Advances in Treatment of Pediatric Brain Tumors

Histopathology 

Assessment of the histologic characteristics of a tumor by light and sometimes electron microscopy, is necessary for their diagnosis/classification, and aids in grading the degree of anaplasia for those tumors in which grade is useful for prognosis. The number of tumor mitoses in a microscopic field, the traditional measure of cellular proliferation, also helps assign tumor grade. However, a newer approach to evaluating cellular proliferation employs immunostaining of Ki-67, a nuclear antigen expressed in certain phases of the cell cycle, using the equivalent monoclonal antibody MIB-1.8 The MIB-1 proliferation index correlates with the biologic behavior and prognosis of low-grade and malignant gliomas in children,9, 10 and in medulloblastomas, where “hot spots” of high proliferation rate were predictive of poor outcome.11

Genetic tumor markers 

Recently, the detection of molecular genetic traits of pediatric brain tumors is providing new information to classify tumors and predict prognosis. This should translate to more accurate risk-stratification than traditional tumor grading and staging, and ultimately, to better treatment outcomes. Some of the genetic alterations of pediatric brain tumors differ from those of adults. For example, p53 mutations were found in 40% of the malignant gliomas of children older than 3 years, a much higher rate than in the tumors of older adults, but comparable to that in young adults.12 Such tumors are known as secondary high-grade gliomas, because they contain p53 mutations, which are present in low grade gliomas, subsequently progress to high-grade gliomas accompanied by mutations of other tumor suppressor genes such as the retinoblastoma (Rb) gene or phosphatase and tensin homology (PTEN) gene.13 Overexpression of p53 in malignant gliomas in children, found in one third of tumors examined, and not always in association with a p53 mutation, was strongly associated with a poor prognosis, independent of tumor histology and clinical prognostic features.14 Adult high-grade gliomas referred to as de novo (primary) malignant gliomas do not show p53 mutations or overexpression, but rather tend to show amplification of the epidermal growth factor receptor (EGFR) gene or inhibitor of cyclin-dependent kinase 4 tumor suppressor gene as probable mechanisms of tumorigenesis. However, in high grade gliomas in children, EGFR amplification is rarely seen, although overexpression of EGFR is very common.15 Because of these differences in the genetic alterations between pediatric and adult pediatric brain tumors, age-specific analysis will be essential in future studies.

For medulloblastomas, genetic markers which have correlated with prognosis in small studies include Trk C (neurotrophin 3 receptor) expression, of which a high level correlated with good outcome16 and C-myc oncogene amplification, which had very high predictive power of a poor prognosis, though it was only present in 7% of the tumors tested.17 Other genetic alterations found in medulloblastomas, which may have prognostic predictive value are those of the N-Myc and ERRB2 oncogenes,18 loss of caspase-8 expression,19 and mutations of several signal transduction pathways including the PTCH1/”Sonic Hedgehog”, Wingless” (WNT/WG)/beta catenin, and PDGF-a/RAS/MAP tyrosine kinase pathways.20, 21 At this time, the best independent predictor of medulloblastoma outcome may be the overall pattern of gene expression. DNA microarray gene expression analyzed in tumor tissue from a large cohort of medulloblastoma patients showed that clinical outcome is predicted by the profile of gene expression, and is a better predictor than the expression or amplification of any single gene or any single gene mutation.22 The results of the gene expression patterns and other genetic alterations are now being validated in the current Childrens Oncology Group (COG) protocols for medulloblastoma, and may complement clinical and histologic criteria for risk-stratification in future studies.

Mutation of the gene INI-1 on chromosome 22, has recently been detected in a high percentage of patients with the intracranial tumor, the atypical teratoid/rhabdoid tumor (AT/RT).23 These tumors, which predominate in young children, are frequently mistaken for medulloblastoma or primitive neuroectodermal tumors (PNET) based on histological appearance, but are much more aggressive with a worse prognosis and require more intensive treatment. Therefore, the malignant tumors of children less than 3 years of age should probably be tested for this mutation.

