Open surgical removal of brain tumors has been a mainstay of glioma management for decades.[3-6] The goals of surgery are threefold: (1) to alleviate mass effect and compression of brain structures, (2) to restore normal cerebrospinal fluid pathways, and (3) to reduce the tumor burden for other therapies. Unfortunately, total excision of glioma is rarely possible. Glioma cells are highly invasive, and have been demonstrated 4 cm or more away from the primary tumor mass. Most of these cells are interdigitated with normal functioning brain parenchyma, and resection of these regions can result in unacceptable neurologic deficits. Further, while gliomas rarely metastasize outside the central nervous system, they can disseminate widely in both hemispheres of the brain. Tools to aid surgeons in differentiating normal tissue from glial cells at the periphery of tumors can improve the extent of microscopic removal. These methods include the use of tumor fluorescence,[8,9] infrared imaging,[ 10] and diffusion-weighted tensor imaging of white matter pathways.[ 11] The utility of intraoperative magnetic resonance imaging in this setting has not been validated. Gross total removal of all enhancing components of glioma has never been clearly correlated with a higher cure rate. However, cytoreductive surgery to remove all of the enhancing volume of high-grade glioma has been correlated with improved length of survival and improved quality of life compared with biopsy alone.[3,6,12,13] Despite its limitations, surgical resection remains the most effective single therapy for gliomas, especially when tumors recur or progress. Radiation Therapy
Gliomas are not particularly radiosensitive tumors. Radiation doses of approximately 60 Gy and greater are effective in retarding glioma progression but have not been demonstrated to provide long-term control. To some degree, increasing doses of radiation have increasing efficacy on higher-grade tumors, although most dose-escalation studies have failed to demonstrate improved survival. Unfortunately, normal brain parenchyma is also sensitive to radiation effects. The tolerance against brain injury decreases quite rapidly above 70 Gy, effectively limiting total radiation doses to the 60-70 Gy range. The use of radiosensitizers has not proven to significantly improve the effects of radiation or long-term outcome. Chemotherapy
Systemic chemotherapy has proven to be quite disappointing in the treatment of gliomas. This is partially attributable to the poor distribution of drug in the brain due to the bloodbrain barrier. Many agents, including carmustine(Drug information on carmustine) (BCNU [BiCNU]), lomustine (CCNU [CeeNU]), procarbazine(Drug information on procarbazine) (Matulane), and temozolomide(Drug information on temozolomide) (Temodar) have demonstrated response against high-grade gliomas. However, these agents tend to produce fairly limited or partial responses in both upfront and recurrent settings. Most recently, the combination of temozolomide with radiation therapy in patients with newly diagnosed glioblastoma multiforme, followed by a course of temozolomide, has been found to increase median survival by 2 months compared to radiation alone, with approximately 35% of patients surviving beyond 18 months.[16,17] Locoregional Methods The limitations of these traditional locoregional treatment methods, coupled with the disappointing results of systemic chemotherapy have led to the development of alternative locoregional treatment approaches. These methods are outlined below. Locoregional Radiotherapy
Improved delivery of high doses of radiation to the main tumor volume and immediate surrounding tissue where infiltrating cells are most numerous has been widely tested. Brachytherapy with implantable I-125 seeds was frequently employed in the 1980s and 1990s, with good results in younger patients but a high rate of symptomatic brain necrosis.[18,19] In more recent years, brachytherapy has been largely supplanted by stereotactic radiosurgery, with very similar results.[20,21] These methods are generally limited to small tumor volumes of 3 cm or less. The addition of hyperthermia has not proven beneficial in this setting. Most recently, an I-125-filled balloon (Gliasite) has been approved by the US Food and Drug Administration (FDA) for delivery of brachytherapy in resected tumor cavities. Early results with this method appear similar to those seen with more traditional brachytherapy seeds. The high rate of brain necrosis and general failure of these methods to affect the overall survival of patients with glioma has limited enthusiasm for these approaches. Targeted Radioimmunotherapy and Radiopeptide Therapy
