INTRODUCTION — The malignant gliomas are rapidly progressive brain tumors that are classified as anaplastic astrocytoma (AAs), oligodendroglial tumors, and glioblastoma (GBM), based upon their histopathologic features. The prognosis for these patients is poor, even after aggressive treatment that includes surgery, adjuvant radiation therapy (RT), and adjuvant chemotherapy.

The development of new approaches to the treatment of malignant gliomas proceeds through clinical trials in the same way as other malignancies. Issues of particular importance in developing agents to treat these patients include the following:

Malignant gliomas include multiple tumor types, with associated differences in natural history and response to treatment (GBMs versus AAs versus oligodendroglial tumors).

Concomitant medications (particularly corticosteroids and antiepileptic drugs) can interact with the cytochrome P450 system, thus potentially altering the metabolism of systemic antitumor agents.

Accurate assessment of tumor response can be difficult, even with modern imaging modalities. Time to progression and the percentage of patients free from progression at six months are used as surrogate end points that more accurately reflect the antitumor activity of new agents.

CYTOTOXIC CHEMOTHERAPY — New cytotoxic agents to treat malignant gliomas generally have been evaluated in patients with recurrent or progressive disease. Temozolomide, now the preferred agent for adjuvant chemotherapy, was developed in this setting.

Alternative treatment schedules with temozolomide are being studied and may have utility in patients who have failed the standard temozolomide regimen. A number of other agents have demonstrated some activity in phase II studies, which was manifested by partial responses on imaging studies. These include procarbazine, etoposide, tamoxifen, pegylated liposomal doxorubicin, carboplatin, cyclophosphamide, paclitaxel, and the topoisomerase inhibitors irinotecan and topotecan.

In general, response rates (including stable disease) for monotherapy using these agents average 40 percent, with a median response duration of 26 to 30 weeks. Response rates are usually higher in patients with AA than GBM (50 to 60 versus 25 to 35 percent, respectively) and the median duration of response is longer (30 to 40 weeks versus 20 to 24 weeks). The use of various combinations of these agents may increase the objective response rate somewhat but has not been shown to increase survival.

These efforts to improve cytotoxic chemotherapy for malignant gliomas have not increased survival, and the focus of research has shifted to targeted therapies.

TARGETED SYSTEMIC THERAPY — An increased understanding of the molecular pathways involved in signal transduction, angiogenesis, and cell growth has led to the development of a number of targeted agents. Such molecularly targeted agents are under active evaluation, alone and in various combinations, for patients with malignant gliomas and other tumors.

Specific molecular targets and agents that are most promising or have been extensively evaluated are reviewed here, although all remain experimental.

Vascular endothelial growth factor pathway — Vascular endothelial growth factor (VEGF) plays a critical role in the development of the abnormal vasculature observed in malignant gliomas and other tumors. Antiangiogenic strategies targeting this pathway include the use of monoclonal antibodies that bind VEGF, small molecule inhibitors of the tyrosine kinases (TKs) activated by the VEGF receptors, and inhibitors of protein kinase C, which is an important downstream component of the VEGF cascade.

These approaches have promising activity in patients with malignant gliomas and are under continuing evaluation. The normalization of tumor vasculature induced by these agents can decrease tumor-associated edema and may facilitate the delivery of other agents to the tumor.

Bevacizumab (Avastin) — Bevacizumab is a monoclonal antibody that binds VEGF and prevents its interaction with VEGF receptors on the cell surface. Phase II results indicated that the combination of bevacizumab, either alone or in with irinotecan, may have an unusually high level of activity against recurrent malignant gliomas.

Cediranib (Recentin)— Cediranib is an oral inhibitor of all of the VEGF receptor TKs, as well as the TKs associated with platelet derived growth factor (PDGF) receptors. In preclinical studies, cediranib decreased tumor vessel permeability and controlled edema despite ongoing tumor growth

In a study of 16 patients with recurrent GBM, cediranib caused a structural and functional normalization of tumor vasculature within 24 hours in all patients, as documented by MRI These vascular changes required continuation of therapy, and interruption of drug resulted in a recurrence of vascular abnormalities. Partial responses were documented by imaging criteria in 9 of the 16 (56 percent), and the progression-free survival at six months was 26 percent. Hypertension due to treatment was observed in 15 patients (93 percent), 12 of whom required therapy. The normalization of tumor vasculature was accompanied by a reduction in tumor-associated vasogenic edema. Corticosteroid therapy was reduced or discontinued in all 11 patients who were steroid-dependent at the initiation of cediranib therapy.