Prodrug activation 

Prodrug activation is also termed “suicide therapy,” in which a vector carrying a gene coding for a protein that sensitizes cells to a specific drug is delivered to a tumor. The largest experience with this type of therapy uses the herpes simplex virus thymidine kinease-gancyclovir regime to treat malignant gliomas. In this model, a modified herpes virus vector expressing the enzyme, thymidine kinase, after stereotactic intratumoral injection, is preferentially taken up and encorporated by rapidly dividing tumor cells.42 The patient is given gangcyclovir systemically, which is then converted by the thymidine kinase to a toxic metabolite, killing the tumor cells. Unfortunately, this regimen, which has been tried in children with high-grade gliomas, has so far met with limited clinical success, mainly because of poor transfection efficiency of tumor cells.43 There are ongoing studies with several other prodrug activator proteins and their corresponding prodrugs that may provide more powerful combinations for gene therapy in the future.44

Tumor suppressor gene therapy 

Tumor suppressor genes code for proteins that inhibit cell growth. The loss of tumor suppressor function by mutation or deletion of a tumor suppressor gene has been implicated as the mechanism of oncogenesis of glioma and medulloblastoma brain tumors.45 Tumor suppressor gene therapy replaces the missing or mutated tumor suppressor protein, theoretically restoring the normal pre-tumor cell phenotype. Preclinical animal studies applying this therapy for the p53 and PTEN tumor suppressor genes encouraging, but tumor heterogeneity in human tumors, with cells in the same region expressing different genes, is an obstacle to using this paradigm, clinically.46

Antisense gene therapy 

Antisense gene therapy prevents the expression of proteins that facilitate tumor progression. Knowledge of the oncogene DNA sequence responsible for a protein important for tumorigenesis, allows design of a vector with the “antisense” gene that will produce a single-stranded RNA molecule with a sequence complementary to the mRNA produced by the abnormal tumor target gene. When the antisense gene is taken up and expressed in the tumor, the complementary RNA produced then binds to mRNA of the tumor target gene, hindering the ribosome apparatus, preventing mRNA translation and synthesis of the pathogenetic protein. Using this method, several genes have been inhibited in animal models, such as urokinase plasminogen activator receptor and cathepsin B, resulting in some tumor regression.47 This methodology should have several advantages over other gene therapy methods, including not being hindered by low transfection rates and having the possibility of inhibiting multiple pathogenetic genes within the same vector. A disadvantage is the relative instability of the single stranded antisense RNA.

Medulloblastoma/PNET therapy 

Surgical resection of the tumor is first line of therapy for medulloblastoma/PNET. Surgery plays an important immediate role in reducing tumor burden, and often in opening the ventricular system, re-establishing normal CSF flow and ultimately obviating shunt placement. Moreover, among medulloblastoma and supratentorial PNET patients with no metastatic disease, there is evidence suggesting that extent of residual tumor after surgery correlates with outcome.51, 52 However, surgical resection, alone, cannot cure medulloblastoma, and radiation therapy to only the primary tumor site, or at a dose of <5000 cGy to the primary site, is associated with poor outcomes. Thus, for some time, standard therapy for medulloblastoma after surgical resection, had been craniospinal irradiation (35-36 Gy) and a boost to the posterior fossa (54-56 Gy), with 5-year progression-free survival (PFS) of about 60%.18 Although this approach had a major impact on survival, it was far from satisfactory, considering the significant morbidity of craniospinal RT, death from progressive disease in 30-40 % of average-risk patients, and less than 30% survival of those with high-risk disease. Supratentorial PNETs may have an even poorer prognosis after surgery and RT, with a 5-year PFS in one series, of only 12%.50

High-risk medulloblastoma/PNET therapy 

All these factors were an impetus for medulloblastoma/PNET therapy trials combining RT and chemotherapy. Several trials demonstrated efficacy of adjuvant chemotherapy for high-risk metastatic or residual disease.53, 54, 55 Especially promising results were reported for high-risk medulloblastoma patients using an adjuvant regimen of cisplatin, CCNU and vincristine after standard neuroaxis irradiation, with a 5-year PFS of 85%.56 Patients who were high-risk due to residual disease did better than those with metastatic disease, with 90% 5-year survival vs. 67%, respectively. Another approach to high-risk medulloblastoma/PNET is the administration of high-dose chemotherapy after craniospinal irradiation. A small international consortium treated 19 high-risk medulloblastoma/PNET patients after RT, with 4 tandem cycles of high-dose chemotherapy with stem cell support, and reported 84% 2-year PFS from the start of radiation therapy.57

