Vascular neogenesis is a pathologic hallmark of glioblastoma, and has been a major focus of study in neuro-oncology for decades (Tables 1, 2). Angiogenesis is controlled through a complex network of pro- and antiangiogenic blood vessels, and chief among these is vascular endothelial growth factor (VEGF) and its receptor, vascular endothelial growth factor receptor (VEGFR).

Capitalizing on this translational knowledge, bevacizumab, a recombinant humanized monoclonal antibody against VEGF-A, came into the landscape as one of the first antiangiogenic agents. Based on reported successes in other systemic cancers, two early phase II studies for recurrent glioblastoma used bevacizumab and irinotecan (a topoisomerase I inhibitor) and demonstrated 60 % radiographic response and a 6-month progression-free survival (PFS) of 38–46 % [3, 4]. The larger randomized phase II BRAIN trial was later performed, evaluating the activity of bevacizumab with or without iriniotecan in recurrent glioblastoma, and showed radiographic response in 38 and 28 %, respectively, and 6-month PFS was 50 and 42 %, respectively [5]. Around the same time, a single-arm phase II study of bevacizumab alone was published for recurrent glioblastoma and showed radiographic response in 35 % of patients, and a 29 % 6-month PFS [6]. Based on these studies, the FDA granted accelerated approval of bevacizumab for recurrent glioblastoma, contingent upon subsequent phase III study [7].

In response to this requirement by the FDA, two large phase III, randomized trials were subsequently published in newly diagnosed glioblastoma. The AVAglio trial, an international, industry-sponsored, randomized trial, evaluated bevacizumab versus placebo in addition to standard chemoradiation with temozolomide [8]. With co-primary endpoints of PFS and overall survival (OS), the predefined PFS significance level was met (10.6 vs. 6.2 months), though the target OS significance endpoint was not met (16.8 vs. 16.7 months) [8]. Simultaneously published, the RTOG 0825 study was also as a phase III, randomized trial in newly diagnosed glioblastoma patients. Similar to AVAglio, this was designed to compare bevacizumab versus placebo in combination with standard chemoradiation with temozolomide, with a planned crossover [9]. With co-primary endpoints of PFS and OS, the bevacizumab arm did not meet either predetermined PFS (10.7 vs. 7.3 months) or OS (15.7 vs. 16.1 months) levels of significance. The studies also evaluated quality of life measures and durability of functional status, and AVAglio found benefit from bevacizumab whereas RTOG 0825 found detriment by the addition of bevacizumab. Following these mixed results, it is unclear whether the FDA will grant full approval of bevacizumab and the field anxiously awaits further developments.

Aside from FDA-required study, many additional phase II bevacizumab-based combination trials have been performed for recurrent glioblastoma [10]. Bevacizumab with irinotecan has been reported by several groups, with PFS ranging from 20 to 24 weeks, and OS from 31 to 42 weeks [3–5, 11]. Bevacizumab and irinotecan combined with cetuximab (an EGFR inhibitor, see below) was reported with a PFS of 16 weeks and an OS of 29 weeks [12], and combined with carboplatin had a PFS of 23 weeks and an OS of 33 weeks was reported [13]. Combinations of bevacizumab and sorafenib (a multi-targeted tyrosine kinase inhibitor, see below) had a PFS of 21 weeks and an OS of 36 weeks [14], bevacizumab and temozolomide had a PFS of 16 weeks and an OS of 37 weeks [15], bevacizumab and erlotinib (an epidermal growth factor (EGFR) inhibitor, see below) had 18 weeks PFS and 45 week OS [16], and bevacizumab and temsirolimus (an mTOR inhibitor, see below) had 8 week PFS and 15 week OS [17]. Finally, employing older chemotherapy drugs, studies of bevacizumab with etoposide had a PFS of 18 weeks and an OS of 46 weeks [18], and with fotemustine had a PFS of 21 weeks and an OS of 36 weeks [19]. Although these prospective studies are not easily compared due to their small size, varying regimens and patient populations, the general consensus has been that no combination reported to date surpasses outcomes of bevacizumab monotherapy for recurrent high-grade gliomas [20].

Aside from bevacizumab, another means of removing the angiogenic effect of VEGF is to remove the ligand itself. Aflibercept is a fusion protein that scavenges both VEGF and placental growth factor from patients, inhibiting angiogenesis. One trial of aflibercept has been published in adults with recurrent high-grade gliomas and showed moderate toxicity but minimal activity with a PFS of 24 weeks for patients with anaplastic tumors and only 12 weeks for patients with glioblastoma, suggesting limited utility as a monotherapy in these patients [21]. Further study has, therefore, not been sought.

Integrins regulate cell adhesion and are important in tumor growth, invasion, and angiogenesis [22, 23 ]. Cilengitide is a selective integrin inhibitor, targeting αvβ3 and αvβ5 integrins, and has been studied in multiple trials for recurrent gliomas. A phase I study of cilengitide monotherapy for recurrent high-grade gliomas demonstrated adequate drug tolerability and safety, with sustained tumor responses that correlated with decreased MRI perfusion [24]. A phase II trial in recurrent glioblastoma showed modest antitumor results, with an overall survival of 9.9 months and 6-month progression-free survival of 15 %, trending better with higher doses of medication [25]. In a surgical trial, designed to prove drug delivery to recurrent glioblastoma, patients received cilengitide for 3 doses before tumor resection, then continued for up to 2 years after surgery [26]. Cilengitide was detected in all tumors, demonstrating adequate delivery, but led to a 6-month progression-free survival of only 12 % [26]. From this, the authors suggested that combination studies with cilengitide might be more efficacious.

