Changes in the Approach to Central Nervous System Tumors in Childhood

PEDIATRIC NEUROLOGY 0031-3955/92 $0.00 + .20 CHANGES IN THE APPROACH TO CENTRAL NERVOUS SYSTEM TUMORS IN CHILDHOOD Patricia K. Duffner, MD, and Michael E. Cohen, MD The number of children who survive brain tumors has increased over the past 20 years because of advances in surgery, radiation, and possibly chemotherapy. Previously, minimal concern existed about possible adverse effects of these types of therapies because the number of long-term survivors was so limited. Today, 50% of children with all types of brain tumors may be expected to survive 5 years. The goals of neuro-oncology have broadened to include not only improved survival rates but also improved quality of life. In this article, we discuss both of these areas: (1) changes in therapy that have impacted survival rates and (2) changes in therapy as a consequence of complications of treatment. CHANGES IN THERAPY THAT HAVE IMPACTED ON SURVIVAL RATES Two tumors in which changes in therapy have impacted on survival rates are medulloblastomas and brain-stem gliomas. Perhaps the most remarkable improvement in survival has been in those children with medulloblastomas. Medulloblastomas Medulloblastomas represent 20% to 30% of all childhood brain tumors. Survival rates of children with these tumors have changed dramatically over the last 60 years. In 1930, Cushing reported that 1 of 61 children operated for From the Department of Neurology, State University of New York at Buffalo School of Medicine and Biomedical Sciences; and the Division of Child Neurology, Children's Hospital of Buffalo, Buffalo, New York PEDIATRIC CLINICS OF NORTH AMERICA VOLUME 39 • NUMBER 4 • AUGUST 1992 859 860 DUFFNER & COHEN medulloblastoma was alive at the end of 3 years. 18 A review of survival rates of children with medulloblastomas in Britain has showed a further trend toward increasing survival over time. In 1962 to 1970, the 5-year survival rate for a large cohort of patients was 18%; from 1971 to 1974, the survival rate was 27%; and from 1975 to 1978 38% survived.1O Most recently, based on staging and stratification of patients before treatment, survival rates of 60% are being reported. Most studies have found that those patients with medulloblastoma who are older at diagnosis, who have had bulk removal at time of surgery, and who have no evidence of extension into· the brain-stem and no dissemination through the neuraxis, have the best prognosis. Those with significant postoperative tumor burden, young age at the time of diagnosis (under 5 years of age), dissemination through the neuraxis, and possibly evidence of cellular differentiation, have the worst prognosis. 2,22,34,49 Of all the putative prognostic factors, the value of histologic criteria has been the most equivocal. The Children's Hospital of Philadelphia group has suggested that undifferentiated tumors have the best prognosis. They speculated that these counterintuitive results occur because less differentiated tumors are probably more radiosensitive or because the more differentiated the tumor, the more extensive it is at the time of diagnosis, or both.>1 Others suggest that tumor differentiation is a reflection of maturation and correlates with a longer recurrence-free period. In a recent study from Israel, the prognostic significance of glial fibrillary acidic protein (GFAP) in histologic sections of patients with medulloblastomas was evaluated. In this study, those patients who were GFAP positive had a survival rate of 82% compared with 30% in the GFAP negative group. These results were not influenced by tumor location, patient age, or different treatment regimens. 42 Thus, it can be seen that staging criteria are variable, and when one is assessing prognosis, one needs to consider multiple factors.44 Survival statistics can be weighted depending on how the investigator chooses to separate the study groups. Despite these complexities, it is increasingly apparent that in a certain subset of patients, prognosis, which was dismal at the turn of the century, has improved remarkably. Radiation therapy has been responsible for the major improvements in the survival of children with medulloblastoma. ll As early as 1953, Patterson and Farr, using orthovoltage equipment, demonstrated that radiation of the tumor bed and the craniospinal axis favorably affected cure rate. 74 Theirs was the first report documenting the value of neuraxis radiation, and it has become the minimum standard for treatment of this illness. More recently, two large studies evaluating the effectiveness of adjuvant chemotherapy became available. Treatment results and long-term prognosis of children with brain tumors from the Royal Marsden Hospital (1950-1981) were analyzed. Of this group, 143 children had medulloblastomas. Those completing radiotherapy had survival rates of 42% at 5 years, 33% at 10 years, and 29% at 15 years. The five-year survival rate for 19 patients with total excision was 62% compared with a 5-year survival of 41 % of 61 patients in whom a subtotal excision was obtained. Between 1970 and 1981, 38 of 53 patients received adjuvant chemotherapy. The survival rates in this group were compared with those of 105 cases between 1950 and 1981 who did not receive adjuvant chemotherapy. The 5- and 10-year survival rates for those receiving chemotherapy were 65% and 56%, respectively, suggesting a much greater survival for patients receiving chemotherapy than for historical controls. The use of historical controls, however, must be viewed with caution because improvements in surgery and radiotherapy have had a major impact on survival results. 10 CHANGES IN THE APPROACH TO CNS TUMORS IN CHILDHOOD 861 The International Society of Pediatric Oncology (SlOP), consisting of 44 centers from 15 countries, evaluated 286 patients with medulloblastomas randomized to either radiotherapy alone or radiotherapy plus chemotherapy (vincristine and CCNU). As of September 1988, their results suggest a small but insignificant difference in disease-free survival in favor of chemotherapy. When certain subgroups were evaluated as independent variables, however, i.e., total compared with subtotal resection, age greater or less than 2 years, and evidence of brain-stem and neuraxis dissemination, adjuvant chemotherapy was found to have a significant effect on survival. A similar study, reported in 1990 by members of the Children's Cancer Study Group (CCSG) and the Radiation Oncology Group (RTOG), evaluated the results of postoperative radiation and vincristine, CCNU, and prednisone compared with surgery and radiation therapy alone. 34 The 5-year survival rate for both groups was approximately 65%. As in the Royal Marsden Study, those patients with more extensive tumors who had been treated with chemotherapy had progression-free survivals that were better but differences were not statistically significant. Both these studies suggest that patients with more extensive disease and young children are at greatest risk. In only one (the SlOP study) was the degree of surgical resection correlated with response to chemotherapy. In that study, patients with partial and subtotal resection benefited from chemotherapy and those with total resection did not. Based on the concept that chemotherapy has a role in the treatment of medulloblastomas, Packer treated a group of 26 patients with cisplatin, CCNU, and vincristine. 