|Home | About | Journals | Submit | Contact Us | Français|
Median survival of patients with malignant glioma (MG) from time of diagnosis is approximately 1 year, despite surgery, irradiation and conventional chemotherapy. Improving patient outcome relies on our ability to develop more effective therapies that are directed against the unique molecular aberrations within a patient’s tumor. Such molecularly targeted therapies may provide novel treatments that are more effective than conventional chemotherapeutics. Recently developed therapeutic strategies have focused on targeting several core glioma signaling pathways, including pathways mediated by growth-factors, PI3K/Akt/PTEN/mTOR, Ras/Raf/MEK/MAPK and other vital pathways. However, given the molecular diversity, heterogeneity and diverging and converging signaling pathways associated with MG, it is unlikely that any single agent will have efficacy in more than a subset of tumors. Overcoming these therapeutic barriers will require multiple agents that can simultaneously inhibit these processes, providing a rationale for combination therapies. This review summarizes the currently implemented single-agent and combination molecularly targeted therapies for MG.
The term glioma refers to a group of cancers that includes astrocytomas, oligodendro-gliomas and mixed gliomas. Astrocytomas constitute the largest group of CNS tumors and can be further subdivided. The WHO classifies tumors into four grades based on cellular differentiation. Low-grade glioma resembles its tissue of origin (differentiated), while high-grade glioma does not (undifferentiated). Glioblastoma multiforme (GBM), which is a high-grade or malignant glioma (MG), is the most common form of astrocytoma and is defined by histologic criteria that include hyper-cellularity, pleomorphism, necrosis, pseudopallisading and vascular proliferation . GBMs exhibit abnormal cell proliferation and tumor angiogenesis  and can be categorized into two groups based on clinical presentation as primary or secondary. Secondary GBMs, which represent 5% of all GBMs, show malignant progression from an antecedent lower grade tumor, whereas primary GBMs, which represent 95% of all GBMs, present as advanced cancers . These two presentations of GBM are also associated with different sets of molecular genetic alterations .
Despite aggressive surgical approaches, optimized radiation therapy regimens and the application of cytotoxic chemotherapies, the median survival of patients with GBM from time of diagnosis is approximately 14.6 months, which has changed little in decades . The poor response of MG, which includes WHO grade III and IV tumors, to conventional therapies reflects a resistance to undergoing apoptosis in response to DNA damage; this may result from mutations of tumor suppressor and cell-cycle control genes and aberrant activation of growth and survival signaling pathways. Understanding the molecular genetic alterations in cancer may provide more effective therapeutic strategies to target the specific mutations in a patient’s tumor.
Completion of the human genome project, integrated genomic analyses of primary and secondary GBM [3,6], and the emergence of molecularly targeted therapeutics have provided insight into a large number of neoplastic molecular alterations, which may ultimately offer novel drug targets. Molecularly targeted therapies may provide novel treatment modalities that are more effective and less toxic than conventional chemotherapies. Development of these therapies has recently focused on targeting several signaling pathways, such as pathways mediated by growth factors, PI3K/Akt/PTEN/mTOR, Ras/Raf/MEK/MAPK and others. However, given the molecular diversity and heterogeneity of MGs, it is unlikely that any single agent will have efficacy in more than a subset of tumors. Parallel, redundant and converging signaling pathways associated with gliomagenesis have required the use of multiple small molecule inhibitors that can simultaneously target and inhibit these processes. In this review, we discuss the currently available single-agent and combination molecularly targeted therapies for the treatment of MG.
During the past decade, our understanding of the molecular aberrations occurring in MG has deepened [7–10]. Recent genome-wide molecular characterizations have provided a novel view of the molecular genetic backdrop of GBM [3,6], and research efforts have focused on molecularly targeted therapies with the capacity to specifically target unique tumor aberrations while leaving normal brain cells unharmed .
Despite the variability and heterogeneity of these tumors, common alterations in specific cellular signal transduction pathways occur within most MGs (Figure 1). These genetic alterations drive invasiveness, proliferation, cell survival, evasion of apoptosis, avoidance of immune surveillance and ability to form and sustain new blood vessels [12–14]. Our understanding of these aberrations in glioma are being translated into novel brain tumor treatments, and the spectrum of available molecularly targeted agents can be categorized into several broad groups inhibiting growth factor receptors, signaling pathways, angiogenesis, gene transcription and protein processing, among others (Figure 1). Box 1 summarizes many of the currently implemented molecularly targeted agents. Although additional agents for these and many other targets of interest are in various stages of development, Box 1 provides an overview of the diversity of targeted options and the multiplicity of agents for many of these targets in glioma.
Growth factor pathways provide the proliferative drive in most MGs , and rational therapy has focused on abnormalities within these pathways. Several growth factor receptor pathways are activated in MG, including EGF receptor (EGFR), PDGF receptor (PDGFR) , IGF receptor , FGF receptor , VEGF receptor (VEGFR)  and the TGF-β receptor [20,21]. These receptors and their ligands are overexpressed or mutated to varying degrees on tumor cells, and overactivation of these pathways has lead to aberrant intracellular signaling, tumor development, tumor proliferation, apoptotic resistance, motility, invasion and neoangi-ogenesis . Thus, growth factor receptors, particularly EGFR and PDGFR, have constituted a major target for drug development.
The EGFR is one of a family of receptors that is amplified in approximately 50% of GBMs and overexpressed in many MGs independent of amplification status [14,22]. In approximately 40% of tumors with EGFR amplification, the gene has a deletion of exons 2–7 that causes a defect in the extracellular ligand-binding domain, leading to ligand-independent constitutive activation [23,24]. This mutant is known as EGFRvIII and may be an independent prognostic factor for poor survival outcome [14,25]. Since activated EGFR induces tyrosine phosphorylation of substrates that contribute to cell proliferation, excessive activation of this protein, either by ligand binding or mutation-induced constitutive signaling, may provide cells with a growth advantage under certain conditions.
Gefitinib (Iressa®, ZD1839, AstraZeneca) is a molecularly targeted agent that has been tested in MG. This agent is a competitor for the intracellular ATP binding site of EGFR and is effective in blocking EGFR-dependent cell signaling in cell lines that rely heavily on EGFR activation for proliferative stimulation and tumor growth in EGFR-dependent xenograft models . Initial clinical studies reported that gefitinib was well tolerated at effective doses , with common toxicities being an acneiform skin rash and diarrhea. Results from several clinical studies for non-CNS solid tumors demonstrated activity of gefitinib as a single agent [28,29], although Phase III clinical studies for advanced small-cell lung cancer failed to demonstrate a convincing benefit of adding this agent to conventional treatments . Not surprisingly, recent reports have demonstrated that response to gefitinib is strongly influenced by tumor EGFR status, with a high percentage of objective responses among patients with EGFR mutations .