Nonviral targeted cancer therapies principally depend on receptor-mediated selective binding of drugs to tumor cells. The success of this approach requires high specificity or selectivity of binding to tumor cells. This can be accomplished by either overexpression of receptors by tumor cells or preferential expression of receptors not found on normal brain tissues. Because the blood-brain barrier effectively limits the size of molecules entering the brain, direct delivery of these targeted therapies via a locoregional delivery system is imperative for most agents. One approach for receptor-based targeted therapy of glioma is the utilization of tumor-specific antibodies to selectively bind gliomas and deliver various cytotoxic agents to the tumor. Examples of this approach in glioma therapy include targeting the epidermal growth factor receptor (EGFR), fibronectin, and the extracellular matrix molecule tenascin- 81C6 with I-131-radiolabeled ligands. A phase II trial of I-131- antitenacin antibodies injected into a postresection surgical cavity (doses of up to 100 mCi) followed by radiotherapy and chemotherapy was completed in 33 patients with glioblastoma multiforme. A median survival of 79.4 weeks was reported. However, imaging studies demonstrate that the antibodies were too large to penetrate beyond a few millimeters into the surrounding brain parenchyma, essentially neutralizing any potential targeting advantage. Furthermore, tenascin is widely expressed throughout the nervous system, and epidermal growth factor receptor, while upregulated in some gliomas, is also widely expressed by normal brain cells, limiting the specificity of these approaches. Both of these compounds are now in phase II clinical trials. Targeted therapy of glioma with smaller molecules has also gained attention in recent years. Chlorotoxin, a 36-amino acid peptide identified as a neurotoxin in a scorpion venom, has been found to bind to a variety of malignancies with minimal to no binding on normal cells.[28,29] Although the exact binding site for chlorotoxin has not been clearly identified, a synthetic analog of this compound has been conjugated to I-131 and has undergone phase I testing for recurrent glioma. The compound, I-131- TM-601, was injected into the resection cavity of 18 patients with recurrent high-grade gliomas via an Ommaya reservoir. Exquisite and long-term binding of the radioligand to the tumor was observed for up to 8 days postinjection. Median survival in the trial was 5.7 months, with three long-term survivors and two complete responses. The main advantages of this approach appear to be that (1) the small size of the protein permits larger scale diffusion in the brain and the potential to cross the blood-brain barrier, (2) no toxicities have been observed, and (3) it offers ease of delivery. A phase II trial of this agent is currently under way. Utilization of the targeting strategy conjugated to biologic toxins is also under consideration. Locoregional Chemotherapy
Methods of delivering higher concentrations of chemotherapeutic agents directly to the brain tumor have been of great interest for many years, and the strategies employed parallel trends in methods of delivering a variety of agents such as targeted toxins and viral gene therapy. These delivery methods will be reviewed in conjunction with this discussion of locoregional chemotherapy but apply just as well to later topics. The most obvious method for locoregional delivery of drugs to a brain tumor is direct injection of drug into the tumor or tumor resection cavity via an implanted Ommaya reservoir or syringe at the time of surgical resection or biopsy. This approach is simple, can be performed by almost any physician, and has been tested with several drugs, including BCNU, methotrexate(Drug information on methotrexate), cisplatin(Drug information on cisplatin), and cyclophosphamide(Drug information on cyclophosphamide).[ 32] While toxicities have been tolerable, tumor responses have been very disappointing, resulting in an abandonment of this strategy. Lack of diffusion of these compounds into the surrounding parenchyma is a major limiting factor, and methods to increase the tissue penetration of drugs are being tested. A current trial of intratumoral injection of BCNU dissolved in ethanol (DTI-105) to increase tissue penetration is the most prominent example of this modified approach. Intracarotid injection of chemotherapy supplemented by methods to transiently open the blood-brain barrier,[ 34] has been studied for many years. Again, the main goal of this approach is to increase concentrations of drug in the brain utilizing a "firstpass" effect. Cisplatin and etopiside have been most frequently employed,[ 35,36] although many agents have been tested.[37,38] Success with this strategy has been variable, and most studies have failed to demonstrate a clear benefit in survival or time to progression. There have been no randomized prospective trials utilizing this method. Further, it is highly invasive, requiring a cerebral angiogram, and complications such as blindness due to infusion of the ophthalmic artery have been reported.[35,39] Improvements in angiographic technique have reduced complications, but these methods have failed to gain acceptance as a standard method of treatment. Delivery of chemotherapy via timereleased polymer wafers is currently the only FDA-approved form of locoregional chemotherapy. Polifeprosan 20 with carmustine (Gliadel) is a synthetic biodegradable polymer wafer containing a 3.8% concentration of BCNU. Typically, six to eight wafers are implanted directly along the walls of a resected tumor at the time of surgery. Several animal and human studies indicate the effect mimics a 4- to 6-week infusion of BCNU. A phase III trial comparing Gliadel to placebo wafers demonstrated a slight improvement in median survival of about 8 weeks. A main advantage of this form of administration is that it avoids many of the untoward side effects of systemic BCNU such as thrombocytopenia. The cost of the wafers roughly approximates the cost of six courses of systemic BCNU. Efforts to increase the concentration of BCNU in the wafers have led to increased adverse events, although concentrations of up to 20% have been achieved and felt to be tolerable. No agent other than BCNU is commercially available or FDA-approved in this formulation. Various shapes of the polymer are also being tested to ease insertion.