6Y He defined his high-risk group as children at least 5 years of age at the time of diagnosis; tumors that were partially resected or biopsied; tumors that were disseminated at the time of diagnOSis; and tumors that had evidence of cellular differentiation. Radiation dose was reduced in children between the ages of 15 months and 3 years in an effort to decrease side effects. Event-free survival rates were compared with those of 24 children who had similar risk factors treated at the Children's Hospital of Philadelphia between 1975 and 1983. Ninety-five percent of the patients who entered the study were alive and free of disease at a median of 24 months from time of diagnosis. Toxicity included grade 2 to 3 renal toxicity, high-frequency hearing loss from platinum, and myelosuppression. Hematologic side effects were not a limiting factor. Another chemotherapy trial was reported by Kovnar et al. 54 They evaluated a preradiation chemotherapy regimen consisting of cisplatin and VP16 in 11 children with newly diagnosed medulloblastomas, pineoblastomas, and cerebral neuroblastomas. All eight children with medulloblastomas responded. Although the study is small and the results are early, none of his patients with medulloblastoma showed signs of progression 18 to 48 months following diagnosiS. Single-institution studies that include small numbers of patients need to be interpreted cautiously because their results are almost invariably better than those from multi-institutional studies. Thus, the larger series reported by SlOP and CCSG are probably a more accurate reflection of the value of current therapies. Both of these studies indicated an advantage to adjuvant chemotherapy in prolonging event-free survival in certain subgroups, suggesting that future chemotherapy trials are warranted, although perhaps with different agents and increasing dose intensity schedules. Further, these studies suggest that patients with disseminated disease at the time of diagnosis and younger patients have a significantly worse prognosis. The ideal chemotherapy regimen is yet to be identified. Stratification of 862 DUFFNER & COHEN patients into high-risk and low-risk groups will increasingly influence treatment options. Aggressive and potentially toxic therapies will be restricted to those stratified to high-risk groups whereas less toxic strategies will be devised for patients considered to be low risk. For patients with high-risk medulloblastomas, the cancer consortiums in this country are turning to increasing dose intensity schedules, preradiation chemotherapy followed by radiation and postradiation chemotherapy, and exploration of immune modulating methods. For low-risk patients, treatment options may include reduced doses of neuraxis radiation, possibly coupled with adjuvant chemotherapy. Brain-stem Gliomas The tumor with the poorest prognosis is the brain-stem glioma. Several series report 5-year survival rates ranging from 5% to 30%. Most children die within 2 years of diagnosis. Despite this overall gloomy prognosis, this group of tumors is now recognized as being heterogeneous, and prognosis may vary depending on location, duration of symptoms before diagnosis, neuroimaging characteristics, and to some extent histopathologic criteria. 13, 88, 94 Histopathologic diagnosis is hampered by the limited amount of biopsy material that can be obtained at surgery. Because of the fears of damaging vital brain-stem structures and the possibility of brain-stem swelling, most biopsy material is obtained by stereotactic methods. 41 , 43, 46, 53 The finding of a low-grade neoplasm on stereotactic biopsy cannot be viewed as definitive because of the possibility of sampling error and the potential for malignant dedifferentiation. These considerations have led many authors to reject the need for pathologic confirmation. Others have suggested that because most patients have pilocytic tumors that are uniform in structure with little regional variation, sampling error for this histologic grouping may not be a problem and biopsy is worthwhile.36, 80 Several articles in the literature address the relation of pathologic diagnosis to therapy and outcome. In a Japanese series of 23 patients with histologically proven brain-stem gliomas, 11 had glioblastomas and the remaining had low grade gliomas. Patients whose tumors had malignant features survived fewer than 15 months following diagnosis, suggesting that the finding of malignancy was a predictor of poor prognosis. In most patients with low-grade tumors, a 5-year actuarial rate of 50% was reported, suggesting a better prognosis for the more histologically benign-appearing tumors.64 The need for biopsy was further underscored in a report from Marseilles in which 17% of 33 patients with brain-stem tumors had non-neoplastic lesions. These included hemangiomas, hemorrhagic infarcts, arachnoid cysts, telangiectasia of the pons, cavernous hemangiomas, and hamartomas. 36 These reports emphasize the value of pathologic confirmation in arriving at a treatment decision. 6 In addition, traditional concerns regarding surgical morbidity may be overstated. For instance, in the Marseilles group, the mortality rate was 3% and morbidity was temporary in 18% and permanent in only 3%. With the use of stereotactic biopsy, a surgical approach to the brain-stem is becoming more accepted. 36 Because of increasing knowledge of biologic behavior, recent treatment approaches have been tailored to prognostic factors. Sanford et al 83 developed a set of clinical criteria for selecting patients with an adverse prognosis for inclusion in the POG study of hyperfractionated radiation. Children selected for inclusion in the protocol were those who had a duration of symptoms and CHANGES IN THE APPROACH TO CNS TUMORS IN CHILDHOOD 863 signs of less than 6 months; tumors that infiltrated the brain-stem proper, i.e., the mesencephalon, pons, and medulla; and those who had evidence of two or three brain-stem signs, i.e., long tract signs, cranial neuropathies, or cerebellar signs. Lesions that were focal, exophytic, or originated in adjacent anatomic structures such as the cerebellar peduncle or cervical medullary junction were excluded. In this group of high-risk patients, median survival was 11 months with a range of 2 to 28 months. Of the 11 patients available for histologic diagnosis, the majority had either glioblastomas or astrocytomas. One had pilocytic astrocytoma, one, ependymoma, and one, mixed glioma. In five autopsy cases, four had a glioblastoma and one had a low-grade astrocytoma. In only one patient could biopsy results be compared with autopsy results, limiting the ability to determine whether the tumor had been dedifferentiated. Epstein and others, using imaging criteria, suggested that prognosis and treatment decisions can be made based on anatomic localization. Epstein and Wisoff defined four categories of brain-stem tumors, i.e., diffuse, focal, cystic, and cervicomedullary.33 The diffuse brain-stem tumor is the most common and has the worst prognosis. These tumors are characterized by areas of hypodenSity throughout most of the