With these issues in mind, Phase I/II studies of gefitinib were initiated in patients with GBM. In a Phase II study, 12.7% of patients were noted to have partial tumor regression, although the median times to progression were not better than historical controls . In another study for recurrent GBM, the median progression-free survival (PFS) was 2 months, the PFS at 6 months (PFS6) was 13% and the median overall survival (OS) was 10 months, with no observed radiographic response . Another Phase II study by the North Central Cancer Treatment Group (NCCTG) had noted some radiographic responses but no superior survival benefits .
Clinical trials for gliomas have also been completed for erlotinib (Tarceva®, OSI-774, Genentech), which also reversibly inhibits EGFR by competition with the ATP binding site . Phase I studies reported toxicities that were comparable to gefitinib . Phase II studies in patients with non-small-cell lung cancer demonstrated tumor response , and a recently completed trial for MG showed that erlotinib, alone or with the DNA alkylating agent temozolomide (TMZ), demonstrated a PFS6 of 10.5% . However, as with gefitinib, median PFS for these patients was no better than historical control data. Several Phase II trials observed only modest impacts on PFS6 for erlotinib as a single agent [39–42], and although one study found that the combination of erlotinib with TMZ and concurrent radiation therapy was well tolerated, it was unclear if outcomes improved .
Since responses, albeit modest and short-lived, were observed with these agents, efforts were made to determine whether a molecular genetic profile could distinguish responders from nonresponders. Although one analysis failed to note a clear association between EGFR status and response , subsequent studies noted that responses were typically seen in tumors that had amplification of EGFR combined with low levels of Akt  or expression of the EGFRvIII variant and preservation of PTEN , thereby suppressing constitutive Akt activation. These findings may provide a rationale for stratifying patients based on genomic studies that identify the molecular aberrations in individual tumors.
Lapatinib (GW572016, GlaxoSmithKline, Research Triangle Park, NC, USA) is another EGFR inhibitor that is currently in an early phase of trials  and has demonstrated activity against erbB2-expressing tumors . A Phase II study incorporated a molecular biology component for patients in whom a resection was planned. In these patients, lapatinib was administered prior to surgical resection and then a portion of the tumor was analyzed for suppression of the EGFR and other downstream molecular features. A recent Phase II study demonstrated that lapatinib does distribute into glioma tissue, and a formal report of response data is eagerly awaited .
Cetuximab (Erbitux®, ImClone Systems) is a monoclonal antibody against EGFR that has demonstrated preclinical anti-tumor activity in GBM and was effective when administered systemically in an intracranial model [14,50]. However, antibodies have limited penetration into the tumor due to the BBB, which may limit the clinical efficacy of this therapeutic modality. In a recent Phase II study that stratified patients according to EGFR gene amplification status, cetuximab was well tolerated but had limited activity in patients with recurrent high-grade glioma, with a median OS of 5 months .
EGFR: EGF receptor; HDAC: Histone deacetylase; HSP: Heat-shock protein; PDGFR: PDGF receptor; PKC: Protein kinase c; Topo I: Topoisomerase I; VEGFR: VEGF receptor.
Overexpression and activation of PDGFR may be an important alteration contributing to the phenotype of MGs  and represents an important step in the transition from WHO grade II–III gliomas in adults . PDGF was originally identified as a potent mitogen for fibroblasts, glial cells and smooth muscle and was shown to stimulate tumor growth and angiogenesis in preclinical studies . The various PDGF isoforms (AA, AB, BB, CC, DD) bind with differential affinity to two cell-surface PDGFRs [16,55] and concurrent expression of one or more of these ligands and their receptors has been observed in a high percentage of MGs .
A number of PDGFR inhibitors are in various phases of clinical development. One agent that has been extensively tested is imatinib (STI571, Gleevec®, Novartis Pharmaceuticals), which is a kinase inhibitor of PDGFR, c-Kit and BCR–ABL that has exhibited antiglioma activity alone  and in combination with radiation therapy . This compound initially became a focus of clinical interest because of its potent inhibition of BCR–ABL, which is associated with Philadelphia chromosome-positive chronic myelogenous leukemias [59,60], and of c-Kit, which is constitutively activated in a substantial percentage of gastrointestinal stromal tumors . As imatinib was also potent in blocking PDGFR signaling , this agent was tested in tumors with PDGFR-driven proliferation.
Preliminary studies of imatinib in glioma cell lines demonstrated inhibition of proliferation in vitro of GBM cell lines and delay of tumor growth in vivo in heterotopic glioma models ; however, results were less impressive than those noted for c-Kit and BCR–ABL-dependent tumors. Based on these results, dose- escalation studies in MGs were initiated by several cooperative groups, including the North American Brain Tumor Consortium (NABTC) and Pediatric Brain Tumor Consortium. The NABTC trial, however, achieved a PFS6 of only 16% , and an unexpected finding from several of these studies was intratumoral hemorrhage during treatment . Further studies, however, are required to determine if imatinib is associated with an increased risk for intratumoral hemorrhage. The European Organization for Research and Treatment of Cancer (EORTC) conducted a Phase II study of imatinib in recurrent gliomas, and preliminary results showed a modest PFS6 of 15.7% . A recent study by Marosi et al. demonstrated a PFS6 of 32% in patients who had expression of PDGFR detected by immunohistochemistry, indicating that patient stratification based upon tumor molecular phenotype may enhance therapeutic efficacy .
The response of cells to growth factors is mediated by cell-surface receptors that interact with downstream signaling components. Upon binding to ligand, these growth factor receptors transduce a signal through several common pathways, including the Ras/Raf/MAPK and PI3K/Akt pathways (Figure 1) [67–70]. These signaling molecules ultimately relay information to various parts of the cell to modulate cell growth, differentiation, protein trafficking, secretion of angiogenic factors, membrane activity and apoptosis. Inhibition of intermediate and downstream components of growth factor signaling pathways is, therefore, a promising strategy for interfering with the proliferation of MG and other brain tumors .