High-risk supratentorial PNETs, generally, have not done as well as high-risk medulloblastomas. Patients with these tumors treated on the cisplatin/CCNU/vincristine regimen, with craniospinal RT up-front, fared more poorly, with 37% 5-year PFS compared with 67% for high-risk medulloblastoma.58 Some studies have demonstrated a benefit from chemotherapy for supratentorial PNETs. Dirks et al showed a trend to longer survival in those supratentorial PNET patients who received chemotherapy vs. those who did not.50 Infants with non-pineal PNETs treated on the “Baby POG” studies had very good 5-year PFS of 55%, but with no benefit for pineal PNET.25 Recently completed CCG trials for high risk PNET/medulloblastoma studied the use of further dose-intensification of chemotherapy in 3 tandem cycles, with stem cell support after RT, or of chemoradiotherapy using carboplatin as a radiosensitizer for high-risk medulloblastoma/PNET. The results of these studies are pending.

Average-risk medulloblastoma therapy 

The improved outcome of high-risk medulloblastoma patients gave momentum to efforts aimed at minimizing treatment-related neurotoxicity in children with average-risk medulloblastoma by reducing craniospinal radiation dose and supplementing the treatment regimen with chemotherapy. Previous attempts at neuroaxis radiation dose-reduction without other adjuvant therapy, had failed when the dose was reduced to 18 Gy, or even to 24 Gy.59, 60 Several subsequent studies showed outcomes of at least 70% PFS with the addition of chemotherapy to craniospinal axis dose of 24 Gy.61, 62 From a pilot study using the regimen of cisplatin/CCNU/vincristine after 24 Gy RT dose to the neuroaxis, children with average-risk medulloblastoma had a 3-year PFS of 85%.56 In the most recent COG study, after treatment with 24 Gy craniospinal RT and concurrent weekly vincristine, children with average-risk disease were randomized to two different adjuvant chemotherapy regimens (cisplatin/CCNU/vincristine vs. cisplatin/cyclophosphamide/vincristine), each of which resulted in 4-year PFS rates of greater than 80%.63

Infant medulloblastoma therapy 

Infants and very young children with medulloblastoma have poorer survival than older children when treated with standard RT, and even more significantly, they sustain much greater treatment-related neurotoxicity. Therefore, in the mid-1980s, Duffner et al.,25 through the Pediatric Oncology Group, developed and implemented a novel approach in which prolonged postoperative multi-agent chemotherapy was given in an attempt to delay radiation therapy for infants less than 3 years of age with medulloblastoma (other infant malignant brain tumors were also included). This approach proved to be successful in delaying RT for as long as 2 years in those youngest at diagnosis, and even in reducing the dosage of neuroaxis irradiation to 24 Gy in children with no evidence of disease after chemotherapy. Although the overall survival rate of 40% at 5 years was poorer than for older children with average-risk medulloblastoma who receive RT upfront, among infants who had a surgical gross total resection (GTR), 5-year survival was 60%, and in those with no metastases at diagnosis and a GTR, it was 69%, comparable to that average-risk patients treated initially with RT. Children’s Cancer Group conducted an infant brain tumor study that similarly delayed RT in infants < 3 years, using prolonged “8-in-1” chemotherapy, but with a 3-year PFS of only 22%, (likely because this chemotherapy was less intensive), although most of these long-term survivors, in fact, never received any RT.64

A different approach with the goal of deferring RT until relapse was pursued in several small trials using various different chemotherapy regimens; outcomes ranged from 5-year survival of 67%, to 2-year survival of 43%.65, 66, 67 The results of the French Oncology Group (SFOP) study of children < 5 years old treated for 16 months with multiagent chemotherapy and deferring RT until relapse, emphasized the importance of radical surgery; the 5-year PFS of the 34 children with no metastatic disease (M0 stage) and GTR was 41% vs 0% for those with a subtotal resection.68 Taken together, these studies suggest that infant medulloblastomas are responsive to chemotherapy, and that a gross total resection and M0 staging are significant prognostic factors. Such infants could reasonably be treated with relatively-prolonged chemotherapy, withholding RT until relapse; those with metastatic or residual tumor require a different approach.

