|Home | About | Journals | Submit | Contact Us | Français|
Adults with malignant glioma, especially the most common subtype, glioblastoma multiforme, have an unacceptably poor outcome with current therapies. Malignant gliomas are amongst the most angiogenic of cancers, and VEGF is the dominant angiogenic mediator in these tumors.
To summarize the clinical experience of VEGF-directed treatment for malignant glioma.
We reviewed the completed, ongoing and planned clinical trials evaluating anti-VEGF strategies for malignant glioma patients.
Recent studies incorporating anti-VEGF agents plus cytotoxic therapy among recurrent malignant glioma patients have achieved unprecedented improvements in radiographic response, time to progression and survival. Furthermore, acceptable toxicity was observed. Hence, a major current focus in neuro-oncology is to further develop antiangiogenic strategies for this desperate patient population.
Effective therapy for patients with malignant glioma, the most common primary tumor of the CNS, remains elusive. Median survival for parients with glioblastoma multiforme (GBM), the most common malignant glioma, is under 15 months following standard-of-care therapy with surgery, radiation and temozolomide , and there is no effective therapy following recurrence . Heterogeneity within and across tumors, high rates of de novo and acquired resistance, and limited delivery are major impediments to cytoxins directly targeting GBM cells. Hence, therapeutics that indirectly attack these tumors by targeting vital components of the supporting extracellular matrix, including neovasculature, are increasingly being explored.
Growth of all tumors is dependent on angiogenesis, the formation of new blood vessels from pre-existing vasculature . Tumor angiogenesis is orchestrated by a simultaneous increase in expression of genes, including VEGF, acidic and basic fibroblast growth factor (FGF), IL-8 and -6, hypoxia-inducible factor 1 alpha (HIF-1α) and the angiopoietins, with downregulation of endogenous angiogenesis inhibitors, such as thrombospondins, angiostatin, endostatin and interferons . VEGF is a paramount common denominator required for tumor angiogenesis and pathogenesis [5,6]. Therapeutic exploitation of the VEGF axis has achieved substantial clinical benefit across many cancer subtypes [7,8]. However, enthusiasm for evaluating these agents among patients with CNS tumors has been dampened by safety concerns, including the risk of intracranial hemorrhage. Nonetheless, recent trials among malignant glioma patients treated with VEGF- or VEGF receptor (VEGFR)-targeting therapeutics plus chemotherapy report unprecedented rates of antitumor benefit, as well as acceptable safety profiles. Specifically, the regimen of bevacizumab (BV), a humanized anti-VEGF monoclonal antibody, plus irinotecan, achieved a 10-fold improvement in radiographic response as well as significant increases in progression-free and overall survival among recurrent GBM patients [9,10], Several additional clinical trials are ongoing, or starting soon, to validate and expand these efforts, including multiple studies to evaluate a variety of VEGF as well as non-VEGF antiangiogenic strategies for malignant glioma patients (Table 1). In this review, we briefly describe angiogenesis in malignant glioma and then detail clinical activities targeting VEGF/VEGFR for malignant glioma patients.
Proliferation, survival and invasion of malignant gliomas critically hinge on an adequate blood supply. Malignant gliomas are among the most angiogenic of cancers , primarily due to a tumultuous and somewhat redundant constellation of genetic and cellular signaling cues culminating in a remarkably prolific capability for neovascularization. Angiogenesis is fueled by several pro-angiogenic factors in malignant glioma, among which VEGF is dominant Six VEGF isoforms (VEGF-A, VEGF-B, VEGF-C, VEGF-D, VEGF-E and placental growth factor) and several additional, biologically active VEGF variants generated by alternative gene splicing or protease cleavage, are secreted by tumor cells, infiltrating inflammatory cells and platelets and can be sequestered in the extracellular matrix [12–18]. Glial tumors exhibit a prototypic ‘angiogenic switch’, in that a hallmark of transformation from low-grade to high-grade gliomas is the induction of pro-angiogenic mediators and new blood vessel formation . Furthermore, increased VEGF expression predicts glioma aggressiveness and poorer outcome . VEGF expression in malignant gliomas is most concentrated adjacent to areas of necrosis and hypoxia, including cellular pseudopalisades at the tumor leading edge [19,21–28]
Several hypoxia-dependent and independent mechanisms converge to produce an abundance of VEGF in the micro-environment of malignant gliomas [5,19,29–32]. Hypoxia, a prominent feature of malignant gliomas [28,33], enhances expression and stabilization of HIF-1α, which acts as a transcription factor to activate myriad target genes regulating tumor angiogenesis, migration and survival, including VEGF and VEGFRs [33–36]. Aberrant activation of multiple growth factor receptors in malignant glioma, including EGFR [37,38], platelet-derived growth factor receptor (PDGFR) [39–41], scatter factor/hepatocyte growth factor receptor (MET) , IGF receptor (IGFR) [43,44], stem cell factor receptor (c-Kit) , and FGF receptor (FGFR) [46–48], also increases VEGF activity. Dysregulated signaling of the phosphatidylinositol 3-kinase (PI3K/Akt) and Ras/MAPK pathways, which occur frequently among malignant gliomas, also augment VEGF expression [37.49].