In certain tumor types, mutation of one of the Ras genes to a constitutively active protein has been associated with tumorigenesis . Although such mutations are rare in MG, Ras activity may be markedly elevated as a result of deregulation of signaling through growth factor receptors [73,74]. These observations suggest that targeted inhibition of Ras-dependent signaling may constitute a therapeutically useful strategy.
Before translocation to the cell membrane and subsequent activation, Ras undergoes a post-translational lipid modification catalyzed by farnesyltransferase, and one strategy for interfering with Ras activity has involved targeting this process [75,76]. Although farnesyltransferase inhibitors (FTIs) inhibit Ras farnesylation, their antiproliferative effects are not specific, as they target other proteins, including the centromere-associated proteins CENP-E and CENP-F, RhoB and E, the nuclear lamins and Rap2 [77,78]. Previous studies have demonstrated that astrocytomas are sensitive to growth inhibition by FTIs at low micromolar concentrations .
Tipifarnib (R115777, Zarnestra®; Johnson and Johnson) is a nonpeptidomimetic FTI with activity against many human cancer cell lines and xenograft models . Although some activity was observed in patients with hematological malignancies , the results of Phase II and III trials in adults with solid tumors have been less promising [82,83]. In a recent Phase II trial, tipifarnib showed only modest activity, with a PFS6 of 12% in recurrent GBM . Another study was stopped early due to progression of disease in 48% of patients, where administration of tipifarnib prior to irradiation in patients with newly diagnosed GBM and residual enhancing disease did not result in improvement in survival .
Lonafarnib (SCH66336, SARASAR, Schering-Plough) is another FTI that was examined in a Phase I study . This agent has previously been evaluated in a variety of tumor types alone and in conjunction with conventional chemotherapeutic agents [87,88], and further studies will be needed to determine the efficacy of this agent in patients with glioma.
Another important signaling pathway in glioma involves activation of protein kinase C (PKC), which constitutes an element of the signaling cascade of several growth factors, such as EGF and PDGF, that stimulate glioma cell proliferation . Activation of PKC induces phosphorylation of other effectors, such as Raf and MAPK, as well as activation of Ras [90,91]. Levels of PKC activity have been noted to correlate with proliferative status in neoplastic astrocytes , and astrocytoma cells express levels of PKC that are up to tenfold higher than in normal astrocytes. In addition, pharmacological and antisense agents that target PKC isoforms that were overexpressed in astrocytomas diminished glioma proliferation in vitro .
Tamoxifen is a well-known anti-estrogen drug that also inhibits PKC. High doses of tamoxifen have demonstrated anti-tumor activity in glioma xenografts , but failed to show benefit in clinical trials [95,96]. In an attempt to increase clinical efficacy, tamoxifen has been combined with many other standard chemotherapeutic regimens, such as low-dose TMZ, without clear additional benefit to date . One study used a thyroid function suppressor, propylthiouracil, to modulate the effect of IGF-1, which is a naturally occurring antagonist of tamoxifen. In this study, median OS was significantly extended in hypothyroid patients , although further studies are needed to evaluate this effect.
Enzastaurin (LY317615, Eli Lilly and Company) is a PKC inhibitor that prevents substrate phosphorylation by competing with the enzyme’s ATP binding site. This agent has demonstrated antiangiogenic activity , as well as anti-tumor activity, in U87 glioma xenografts and a variety of other solid tumors . Additionally, this compound inhibits signaling through the PI3K/Akt signaling pathway, suppresses phosphorylation of glycogen synthase kinase 3β, induces apoptosis and inhibits proliferation in cultured cell lines from a variety of human cancers, including gliomas, and has antiangiogenic activity. Enzastaurin was well tolerated as a single agent  and preliminary results in adult patients with recurrent MG reported that 22% of patients achieved objective radiographic responses and 5% achieved stable disease . However, intratumoral hemorrhages were noted in seven patients. Based on these encouraging results, a Phase III study had been initiated in adult patients with GBM but was terminated prematurely owing to clinical inefficacy.
One of the most critical downstream signals to reach the nucleus involves the MAPK cascade. This cascade involves multiple separate protein kinases. The most proximal kinase is the Raf (MAPKKK) family, which includes at least three members. Raf, a serine/threonine kinase, is recruited to the cell membrane, stabilized by interaction with other proteins and phosphorylated to an active form . Activated Raf phosphorylates and activates MAPK/ERK kinase (MEK, also known as MAPKK), which subsequently activates MAPK (also known as ERK) . These enzymes subsequently activate a number of downstream mediators that regulate transcription, protein translation and cytoskeletal rearrangement.
Several agents have been developed that function as Raf kinase inhibitors. Sorafenib (BAY 43-9006; Bayer) has been a focus of combinatorial studies for MGs in preclinical and clinical studies [104,105]. Since this agent also inhibits VEGFR activation, it may have applicability as an antiangiogenic agent (discussed later). Additionally, the novel inhibitor AAL881 (Novartis), which is a dual inhibitor of Raf and VEGFR, induced profound tumor responses in gliomas in a preclinical study but is not currently being used clinically .
The activation of Akt via PI3K is another major pathway implicated in growth factor receptor-mediated signaling. PI3K is a phospholipid kinase that regulates several cancer phenotypes, including cell growth, proliferation and antiapoptosis, and activation of this pathway is associated with a poor prognosis in patients with glioma [14,107]. Upon cell-surface activation, PI3K phosphorylates phosphatidylinositol 4,5-biphosphate (PI4,5P2, PIP2) to form PI3,4P2 and PIP3. PIP3 leads to translocation and activation of Akt. Activated Akt phosphorylates several proteins involved in cell survival and growth signaling, such as Bad, mTOR, forkhead transcription factor, glycogen synthase kinase and mTOR [108,109].
Under normal conditions, Akt activation is inhibited by PTEN, a phosphatase that converts PIP3 to PIP2 . However, PTEN is mutated in at least 40% of GBMs , particularly in primary GBMs, which leads to constitutive activation of the PI3K pathway. The importance of this pathway in gliomagenesis is highlighted by the fact that transfer of a wild-type PTEN gene to the PTEN-deleted U87 glioma cell line suppressed tumor growth, leading to cell cycle arrest . Conversely, increased PI3K/Akt activity has been associated with resistance to radiation therapy .