Protein kinase C — Protein kinase C beta is an important component of the VEGF cascade, which appears to be the predominant angiogenic factor for gliomas. Enzastaurin is a potent and selective inhibitor of protein kinase C beta. Although a phase II study showed promising results, a phase III trial in which 266 previously treated patients with recurrent malignant glioma were randomly assigned to either enzastaurin or lomustine]. At the interim analysis, statistically nonsignificant differences in the objective response rate, progression-free survival (PFS), and overall survival all were inferior with enzastaurin compared to lomustine (3 versus 4 percent, hazard ratio [HR] for PFS 1.28, 95% CI 0.97-1.70, and HR for death 1.20, 95% CI 0.88-1.65).

Epidermal growth factor pathway — The epidermal growth factor receptor (EGFR) is dysregulated through overexpression, amplification, or activating mutations in the majority of patients with malignant gliomas. Although early studies suggested that small molecule inhibitors of the EGFR tyrosine kinase had activity in patients with malignant glioma, two multicenter phase II studies with erlotinib (Tarceva) have failed to confirm substantial activity for erlotinib in patients with malignant glioma, either for recurrent disease or as part of the initial treatment regimen.

In a phase II from the North Central Cancer Treatment Group (NCCTG), erlotinib was combined with temozolomide plus RT in 97 patients with GBM. There was no improvement in survival compared to other studies in which temozolomide was combined with RT. Sensitivity to erlotinib could not be predicted based upon analysis of molecular subsets. In a phase II trial conducted by the European Organisation for Research and Treatment of Cancer (EORTC), 110 patients with recurrent GBM were randomly assigned to erlotinib or a control arm of either temozolomide or carmustine. There was a lower six-month progression-free survival with erlotinib (11 versus 24 with conventional chemotherapy).

Monoclonal antibodies directed against the EGFR also are under evaluation. Nimotuzumab has shown activity in the pediatric population for the treatment of high-grade gliomas and is currently being evaluated in conjunction with RT in a phase III trial in children with diffuse pontine glioma. Another monoclonal antibody targeting the EGFR, a radioimmunoconjugate using I-125 Mab 425, has been evaluated in adults with GBM or malignant glioma

Platelet derived growth factor — Platelet derived growth factor receptor (PDGFR) is overexpressed in malignant gliomas. Inhibition of PDGFR has given only limited evidence of activity in treating patients with malignant gliomas. Imatinib, a potent inhibitor of multiple tyrosine kinase receptors including PDGFR, showed evidence of activity in a phase II study when combined with hydroxyurea. However, more extensive phase II studies have shown only minimal evidence of single agent activity.

Integrin inhibitors — Integrins are involved in endothelial cell proliferation and migration. Cilengitide is a cyclic pentapeptide that acts as a selective integrin antagonist, has direct antitumor activity, and inhibits angiogenesis. Cilengitide has been evaluated in phase I and phase II studies, which have confirmed that this agent is well tolerated, and suggested that it may improve survival in patients with newly diagnosed GBM when used in combination with temozolomide plus RT. A phase III randomized trial in patients with newly diagnosed GBM has been initiated, in which patients are randomly assigned to standard radiation therapy plus temozolomide or the same treatment plus cilengitide (NCT00689221) The trial is limited to patients whose tumors have a methylated promoter of the methylguanine-DNA methyltransferase (MGMT) gene.

Histone deactylase inhibitors — Histone deacetylase inhibitors are a novel class of agents that can induce tumor cell growth arrest, differentiation, or apoptosis. Vorinostat is an orally active agent in this class that is used in the treatment of cutaneous T-cell lymphomas. ( Vorinostat is under evaluation for patients with recurrent GBM. In a phase II study, 66 patients were with treated vorinostat. Overall, nine of the first 52 patients (17 percent) were progression-free at six months, which met the predetermined requirement for further evaluation. The most common severe side effects were thrombocytopenia and fatigue.