A recently conducted trial, CCG 99703, tested an hypothesis that intensifying the chemotherapy while deferring RT, would improve outcome for infant medulloblastoma patients < 3 years of age. This study employed three cycles of chemotherapy similar to the initial “Baby POG” regimen, followed by three cycles of dose-intensified (but subablative) chemotherapy, with carboplatin and dose-escalation of thiotepa, supported with autologous stem cells collected after the initial cycles of chemotherapy; no RT was prescribed or specifically recommended. The study closed to accrual in late 2004 so that outcome results are pending.

Also, because of concern for neurotoxicity of RT in young children, the use of high-dose ablative chemotherapy with autologous bone marrow transplant (ABMT) or stem cell rescue for children with recurrent or newly diagnosed tumor is being explored. In view of the chemo-sensitivity of medulloblastomas there have been preliminary studies in which a small number of newly-diagnosed infants were successfully treated in consolidation with high-dose chemotherapy followed by bone marrow or stem cell support; this approach may have a larger role in treating young children, as it may for patients who relapse after standard therapy.69, 70

Because of the high risk for development of neurocognitive deficits and the moderate success using chemotherapy, there has been little enthusiasm for administering RT, especially in a craniospinal volume, to infants and children < 3 years of age. However, the possibility of using RT for local tumor control is being re-considered for infants, in view of the availability of the newer focused or 3-D conformal radiation. This approach is currently being used for infants with M0 medulloblastoma in COG study 9934; infants receive 3-D conformal RT after multiagent induction chemotherapy (and a second surgery for residual after chemotherapy), followed by 8 additional months of maintenance chemotherapy. Moreover, this conformal approach is being applied to primary site radiation therapy for average-risk medulloblastoma in older children, also aimed at reducing the neurotoxicity of brain irradiation. The methodology facilitates the accurate delivery of the tumor bed-radiation boost, with significant reduction in the radiation scatter to normal brain structures including the cochlea and auditory nerve in the posterior fossa, as well as to the diencephalon and other portions of supratentorial brain. Two single institution pilots have already demonstrated safety of this technique, with no increased posterior fossa recurrences among patients.71, 72 These findings are being validated in the current COG average-risk medulloblastoma study, which randomizes patients to conformal tumor bed with margins volume vs standard posterior fossa volume.

Medulloblastoma/PNET investigational therapy 

Better therapy for medulloblastomas undoubtedly will have its basis in clarification of tumor molecular biology, and could evolve in at least two ways. Improved understanding of the molecular signature of individual tumors will help in determining prognosis and more accurate tumor risk-stratification, permitting children at lower risk for recurrence to safely receive less toxic therapy, and reserving more intensive treatment for those at higher-risk. Knowledge of the molecular defects critical in tumorigenesis could also provide the means to use them as targets for novel therapeutic approaches.

A number of studies have identified several possible molecular traits that could serve as prognostic factors, as well as potential targets for therapy of medulloblastoma. Among these are the amplification or overexpression of several oncogenes, including ERBB2, C-Myc, and N-Myc,18 loss of caspase-8 expression,19 and mutations in several other signal transduction pathways including the PTCH1/“Sonic Hedgehog” pathway, “Wingless” (WNT/WG)/beta catenin pathway,37 and platelet-derived growth factor-alpha (PDGF-a) and RAS/MAP tyrosine kinase pathway.21 The best outcome predictor, to date, appears to be tumor gene expression profiling. A pilot study of medulloblastoma DNA microarray gene expression was able to demonstrate that clinical outcomes of children with medulloblastoma was highly predictable on the basis of the gene expression profile of their tumors at diagnosis, and independent of clinical or other criteria.22 The prognostic accuracy of any of these molecular markers in medulloblastoma is not yet firmly established, as they have all been evaluated in somewhat underpowered studies or have included heterogenous patient groups. However, validation of the most promising of these markers is ongoing in current COG and other international consortia medulloblastoma studies.