pons and may extend into the midbrain or into the medulla. They may or may not enhance with contrast agents. He and others have emphasized that the magnetic resonance imaging (MRI) scan may demonstrate a more diffuse nature of the tumor when compared with the computed tomography (CT) scan. Focal brain-stem tumors are identified by areas of contrast enhancement on CT and absence of associated hypodensity. The MRI scan may either substantiate these findings or demonstrate that the tumor, rather than focal, is more diffuse and thus has a more malignant potential. Cystic tumors behave in the same manner and have imaging characteristics similar to those of cerebellar astrocytomas. CT may identify a mural nodule that enhances with contrast and is associated with a large cyst. Tumors of this nature may be found in the pons or extend into the cerebellar peduncle. In four of Epstein's cases in which the CT scan showed enhancement of both the wall of the cyst and the cellular components of the neoplasm, the MRI scan revealed more extensive involvement than that suggested by CT. Tumors of the cervicomedullary junction extend to the medulla and caudally into the cervical cord. These tumors rarely extend above the pontomedullary junction. The prognosis for each group varies. Those that present as dorsally exophytic, are cervicomedullary in location, or are cystic with mural nodules are amenable to surgery. The tumors that benefit most from surgery are those that are dorsally exophytic in the brain-stem and bulge into the fourth ventricle. Cystic, cervicomedullary tumors, and less frequently, focal tumors, may respond in a fashion similar to that of the dorsal exophytic brain-stem tumor. In these groups, radical surgery offers the possibility of long-term remission and even cure. Laser and cavitron surgery are used to debulk the neoplasm. Surgical approaches can be more radical in the upper cervical cord but must be more conservative in the medulla because the risk of injuring adjacent gray matter is greater. Ultrasound is used to monitor the dissection to identify cysts that may be overlooked as surgery proceeds. 32• 33 Another group of patients with brain-stem gliomas who have a good prognosis are patients with neurofibromatosis. Brain-stem lesions in these patients may have a different biologic expression than those in patients with brain-stem tumors without neurofibromatosis. In several recent reports, patients with brain-stem tumors and neurofibromatosis have had long, progression-free survival rates. Of interest, some of these children were not treated and have had prolonged freedom from progression. 62. 70 864 DUFFNER & COHEN For those children in whom surgery is not an option, hyperfractionated radiotherapy has been suggested. Hyperfractionation allows a higher dose of radiation to be given over a period similar to that used for standard radiation. Rather than giving 180 cGy of radiation once a day for 5 to 6 weeks, radiation doses per fraction of 110 to 120 cGy are given twice a day. This technique allows the total dose of radiation to the tumor to be increased 10% to 20% over the standard radiation dose of 5500 cGy. Decreasing the dose per fraction and increasing the total dose of delivered radiation theoretically produces a greater tumoricidal effect without increasing side effects. The theory of hyperfractionatton rests on the concept that neoplastic cells are more sensitive to ionizing radiation in the proliferating phase of the cell cycle. Those cells that have not entered the proliferating phase of the cell cycle may not be affected by a given dose of radiation. Thus, the likelihood of increasing cell kill would be enhanced by delivering multiple fractions of radiation over a given period of time. Further, normal glial tissue that is not proliferating has considerably more capacity to repair sublethal radiation damage from smaller doses per fraction. Theoretical and experimental evidence suggests that an inverse relationship of radiation damage to oxygen content decreases with dose per fraction. This observation suggests that radioresistant hypoxic cells may be more susceptible to multiple daily fractions of radiation than to single doses. Based on this rationale, several groups have entered into phase 2 trials of hyperfractionated radiation. Over the last several years, several series have escalated hyperfractionation from 6600 cGy to 7800 cGy. 30, 37, 38, 68 These studies have documented the ability of normal tissue to tolerate increasing doses. The cell kill effect of 6600 cGy given by hyperfractionation has been determined to be the same as that achieved by standard doses of 5500 cGy. Larger doses of radiation have, however, been associated with increasing progression-free survival. In Edwards' group30 of 34 patients treated with 7200 cGy, the median time to tumor progression waS 44 weeks with a median survival of 64 weeks. In a multi-institutional study using 7200 cGy, a 30% progression-free survival was noted at 20 months. Published Pediatric Oncology Group (POG) studies using 6600 cGy and 70.2 cGy revealed similar results. The median time to progression was 6 months with the median survival of 10 months. Toxicity at the higher dose levels was greater than at the lower dose levels. 30 In the POG study, approximately 50% of the patients remained on steroids for more than 3 months following the beginning of radiotherapy. Complications included opportunistic infections, increased glucose intolerance, hypertension, osteoporosis, and mood swings. At higher dose levels, cystic degeneration and intralesional necrosis were identified in a few patients. Most of these complications could be managed conservatively, but several required surgical decompression. Tissue removed from the necrotic lesions was benign in three of four patients. Results are now near completion in several groups using doses of 7560 cGy and 7800 cGy.37.38 These studies have accrued large numbers of patients and have demonstrated the feasibility and relative safety of using twice-daily fractionation. Although progression-free survival rates have improved with increasing doses of hyperfractionation, the overall survival of children with brain-stem gliomas remains a therapeutic challenge. Chemotherapy has not had a major role in treatment. Future brain-stem protocols are apt to emphasize increasing the dose intensity of chemotherapy coupled with aggressive radiotherapy such as hyperfractionation, accelerated fractionation, or radiosurgery. In summary, despite an overall poor prognosis, a subgroup of patients with brain-stem gliomas have a favorable prognosis. These include those CHANGES IN THE APPROACH TO CNS TUMORS IN CHILDHOOD 865 patients with focal, cystic, or cervicomedullary lesions and those with neurofibromatosis. These groups should be stratified to less aggressive treatment protocols because their outlook is more hopeful. In contrast, the child with diffuse abnormalities on MRI and CT scans and a short duration of symptoms and signs carries a poor prognosis and merits aggressive therapy. CHANGES IN THERAPY AS A CONSEQUENCE OF COMPLICATIONS OF TREATMENT The long-term effects of central nervous system (CNS) radiation were initially identified in the mid to late 1970s. 45, 75, 84 These early