Further, several newly diagnosed glioblastoma trials have been performed with cilengitide. In a phase II dose-randomized trial of cilengitide plus temozolomide and radiation in newly diagnosed glioblastoma, comparing to historical controls, adequate tolerability was found with an average overall survival of 19.7 months, which was better than 14.6 months in their historical control [27]. Notably, this benefit was seen regardless of MGMT methylation status. Another single-arm phase I/II study of cilengitide with concurrent radiation and temozolomide, followed by adjuvant cilengitide and temozolomide demonstrated good tolerability with an overall survival benefit in those patients with MGMT promoter methylation [28]. The follow-up phase III CENTRIC study, however, demonstrated no statistical benefit of adding cilengitide to standard chemoradiation for MGMT-methylated patients, with PFS of 10.7 months for control and 13.5 months for cilengitide, and 26.3 month OS in both arms [29]. For newly diagnosed glioblastoma patients without MGMT methylation, the phase II CORE study showed a modest survival advantage in adding cilengitide to standard radiation and temozolomide, with PFS of 4.1 months for control, 5.6 months for twice weekly cilengitide, and 5.9 months for five times weekly cilengitide during radiation, then twice weekly cilengitide thereafter [30]. These poor results ended the development of cilengitide for treatment of high-grade gliomas.

Among the largest and most promising classes of antineoplastic agents are tyrosine kinase inhibitors (TKI) which target receptor tyrosine kinases (RTKs), the phosphorylating enzymes that control signaling cascades for cell growth and survival. The most commonly upregulated or activated RTKs in glioblastoma include epithelial growth factor receptor (EGFR), VEGFR, and platelet-derived growth factor receptor (PDGFR). Downstream, two of the most constitutively activated pathways in glioblastoma signal through rat sarcoma (RAS) and phosphatidylinositol 3-kinase (PI3K), which have activating mutations in up to 88 % of cases [31]. Further, mutations in inhibitory signaling proteins, such as phosphatase and tensin homolog (PTEN) and neurofibromatosis 1 (NF1), are commonly seen in glioblastoma and promote RTK signaling activation and cell survival [31]. By pharmacologically inhibiting particular RTKs in the aforementioned signaling cascades, downstream proteins and cell functions may also be inhibited. Many such therapeutics have been studied and are summarized below (Table 3).

The small molecule inhibitors that make up TKIs are generally orally bioavailable, but many have poor penetration across the blood–brain barrier (BBB) or are substrates of drug efflux pumps. Additional difficulties with the class include off target effects, leading to unwanted side effects, as well as drug resistance due to upregulation of parallel signaling cascades (Table 3).

With the advantage of oral bioavailability, a large number of VEGFR-targeting TKIs have been examined in patients with glioblastomas. The hope has been for a robust antiangiogenic effect, translating into targeted antitumor efficacy, and improving upon the success of bevacizumab.

Cediranib, a potent VEGFR-targeting TKI, is the most extensively studied of these. A phase II clinical trial of cediranib alone for recurrent glioblastoma was initially encouraging, showing a 6-month PFS of 26 %, and radiographic response in 57 % of patients when measured as >50 % reduction in contrast enhancing volume [32]. However, in employing the Macdonald criteria, only 27 % radiographic response was seen [32]. A phase III trial of cediranib monotherapy versus cediranib and lomustine versus lomustine monotherapy showed that the addition of cediranib to lomustine provided no benefit [33]. Further study of cediranib using MRI perfusion showed that only a small subset of patients receiving this drug demonstrated a decrease in perfusion, which was ultimately associated with better clinical responses [34]. This implies that some patients may respond well to cediranib, but identifying those individuals up-front remains a challenge.

Vandetanib is a multi-targeted TKI, which mainly affects VEGFR-2, as well as EGFR. This drug has been evaluated in one small phase I/II trial of recurrent high-grade gliomas and showed a 6-month PFS of 6.5 % in glioblastomas and 7 % in anaplastic gliomas, with overall survival of 6.3 months and 7.6 months, respectively [35]. Combinations of vandetanib with sirolimus and cilengitide have also failed to show any significant activity.

Sorafenib, another TKI-targeting VEGFR, as well as PDGFR and rapidly activated fibrosarcoma (RAF), has been investigated in several prospective trials. In two phase II studies of sorafenib and temozolomide for recurrent glioblastoma, 6-month PFS ranged from 9 to 26 %, with tolerable side effects, but less than robust antitumor effect [36, 37]. Similarly, when combining sorafenib with bevacizumab in recurrent glioblastoma, there was no survival advantage above historical controls of bevacizumab alone [14]. Finally, for recurrent glioblastoma, a phase II study of sorafenib and erlotinib (EGFR TKI, see below) did not meet its primary objective of 30 % increase in OS compared with historical controls [38]. In newly diagnosed glioblastoma, adding sorafenib to temozolomide after standard chemoradiation did not appear to improve efficacy of treatment compared to standard therapy [39]. Finally, a phase I/II trial combining sorafenib and temsirolimus (mTOR TKI, see below) for recurrent glioblastoma was terminated after the stage I portion due to lack of activity [40].

Pazopanib, another multi-targeted TKI, inhibits VEGFR, PDGFR, and c-KIT. A phase II study of pazopanib monotherapy in 35 recurrent glioblastoma patients was reasonably well tolerated, but did not prolong progression-free survival despite a moderate amount of radiographic response [41]. Also, a phase II evaluation of pazopanib and lapatinib (EGFR TKI, see below) was unsuccessful due to excessive dose-limiting toxicity and lack of efficacy [42].

Tivozanib is another VEGFR-targeting TKI, currently under investigation with no reported outcomes to date (ClinicalTrials.gov NCT01846871).