Based on these observations, PI3K and Akt appear to be rational therapeutic targets for GBM. Perifosine (Keryx) is an oral Akt inhibitor that is undergoing assessment in MG . Inhibitors of PI3K, such LY294002, have demonstrated promising activity in preclinical models, but their toxicity profile has precluded clinical use .
Activated by both Akt and Ras pathways, mTOR transduces proliferative signals mediated through the PI3K/Akt pathway by activating the downstream ribosomal S6 kinase and inhibiting the eukaryotic initiation factor 4E-binding protein-1, which are required for translation of mRNAs necessary for progression from G1 to S phase [14,114]. Overexpression of growth factors or deletion of PTEN increases activation of mTOR. Several mTOR inhibitors are being investigated in clinical trials, including rapamycin (sirolimus, Rapamune®, Wyeth), temsirolimus (CCI-779, Wyeth), AP23573 (Ariad) and everolimus (Rad-001, Certican®, Novartis). All of these agents inhibit GBM proliferation in culture and in intracerebral xenografts.
Temsirolimus binds to the immunophilin FK-506 binding protein 12 (FKBP12) and forms a complex that inhibits mTOR. It has demonstrated modest activity in recent Phase II studies , both as a single agent and in combination with cisplatin for human medulloblastoma and glioma cell lines . In a recent Phase II study, administration of temsirolimus to patients with progressive GBM was well tolerated, with some objective responses . High levels of phosphorylated ribosomal S6 kinase at baseline seemed to correlate with response to treatment. Radiographic improvement was evident in some patients, although the PFS6 was only 7.8%  and 2.5%  in two studies.
NVP-BEZ235 is a novel dual PI3K/mTOR inhibitor that has shown inhibitory effects on tumor vasculature , breast cancer growth in cells with activating PI3K mutations  and primary human pancreatic cancer growth as xenografts . In vivo glioma models treated with this agent showed antiproliferative activity , and a Phase I trial in patients with solid tumors, particularly breast cancer, is currently underway . While this agent is not being applied in glioma clinical trials, studies with other PI3K/mTOR inhibitors, such as XL765, are in progress.
Glioblastoma multiformes are among the most highly vascularized tumors [122,123], where genetic aberrations result in upregulation of proangiogenic factors and downregulation of angiogenic inhibitors. Glioma cells also have remarkable invasive capabilities that allow infiltration into the surrounding brain . Proangiogenic cytokines, such as VEGF that is upregulated in most patients with GBM and is the most potent endothelial cell mitogen, drive an increase in vascularity and endothelial cell proliferation . Several therapeutic approaches that inhibit tumor cell migration and endothelial cell proliferation are being evaluated, including VEGFR inhibitors, integrin antagonists, endothelin receptor antagonists and PKC inhibitors.
Secretion of VEGF by tumor cells depends heavily on EGFR-mediated signaling  and a significant component of the therapeutic efficacy of EGFR-targeting agents reflects this secondary effect on tumor angiogenesis [127,128]. Similarly, PDGF has been demonstrated to stimulate tumor angiogenesis  in addition to supporting the growth and survival of vascular pericytes and promoting VEGF secretion by glioma . Thus, an important consideration in evaluating the therapeutic utility of growth factor receptor inhibition must focus on the effects of these agents on the surrounding tumor vasculature [64,127].
As VEGF represents a major stimulatory factor for the initiation of angiogenesis, inhibition of ligand–receptor interactions has been a focus of recent attention for MG. A number of agents have been examined, including vatalanib (PTK787/ZK 222584, Novartis and Schering AG), ZD6474 (Zactima™, AstraZeneca), and CEP-7055 (Sanofi-Aventis). In Phase I/II trials, vatalanib, a VEGFR kinase inhibitor, achieved 4% partial response (PR) and 56% stable disease . Although combinations of vatalanib with either TMZ or lomustine were well tolerated in another Phase I/II trial, anti-tumor efficacy was modest . In a trial of 16 patients, AZD2171 sustained reduction in permeability with a reduction in tumor-associated vasogenic brain edema and clinical benefit in most patients . Clinical trials of other multi-targeted kinase inhibitors, such as sorafenib and sunitinib (SU11248), have also been initiated in MG and results are eagerly awaited (Table 1).
Bevacizumab (Avastin®, Genentech) is a recombinant, humanized neutralizing monoclonal antibody to VEGF that has demonstrated encouraging radiographic results in patients with recurrent MG [14,134] and has recently been approved by the US FDA for treatment of recurrent GBM. This agent has been shown to decrease vascular permeability and increase apoptosis in intracranial xenografts of human glioblastoma, and synergism has been observed with several chemotherapeutic agents . In a Phase III placebo-controlled trial of 815 patients with metastatic colorectal cancer who were randomized to receive combination bevacizumab therapy, bevacizumab produced a significantly better rate and duration of response compared with placebo . Bevacizumab also prolonged PFS compared with placebo for patients with metastatic renal cell cancer . On the basis of the preclinical and clinical efficacy demonstrated in adult studies of recurrent solid tumors, a Phase II study was initiated for patients with recurrent MG , where the PFS6 was 46%, OS6 was 77% and PR was 57%. In another Phase II study of bevacizumab alone or in combination with irinotecan, the PFS6 for bevacizumab alone was 35.1% and for the combination was 50.2%, both providing encouraging activity in patients with recurrent GBM . These data suggest that there may be a meaningful difference in response rates, but the limited size of the study precluded a determination of whether there was a statistically significant improvement with the combination. Additionally, a recent Phase II trial reported that single-agent bevacizumab had significant biologic activity in patients with recurrent GBM, with a PFS6 of 29% and median OS of approximately 8 months .
VEGF Trap (Regeneron) is a soluble decoy receptor of VEGF that has demonstrated efficacy in several preclinical cancer models . It is under clinical development for treatment of MG by the NABTC .
Thalidomide (Thalomid®, Celgene) has been investigated for the treatment of GBM because of its antiangiogenic effects. Although the exact mechanism of action is unclear, thalidomide probably acts as an inhibitor of VEGF and bFGF and may interfere with integrin receptors . This agent has had modest activity in this setting, although changes in serum levels of bFGF correlated with overall survival . More recent studies have examined lenalidomide (CC-5013, Revlimid®, Celgene), a more potent analog of thalidomide. Long-term stable disease has been observed in several patients with gliomas, and a Phase II study of irinotecan plus lenalidomide in patients with recurrent GBM is underway . The combination of thalidomide and chemotherapy appears to be more active in patients with recurrent gliomas than either approach alone .