studies, which warned of leukoencephalopathy and adverse effects on learning and growth, were followed by a large number of articles in the 1980s that documented, first in retrospective and then later in prospective studies, the potential consequences of CNS treatment. Unfortunately, only a small percentage of these research articles made their way into the pediatriC literature, with most published in journals of Neurology, Neurosurgery, Oncology, and Radiation Therapy, Consequently, pediatricians, who provide primary care for many of these children, may be unaware of adverse effects of treatment. Recognition of these effects has recently led to several attempts by the cancer cooperative groups to alter therapy in an effort to reduce neurotoxicity. Intelligence Most of the initial work on the adverse effects of CNS therapy on intelligence focused on children with acute lymphoblastic leukemia (ALL). In the 1970s and 1980s most children with ALL received cranial radiation (18002400 cGy) as part of CNS prophylaxis. As increasing numbers of children with ALL survived, effects of CNS prophylaxis on neuropsychological function became more relevant. 61 One early study retrospectively evaluated children with ALL who had received different forms of CNS prophylaxis, i.e., cranial irradiation, intrathecal plus intravenous methotrexate, and intrathecal methotrexate. 82 All the children superficially performed in a normal fashion and were attending regular classes. Detailed neuropsychologic testing, however, revealed significantly worse results on tests of intelligence and achievement in those children who had received cranial irradiation in a dose of 2400 cGy. In addition, the children were described as being impulsive and having attentional difficulties. None of the children in that study had a history of CNS disease that might have contributed to their learning difficulties, and all had been in <;ontinuous complete remission for 1 year. Therefore, cranial irradiation was implicated as the cause of the children's learning problems. Later studies confirmed these results and amplified the specific learning disabilities these patients developed, i.e., difficulty with math, visual motor function, and spatial concepts. I? In 1969, Bloom suggested that 82% of 22 children with brain tumors had no long-term disability following treatment with surgery and radiation therapy.n It was not until the results of the early ALL late-effects studies were published in 1976 that a number of investigators began retrospectively to evaluate children with brain tumors.61 The majority of the early studies consisted of small numbers of patients and, as such, could not be analyzed from a statistical point of view. Nonetheless, when repeated studies reported the same 866 DUFFNER & COHEN results, i.e., 10% to 20% of children having intelligence quotients (IQs) above 90 with 30% to 50% of children having IQs below 70,28,45,77.90 it became clear that the higher dose of CNS radiation used to treat brain tumors was associated not only with learning disabilities but also with frank mental retardation. Of interest, the limited number of prospective studies published since the late 1980s have not revealed the same frequency of mental retardation except in certain subsets, such as children radiated at a young age. 27 Nonetheless, these studies have demonstrated a clear-cut decline in intelligence by 2 years following completion of radiation and at least one has demonstrated continued deterioration for at least 10 years following treatment. 47 All studies to date have reported that most patients radiated for brain tumors develop learning disabilities and attention deficit disorders. 27, 70 Endocrine System The adverse effects of cranial irradiation on growth have been studied in children with leukemia as well as in those with brain tumors. Radiation-induced growth hormone deficiency occurs with either hypothalamic or pituitary damage, but in most cases the defect is hypothalamic. Radiation produces an immediate suppressive effect on the hypothalamic pituitary axis, and frank growth hormone deficiency can be identified as early as 3 months following completion of radiation. 19, 24, 29 Growth hormone deficiency relates to both the size and number of fractions. Thus, a larger dose of radiation given over a shorter period of time is more likely to cause growth hormone deficiency than the reverse. Shalet has suggested that a minimum dose of 2500 to 2900 cGy is necessary to cause radiation-induced growth hormone deficiency.8s Studies of children with leukemia treated with varying forms of CNS prophylaxis revealed that although radiated children had a significantly higher incidence of growth hormone deficiency than children treated with chemotherapy alone, some children who were not radiated also developed biochemical growth hormone deficiency.96 Despite biochemical deficiencies, longitudinal growth is generally normal in the leukemic population. In contrast, children with brain tumors who develop radiation-induced growth hormone deficiency have clinical growth failure as well. Most studies have suggested that approximately 80% of children radiated for brain tumors have both biochemical and clinical growth failure. 24,65, 73, 86 Clinical growth failure has multiple causes. Some children with brain tumors do not grow well either because of poor nutrition or coincident radiationinduced hypothyroidism. 52 Moreover, precocious puberty is common following cranial irradiation, leading to an initial growth spurt but ultimately short stature because of premature closure of the epiphyses. '2 Another factor contributing to the poor growth of children with brain tumors is spinal radiation. Because many childhood tumors seed the neuraxis, radiation must be delivered to the full spine. This radiation, particularly if delivered to a young child, adversely impacts on vertebral body growth. 87 Because so many factors contribute to the poor growth of children with brain tumors, it is hardly surprising that treatment with exogenous growth hormone is less effective than in children with idiopathic growth hormone deficiency. One reason for this discrepancy is that the duration of growth hormone treatment is often shorter in children with brain tumors than in children with idiopathic growth hormone deficiency. Children with idiopathic growth hormone deficiency often have a delayed puberty. As such, they CHANGES IN THE APPROACH TO CNS TUMORS IN CHILDHOOD 867 potentially have several years in which to receive treatment before fusion of their epiphyses. In contrast, because of the early onset of puberty, radiated children have less time to receive treatment because growth hormone therapy is ineffective once the epiphyses are fused. In addition, many endocrinologists do not treat children with brain tumors until the children have been tumorfree for 2 years. As such, the window of opportunity for treatment is often limited in this patient population. Perhaps most important, exogenous growth hormone does not improve the growth of vertebral bodies that have been damaged by radiation, and, even with optimal response to growth hormone therapy, the children do not achieve a normal adult height. Despite these caveats, new approaches to the treatment of patients with radiation-induced growth hormone deficiency are being developed. By using different doses and dosing intervals of synthetic growth hormone, some investigators have reported better responses. 