[41,42] Several other polymer-based methods of cancer treatment are also being tested.[ 43] In addition, several phase I trials investigated the effects of combining intracavitary chemotherapy with systemic chemotherapy. FDA-approved in this formulation. Various shapes of the polymer are also being tested to ease insertion.[41,42] Several other polymer-based methods of cancer treatment are also being tested.[ 43] In addition, several phase I trials investigated the effects of combining intracavitary chemotherapy with systemic chemotherapy. Most recently, significant interest has developed around the use of convection- enhanced delivery (CED) of chemotherapies to brain tumors. CED refers to the process of applying uniform positive pressure to overcome the natural resistance of the surrounding tissues and essentially "push" the drug into the parenchyma. Using very slow infusion rates (0.5-4 μL/min) over long time periods (3-5 days), one can deliver relatively uniform concentrations of drug up to 4 cm away from the tip of the infusion catheter, with well-tolerated side effects and a reasonable risk profile. CED is a rapidly advancing field, with improvements in catheter placement, drug delivery techniques, and drug preparation all having an impact on the potential efficacy of this method. One phase I/II study of CEDdelivered paclitaxel(Drug information on paclitaxel) has been reported.[ 13] This study of 15 patients with recurrent high-grade glioma demonstrated a 73% response rate (five complete responses, six partial responses) but also a high rate of complications. Other drugs such as docetaxel(Drug information on docetaxel) (Taxotere) have been tested, with more preliminary results available. Animal models have suggested that this technique may also be very promising for brain stem gliomas. Several other studies are either under way or being planned. CED seems particularly promising for unresectable tumors and as an adjunctive therapy before or after radiotherapy. Biologic Therapies Viral Gene Therapy
The promise of viral-based gene therapy for glioma has prompted extensive investigation. The basic concept employed is that a virus acts as a genetic "carrier" to deliver a transgene to tumor cells. The transgene either integrates into the host DNA or uses the host cell replication mechanism to produce a gene product, which then effects tumor-killing. The exact mechanism of tumor cell death depends on the transgene product. The viruses are usually packaged into murine cells as a delivery vehicle and injected directly into the tumor cavity. The most widely utilized trangene construct is the herpes simplex virus- thymidine kinase (HSV-TK)/ganciclovir system.[48-50] Transduction of tumor cells with HSV-TK makes these cells over 5,000-fold more sensitive to the antiviral drug ganciclovir(Drug information on ganciclovir). When cells infected with HSV-TK are exposed to orally administered ganciclovir, they are terminally phosphorylated and die. In addition, a prominent bystander effect occurs in cells not infected with HSV-TK-presumably due to cell-cell interactions- that also kills noninfected cells. Alternative prodrug strategies such as the cytosine deaminase/fluorocytosine (5-FC) system have been tested,[ 51] although not clinically. In this system, viral particles containing the cytosine deaminase transgene infect tumor cells, leading to the production of the enzyme cytosine deaminase. When cells containing the enzyme are exposed to 5-FC, they convert the 5-FC to fluorouracil(Drug information on fluorouracil) (5-FU), a potent antitumor drug. Other strategies include production of cytokines such as interleukin (IL)-2 for immune modification.
- Retroviruses-The choice of a viral vector is of critical importance for gene therapy. Retroviruses have been most commonly used for brain tumor therapy. Retroviruses are RNA viruses that can integrate viral DNA into the host cell genome. Because most retroviruses preferentially infect dividing cells, they are ideally suited for use in tumors, especially brain tumors, where the majority of neurons are postmitotic and normal glia have a low mitotic index (< 0.5%). Adenoviruses-particularly replication- selective "oncolytic" viruses- have also been utilized for glioma therapy. The main advantage of this approach is that the virus can reproduce and spread to other cells, thereby increasing cellular killing. A replication-specific ("oncolytic") variant of the DNA virus HSV-1 has also been developed and tested in glioma. The main advantages of the HSV-1 vector are its large transgene capacity, high titer level, sensitivity to ganciclovir, and lack of insertional mutagenesis in the host genome. A detailed discussion of viral and gene therapy strategies that have been tested for brain tumor therapy is beyond the scope of this article, but excellent review articles are available.