Another agent being examined as a potential angiogenesis inhibitor and inhibitor of cell invasion is the integrin inhibitor cilengitide (EMD121974, EMD). Integrins are often over-expressed in glioma and mediate cell adhesion, migration and invasion into the surrounding tissue . In a recent Phase II trial, cilengitide monotherapy was well tolerated and exhibited modest anti-tumor activity in patients with recurrent GBM, with a PFS6 of 15% and median OS of 9.9 months .
Other angiogenic inhibitors of interest include COX-2 inhibitors, based on the association between COX-2 overexpression in glioma and angiogenesis . Both celecoxib (Celebrex®, Pfizer) and rofecoxib (Vioxx®, Merck) have been combined with conventional chemotherapeutic agents in studies of patients with recurrent brain tumors [147,148]. COX-2 inhibitors may also have a role in treating peritumoral edema . Furthermore, endostatin and angiostatin, which are natural product inhibitors of angiogenesis, and atrasentan (Xinlay™), which is a selective inhibitor of the endothelin A receptor, are agents that may inhibit glioma-induced angiogenesis [150,151]. However, the median PFS was only 1.5 months in a Phase I study of atrasentan. The previously described enzastaurin has also been noted to decrease VEGF levels in a mouse tumor model [98,152].
Recent studies have indicated that alterations of gene expression by epigenetic modifications may influence the growth-promoting phenotype of MG. Histone proteins organize DNA into units called nucleosomes , and the acetylation (by histone acetyltransferases) and deacetylation (by histone deacetylases [HDACs]) of these proteins play an important role in the regulation of gene expression . Acetylation of histone lysine residues is associated with relaxation of the DNA wrapped around the core histones, enhancing access by the transcriptional machinery; conversely, deacetylation condenses the nucleosome structure, restricting access to the DNA . There is evidence that histone processing is altered in MG, and treatment of cells with inhibitors of HDACs may result in increased expression of a variety of genes that promote cell cycle arrest, inhibit cell growth, induce terminal differentiation, induce apoptosis and prevent formation of tumors [153,155].
Phenylbutyrate was one of the first HDAC inhibitors to be evaluated clinically. In a Phase I/II study in adults with recurrent solid tumors, 25% of patients had stable disease for over 6 months  and, in a study of recurrent multicentric MG, Baker et al. reported a complete response . Clinical trials of several HDAC inhibitors, including valproic acid (Depakote), depsipeptide (FK228) and vorinostat (suberoylanilide hydroxamic acid), are currently underway or planned . A preclinical study of vorinostat suggested that HDAC inhibition could enhance radiation-induced cytotoxicity in human prostate and glioma cells , and a recent Phase II trial of vorinostat in recurrent GBM reported that monotherapy was well tolerated in patients with recurrent GBM and had modest single-agent activity, with a median OS of 5.7 months . Trials of combination therapies with the HDAC inhibitors vorinostat or depsipeptide are currently ongoing.
The proteasome is a proteolytic complex involved in numerous cellular functions, including protein homeostasis, the cell cycle, apoptosis, inflammation and resistance to antineoplastic therapy . Disruption of the degradation of regulatory molecules (e.g., proapoptotic proteins) by proteasome inhibitors can induce cell growth arrest and apoptosis, and may increase sensitivity to chemotherapeutics . In addition to these effects, inhibition of the proteasome may also result in degradation of nuclear factor-κB, which is involved in counteracting downstream mediators of apoptotic signaling. Bortezomib (Velcade®, PS-341, Millennium) is a recently approved proteasome inhibitor for multiple myeloma and mantle cell lymphoma that has induced cell cycle arrest and apoptosis in glioma cell lines . However, a Phase I/II study of bortezomib in MG was prematurely terminated owing to limited activity , possibly due to limitations in activity as a single agent or an inability to penetrate into the tumor at therapeutically active concentrations.
Inhibiting members of the heat-shock protein (HSP) family is another strategy for modulating cell resistance to apoptosis. HSPs are involved in the conformational maturation, stability and function of a variety of key growth-stimulating, apoptosis-inhibiting and cell survival proteins. Inhibitors of HSP90, such as geldanamycin and 17-allylaminogeldanamycin (17-AAG), have potentiated the efficacy of conventional cytotoxic chemotherapeutic agents , as well as other signaling modulators , against gliomas. However, their clinical applicability for gliomas has been limited by their poor penetration through the BBB.
Given the biological diversity and heterogeneity of tumors, it is not surprising that single-agent molecularly targeted therapies have been unable to cure patients with GBM , and results of first-generation clinical trials with these agents have been disappointing. No major survival advantage has been observed in the population to date, although transient responses to some antiangiogenic agents, such as bevacizumab, have been noted. Despite promising in vitro and in vivo data, intrinsic or acquired resistance to these therapies has been a major therapeutic obstacle, and understanding the rationale for combinatorial strategies will therefore rely on a review of the major resistance mechanisms to targeted therapies.
The most notable mechanisms of in vitro and in vivo drug resistance are tumor heterogeneity, redundancy and parallel processing of intracellular signaling pathways, inactivating metabolism, loss of negative inhibition, mutations leading to constitutive activation, coactivation of receptor tyrosine kinase receptors (RTKs) and limited drug delivery. For instance, coactivation of families of RTKs appears to be common in most human cancers, including GBM . Multiple RTKs are coactivated through redundant inputs, which maintain downstream signaling and limit the efficacy of single-targeting treatments. Combinations of RTK inhibitors or single drugs with activities against multiple RTKs may decrease intracellular signaling, increase cytotoxicity and limit anchorage-independent growth in glioma cells . Signaling through these pathways may also be increased by constitutive activation of intracellular signaling molecules that are downstream of the targeted protein, or by constitutive signaling through a receptor even in the absence of ligands . Additionally, tumors may become resistant to targeted therapy over time through acquired secondary mutations within the targeting agent’s binding site, which has been noted in the treatment of non-small-cell lung cancer with gefitinib [169,170]. Drug combinations that target multiple binding sites may be less susceptible to resistance via this mechanism. Furthermore, targeted therapies are also susceptible to mechanisms of resistance that affect their specific targeted pathway, as well as compensatory activation of alternative signaling pathways.