Expression of VEGFRs (VEGFR-1, VEGFR-2 and VEGFR-3) and co-receptors, including the neuropilins, although very low in the normal brain, are also markedly increased in malignant gliomas [21,23–25,50,51]. Ligand binding activates VEGFRs, triggering a downstream intracellular signaling cascade that promotes endothelial cell proliferation, survival, activation, invasion, migration and permeability [5,52]. VEGF signaling further fortifies the angiogenic response by triggering endothelial cell nitric oxide synthesis , and by mobilizing bone-marrow-derived endothelial cell progenitors [54–57].
Finally, malignant gliomas robustly express several critical mediators of endothelial cell invasion, a critical step in new blood vessel formation, including MMP-2 and -9, urokinase-type plasminogen activator (uPA) and its receptor (uPAR), cathepsin-B, integrins αvβ3 and αvβ5, and tenascin-C [58–63].
Although the precise antitumor mechanism of VEGF-targeting agents is unknown, increasing amounts of data suggest that effective VEGF inhibition exerts multiple detrimental effects on malignant glioma viability and survival. In addition to ‘choking off’ the blood supply and depriving tumors of vital nutrients and oxygen, preclinical and clinical studies confirm that VEGF-targeting agents are particularly effective when combined with cytotoxins [8–10,64–67]. Potential underlying mechanisms of this enhanced activity include the ability of antiangiogenic agents to sensitize tumor endothelial cells to cytotoxic agents [68,69] and to block a compensatory VEGF surge and/or accelerated tumor cell repopulation induced by rapid cell killing [70,71]. Additionally, appropriately dosed antiangiogenic agents may selectively ‘prune’ tumor vasculature, thereby transiently normalizing perfusion and improving chemotherapy delivery [72–75]. In support of this model, lowered interstitial fluid pressure, higher oxygen content, and decreased permeability were observed in an orthotopic GBM model, following judiciously administered antiangiogenic therapy . Further support for this mechanism of action is emerging from recent clinical studies conducted among patients with colon cancer and GBM undergoing anti-VEGF therapy plus chemotherapy [77,78].
The recent discovery that VEGF-targeting agents can effectively inhibit activity of malignant glioma stem cells provides yet another highly intriguing antitumor mechanism of action for these agents . The self-renewal and tumor-forming abilities of malignant glioma stem cells have recently been shown to be critically dependent on a bi-dimensional interaction with endothelial cells within the immediate microenvironment, referred to as the perivascular niche, which can also be blocked by anti-VEGF agents . In addition, metronomic chemotherapy, felt to block angiogenesis by inducing endothelial cell apoptosis, can also target GBM stem cells . The cumulative findings of these studies suggest that effective antiangiogenic therapy may target GBM stem cells directly and may also critically perturb the perivascular niche required for stem cell well-being .
Following observations that BV improved outcome when administered with chemotherapy to patients with colorectal, breast, lung and pancreatic cancer patients , an initial evaluation of BV plus the topoisomerase-1 inhibitor irinotecan (Camptosar; Pfizer) reported a dramatic rate of radiographic response among recurrent primary CNS tumor patients , These results led to a formal Phase II study of this regimen for recurrent malignant glioma patients [9,10,84]. Irinotecan was included in this regimen because it has modest activity as a salvage agent among recurrent malignant glioma patients, including a 5 – 15% radiographic response rate and a median progression-free survival (PFS) of ~ 12 weeks [85–89].