Thalidomide — The antiangiogenic activity of thalidomide is thought to result from inhibition of cell migration by decreased expression of leukocyte beta integrin subunits, which are crucial for cell-matrix interactions. Thalidomide has been evaluated as a single agent and in combination with carmustine in patients with recurrent malignant glioma. It has also been assessed in combination with temozolomide and celecoxib in patients who had just completed their radiation therapy. Although these trials have shown hints of activity, the contribution of thalidomide is uncertain and thalidomide does not have an established role in these patients.

BIOLOGIC APPROACHES — Various biologic approaches have been evaluated to treat patients with malignant glioma. Those that have been most extensively evaluated or show promise are discussed here.

Interferons — Interferons are immune modulators that stimulate antiviral and cytotoxic activity by inducing the expression of several cellular genes, a number of which participate in regulation of cell growth and apoptosis. Both interferon alpha (IFNa) and interferon beta (IFNb) have been evaluated in phase I and II clinical studies. Encouraging response rates and survival times were reported when interferon was combined with systemic chemotherapy. However, IFNa was ineffective in a phase III trial, in which 383 patients with newly diagnosed malignant glioma were randomly assigned to receive carmustine with or without IFNa (12 mU/m2 on days 1 to 3, during weeks 1, 3, and 5, every seven weeks) following RT [58]. IFNa did not significantly improve either the time to progression or overall survival, and treatment-related toxicity was increased.

Vaccines — Anticancer vaccines stimulate the host immune system to recognize the cancer as foreign. Because brain tumors do not elicit an effective immune response, vaccine therapy requires that the tumor sample be exposed to an agent that enhances the ability of antigen-presenting cells to stimulate an immune response. Several different approaches are under exploration: One approach has focused on dendritic cells. In two pilot studies, tumor lysates were used to stimulate dendritic cells, which were then used as a vaccine. Systemic cytotoxicity was elicited in the majority of treated patients, and modest antitumor activity was shown in two patients.

In an alternative approach, an autologous vaccine was prepared from tumor cell cultures that were infected with Newcastle disease virus [62-64]. Two pilot studies showed some evidence than an immune response was stimulated, although only one suggested a survival benefit. A vaccine (CDX110) uses a peptide based upon the mutated sequence in the epidermal growth factor receptor variant III (EGFRvIII) [65]. Studies with this vaccine suggest that it generates a humoral response, even when given in conjunction with temozolomide.

Gene therapy — Gene therapy involves the transfer or modification of a gene in tumor cells to stimulate a local immune response, replace a lost gene, block transcription, increase the sensitivity of a tumor to administered drugs, provide an enzyme that catalyzes the production of a cytotoxic product, or to transfect a tumoricidal virus.

A variety of approaches have been used for gene delivery, including the direct introduction of a gene and the use of various viral vectors. The most common experimental approach has been the introduction of the herpes simplex virus thymidine kinase gene into the tumor using a viral vector [66]. If ganciclovir is subsequently administered, it is phosphorylated by the thymidine kinase gene, leading to the production of toxic triphosphates that block DNA replication [67]. The diffusion of phosphorylated nucleosides away from dying cells to adjacent non-transduced tumor cells causes a "bystander effect", resulting in their death. Potential limitations of this strategy include lack of transduction of tumor cells, transduction of endothelial cells, and immunologic rejection of the murine vector cells.

BLOOD-BRAIN BARRIER — The most important factor affecting the delivery of drug to the brain and brain tumors is the transport of the agent across the blood-brain barrier (BBB) and the blood tumor barrier (BTB). The effectiveness of getting drugs across the BBB and BTB is also influenced by the regional blood flow and drug pharmacokinetic profile.

Strategies to increase drug delivery to brain tumors have included the following:

Intraarterial drug administration. Disruption of the BBB by hyperosmolar solutions or biomolecules High-dose chemotherapy

Direct intratumoral injection of free drug, or the use of drug embedded in a controlled-release, biodegradable matrix delivery system. Even if the BBB is overcome, drug access to tumor cells may be hindered by increased intercapillary distances, greater interstitial pressure, lower microvascular pressure, and the uptake of drug by normal brain tissue (the so-called "sink effect"). BBB characteristics — The capillary endothelium comprising the BBB consists of a sheet of epithelium-like cells connected by tight junctions on a basement membrane that lacks fenestrations. The physiologic role of the BBB is assumed to include maintenance of a constant biochemical environment in the interstitium; the BBB forms a barrier that is relatively impermeable to many water-soluble compounds and protects the brain from foreign molecules.  Although some drugs may be transported across the BBB using specific endothelial cell transport mechanisms, most cytotoxic drugs can gain access to the central nervous system (CNS) by passive diffusion. Drug properties associated with an increased ability to cross the BBB include the following: Lipid-solubility/ Neutral at physiologic pH/ Not highly bound to plasma proteins/ Molecular weight <200