Given these therapeutic challenges and the multiple mutations leading to gliomagenesis, tumor cells will need to be targeted with several agents simultaneously to ensure a cure or long-term survival. Combination therapies that target multiple signaling pathways or different constituents in the same pathway may over-come resistance mechanisms and widen the therapeutic window, ultimately enhancing the effect on tumor cells without increasing toxicity on normal cells. However, therapeutic combinations are limitless and a strategy is necessary to choose only the most effective and synergistic of combinations. These therapies include multi-targeting single-agent therapies or multiagent therapies, both of which are being tested in MG. There is a particularly strong rationale for targeting growth factor receptor pathways and tumor angiogenesis, owing to promising preclinical and clinical studies.
Multi-targeting kinase inhibition is a promising strategy for the treatment of glioma, as discussed earlier. Unlike pathway signaling in chronic myelogenous leukemia that depends on nonredundant single pathways or ‘oncogene addiction’ for maintaining malignancy , MG may have numerous constitutively activated growth and survival signaling pathways. Accordingly, targeting multiple signaling pathways with multi-targeted single-agent kinase inhibitors may prove more efficacious than highly specific agents. Table 1 summarizes several of these agents, their targets and the clinical models being used [171–173]. Of note, sunitinib dually inhibits VEGFR and PDGFR and is currently being investigated in glioma. Unfortunately, several studies in recurrent glioma have been stopped because of the lack of efficacy of sunitinib as a monotherapy [174,303]. NVP-BEZ235 is another dual inhibitor that has shown activity in various cancers, including breast cancer, pancreatic cancer, multiple myeloma and melanoma, as well as in preclinical studies in gliomas [121,175], although this agent is not being used in GBM clinical trials. AZD2171 is a pan-VEGFR tyrosine kinase inhibitor that has shown promise in the preclinical  and clinical settings [133,177], and may have a function in normalizing tumor vasculature and alleviating edema in patients with GBM or non-small-cell lung cancer.
There is a strong rationale for targeting growth factor receptors. EGFR is overexpressed or amplified in the majority of gliomas , and the EGFRvIII variant is a constitutively active, ligand-independent receptor that may have a distinct pattern of response or resistance to several RTK inhibitors, including erlotinib and gefitinib . One rationale for combination therapies is to target downstream intracellular effectors of the EGFR signaling pathway, including PI3K and mTOR. Combined and simultaneous inhibition of receptor and intracellular effectors may confer improved outcomes in patients with GBM. Additionally, activation of the EGFR pathway may be accompanied by activation of parallel inputs through PDGFR and VEGFR pathways. Coactivation of alternative RTKs may lead to activation of parallel, convergent and divergent intracellular signals that may over-come upstream blockade by a single targeting therapy. Therefore, simultaneous targeting of multiple activated receptors or intracellular signaling effectors may be necessary to block downstream signaling pathways.
Several clinical trials have combined EGFR and mTOR inhibitors and are summarized in Table 2. Preclinical studies have reported a chemosensitizing effect of the mTOR inhibitor rapamycin on the effect of EGFR inhibitors in PTEN- and PTEN-intact GBM , and the addition of an mTOR inhibitor to EGFR blockade may augment downregulation of Akt . However, a recent study suggests that EGFR signals to mTOR through PKC and independently of Akt, and inhibition of PKC may lead to decreased viability of glioma cells regardless of PTEN or EGFR status . A preliminary clinical trial with the combination of gefitinib and sirolimus suggested only a modest effect when combining these agents , although studies are still in progress. In another study that combined everolimus and gefitinib, 26% of patients had a PR and 11% had disease stabilization, although the PFS was only 2.6 months, which was no better than historical controls . Doherty et al. reported a PR of 19% and PFS6 of 25% in a pilot study combining gefitinib or erlotinib and sirolimus . Studies combining mTOR inhibitors with other targeted agents, including AEE788 (EGFR and VEGFR inhibitor), EKB569 (EGFR inhibitor) and sorafenib (VEGFR, PDGFR and Raf kinase inhibitor) are also in progress. In Phase I/II studies of erlotinib and temsirolimus in patients with recurrent GBM, the authors concluded that the combination had a higher than expected incidence of toxicities and had minimal activity in recurrent MG (Table 2) . The combination of sorafenib and temsirolimus was also examined  and, although moderately well tolerated, the investigators concluded that it did not exhibit sufficient activity in recurrent GBM to warrant further investigation.
Another strategy for improving the clinical efficacy of EGFR inhibitors is to suppress parallel signaling pathways, such as PDGFR, Raf or VEGFR. Several such trials are in progress and their results are eagerly anticipated (Table 2). In particular, a Phase II trial in progressive or recurrent MG is currently assessing the combined efficacy of erlotinib and sorafenib, which together inhibit EGFR, PDGFR, VEGFR and Raf . Results from this trial will provide insight into the effect of inhibiting parallel growth factor signaling pathways in glioma. In a recent Phase I/II study of sorafenib and erlotinib for patients with recurrent GBM, response rates were modest and the combination was not without toxicity . EGFR-targeted monoclonal antibodies have also been noted to enhance the effects of cisplatin , topotecan, gemcitabine  and taxol. Recent studies also suggest that small-molecule inhibitors of EGFR kinase activity, such as gefitinib [128,189], and of PDGFR kinase activity, such as imatinib [190,191], may achieve similar potentiation of conventional therapies.
Bevacizumab has shown clinical promise as an inhibitor of tumor angiogenesis, and favorable outcomes using the combination of bevacizumab with irinotecan have inspired further interest in the inhibition of angiogenesis/VEGFR pathway with combination therapies (Table 2) [134,139]. The combinations of these therapies have achieved more favorable outcomes compared with historical controls, although it remains uncertain whether patients receiving combination therapy have better response rates than patients receiving bevacizumab alone [134,139]. In a noncomparative Phase II trial of bevacizumab alone or in combination with irinotecan in recurrent MG, median PFS was 23 weeks, OS6 was 72% and PFS6 was 38% for the combination. However, in a Phase II trial comparing the effect of bevacizumab alone or in combination with irinotecan, there was a nominal benefit in PFS6 in patients with recurrent GBM who received combination therapy, although this difference did not reach statistical significance . These results provide encouraging evidence of efficacy with VEGFR inhibition, although selection of synergistic combinations will require further clinical trial developments. Clinical trials assessing the use of this agent in newly diagnosed MG and in combination with other agents are in progress.