- Clinical Trials-Several clinical trials have been carried out utilizing viral gene therapy in recurrent highgrade gliomas. In the majority of these trials, murine virus-producing cells (VPCs) carrying a retrovirus producing the HSV-TK transgene were used. A phase I/II study in France with 12 patients that received VPCs injected into the walls of the tumor cavity followed by 14 days of ganciclovir reported no treatment-related adverse events, a median survival of 206 days, and one long-term survivor. A multicenter international phase II trial of 48 patients demonstrated a median survival of 8.6 months, with a 12-month survival rate of 27%. A phase III randomized trial of VPCs containing the HSV-TK was performed in 248 patients with newly diagnosed glioblastoma. A total of 128 patients received the HSV-TK VPCs injected into the tumor cavity wall, followed by 14 to 27 days of ganciclovir, and then radiation therapy, while the control group received surgery and radiation alone. Although the trial demonstrated the method to be feasible and safe, there was no change in median survival or progression-free survival in the group receiving the gene therapy, indicating a lack of efficacy. Several phase I or I/II trials have been carried out with an adenoviral HSV-TK. Results have been similar to those seen in the retroviral studies, although one study suggested that the adenoviral approach may have a higher rate of infectivity. Finally, two phase I trials of oncolytic mutant HSV-1 have been performed, with a total of 30 patients.[56,57] In all, these studies suggest that the viral transgene approach is appealing but currently very limited due to inability to infect a sufficient number of cells to have an impact on disease control. Of note, all of these studies have been performed utilizing direct injection of VPCs into the tumor cavity. Increased distribution of VPCs utilizing CED may also improve results in future studies.
Mechanisms of cell-specific killing other than radiation or chemotherapy have been developed and tested. Utilization of toxins produced by bacteria is of current interest. These toxins are usually taken up by cells via active transport, resulting in inhibition of protein synthesis and subsequent cell death. Tumor-selective targeting is accomplished by joining these toxins with cell-specific antibodies or ligands. One example of this approach is a chimeric fusion molecule made from modified diphtheria toxin (DT) conjugated to human transferring receptor (Tf-CRM107). Transferrin is ubiquitously expressed on many cells but is highly upregulated in rapidly dividing cells, especially glioma. Conjugation of DT to the transferrin receptor mediates a dramatic increase in sensitivity to cellular death. Tf- CRM107 has been tested as an antiglioma agent in several phase I and phase II trials.[58,59] In these trials, the agent was given via direct convection due to its large size (128 kD) and poor diffusability. In a phase I trial involving 32 patients, Tf- CRM107 was given intratumorally via CED for a period of 3 to 45 days, at rates of 0.5 to 4 μL/min. The drug produced 2 complete responses and 8 partial responses, with 19 nonresponders and many adverse events, mostly neurologic. A phase II trial of this same agent in 44 patients produced an 11% complete response rate and 16% partial response rate. Median survival was 37 weeks. The most severe drug-related adverse event was brain edema, which occurred in six patients. A phase III trial of this agent compared with the best standard of care is currently under way. Another approach targets IL-13 receptors shown to be preferentially expressed on gliomas.[7-9] Specifically, a mutant form of IL-13, termed IL-13α, has very high specificity for overexpression in high-grade glioma. Recombinant forms of IL-13 receptor antibody coupled to Pseudomonas exotoxin (PE) and delivered by CED via a catheter, have completed earlystage clinical development.[9,10] Several phase I trials demonstrated very promising tumor responses when the CED catheters were placed at least 2 cm from any sulcal or ependymal surface, but minimal responses otherwise. These studies emphasize the importance of accurate delivery of the compound to the tumor parenchyma if targeted therapy of large-molecule toxins is to be effective. A phase III trial comparing IL-13-PE to BCNU wafers after resection of recurrent glioma is currently under way. Prominent enhancement of the tumor periphery that may mimic progressive tumor is commonly seen following administration of toxins but will often resolve within a few months if corticosteroid therapy is initiated. The delivery of targeted toxins to the brain via CED is a very promising and rapidly advancing treatment option for locoregional therapy.