Each year in the United States, an estimated 11,000 children between the ages of 0 and 15 years are diagnosed with cancer, of which 2,200 suffer from invasive central nervous system (CNS) tumors.[ 1,2] Brain tumors are second only in frequency to acute lymphoblastic leukemia in children. The incidence of CNS tumors in children under age 20 years in this country is 3 to 4 cases per 100,000/year. Just over half of all pediatric brain tumors (52%) are low-grade cerebellar astrocytomas, followed by primitive neuroectodermal tumors (PNETs), other gliomas (21%), and ependymomas (9%), as illustrated in Figure 1.[1] The incidence of low-grade astrocytomas, PNETs, and ependymomas is inversely proportional to age, whereas that of malignant glioma is relatively constant between birth and age 20 years.[2] The incidence of CNS tumors in children has been increasing in recent years, which is partlyattributed to the advent of better imaging technology including magnetic resonance imaging (MRI).[3] The risk factors for the development of brain tumors in children are mostly unknown. Less than 10% of CNS malignancies are related to distinct genetic syndromes (Table 1).[4] While radiation exposure is a recognized risk factor for brain tumors, the role of other environmental toxins is relatively unclear.[1] Although the annual mortality rate of pediatric cancers has steadily decreased over the past 2 decades, the proportion of deaths from CNS tumors in the same population has increased from 18% to 30%.[2] These figures clearly highlight the less than optimal outcomes in children with CNS malignancies, compared to other pediatric tumors. Factors That Contribute to Treatment Failure While surgery and radiotherapy have long been established as treatment modalities for patients with brain tumors, the role of chemotherapy in this population is far from clear. Nevertheless, chemotherapy has continued to play a significant part in the therapeutic armamentarium against these tumors and may contribute to disease stabilization and cure in some patients. Concurrently, it has also become obvious that a significant proportion of patients with brain tumors suffer progressive disease during or following chemotherapy. The causes for such therapeutic failure have been extensively explored and reported. Lack of response to chemotherapy has invariably been due to the presence of the blood-brain barrier and drug resistance.[5-7] The refractoriness of brain tumors to chemotherapy stems from a multitude of factors that can be broadly classified as those caused by apparent or inherent cellular resistance.[5,8] Apparent drug resistance to a chemotherapeutic agent is usually due to the presence of the blood-brain barrier,[6,7] or the cell kinetics of a large tumor that has a smaller growth fraction (larger number of cells in the G0 fraction of the cell cycle) and hypoxic areas that limit the effect of chemotherapy.[8] Inherent drug resistance can be either de novo or acquired. Mechanisms of resistance to chemotherapeutic agents that are typically used in brain tumors are listed in Table 2. Blood-Brain Barrier and Its Disruption
The blood-brain barrier is composed of endothelium and covers almost the entire capillary network supplying the brain. The endothelium in the blood-brain barrier is nonfenestrated and has high-resistance tight junctions. Additional components of the blood-brain barrier include the astroglial processes, basement membrane, and pericytes (Figure 2).[4,5] The proliferation and invasion of tumor cells in the brain generally results in disruption of the brain microvasculature, breach of the blood-brain barrier, and development of vasogenic edema, even in small tumors.[4] The interstitial edema resulting from this increased capillary permeability can in turn influence cerebral blood flow, brain metabolism, and intracranial pressure. Tumor cells also secrete proangiogenic factors including basic fibroblast growth factor (b-FGF) and vascular endothelial growth factor (VEGF), resulting in the influx of new blood vessels into the tumor-a process called tumor angiogenesis.[9,10] These tumor capillaries are differentfrom the capillaries of the normal brain in that they are hyperplastic, have frequent fenestrations, lax intercellular junctions, and less well-developed glial processes abutting on the abluminal surface of the endothelium.[6] Thus, the continuing proliferation of tumor cells in the brain actually results in disruption of the blood-brain barrier.[7] However, it is possible that such disruption can vary between tumors and even within a given tumor. Also, it is likely that small tumors (eg, the infiltrative edge of a malignant glioma) might have a relatively intact blood-brain barrier that may lead to chemotherapy failure.[7] Blood-Brain Barrier and Chemotherapeutic Efficacy