Combination therapies targeting VEGFR and EGFR also hold promise for the treatment of MG. It is becoming increasingly evident that the EGFR pathway is involved in tumor angiogenesis . Data from recent preclinical studies have shown upregulation of VEGF and matrix metalloproteinases when EGFR is activated in various cancers [193–195], including human glioma, which may confer a subsequent resistance to EGFR inhibition [196,197]. Interestingly, inhibition of EGFR resulted in down-regulation of several angiogenic signaling molecules, including VEGFR, IL-8, bFGF and matrix metalloproteinase-9 [188,198–201]. Thus, combined inhibition of these two pathways may interfere with a molecular feedback loop responsible for acquired resistance to EGFR inhibitors, promote cell death through apoptosis and ablate tumor angiogenesis .
Studies have also suggested that combined inhibition of these receptors may be clinically efficacious in tumors where targeting either EGFR or VEGFR alone is ineffective [172,203]. Given the favorable effect of bevacizumab and irinotecan against glioma and the frequent mutation/amplification status of EGFR in MG, one study assessed the efficacy of combining cetuximab with irinotecan and bevacizumab . Results from this study and several ongoing antiangiogenesis combination therapy trials are eagerly awaited (Table 2). In a preliminary study, the combination of bevacizumab and erlotinib was reasonably well tolerated in recurrent MG, with radiographic responses reported in 12 out of 25 patients . However, it remains uncertain whether the combination offers any benefit compared with therapy with bevacizumab alone. Additional studies will provide further direction for the use of this combination therapy .
Furthermore, the combination of thalidomide and chemotherapy appears to be more effective in patients with recurrent gliomas than either approach alone . Patients with recurrent GBM who received thalidomide and carmustine had a response rate of 24%, which compared favorably with carmustine alone 143]. The combination of thalidomide and TMZ in patients with GBM was also more effective than thalidomide alone with respect to survival and response . Additionally, the combination of endostatin and semaxanib (SU5416) has been reported to achieve superior tumor growth inhibition in preclinical models compared with treatment with either agent alone .
Several preclinical and clinical studies are currently testing the efficacy of combining other molecularly targeted agents. Preclinical studies have reported potentiation of proteasome inhibitor-induced apoptosis by HDAC inhibitors in a panel of GBM cell lines , which has provided a rationale for clinical trials with these agents . Furthermore, a preclinical study found that the combination of irinotecan and a PKC inhibitor led to a decrease in proliferation and an increase in apoptosis in MG cells , which may provide a rationale for future clinical studies. As our understanding of the molecular underpinnings behind gliomagenesis deepens, we will develop novel combinations of molecularly targeted therapies that target these nodes of chemosensitivity.
Combination therapies of molecularly targeted agents with surgery and radiation therapy have also been developed in MG. In general, conventional therapies for MG involve tumor resection, irradiation and systemic chemotherapy [14,209]. Unfortunately, most patients develop recurrence or progression after radiation treatment, and tumor radioresistance makes re-irradiation treatment less effective and potentially more toxic . Thus, molecularly targeted therapies that can enhance radiation sensitivity may improve outcomes in these patients and may help reduce the antagonistic effects of conventional combination therapies. For instance, increased signaling through the EGFR pathway may confer glioma resistance to the combination of radiotherapy and chemotherapy. Thus, a rational strategy for overcoming this resistance focuses on the inhibition of the EGFR pathway  and marked radiosensitization has been achieved by EGFR-specific monoclonal antibodies [211,212] and dominant negative transfection . In recent Phase I/II studies of erlotinib and TMZ with radiation therapy in newly diagnosed GBM, patients treated with the combination did not have signs of benefit compared with TMZ controls in one study , although patients treated with the combination had better survival than historical controls in another study . Additional studies with this combination are warranted. Furthermore, in a Phase I/II trial of gefitinib with radiotherapy in newly diagnosed GBM, there was no improved outcome compared with historical controls (Table 2) .
Therapies targeting angiogenesis may also sensitize tumors to radiation therapy, and combinations of VEGFR pathway inhibitors with irradiation have consistently demonstrated improved tumor growth delay . In a recent preclinical study, the combination of antiangiogenic therapy with radiation therapy achieved better results in mice bearing human glioblastoma xenografts . Combination studies of vandetanib or bevacizumab with radiation therapy are currently ongoing in patients with MG [308,309], and future studies will continue to evaluate the use of targeted therapies with conventional multimodality treatments in MG.
Owing to the poor responsiveness of MGs to conventional therapies, there is a pressing need to implement new treatment approaches to improve the outcome of patients with these tumors. Studies have demonstrated that the proliferation and survival of MG cells is strongly influenced by several molecular pathways that stimulate tumor growth, promote angiogenesis and inhibit apoptosis. Therefore, these pathways have emerged as promising targets for therapy and have been effectively addressed by small-molecule inhibitors. However, MG arises from a culmination of multiple molecular genetic alterations that produces significant heterogeneity between and within these tumors. These mutations may confer an intrinsic or acquired resistance to specific agents, explaining why most single-targeted monotherapies have failed to demonstrate improvement in survival in unselected patients. Given these therapeutic challenges, it is unlikely that any single agent will have long-term efficacy in more than a subset of tumors. Accordingly, appropriate use of these agents will probably require combinations of molecularly targeted agents and conventional therapies or combinations of several molecularly targeted agents administered as a ‘cocktail’, incorporating a tumor-tailored combination regimen based on the unique molecular features of the individual tumor.
Identification of tumor genotypic and phenotypic features that predict response to specific agents holds future promise for selecting patients who are most likely to respond to specific molecularly targeted approaches. In fact, response to therapies may correlate better with genetic characteristics than with histopathology . Genome-wide characterization studies of MG have begun to identify potentially useful and heretofore unrecognized genetic alterations for the classification and targeted therapy of GBMs . The sensitivity of a tumor to a molecularly targeted therapy may be dependent on its specific molecular abnormalities within the tumor, and it is critical to identify markers of response to certain therapies through genomic characterization. For instance, GBM with loss of PTEN and tumors with increased expression of phospho-Akt may be resistant to EGFR inhibitors. However, such tumors may respond to mTOR inhibitors due to activation of the PI3K/Akt pathway. Ultimately, genotyping each patient’s tumor through genomescale characterization studies will increase our understanding of the molecular changes that drive malignant growth in a patient’s tumor and will guide the rational selection of combination molecular therapies for each patient.