Adult patients with measurable, recurrent Grade 3 or 4 malignant glioma and a Karnofsky performance status of ≥ 60% were eligible. Patients were also required to have adequate bone marrow, hepatic and renal function and to be at least 6 weeks from prior surgery and 4 weeks from prior radiation therapy or chemotherapy (6 weeks for nitrosoureas). Patients with more than three prior episodes of progression, evidence of blood on pretreatment imaging, requirement for warfarin anticoagulation, or previous BV treatment were not eligible. The study primary end point was 6-month PFS.
An initial cohort of 32 patients received BV (10 m/kg) and irinotecan every 2 weeks followed by a second cohort of 36 patients treated with BV every three weeks (15 mg/kg) and irinotecan on weeks 1, 2, 4 and 5 of each 6-week cycle (Figure 1). Patients not on concurrent CYP-3A enzyme-inducing antiepileptic drugs (EIAEDs, phenytoin, carbamazepine, oxcarbazepine, phenobarbitol and primidone) received 125 mg/m2 of irinotecan, whereas those on EIAEDs received 340 mg/m2 due to the profound EIAED-induced increase in metabolism of irinotecan and its major metabolite, SN-38 . Patients were evaluated with a complete physical examination and MRI after each cycle. Radiographic response was independently assessed by two study investigators using the modified MacDonald criteria . Of note, for this study, stable or improved T2 and fluid-attenuated inversion recovery (FLAIR; signal abnormalities were also required to define radiographic response in addition to the traditional MacDonald criteria. Treatment was discontinued for progressive disease, unacceptable toxicity or the need to initiate systemic anticoagulation.
Patient characteristics for both cohorts are summarized in Table 2. Of note, all had received prior temozolomide chemoradiation and had a median of two prior episodes of progressive disease (range, 1 – 3).
The study regimen was overall adequately tolerated. Two patients (3%) developed a grade 2 CNS hemorrhage, including a patient in cohort 1 after ten cycles of therapy and a patient on enoxaparin in cohort 2 after nine cycles of therapy. Eight patients (12%), four from each cohort, developed grade 3 – 4 thromboses, including one patient with an arterial cerebrovascular ischemic event, suggesting that BV may further heighten the known risk of thrombosis among malignant glioma patients [92–94], Therapy was discontinued in four patients (6%) due to grade 2 fatigue and in two patients (6% of cohort 1) for grade 2 proteinuria. Four patients (6%) in cohort 2 discontinued therapy due to grade 3 nausea/emesis or diarrhea, most probably due to the more intensive irinotecan schedule.
The rates of radiographic response are summarized in Table 3. For comparison, radiographic response rates among patients treated with temozolomide at first recurrence are included [95,96]. The rate of radiographic response was considerably higher among patients treated with BV plus irinotecan, even though they were more heavily pretreated and had failed prior temozolomide. Overall, 40 of the 68 patients (59%) achieved a radiographic response, including 22 (65%) with grade 3 tumors and 18 (53%) with grade 4 tumors (Figure 2).
Salvage therapies for GBM patients typically achieve only single-digit rates of radiographic response, while the majority of patients progress as their best response [2,95,97]. In contrast, a majority of patients treated with BV plus irinotecan achieved a radiographic response while only a single-digit rate of progressive disease (6%) was observed. Finally and most encouragingly, clinical and neurological status commonly reflected radiographic findings in that most patients who achieved a radiographic response also noted neurological improvement and were able to taper chronic dexamethasone dosing. Similar rates of radiographic response to BV plus chemotherapy were recently described in two additional reports. Specifically, 4 of 10 recurrent GBM patients (40%) and 3 of 4 recurrent grade 3 patients (75%) achieved a partial response in one study . In a second study, 8 of 21 heavily pretreated patients with recurrent malignant glioma (38%) achieved a radiographic response, including 7 of 18 with GBM (39%) .
