over 50 years. It is clear that a new avenue of therapeutic exploration is needed that addresses the key mechanisms governing GBM pathogenesis. The information achieved from these molecular genetic and biological studies will develop and improve the next generation of clinical trials and therapeutic development.