Understanding mechanisms of tumorigenesis for future molecular classification and prognosis is also the first step in the development of molecular-targeted therapies. Specific small molecule tyrosine kinase inhibitors could prove effective against targets in some medulloblastoma (and other brain tumors). These include imatinib mesylate (Gleevec), a PDGFa/RAS/MAP tyrosine kinase inhibitor,21 Erlotinib (Tarceva), which inhibits the oncogene, ERBB2 tyrosine kinase, and Iressa (gefitinib), which inhibits the EGFR tyrosine kinase.13 Molecular-targeted therapies still in pre-clinical evaluation include cyclopamine, a plant-derived teratogen that suppresses the membrane protein, SMO, which is activated and stimulates cell proliferation when there are mutations in the PTCH/Sonic Hedgehog pathway of some medulloblastomas;20 cyclopamine has side effects that limits its clinical application, but other agents which inhibit this pathway are also under development.

The retinoid, cis-retinoic acid, is another therapeutic agent soon to be evaluated in a randomized fashion in the upcoming COG protocol for high-risk medulloblastoma/PNET tumors. Retinoids mediate apotosis in medulloblastoma cells in vitro, and suppress tumor growth in xenograft models.73

Low-Grade Gliomas 

Pilocytic astrocytomas are the second most-common tumor in the posterior fossa after medulloblastoma (12-15% of pediatric brain tumors), and when treated with surgical resection alone, have the best prognosis of any pediatric intracranial neoplasm. Cerebral low-grade gliomas can be pilocytic astrocytomas, but are more often non-pilocytic gliomas of various types; they have greater propensity to infiltrate surrounding brain, which makes performing a complete resection more difficult. Nevertheless, among patients who undergo total resection, 10-year survival rates exceed 80% for hemispheric cerebral tumors,74 and 90% for cerebellar tumors.3, 75 For tumors that progress following surgery, radiation therapy is beneficial,76 although there is less experience using it in this setting than there is for the usually unresectable deep midline low grade gliomas discussed below.

In contrast to cerebellar or cerebral low grade gliomas, the midline optic pathway/hypothalamic gliomas (OPG), which constitute 3-5% of pediatric brain tumors and are also frequently low-grade pilocytic astrocytomas, are rarely amenable to complete resection, by virtue of their infiltration of critical structures. These tumors of the visual pathways may involve one or more optic nerves, the optic chiasm (where they may be exophytic into the third ventricle or hypothalamus), the optic tracts, optic radiations, and the occipital cortex (Figure 2). They predominate in young children (75% in those <10 years) and are associated with Neurofibromatosis Type 1 (NF 1); 20-30% of OPGs are in NF 1 patients.77 In lesions that are exophytic from the chiasm into third ventricle or hypothalamus, a subtotal resection may be possible, sometimes obviating a shunt in patients with hydrocephalus.78 However, for most of these tumors, if treatment is required, alternatives to surgery are necessary.

View Large Image Download to PowerPoint

FIG. 2. Axial T1-weighted brain MRI after gadolinium administration, demonstrating enhancing right optic nerve glioma. b. Sagittal T1-weighted brain MRI of an enhancing optic nerve/chiasm glioma that is exophytic into the hypothalamus.

Although OPG neoplasms nearly always have low-grade histology, their biologic behavior and growth characteristics, in vivo, vary tremendously. Good prognostic factors for stability after diagnosis include involvement only of visual structures and not of hypothalamus, older age of the child, and occurrence in a patient with NF1.78, 79 The treatment of these tumors must be individualized; children with minimal and non-progressive symptoms and radiographic stability, are followed expectantly. Although radiation therapy had been the mainstay of treatment of these tumors,80 because of the association of radiotherapy with increased risk for neurocognitive and neuroendocrine late effects, especially in children younger than 5 years,81 currently, RT is usually withheld in favor of administering chemotherapy in an attempt to defer, if not obviate RT altogether. Several chemotherapy regimens including carboplatin alone, carboplatin/vincristine, and procarbazine/CCNU/vincristine/6-thioguanine have been tried for incompletely resectable low-grade gliomas in any location, although the majority are OPG/hypothalamic gliomas. These pilot trials reported regression or stability of tumors in 75-90% of patients.82, 83, 84 At follow-up, children treated on the carboplatin/vincristine regimen before age 5 years, had a 75% 4-year PFS rate.85 The outcome of a recently-closed COG phase III randomized trial comparing the latter two of these regimens is pending.