Median PFS rates for grade 3 and 4 patients reported by Vredenburgh were nearly twofold greater than those achieved with temozolomide at first recurrence (Table 3) [95,96]. In addition, the rates of 6-month progression-free survival (6-PFS) were much higher than those achieved with temozolomide. Furthermore, six patients with recurrent GBM (18%) completed a year of therapy, and five of the six had no hypermetabolic activity noted on companion [18F] fluorodeoxyglucose positron emission tomography (FDG-PET) imaging at therapy completion, suggesting the absence of active tumor. Analogously, seven patients with recurrent grade 3 tumors (21%) completed one year of therapy, and six of these had negative FDG-PET scans at study completion. Similarly, the median overall survival for patients with grade 3 and 4 tumors in this study significantly surpassed the median survival reported for historical temozolomide studies among patients at first recurrence [95,96] and among patients treated on a series of salvage therapies (Table 3) . Although decreased permeability and lowered contrast-agent uptake induced by BV may have contributed to the rate of radiographic response observed among patients treated with BV plus irinotecan, one would predict that an agent/regimen solely capable of decreasing vessel permeability should not meaningfully improve progression-free and overall survival among recurrent malignant glioma patients. Therefore, the improved rates of PFS and overall survival observed among patients treated with BV plus irinotecan, as compared with those achieved historically with other salvage therapies, strongly supports an underlying antitumor action of this regimen.
Following the Vredenburgh study [9,10], additional studies were initiated to determine the contribution of BV compared with the combination of BV plus irinotecan on outcome. Specifically, a randomized Phase II study comparing responses of GBM patients at first or second recurrence to either BV alone or the combination of BV plus irinotecan (CPT-11) recently completed accrual. Preliminary results of this study reveal that patients randomized to BV alone achieved a radiographic response rate of 21% and a 6-PFS of 36%. Patients who received BV plus CPT-11 appeared to fare better in that they achieved a radiographic response rate of 34% and a 6-PFS of 51%. The study results were not sufficiently mature to evaluate overall survival . In addition, single arm studies evaluating BV monotherapy among recurrent malignant glioma patients are underway, including an imaging-intensive trial conducted at the National Cancer Institute. Another ongoing study recently reported preliminary results of 15 recurrent malignant glioma patients treated with 15 mg/kg BV alone every 3 weeks. In this report, only two patients (13%) achieved a partial response, whereas five (33%) were stable, and eight (53%) progressed . Although preliminary, these results suggest an inferior disease control rate for BV alone as compared with that of BV plus irinotecan.
As stated above, most patients with radiographic response reported by Vredenburgh also noted improvement in neurologic status [9,10]. Gonzalez recently reported similar findings among eight patients with CNS tumors and radiation necrosis treated with BV, seven of whom also received chemotherapy. Specifically, a 48% (± 22% SD) average decrease in T1-weighted post-Gadolinium-contrast measurements and a 60% (± 18% SD) decrease in FLAIR changes were reported, while all patients on pretreatment dexamethasone therapy substantially reduced dexamethasone by an average of 8.6 mg (± 3.6 mg) per day .
Based on the highly encouraging, initial results reported for BV plus irinotecan, multiple additional studies are ongoing to evaluate different BV-based combinatorial regimens, including separate single-arm studies at Duke University evaluating BV plus protracted, metronomic dosing of either daily temozolomide or etoposide. In addition, the Radiation Therapy Oncology Group (RTOG) is currently conducting a trial randomizing recurrent GBM patients to receive BV with either protracted temozolomide therapy (75 mg/m2/day for 21 days each month) or irinotecan every 2 weeks (RTOG 0625). A study combining BV plus the EGFR inhibitor erlotinib is also underway, based on synergistic antitumor activity observed in preclinical GBM models treated with dual EGFR/VEGF inhibitor therapy .
At present, four single-institutional studies are underway to evaluate BV in the treatment of newly diagnosed GBM patients. Investigators at Duke are evaluating the addition of BV to chemoradiation with temozolomide, followed by cycles of BV plus temozolomide and irinotecan. In a separate study, neoadjuvant BV plus temozolomide is being evaluated prior to XRT/temozolomide for patients with bulky, unresectable or multi-focal tumors. A study at the University of California, Los Angeles is evaluating the addition of BV to standard temozolomide chemoradiation, while the combination of BV plus erlotinib plus standard temozolomide chemoradiation is being evaluated at the University of California, San Francisco. Finally, a multi-center, randomized Phase III clinical trial is being planned to evaluate the current temozolomide chemoradiation standard of care with or without BV.