Temozolomide and the nitrosoureas fulfill these criteria, and these agents are widely used for the treatment of malignant gliomas . Most other cytotoxic drugs are larger, with a molecular weight greater than 200. The normal BBB also expresses the P-glycoprotein on the luminal surface of capillary endothelial cells. This transmembrane protein is responsible for the active efflux of many cytotoxic agents from the brain and brain tumor cells. Preclinical studies suggest that agents that reverse the function of P-glycoprotein (eg, verapamil) may increase passage of agents such as doxorubicin, vinblastine, and paclitaxel across the BBB, but the clinical relevance of these findings is unknown.

Blood-tumor barrier — In contrast to normal brain, the neovasculature within brain tumors is characterized by open tight junctions, gap junctions, fenestrations, and numerous intracellular vesicles . Junctional clefts in the endothelial cells are also present in adjacent brain capillaries that are not in direct contact with the tumor. The increased permeability of the BTB is responsible for contrast enhancement of brain tumors on CT and MRI, since iodinated, water-soluble contrast agents do not penetrate areas of the brain with an intact BBB . The permeability of the BTB varies widely between patients and within different areas of the same tumor, and this results in large variations in the enhancement patterns of malignant brain tumors on contrast studies. The permeability alterations associated with the BTB are typically found in the central part of large malignant gliomas; the proliferating edges of these tumors seem to have a normally functioning BBB and typically do not enhance with contrast on CT and MRI. However, approximately 30 percent of patients with malignant gliomas have nonenhancing lesions

The clinical implications of these findings are twofold:

Infiltration of neoplastic cells may extend well beyond the tumor margin, as defined by contrast-enhanced CT or MRI, thus contributing to incomplete tumor resection (see "Clinical manifestations and initial surgical approach to patients with malignant gliomas" The presence of a selective, normal BBB near the proliferating edge of a brain tumor may result in variable delivery of water-soluble drugs to this region, contributing to the failure of chemotherapy to control disease.

Intraarterial chemotherapy — Intraarterial (IA) chemotherapy increases the ratio of regional to systemic blood flow, thereby achieving a higher local drug concentration with smaller drug doses. The goal of this approach is to increase the amount of drug passing through the BBB while minimizing systemic side effects. The potential disadvantages of the IA chemotherapy include the risk of catheter-related complications (eg, thrombosis, bleeding, infection), treatment-related neurotoxicity (orbital and cranial pain, retinal toxicity, leukoencephalopathy, cortical necrosis [96-98]), and the inability to use drugs that require systemic metabolic activation. Despite these theoretical advantages, at least two phase III trials failed to show a survival benefit for IA chemotherapy [96,97]. The lack of efficacy and the potential toxicity were illustrated by a Brain Tumor Study Group (BTSG) trial in which 315 patients with newly diagnosed malignant glioma were randomly assigned to IA or intravenous carmustine with or without IV 5-fluorouracil (5-FU) and radiation therapy (RT) . IA chemotherapy was associated with significantly shorter survival and severe treatment-related toxicity (including irreversible leukoencephalopathy and ipsilateral blindness in 10 and 15 percent, respectively).

BBB disruption — Drug delivery to brain tumors can potentially be improved by increasing the permeability of the BBB with hyperosmolar solutions (most often mannitol) that induce an osmotic opening of the BBB and BTB. The technique of BBB disruption is complex, requiring transfemoral angiography and general anesthesia. Potential complications include stroke and seizures. Although BBB disruption has theoretical advantages, no clinical benefit has been demonstrated with this approach.

High-dose chemotherapy — Increasing the administered dose might drive more drug across the BBB by increasing the peak concentration of unbound drug in the circulation, thus helping to overcome chemotherapy resistance. The major dose-limiting toxicity of alkylating agent-based high-dose chemotherapy (HDT) is myelosuppression, which can be overcome by the use of autologous hematopoietic cell rescue.

Despite its theoretical advantages, HDT has not made a significant impact on survival for patients with recurrent malignant gliomas, although anecdotal cases of long-term survival have been reported. The median survival using HDT with hematopoietic cell rescue as a component of initial therapy is comparable to that with local therapy followed by conventional dose chemotherapy in most studies, and treatment-related morbidity is substantial.