In the ongoing search for more effective chemotherapeutic agents for patients with brain tumors, there is a general bias toward choosing lipophilic agents with a high octanol-water partition coefficient (a measure of the lipidsolubility of the drug) to enable rapid transfer of these drugs from the blood to the tumor cells and overcome the blood-brain barrier. However, as indicated above, it appears that the bloodbrain barrier might be disrupted even in small tumors. In addition, studies have shown that the average concentration of chemotherapeutic agents in brain tumors does not significantly differ from their extracranial counterparts, although the homogeneity of drug distribution varies both within and between brain tumor deposits.[4,7] While lipophilic drugs do penetrate the blood-brain barrier better, this does not necessarily translate into equal efficacy in all patients with brain tumors[7]; there are clearly other reasons for chemotherapy failure in such patients.[4,6] Nevertheless, it is possible that disruption of the blood-brain barrier may not be uniform in brain tumors, and areas of the brain surrounding the main tumor may have a relatively intact blood-brain barrier. This concept has led to an increasing trend toward devising methods thatfurther disrupt the blood-tumor barrier to facilitate entry of chemotherapy into brain tumors. Such increased disruption could potentially increase drug concentration in areas where the barrier has not been completely disrupted by the tumor.[7] Moreover, disrupting the blood-brain barrier might help chemotherapy reach areas of adjacent tumor invasion of the normal brain, wherein the barrier may be relatively intact.[7,11] Angiogenesis and Brain Tumors
In 1971, Dr. Judah Folkman proposed that continued tumor growth after the initial tumor take (up to 2 mm3) is dependent on the growth of blood vessels into the tumor.[12] The influx of new capillaries into the tumor was termed "angiogenesis." This hypothesis has since led to many studies that have resulted in the discovery of several proangiogenesis and antiangiogenesis molecules in the laboratory.[ 10,12,13] The principal proangiogenesis molecules include alpha and beta fibroblast growth factors (b-FGF), VEGF, and angiogenin.[ 10,13] The principal negative regulators of new capillary growth are thrombospondin and angiostatin.[13] Tumor growth is generally dependent on a balance of the positive and negative regulators of angiogenesis. These factors can be produced by tumor cells, mobilized from the extracellular matrix, or released by macrophages attracted to the tumor.[14] The angiogenic process itself leads not only to further tumor growth, but also to tumor invasion and, ultimately, metastasis to other body sites.[9,10] A feature of many brain tumors is the presence of neovascularization. Immunohistochemical studies have demonstrated the presence of angiogenic factors including b-FGF in high concentrations in brain tumors.[15] This peptide stimulates vascular endothelial cell proliferation, and such cells either produce or possess receptors for b-FGF. Li et al have detected b-FGF in the cerebrospinal fluid (CSF) in 62% of children with brain tumors but in none of a group of controls.[15] The CSF specimens with elevated b-FGF increased the DNA synthesis of capillary endothelial cells in vitro, and such activity was blocked by neutralizing antibody to b-FGF. The concentration of b-FGF in CSF was also correlated with density of microvessels in histologic sections of the brain tumors, and the microvessel density was negatively correlated with prognosis. Although b-FGF was the only proangiogenic factor studied in this report, it is possible that other angiogenic peptides could mediate the growth of brain tumors. The demonstration of the role of angiogenesis in sustaining tumor growth has led to the exploration of inhibitors of angiogenesis as a means of curtailing tumor progression. Elegant preclinical studies in mouse tumor xenograft models have shown dramatic tumor regression and cure of animals bearing tumors.[16] Currently, several phase I studies of novel angiogenesis inhibitors in patients with a wide variety of tumors are under way both in the United States and Europe, and should help our understanding of the toxicity, dose schedules, and possibly the usefulness of these agents in these malignancies.[13] Advances in the Treatment of Specific Pediatric Brain Tumors Low-Grade Gliomas