56 Other methods of improving longitudinal growth include (1) pharmacologically delaying puberty and (2) treating children with growth hormone therapy when they first begin to show signs of growth deceleration. Despite theoretical concerns that exogenous growth hormone might increase early tumor recurrence, no evidence to date supports this belief. 52 As more children are being treated with growth hormone for radiation-induced growth hormone deficiency, we are likely to see better long-term studies of treated patients and clarification of the true incidence of this potential complication. Hypothyroidism, either primary, secondary, or tertiary, is another wellknown complication of cranial and craniospinal radiation. Hypothyroidism may contribute to poor longitudinal growth as well as to learning problems. Primary hypothyroidism develops because the thyroid gland lies within the radiation port to the cervical spine, and secondary or tertiary hypothyroidism develops because of radiation to the hypothalamic pituitary axis. 58 Thyroid replacement therapy is indicated not only in children with primary hypothyroidism but in children with compensated hypothyroidism who are euthyroid but have abnormally high thyroid-stimulating hormone levels. Recent studies have suggested that the addition of chemotherapy further increases the incidence of hypothyroidism. 58 Gonadal dysfunction also may occur as a result of CNS therapy. The neuro-oncology literature has not focused on this area in detail. Because we are now seeing more potential "cures" of brain tumors, awareness of this particular problem is rising. Radiation to the spine is associated with scatter to the ovaries, although the testes are relatively spared. Boys are at greater risk than girls from certain chemotherapeutic agents such as cyclophosphamide. In these cases, damage to the Leydig cells may cause oligospermia or even azoospermia. Fortunately, late recoveries are possible. 14 Leukoencephalopathy CT leukoencephalopathy was first described in children with ALL who had received chemotherapy and cranial radiation. 75 The CT findings were characterized by enlarged ventricles and enlarged sulci, hypodense areas, and areas of calcification. Clinical accompaniments to these abnormalities included seizures, dementia, focal motor signs, and ataxia. Although the original reports attributed the abnormalities to methotrexate, later analysis suggested that it was the combination of methotrexate and radiation or the presence of CNS 868 DUFFNER & COHEN leukemia and methotrexate that produced the changes on CT scans and the associated clinical abnormalities. 25, 66 More recently, with the advent of MRI, leukoencephalopathy has been found in children and adults treated with cranial irradiation alone in the absence of either methotrexate or any form of chemotherapy.'2,97 The MRI changes are described as periventricular hyperintensity, increased T2 weighted signals in the periventricular region that may extend to the gray matter-white matter border. They are scalloped in shape and do not extend to the cortical surface. Although these imaging changes have been associated with clinical leukoencephalopathy, not all MRI scans that demonstrate periventricular hyperintensity are associated with clinical dysfunction. Conversely, some patients with severe radiation-induced dementia may have normal MRI scans. 16 Risk Factors Three of the most important radiation-induced late effects are dementia, endocrinopathies, and leukoencephalopathy. Because these complications of treatment have already been reported in great detail, it is no longer sufficient for researchers solely to document abnormalities. Rather, neuro-oncologists must now identify risk factors for these complications and, once identified, alter treatment if possible. Radiation Dose The dose of radiation is clearly an important risk factor. Although the exact dose required to produce intellectual deterioration is unknown, a comparison of children with ALL treated with 2400 cGy for CNS prophylaxis and children with brain tumors treated with 3600 to 5500 cGy reveals a clear-cut dose-toxicity relationship. Thus, although leukemic children treated with lower radiation doses develop learning disabilities, they are not nearly as dysfunctional as children treated for brain tumors. Admittedly, it is sometimes difficult to isolate individual risk factors in children with brain tumors because of the complicating factors of the site of the tumor as well as the presence or absence of hydrocephalus or seizures or both. In at least one study, Hirsch compared children with cerebellar astrocytomas and medulloblastomas to determine whether intellectual differences were related to hydrocephalus or postoperative treatment. 45 Both medulloblastomas and cerebellar astrocytomas are located in the posterior fossa and cerebellum and are associated with obstructive hydrocephalus at the level of the fourth ventricle and aqueduct of Sylvius. Hirsch retrospectively compared IQs in these two patient populations. Children with cerebellar astrocytomas had only surgery whereas those with medulloblastomas had postoperative radiatjon as well as chemotherapy. A significant IQ difference existed between the two groups: Children with medulloblastomas performed less well than children with cerebellar astrocytomas. In this study, doses of 3500 cGy to the brain and 5500 cGy to the posterior fossa were associated with more intellectual dysfunction than no radiation, even when such parameters as hydrocephalus and tumor location were equivalent. The actual dose of radiation required to produce endocrine deficits is unknown, although Shalet et al85 have suggested that 2500 cGy is needed to produce growth hormone deficiency. The dose of radiation also correlates with the degree of leukoencephalopathy. 16 Thus, children and adults treated with higher doses of radiation are more likely to develop periventricular hyperintensity than patients who receive lower doses. CHANGES IN THE APPROACH TO CNS TUMORS IN CHILDHOOD 869 Change in Treatment as a Response to Radiation Dose It is likely that if the dose of radiation to the brain could be reduced to less than 2400 cGy, adverse effects on intelligence and endocrine function would be lessened and the incidence of leukoencephalopathy would be lower. A recent POG/CCSG medulloblastoma study compared two different radiation regimens in children greater than 3 years of age considered to have "good risk" medulloblastoma, i.e., patients with gross total or subtotal «1 cm3 residual) tumor resection, no evidence of metastases at diagnosis, and tumor confined to the posterior fossa. Because this subgroup, based on historical controls, had an anticipated 5-year survival rate of 70%, the goal of the study was to determine whether reduced radiation to the brain and spinal cord could provide equivalent survival rates and less neurotoxicity than full dose craniospinal radiation. Results have not been published other than in abstract form, but the study was closed prematurely because of the higher incidence of neuraxis