Although there are no prospective randomized trials evaluating radiotherapy for low-grade gliomas in children, benefit was reported for 24 children with progressive optic chiasm gliomas, who had 88% 6-year PFS.80 When radiotherapy is indicated for the treatment of an OPG (or a low-grade glioma in any location), the current strategy of stereotactically-guided conformal radiotherapy is optimal. Results of two studies using conformal RT to treat children with low grade gliomas are encouraging (in 38 children, a 2-year PFS of 88%; in 14 children, a 3-year PFS of 87%).86, 87

Malignant Pontine Gliomas 

The majority of brainstem gliomas are malignant, diffusely infiltrative tumors centered in the pons, and are not resectable. They present with the triad of rapidly progressive cranial nerve deficits, long-tract signs and ataxia, but without hydrocephalus, at a peak age of 5-14 years. Taken together with this typical clinical history, their appearance, with diffuse enlargement of the pons on MRI, is diagnostic, and biopsy is no longer indicated.3 Unfortunately, the prognosis of these rapidly growing, highly malignant tumors is very poor. The foundation of treatment has been radiation therapy, but the median survival is less than one year. Attempts to improve survival using hyperfractionated RT 88, or chemotherapy 89 have not been successful. A current therapy strategy is the application of chemotherapeutic agents as radiosensitizers, to try to improve the outcome for these tumors.90

Supratentorial High-Grade Gliomas 

Supratentorial high grade gliomas in children include mainly anaplastic (malignant) astrocytomas and glioblastoma multiforme, and comprise 10-12% of CNS tumors in children. These include cerebral hemispheric tumors, as well as deep midline thalamic tumors.49

Supratentorial High-Grade Glioma Therapy 

Conventional treatment for cerebral high-grade gliomas has been maximal resection followed by involved-field radiation therapy. Greater extent of tumor resection has been shown to correlate with longer PFS and is the most important clinical prognostic factor for these tumors.91 Radiation treatment of high-grade gliomas in children has been part of all published treatment regimens of the last two decades, although the data supporting its prognostic value are derived mainly from adult studies. Evidence for improved outcome with adjuvant chemotherapy for high-grade gliomas in children is modest, but it is routinely used. From one phase III CCG 921 study of children with high-grade gliomas, the 5-year event-free survival (EFS) for those who received adjuvant chemotherapy with the nitrosourea-based regimen of CCNU/vincristine/prednisone plus RT after surgery was 46%, vs.18% for those who received radiotherapy alone. The benefit of the chemotherapy was demonstrated in the 33 children with glioblastoma in whom the 5-year EFS was 42 % for those treated with chemotherapy plus RT vs. 6% for those treated with RT alone, whereas the number of anaplastic astrocytomas in the study was too small to demonstrate an effect of chemotherapy.92 The other phase III study, CCG 945, for pediatric high-grade gliomas compared two chemotherapy regimens, the prior CCNU/vincristine/prednisone regimen vs. the eight-drugs-in-one-day regimen, for which the overall 5-year EFS rates of 26% vs. 33%, respectively, did not differ.93 A phase II German trial for pediatric high-grade gliomas determined that the pre-radiation and post-radiation chemotherapy significantly benefited those patients who also had a gross total resection, with 5.2 year vs. 1.9 year median survival.94

Other chemotherapeutic agents, such as temozolomide, an orally administered alkylating agent, and irinotecan, a topoisomerase-I inhibitor have been tried for newly diagnosed high-grade gliomas with modest success.95 A very recently completed phase II COG trial used temozolomide administered daily, concurrent with radiation therapy, and then afterward, in monthly cycles. Other studies combining various agents are also ongoing or in development in the COG.