Pediatric low-grade gliomas are a heterogenous group of tumors that constitute the most frequent CNS neoplasia encountered in children (30%- 40% of all CNS tumors diagnosed in the United States).[17] They can be classified based on histology or location (Table 3).[16] Patients with neurofibromatosis type I frequently develop low-grade gliomas (typically piloyctic astrocytoma, WHO grade I), particularly of the optic pathways.[18] Neurofibromatosis type I patients with low-grade glioma have a more favorable prognosis than patients without the genetic disease.[19]

• Surgery-Because these tumors only rarely metastasize through the neuraxis, local control using surgery and/or radiotherapy have been the traditional components of therapy for patients with low-grade glioma. Due to a direct positive correlation between extent of resection and progression-free and overall survival, aggressive surgery should be attempted in tumor locations where feasible.[17] The cystic cerebellar astrocytoma is an example of a low-grade glioma that can be resected completely without causing neurologic deficits, resulting in cure rates of over 90% (Figure 3).[17] Similarly, exophytic tumors of the brain stem including the cervicomedullary region are typically pilocytic astrocytomas, amenable to complete surgical removal with an excellent prognosis.[17] Tectal plate gliomas usually present with hydrocephalus due to early compression of the cerebral aqueduct. Patients with these tumors are observed without any treatment after controlling hydrocephalus with a third ventriculostomy. The operating microscope and ultrasonic surgical aspirator allows the surgeon to perform an adequate tumor resection without compromisingthe adjacent normal brain. Preoperative imaging can be correlated with intraoperative observation using a frame-based or frameless system that also helps in guiding the surgeon with trajectories to deep-seated lesions. Functional MRI, positron-emission tomography, and electrode grids for mapping areas of the cerebral cortex have helped the surgeon to accurately delineate tumor margins and improve surgical morbidity in these patients. However, about 30% of all lowgrade gliomas-including those in the optic pathway, diencephalon (hypothalamus and thalamus), and intrinsic portion of the brain stem-remain unsafe for surgical resection. For tumors in such locations as well as those that recur after initial surgical resection and are associated with functional impairment, tumor control can be obtained with chemotherapy and/or radiotherapy.
• Radiotherapy-There are no reported prospective randomized trials assessing the benefits of adjuvant radiotherapy in children with low-grade gliomas.[17] Extrapolating data from adult phase III trials, it seems appropriate to use focal radiotherapy in doses of 45 to 54 Gy in 1.8- to 2.0-Gy fractions. Three-dimensional (3D) conformal radiotherapy, intensity-modulated radiotherapy, and proton-beam therapy are new radiation delivery techniques that are designed to minimize damage to normal brain but await validation in larger studies.[17]
• Chemotherapy-In view of the deleterious effect of radiotherapy on the growing brain in young children, chemotherapy agents including vincristine, dactinomycin(Drug information on dactinomycin) (Cosmegen), cyclophosphamide(Drug information on cyclophosphamide) (Cytoxan, Neosar), carboplatin(Drug information on carboplatin) (Paraplatin), lomustine(Drug information on lomustine) (CCNU [CeeNu]), and etoposide(Drug information on etoposide) have been employed individually or in combination in patients with progressive low-grade gliomas to delay or avoid irradiation.[19] Following preliminary clinical evidence of activity, carboplatin, either alone or in combination with vincristine, has been used as a front-line therapy for children with low-grade glioma, particularly thosewith optic pathway tumors.[19-21]

Objective responses and disease stabilization have been observed in 25% to 58% and 80% to 90% of patients, respectively, in these studies. The 3-year progression-free survival in patients with recurrent low-grade glioma following this chemotherapy regimen has been reported to be 64% to 68%.[19,21] Myelosuppression is the main toxicity related to carboplatin followed by hypersensitivity reactions in about 10% to 30% of patients. An alternative nitrosourea-based chemotherapy regimen reported by Prados et al was shown to produce prolonged disease stabilization in children with progressive low-grade gliomas.[ 22] This treatment combination is being evaluated in comparison with the carboplatin-based regimen in an ongoing Children's Oncology Group (COG) phase III trial in recurrent lowgrade gliomas. Recently, we and others have also reported on the efficacy of temozolomide(Drug information on temozolomide) (Temodar) in both adults and children with low-grade gliomas.[23,24]