In contrast to large, growth factor ligand antagonists such as BV that act in the extracellular and intravascular compartments, competitive inhibitors of the intracellular tyrosine kinase domain of VEGFRs are also being evaluated in ongoing clinical trials for malignant glioma patients. Vatalanib (PTK787/ZK222584; Novartis), a potent inhibitor of VEGFR1-3, has been evaluated as a single agent and in combination with either temozolomide or lomustine chemotherapy. Only modest rates of radiographic response and progression-free survival were reported, which may have been influenced by suboptimal, once-daily dosing [104,105]. Subsequent studies have demonstrated more effective pharmacokinetic exposures following twice daily vatalanib dosing . A Phase I study of vatalanib plus the PDGFR inhibitor imatinib mesilate for recurrent malignant glioma patients has recently been completed, and a clinical trial of vatalanib plus temozolomide chemoradiation is ongoing for newly diagnosed GBM patients.
Preliminary results were recently reported for a Phase II trial with cediranib (AZD2171, AstraZeneca, UK), a potent, oral pan-VEGFR, PDGFR, and c-kit inhibitor. Out of 16 patients, 9 (56%) achieved at least a partial response, and 3 additional patients achieved stable disease. The median time to progression was 15.8 weeks. Treatment was associated with meaningful improvements in tumor-associated edema, including the ability to reduce pretreatment corticosteroid dosing in eight of 11 patients (73%). Elegant collaborative imaging studies revealed that decreased contrast enhancement was accompanied by significant decreases in tumor vessel size, permeability, blood volume and blood flow, consistent with vascular normalization, which was also noted to reverse following drug interruption . Although intriguing, this report reflects a small study population and limited follow-up assessments. A multi-center, randomized clinical trial is planned to further evaluate cediranib plus lomustine among recurrent GBM patients.
Additional receptor tyrosine kinase inhibitors targeting VEGF are also under evaluation in ongoing clinical trials for malignant glioma (Table 1).
VEGF-Trap (Regeneron), consisting of portions of human VEGFR1 and VEGFR2 extracellular domains fused to the constant region of human IgG1, acts as a soluble decoy VEGF receptor to bind VEGF and thus prevent it from interacting with VEGFRs on tumor endothelial cells [107,108]. VEGF-Trap potentiates radiotherapy in GBM xenografts . A single-arm Phase II trial of VEGF-Trap for recurrent malignant glioma patients by the North American Brain Tumor Consortium recently completed accrual, and a multi-center clinical trial incorporating VEGF-Trap with temozolomide chemoradiotherapy is planned for newly diagnosed GBM patients.
Currently, malignant gliomas are assessed by measuring the largest bidimensional enhancing product following intravenous contrast administration on magnetic resonance imaging (MRI) . However, such assessments may be misleading in that potent anti-VEGF agents can decrease permeability and may lessen contrast enhancement with or without a true underlying antitumor effect. Modification of traditional MRI assessment criteria will probably be necessary to more accurately measure underlying tumor activity following anti-VEGF therapy. Additional imaging approaches may also increase the accuracy of response assessments including positron emission tomography (PET) [110–112], magnetic resonance spectroscopy , and complementary MRI techniques, such as dynamic contrast-enhanced MRI [114–116], dynamic susceptibility MRI [117–119], arterial spin labeling  and high-resolution magnetic resonance angiography [121,122].
Batchelor et al. recently reported that changes in mean blood vessel size and permeability detected by correlative imaging methods were associated with response among recurrent GBM patients treated with AZD2171 (cediranib) . Similarly, Chen et al.  recently reported that recurrent GBM patients who achieved a metabolic response using [18F] fluorothymidine positron emission tomography (FLT PET) following BV plus irinotecan had a threefold longer survival than non-responders (10.8 versus 3.4 months, p = 0.003). Furthermore, early (1–2 weeks after starting therapy) and late (6 weeks after starting therapy) metabolic response correlated more strongly with survival than MRI response.
In addition, circulating biomarkers offer promise to help predict and monitor response to therapy. Clinical evidence is emerging for several potentially valuable circulating biomarkers, including plasma VEGF [78,123–127], basic fibroblast growth factor (bFGF) [78,127], and tumor stromal-derived factor-1 (SDF1α) , as well as viable circulating endothelial cells [56,78,124,128].