Supratentorial High-Grade Glioma Investigational Therapy 

Some investigational approaches to treatment of high-grade gliomas with chemotherapy focus on circumventing mechanisms of drug resistance that develops to alkylating agents. One of the major mechanisms of resistance to the nitrosoureas is the overexpression of the cellular enzyme in tumor cells, O6 guanine alkylguanine alkyltransferase (AGAT), which is responsible for removing alkyl groups from the O6 position of deoxyguanosine and preventing DNA cross-links.96 The drug O6 benzyl guanine (O6 BG) which has high affinity for AGAT, enhances the activity of nitrosoureas in vitro and in vivo.96 Several phase II trials with O6 BG and temozolomide, which also has the AGAT drug-resistance mechanisms, are ongoing.

Because the blood brain barrier (BBB) may be an obstacle to delivery of effective chemotherapy to brain tumors, various tactics have been developed to circumvent this problem. One strategy has been the use of agents, such as mannitol, prostaglandins, histamine and bradykinin, capable of transiently disrupting the BBB for enhanced chemotherapy delivery. Most recently, promising results were reported from a preliminary study using a bradykinin analogue, labridimil (Cereport) children with brain tumors to disrupt the BBB prior to administration of carboplatin.97 Convection-enhanced delivery is another new method of delivering tumor-targeted large molecule toxins directly into tumor through a surgically-placed catheter. Adult phase I trials using this methodology are currently underway to deliver recombinant chimeric proteins, such as interleukin-4, linked with a pseudomonas exotoxin. The tumor-targeted toxins are then taken up by glioma cells, causing cytotoxicity after internalization.98

Increased knowledge of the tumor biology and molecular traits of high-grade gliomas, just as with PNET tumors, points the way to their use as potential targets for therapy. One of the most common molecular alterations so far identified in high-grade gliomas involves the epidermal growth factor receptor (EGFR). Overexpression of EGFR is very common in pediatric high grade gliomas,15 although gene amplification, the most common abnormality in adult, so-called “de novo” (primary) glioblastomas, is rarely seen in high-grade gliomas in children. Amplification of the platelet-derived growth factor alpha receptor (PDGFaR) gene and signaling by the PDGFaR/RAS/MAP tyrosine kinase pathway have been demonstrated in adult glioblastoma and play a role in the tumorigenesis of some pediatric medulloblastoma. This molecular pathway appears not to be as important for most pediatric high-grade gliomas, although PDGFR gene amplification was found in one of 14 high-grade gliomas in a recent study.99 The phosphatidylinositol-3 kinase (PI3K)/ras/AKT pathway and downstream target, mammalian target of rapamycin (mTOR), mediates functions of cell survival and migration, for which the PTEN (phosphatase and tensin homology) gene is an important regulator. Mutations of the PTEN gene have been identified in 6% of pediatric malignant gliomas and 20% of glioblastomas.100 Alterations in the p53 gene are fairly common in pediatric high-grade gliomas. Mutations of p53 were found in 40% of malignant gliomas in children > 3 years, similar to the rate seen in the tumors of young adults. However, overexpression of the p53 protein, found on one third of a series of uniformly-treated pediatric malignant gliomas, did not correlate with the presence of a mutation but did correlate with a poor prognosis.14

Small molecule inhibitors of several of these targets have been developed and are in early clinical trials for adult high-grade gliomas, including Iressa (gefitinib), targeting the EGF receptor tyrosine kinase,13 Gleevec (imatinib), targeting the PDGF receptor tyrosine kinase,21 and Sirolimus (rapamycin) and Everolimus (rapamycin derivative), targeting the PTEN oncogene-activated PI3K/ras/AKT/mTOR pathway.13 Phase I trials of Iressa and Gleevec are ongoing for recurrent and newly diagnosed high-grade gliomas in children. Just as with the anti-angiogenic agents, it seems likely that these drugs will not be as effective as single agents, and in future studies they will probably be combined with other small molecule inhibitors, chemotherapeutic or anti-angiogenic agents to assess their additive or synergistic potential.