failure in children on the reduced radiation arm.22 Although the study failed, it was the first major foray into altering treatment to reduce neurotoxicity. Future studies are likely to combine lower dose radiation with chemotherapy to accomplish the same end. An even more radical approach has been the recent development of a CCSG/POG study for children with low-grade astrocytomas. In the past, children with low-grade astrocytomas were treated with postoperative local radiation to the brain. Because we now have the ability accurately to follow children with brain tumors using noninvasive studies such as MRI, it has been demonstrated that many patients with untreated low-grade astrocytomas continue to have stable disease for years, without evidence of progression. IS As such, it is uncertain whether radiation is either necessary or advisable, particularly in those children who have had total surgical resection. In this study, children with low-grade astrocytomas that have been totally resected are followed longitudinally without further treatment, and those with residual disease will be randomized to either radiation or observation. This study will provide invaluable information regarding whether radiation can be delayed or eliminated in some of these patients. In another study, radiation is either eliminated or delayed in children with visual pathway tumors. Children with tumors of the visual pathway historically have been either operated, radiated, or both. The advent of CT and MRI has increased our awareness of the number of asymptomatic optic gliomas in children with neurofibromatosis. Without treatment, many of these children may be stable for years even without treatment. Therefore, in the POG study, children are not treated until evidence of progressive disease occurs (M.E. Cohen, MD, personal communication, 1992). Radiation to the midline of the brain, which is necessary in children with chiasmatic gliomas, is associated with significant effects on intelligence and endocrine function and is probably one of the causes of moyamoya disease in some of these children.>9 As such, by altering therapy, quality of life should be measurably improved. Radiation Volume Volume of radiation is also an important risk factor for reduced intelligence, endocrine dysfunction, and leukoencephalopathy. Large-volume radiation is associated with intellectual deterioration. Ellenberg reported a statistically significant difference in IQ scores between children with brain tumors treated with whole-brain radiation compared with children treated with local or no radiation. 31 870 DUFFNER & COHEN Whole-brain radiation is more likely than local radiation to cause endocrinopathies. Because whole-brain radiation includes the hypothalamic pituitary axis, growth hormone deficiency, secondary and tertiary hypothyroidism, and, in some cases, cortisol deficiency can be anticipated. Spinal cord radiation, particularly in young children, is associated with failure of vertebral body growth and subsequent short stature that is unresponsive to exogenous growth hormone treatment. 57 If the volume of radiation includes the spinal cord, primary hypothyroidism is also a likely sequela. Finally, large-volume radiation is implicated as being more often associated with leukoencephalopathy than is local radiation. Changes in Treatment as a Response to Volume of Radiation Two approaches are used to alter treatment as a response to concerns over large-volume radiation. The first has been to try to eliminate neuraxis radiation in some patients. Because ependymomas seed the cerebrospinal fluid, some radiation therapists advocate craniospinal radiation to reduce neuraxis dissemination. Others have thought that, because most patients with posterior fossa ependymomas fail at the primary site, it may be unnecessary to provide craniospinal radiation. The POG recently completed a study in which children with ependymomas were evaluated for site of failure. 55 If the results demonstrate that most patients fail in the primary site, then craniospinal radiation can be eliminated in those patients in whom a metastatic workup at diagnosis is negative. If craniospinal radiation in young children is eliminated, better spinal growth can be anticipated as well as a lower incidence of primary hypothyroidism. Additional results may be less adverse effects on intelligence and a lower incidence of leukoencephalopathy. Unfortunately, limiting radiation to the posterior fossa does not protect against growth hormone deficiency because the radiation port extends to the posterior clinoids and, as such, includes the ventromedian nucleus, site of growth hormone releasing hormone. Another method of reducing volume of radiation has been recommended in children with germinomas. 3 Historically, these tumors have been treated with craniospinal radiation. Because germinomas are extremely sensitive to chemotherapy, it has been suggested that they can be effectively treated by combining chemotherapy and reduced volume of radiation. To date, results have been excellent. Group-wide studies are planned to further test this concept. Methotrexate Treatment with methotrexate is a major risk factor for leukoencephalopathy and associated intellectual deterioration. The route of administration of methotrexate correlates with the degree of deficit. Patients with brain tumors treated with either intraventricular or intrathecal methotrexate are at greatest risk, particularly those patients who have altered CSF dynamics caused by the presence of either obstructive or communicating hydrocephalus. In these cases, administration of intraventricular methotrexate is hazardous because the drug is not cleared at a normal rate. Because toxicity of methotrexate relates both to peak level and to duration of exposure, patients with obstructive hydrocephalus treated with intraventricular methotrexate may develop necrotizing leukoencephalopathy as a consequence of trans ependymal flow! Administration of the drug by the intrathecal route may also cause leukoencephalopathy. In these cases, in which normal CSF dynamics are altered, the methotrexate may back- f CHANGES IN THE APPROACH TO CNS TUMORS IN CHILDHOOD 871 diffuse into the ventricles. Because of delayed reabsorption of the CSF, methotrexate remains exposed to the lining of the ventricles for an inordinate period, permitting transependymal egress and resultant leukoencephalopathy. Although the intrathecal and intraventricular routes are most commonly associated with leukoencephalopathy, high-dose intravenous methotrexate has also been associated with leukoencephalopathy .. Methotrexate has been associated with dementia in the absence of leukoencephalopathy. Even patients who appear to function normally do less well than children treated with radiation alone. In a recent study, children with medulloblastoma who had IQ testing 2 years after treatment with radiation and intrathecal methotrexate' had an p.verage IQ of 7081 compared with an average IQ of 90 in two other studies in which children had received postoperative craniospinal radiation alone. 50. 