In addition to cancer genetic characteristics, targeting various other aspects of tumor biology with combination therapies also holds promise for the future treatment of GBM. For instance, disruption and distortion of the tissue microenvironment is necessary for cancer progression, cancer growth, recruitment of nonmalignant cells, invasion and metastasis [219,220]. Thus, controlling cancer may be achieved indirectly by controlling blood vessels, immune and inflammatory cells, growth factors and the extracellular matrix, which all constitute the tumor micro-enviroment . Targeting these aspects of tumor biology in concert with molecularly targeted therapies may provide deeper insight into the interplay between genes and environment in the formation and progression of GBM.
The presence of the BBB has complicated the development and implementation of treatments for GBM. Many small-molecule inhibitors and currently available anticancer therapies for GBM penetrate poorly into the CNS and into tumors. Since most current therapies are delivered systemically, therapeutic concentrations within the brain may be been difficult to attain. For agents with poor brain-tumor penetrance following systemic administration, improved efficacy of these therapies may require novel drug delivery methods to reach the site of disease. Convection-enhanced delivery may allow a more localized drug delivery to a greater volume of tissue  and nanobiotechnology using nanoparticles holds promise for improved systemic therapeutic delivery without interfering with the normal function of the brain [222,223]. Improvement in treatment efficacy will rely on future research efforts focusing on these and other avenues of treatment delivery.
During the development and initial applications of molecularly targeted therapies, it was expected that because of their potential selectivity for cancer cells and specificity of targeting, these agents would be safer and less toxic than traditional chemotherapeutic agents. However, these expectations have been tempered by the observed side effects and toxicities, possibly due to targeting of key pathways. In addition to the common drug side effects (e.g., diarrhea, infusion reactions and others), these targeted agents cause several agent-specific side effects, including proteinuria, hypertension, acneiform rash, dry skin and hair depigmentation [224,225]. The cutaneous and GI tract side effects of gefitinib and erlotinib, fatigue and cardiac side effects from sunitinib, and the need to take antihypertensive medications after bevacizumab may reduce the quality of life and impact activities of daily living in patients. When targeted therapies are combined with conventional therapies, the side-effect profile may further increase its range of toxicities . As with other therapeutic agents, it will be important to balance the impact and possibility of side effects with potential benefits and disadvantages of treatments with targeted agents.
A final and essential consideration is the expense and cost–effectiveness of these molecularly targeted treatments. Although the discovery of these therapies has brought significant breakthroughs in neuro-oncology and related fields, targeted therapies are very expensive and the cost considerations of therapy are not inconsequential. For instance, the cost of initial systemic molecularly targeted therapy for advanced colorectal cancer has increased 340-fold in the past few years [226,227]. As more targeted agents become increasingly available, the issues of cost and cost–effectiveness relative to potential benefit will only continue to intensify. Thus, there is a continued need for evidence-based guidelines for the appropriate use of these agents and for a more comprehensive understanding of the financial impact of these therapies on both our society and our patients.
Improvements in clinical outcomes will depend on the synergistic effects of targeted therapies tailored for specific tumors based on their mutational profile. For many years, we have attempted to discover more effective drug combinations that reduce toxicity and improve chemosensitivity. To date, these combinations have been no more than trial and error, with the best method being rational combinations from known mechanisms of the individual drugs. However, the number of therapeutic combinations is limitless and we will need to develop robust and systematic strategies to rapidly identify promising combinations.
Genome-wide synthetic lethal screening is a promising approach for identifying novel therapeutic combinations. Synthetic lethality defines a genetic interaction where the combination of mutations in two or more genes leads to cell death , and identifying such combinations may confer a synergistic decrease in cancer cell survival. A high-throughput siRNA-based screening strategy has offered simultaneous and systematic genome-wide interrogation of the loss-of-function phenotypes associated with protein suppression without requiring a priori knowledge of gene functions or cellular pathways. In the preclinical setting, this screening strategy has lead to rapid identification of novel synergistic combinations in the context of poly(ADP-ribose) polymerase inhibition , paclitaxel sensitivity  and interactions with the Ras oncogene . Combining these preclinically derived therapeutic combinations with an understanding of the tumor’s mutational profile will facilitate the translation of these results into the clinical setting.
Future studies will also focus on restructuring clinical trials to stratify patients based on tumor genotyping. The integration of clinical and molecular information, which are now becoming available through gene arrays, proteomics and molecular imaging, will lead us to a more effective implementation of targeted treatments. Molecular imaging and mapping of each tumor’s genetic aberrations may ultimately determine the best rational combination of treatments and suggest ways to sensitize tumors otherwise resistant to conventional therapies. Future trials will incorporate tissue analysis to identify the mutations specific to the patient’s tumor, making it possible to determine which drug combinations the patient will respond to best. Correlating tumor genotype with response will provide insight into the efficacy of these molecularly targeted agents and next-generation clinical trials are already beginning to incorporate these design strategies. A recent study demonstrated that patients with GBM containing a methylated MGMT promoter benefited from TMZ, whereas those who did not have a methylated MGMT promoter did not experience such a benefit . Future clinical studies utilizing single- and multi-agent molecularly targeted therapies will incorporate patient stratification based on clinical and molecular information.
Financial & competing interests disclosure
This work was supported in part by National Institute of Health grant NSP0140923 and the Doris Duke Charitable Foundation. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.
No writing assistance was utilized in the production of this manuscript.
Nikhil G Thaker, Doris Duke Clinical Research Fellow, Departments of Neurosurgery, Pharmacology and Chemical Biology, University of Pittsburgh, Pittsburgh, PA 15260 and 6 Oakwood Place, Voorhees, NJ 08043, USA Tel.: +1 856 392 4727 Fax: +1 412 692 5921 ; Email: ude.jndmu@gnrekaht.
Ian F Pollack, Department of Neurosurgery, Children’s Hospital of Pittsburgh, University of Pittsburgh Brain Tumor Center, University of Pittsburgh School of Medicine, Biomedical Science Tower 3, 3501 Fifth Avenue, University of Pittsburgh, Pittsburgh, PA 15213, USA Tel.: +1 412 692 5881 Fax: +1 412 692 5921 ; Email: firstname.lastname@example.org.
Papers of special note have been highlighted as:
• of interest
•• of considerable interest