Several potential mechanisms of resistance to anti-VEGF agents have recently been identified, including compensatory upregulation of alternative angiogenic factors such as PDGF/PDGFR-β, FGF, SDF-1α and angiopoietin-1 (Ang-1)/Tunica interna endothelial cell kinase homolog (Tie-2) following VEGF inhibition [78,129,130], increased mobilization of pericytes , secretion of endothelial cell survival factors , and the ability of glioma cells to induce a more invasive phenotype, accompanied by host blood-vessel co-option and eventual gliomatosis [132–136]. Future exploration of clinical strategies to circumvent these mechanisms of resistance-including agents or combinatorial regimens that target multiple angiogenic mediators and regimens that inhibit key mediators of both tumor angiogenesis and invasion [137–139] may prove highly valuable.
Malignant gliomas have long been appreciated as highly angiogenic. Insights into the biology of tumor angiogenesis, including the central role of VEGF, as well as the successful application of anti-VEGF agents with chemotherapy in other cancers, has led to the evaluation of VEGF as a therapeutic target for malignant glioma. Although initial results are highly encouraging, validation of these findings and further optimization of antiangiogenic strategies are critically required. Nonetheless, targeting VEGF/VEGFR may represent a major therapeutic advance for patients with these deadly tumors.
Malignant gliomas provide an attractive opportunity for antiangiogenic approaches, and recent clinical trials have demonstrated highly promising therapeutic benefit as well as acceptable safety among patients treated with the humanized anti-VEGF monoclonal antibody bevacizumab plus irinotecan. Specifically, in an initial clinical trial of recurrent GBM patients receiving bevacizumab plus irinotecan who had failed standard-of-care temozolomide chemoradiation and had a median of two prior episodes of progressive disease, the rate of radiographic response was 5- to 10-fold higher than that reported in the literature. Of greater importance, radiographic responses were meaningful in that they were associated with improved neurological function and the ability to successfully taper chronic dexamethasone treatment in most cases. Furthermore, responses were associated with greater durations of progression-free and overall survival than those reported for patients treated historically with alternative salvage agents. Furthermore, a recent study utilizing a small-molecule tyrosine kinase inhibitor to block VEGFR activity has also demonstrated similarly exciting therapeutic benefit and safety for recurrent GBM patients. Additional important clinical trials have recently been completed, are underway or are planned to initiate soon in order to build on the promising initial results and to further define the therapeutic potential of anti-VEGF therapy for recurrent malignant glioma patients.
In addition, investigators are highly enthusiastic to extend this approach to newly diagnosed GBM patients, particularly given the overall poor outcome of current standard therapy. A number of exploratory, single-arm studies have recently initiated to evaluate strategies incorporating BV into the current standard of care with temozolomide chemoradiation. Furthermore, a randomized Phase III study is being planned.
Nonetheless, several challenges exist and will require intensive study in the next few years in order for patients to fully benefit from these promising advances. VEGF-targeting agents decrease permeability and hence contrast enhancement, thereby frequently confounding the assessment of radiographic response. Traditional MRI assessment criteria will probably require modification in order to more accurately measure underlying tumor activity following treatment with these agents. Preliminary studies show that correlative MRI and PET approaches may provide imaging biomarkers to improve the reliability of response assessment. Promising circulating biomarkers of response also warrant further investigation. Antiangiogenic agents also have unique toxicities, including the possibility of an increased risk of thromboembolic events, which must be carefully monitored and addressed, particularly among patients with CNS tumors. Costs of newly developed anti-VEGF agents are significant and currently limit access to these agents for brain tumor patients. Finally, although initial results with anti-VEGF agents are very exciting, nearly all patients eventually relapse, which indicates that mechanisms of resistance must be better understood and circumvented. Optimization of anti-VEGF therapy for malignant glioma patients will require answering a number of key questions:
The following key issues have been identified:
This work was supported by National Institutes of Health grant nos. 1-P50-CA108786-01, NS20023, and CA11898 and by grant no. MO1 RR 30 through the General Clinical Research Centers Program, National Center for Research Resources, National Institutes of Health.
Declaration of interest
The authors have no conflict of interest to declare and no fee has been received for preparation of this manuscript.