70 Therefore, although radiation is associated with intellectual dysfunction, the addition of methotrexate may intensify the problem. Change in Treatment as a Response to Methotrexate Methotrexate is an effective agent against certain types of tumors and, via the intrathecal route, may clear the cerebrospinal fluid of malignant cells and bulk tumor. Its severe neurotoxicity has, however, limited its use in recent years. Because the number of active agents in brain tumors is so limited, investigators have been unwilling to give up methotrexate entirely. One attempt to continue to use methotrexate but reduce toxicity was a recent rOG study in which frequent low doses of oral methotrexate were given. Theoretically, this approach would provide good penetration but less neurotoxicity. The study has recently closed and results are not yet available. Concerns about potential neurotoxicity remain, however. If methotrexate is to be used via the intraventricular or intrathecal route, it is necessary to ascertain that CSF dynamics are normal. Radionucleotide CSF studies can determine the flow and reabsorption of spinal fluid over a 24- to 48-hour period. This, in combination with frequent assessment of methotrexate levels, may allow continued use of this agent while limiting the hazards associated with its administration. Young Age Young age at time of cranial irradiation (less than 3-5 years) has been documented as the greatest risk factor for radiation-induced dementia.31. 911, 92 Recent evidence, however, suggests that the majority of infants with brain tumors are abnormal even before treatment, with pretreatment scores in the abnormal range. 63 This finding is hardly surprising, because infant tumors are often extremely extensive and occupy a large percentage of the child's intracranial contents. Whereas recent prospective studies of radiated children with brain tumors have demonstrated that the majority of older children fall into the IQ range of 80 to 90, infants with brain tumors often are frankly retarded. No evidence to date suggests that the young child is at more risk for hypothyroidism or gonadal failure. Adverse effects on growth are, however, more severe in the young child because of the effects of radiation to the spine. Infants who receive radiation to the spine before 1 year of age have a potential loss of ultimate adult height of 10 cm compared with older children who receive radiation to the spine who lose approximately 5 cm. 87 Age less than 3 years has also been more commonly associated with leukoencephalopathy,>o even in children treated with radiation alone. Because 872 DUFFNER & COHEN glial proliferation and myelinization are accelerating at this age, radiation damage to white matter can be severe. Change in Treatment as a Response to Young Age One of the most important risk factors for intellectual deterioration, growth failure, and leukoencephalopathy is young age at the time of radiation. Unlike the risk factors of dose and volume of radiation, in which treatment modifications have occurred solely because of neurotoxicity, changing treatment in young children is of interest not only because of toxic effects, but also because of poor survival rates. Infants and young children have significantly worse survival rates than older children with the same tumor types. 23 , 26, 67, 78 Because infants are often diagnosed late in their course, the tumors are apt to be extremely large and many have already seeded the cerebrospinal fluid by the time of initial diagnosis,2, 5, 21 Diagnostic delays occur because the typical presenting signs and symptoms of brain tumors in infants and young children are relatively nonspecific and are more often thought to represent common pediatric disease than structural disease of the central nervous system (Table 1),5,18,35,39,60,67,89,91,93 Poor prognosis also relates to the limitations on treatment imposed by age, Surgery is the primary treatment modality for patients with brain tumors of all ages, Until recently, surgical mortality was greater than 20% in infants and young children, 1, 39 Infants are extremely susceptible to hypotension and shock caused by blood loss, Even a small amount of blood loss may reflect a significant percentage of the total volume and hence be associated with hypotension and hypothermia,6O With improvements in anesthesia and intensive care monitoring, surgical mortality has dropped considerably in recent years, Nonetheless, morbidity remains significant in infants with extensive tumors, Although aggressive surgical intervention may be associated with improved survival/' 51 it is not curative in children with malignant tumors, Postoperative treatment has traditionally been provided in the form of radiation, Because of fears of eNS toxicity, the dose of radiation is reduced by 10% to 20% and may be insufficient to provide an adequate tumoricidal effect. 21 Unfortunately, even with reduction in the dose of radiation, neurotoxicity is apt to be severe, Because many of the malignant brain tumors in infancy are associated with neuraxis dissemination, radiation must be provided to both the brain and spinal cord,2, 40 As such, adverse effects on intelligence, longitudinal Table 1. REASONS FOR DELAY IN DIAGNOSIS IN INFANTS WITH CENTRAL NERVOUS SYSTEM TUMORS Symptoms Vomiting Nystagmus Hemiparesis FTT Irritability Meningismus Non-Neurologic Etiology GE reflux Formula intolerance Congenital nystagmus Spasmus nutans Cerebral palsy Nonspecific, ? parental deprivation Nonspecific, e,g., colic, psychosocial, sleep disorders Meningitis Neurologic Etiology t ICP Irritation floor 4th ventricle Chiasmatic tumor Hemispheric tumor Diencephalic syndrome t ICP Nonlocalizing sign of mass lesion Incipient herniation Abbreviations: GE = gastroesophageal reflux; ICP = intracranial pressure; FTT = failure to thrive, r CHA0:CES IN THE APPROACH TO CNS TUMORS IN CHILDHOOD 873 vertebral growth, endocrine function, and the development of leukoencephalopathy are to be expected. Nonetheless, until recently, infants with malignant brain tumors were treated with surgery followed by cranial or craniospinal radiation. This treatment was not only relatively ineffective, but the side effects of the radiation were often intolerable. These concerns led to the development of a novel treatment approach for infants with brain tumors. Postoperative chemotherapy was given in an attempt to delay radiation until the child was older and better able to tolerate its effects. In Van Ey's original study, MOPP chemotherapy (nitrogen mustard, vincristine, prednisone, procarbazine) was used in infants with benign and malignant tumors of the brain and spinal cord. 9s This early study was followed in 1985 by a POG pilot in which postoperative cyclophosphamide, vincristine, Cisplatin and VP16 (COPE) were given for 1 to 2 years followed by radiation in infants with malignant brain tumors. Horowitz48 and Loeffler59 also treated infants with malignant tumors with postoperative chemotherapy regimens that were variations on MOPP and COPE. Each of these studies, albeit based on small numbers of patients, demonstrated measurable responses to chemotherapy as well as sustained remissions ranging from 1 to 5 years. Finally, in 1986, the POG opened a study in which children less than 3 years of age with malignant brain tumors were treated with postoperative chemotherapy and delayed radiation. When the study was closed in April of 1990, 198 infants had