Intratumoral drug administration — Direct delivery of a drug into the tumor or a postoperative tumor resection cavity should maximize local drug concentrations while minimizing systemic effects. Two techniques have been used to directly administer drug to the tumor, slow-release systems and direct infusion. Slow-release carrier systems — Controlled-release methods employing various carrier systems permit constant drug delivery into the tumor while protecting the unreleased drug from metabolism. The most extensively evaluated slow-release system is the carmustine polymer wafer, which has been approved for use in patients with malignant gliomas.

Convection enhanced delivery — The feasibility of intratumoral infusion with various chemotherapeutic drugs has been demonstrated in clinical trials. However, no beneficial impact on overall survival was noted in initial studies. Contemporary direct drug administration has utilized convection-enhanced delivery (CED) through surgically implanted catheters. CED optimizes the delivery of antitumor agents to the tumor using a positive-pressure infusion to control drug distribution by adjusting the infusion rate and volume. The potential utility of CED was illustrated by a report of 15 patients with recurrent malignant glioma who were treated with CED of paclitaxel. There were five complete and six partial responses. Complications included chemical meningitis, infections, and transient neurologic deterioration, which was thought to be due to peritumoral edema. CED may be particularly useful for the delivery of large molecules. As an example, CED has been used to deliver cintredekin besudotox, a conjugate of human interleukin-13 (IL-13) with pseudomonas exotoxin . Malignant gliomas express the IL-13 receptor on the cell surface, and this conjugate is used to deliver locally high concentrations of the pseudomonas exotoxin. Multiple phase I studies have established the safety of delivering this agent by CED and suggested antitumor activity in patients with malignant gliomas. However, preliminary data from a phase III trial failed to demonstrate any improvement compared to carmustine wafers.

DIFFERENTIATING AGENTS — Retinoids are differentiating agents that directly induce cell differentiation in vitro and inhibit cell growth, possibly by stimulating transforming growth factor beta. All-trans retinoic acid (ATRA), cis-retinoic acid, and fenretinide have been evaluated in clinical studies in patients with malignant gliomas. Results with this class of agents are illustrated by the following studies:

ATRA was evaluated in 30 patients with recurrent malignant glioma. Although 3 patients had partial responses, and the median time to tumor progression and survival were only 3.8 and 5.7 months, respectively. Cis-retinoic acid had modest activity in a phase II study of 80 patients with recurrent malignant glioma. The median progression-free and overall survival durations after treatment initiation were 10 and 25 weeks, respectively, while the six-month progression-free survival was 19 percent. Cis-retinoic acid has been combined with temozolomide both in patients with recurrent and primary malignant gliomas, without evidence of a synergistic effect. Inhibitors of histone deacetylase have also been evaluated as differentiating agents in patients with malignant gliomas. In an extensive phase II study, phenylacetate showed little if any activity. Sodium phenylbutyrate has been less extensively evaluated, but one prolonged complete remission has been reported with this agent

RADIOIMMUNOCONJUGATES — Direct administration of radiolabeled monoclonal antibodies into surgically created resection cavities can deliver a highly localized radiation dose to residual tumor, with minimal exposure to normal brain. Tenascin is an extracellular matrix glycoprotein that is expressed in malignant gliomas but not in normal brain. Intracerebral delivery of radioimmunoconjugates directed against tenascin into a surgical resection cavity has been evaluated in multiple studies No controlled trials have been conducted, but phase II studies suggested an antitumor effect. In one report that included 33 patients with newly diagnosed malignant glioma, the median survival of those with GBM was 79 weeks. In another series of 43 patients with locally recurrent malignant brain tumors, the median survival for those with GBM was 64 weeks and 63 percent were alive at one year. A phase III trial is being planned.

A radioimmunoconjugate directed against EGFR has also been evaluated as an adjuvant in patients with malignant glioma.

SUMMARY — Malignant gliomas are associated with a poor long-term prognosis. A variety of approaches have been used to improve results in these patients. The most promising of these rely upon an increasing understanding of the molecular pathways involved in tumor growth, and specific therapies designed against these targets. Multiple agents are in phase I, II, and III evaluation. Patients should be encouraged to participate in clinical studies whenever possible.