been enrolled and treated. Although results have not yet been published, it is anticipated that survivals will be as good as historical controls and that the long-term effects of treatment will be less. Whether the use of prolonged chemotherapy in young children has adverse effects on neuropsychological or endocrine function or both remains to be determined. CONCLUSION Current approaches to children with brain tumors are in a state of evolution. The recognition of prognostic factors has allowed treatment to be tailored to both high- and low-risk patients. Moreover, risk factors of neurotoxicity have been identified, and treatment has been altered in some cases to lessen side effects. In the future, the dual goals of improving survival and reducing toxicity will hopefully be achieved. ACKNOWLEDGMENT The authors would like to acknowledge the excellent secretarial support of Mrs. Linda Amaro and Ms. Beverly A. San Filippo. References 1. Albright AL: Brain tumors in neonates, infants and toddlers. Contemporary Neuro- surgery 7:1, 1985 2. Allen Jc, Epstein F: Medulloblastoma and other primary malignant neuroectodermal tumors of the CNS. J Neurosurg 57:446, 1982 3. Allen J, Kim J, Parker R: Neoadjuvant chemotherapy for newly diagnosed germ cell tumors. J Neurosurg 67:65, 1987 4. Allen jC, Rosen G, Mehta BM: Leukoencephalopathy following high-dose IV methotrexate with leucovorin rescue. Cancer Treat Rep 64:1261, 1980 874 DUFFNER & COHEN 5. Ambrosino MM, Hernanz-Schulman M, Genieser NB, et al: Brain tumors in infants less than a year of age. Pediatr Radiol 19:6, 1988 6. Artigas J, Ferszt R, Brock M, et al: The relevance of pathological diagnosis for therapy and outcome of brain stem gliomas: Acta Neurochir Suppl(Wein) 42:166, 1988 7. Asai A, Hoffman H, Hendrick EB, et al: Primary intracranial neoplasms in the first year of life. Childs Nerv Syst 5:230, 1989 8. Baram TZ, Tang R: Atypical spasmus nutans as an initial sign of thalamic neoplasm. Pediatr Neurol 2:375, 1986 9. Bleyer WA, Drake JC, Chabner BA: Neurotoxicity and elevated cerebrospinal-fluid methotrexate concentration in meningeal leukemia. N Engl J Med 289:770, 1973 10. Bloom HJG, Glees J, Bell J: The treatment and long-term prognosis of children with intracranial tumors: A study of 610 cases, 1950-1981. Int J Radiat Oncol Bioi Phys 18:723, 1990 11. Bloom HJG, Wallace ENK, Henk JM: The treatment and prognosis of medulloblastoma in children: A study of 82 verified cases. AJR Am J RoentgenoI105:43, 1969 12. Brauner R, Czernichow P, Rappaport R: Precocious puberty after hypothalamic and pituitary irradiation in younger children. N Engl J Med 311:920, 1984 13. Byrne JV, Kendrall BE, Kingsley OPE, et al: Lesions of the brain stem: Assessment by magnetic resonance imaging. Neuroradiology 31:129, 1989 14. Clayton PE, Shalet SM, Price DA: Gonadal function after chemotherapy and irradiation for childhood malignancies. Horm Res 30:104,1988 15. Cohen ME, Duffner PK, Klein OM: The quandary of longstanding seizures and abnormal neuroimaging. Ann Neurol 24:351, 1988 16. Constine LS, Konski A, Ekholm S, et al: Adverse effects of brain irradiation correlated with MR and CT imaging. Int J Radiat Oncol Bioi Phys 15:319, 1988 17. Copeland DR, Fletcher JM, Pfefferbaum-Levine B, et al: Neuropsychological sequelae of childhood cancer in long-term survivors. Pediatrics 75:745, 1985 18. Cushing H: Experiences with cerebellar medulloblastoma: A critical review. Act Pathol Microbiol Scand 7:1-86, 1930 19. Dacou-Voutetakis C, Xypolyta A, Haidas CM Sr, et al: Irradiation of the head: Immediate effect on growth hormone secretion in children. J Clin Endocrinol Metab 44:791, 1977 20. Davis Pc, Hoffman JC, Pearl GS, et al: CT evaluation of effects of cranial radiation therapy in children. AJR Am J RoentgenoI147:587, 1986 21. Deutsch M: Radiotherapy for primary brain tumors in very young children. Cancer 50:2785, 1982 22. Deutsch M, Thomas P, Boyett J, et al: Low stage medulloblastoma: A Children's Cancer Study Group (CCSG) and Pediatric Oncology Group (POG) randomized study of standard vs. reduced neuraxis irradiation. Proc ASCO #363, 1991, P 124 23. Duffner PK, Cohen ME: Treatment of brain tumors in babies and very young children. Pediatr Neurosci 12:304, 1985-1986 24. Duffner PK, Cohen ME, Anderson S, et al: Long-term effects of treatment on endocrine function in children with brain tumors. Ann NeuroI14:528, 1983 25. Duffner PK, Cohen ME, Brecher ML, et al: Abnormalities of CT scans and altered methotrexate clearance in children with CNS leukemia. Neurology 34:229, 1984 26. Duffner PK, Cohen ME, Myers MH, et al: Survival of children with brain tumors: SEER Program, 1973-1980. Neurology 36:597, 1986 27. Duffner PK, Cohen ME, Parker MS: Prospective intellectual testing in children with brain tumors. Ann NeuroI23:575, 1988 28. Duffner PK, Cohen ME, Thomas P: Late effects of treatment on the intelligence of children with posterior fossa tumors. Cancer 51:233, 1983 29. Duffner PK, Cohen ME, Voorhess MD, et al: Long-term effects of cranial irradiation on endocrine function in children with brain tumors: A prospective study. Cancer 56:2189, 1985 30. Edwards MSB, Wara WM, Urtasun RC, et al: Hyperfractionated radiation therapy for brain-stem glioma: A phase I-II trial. J Neurosurg 70:691, 1989 31. Ellenberg L, McComb JG, Siegel SE, et al: Factors affecting intellectual outcome in pediatric brain tumor patients. Neurosurgery 21:638, 1987 32. Epstein F: Intra-axial tumors of the cervicomedullary junction in children. Concepts in Pediatric Neurosurgery 7:117, 1987 CHANGES IN THE APPROACH TO CNS TUMORS IN CHILDHOOD 875 33. Epstein F, Wisoff JH: Intrinsic brainstem tumors in childhood: Surgical indications. J Neuro Oncol 6:309, 1988 34. Evans AE, Jenkin RD, Sposto R, et al: The treatment of medulloblastoma: Results of a prospective randomized trial of radiation therapy with and without CCNU, vincristine and prednisone. J Neurosurg 72:572, 1990 35. Fessard C: Cerebral tumors in infancy: 66 cIinicoanatomical case studies. Am J Dis Child 115:302, 1968 36. Frank F, Fabrizi AP, Frank-Ricci R, et al: Stereotactic biopsy and treatment of brain stem lesions: Combined study of 33 cases (Bologna-Marseille). Acta Neurochir Suppl 42:177, 1988 37. Freeman CR, Krischer J, Sanford RA, et al: Hyperfractionated radiotherapy in brain stem tumors: Results of a pediatric oncology group study. Int J Radiat Oncol BioI Phys 15:311, 1988 38. Freeman CR, Krischer J, Sanford RA, et al: Hyperfractionated radiation therapy in brain stem tumors: Results of treatment at the 7020 cGy dose level: Pediatric Oncology Group study #8495. Cancer 68:474, 1991 39. Galassi E, Godano U, Cavallo M, et al: Intracranial tumors during the 1st year of life. Childs Nerv Syst 5:288, 1989 40. Geyer R, Levy M, Berger MS, et al: Infants with medulloblastoma: A Single Institution Review of Survival. Neurosurgery 29:707, 1991 41. Giunta F, Grasso G, Marini G, et al: Brain stem expanding lesions: Stereotactic diagnosis and therapeutical approach: Acta Neurochir Suppl 46:86, 1989 42. Goldberg-Stern H, Gadoth N, Stern S, et al: The prognostic significance of glial fibrillary acidic protein staining in medulloblastoma. Cancer 68:568, 1991 43. Halperin C, Wehn SM, Scott JW, et al: Selection of a management strategy for pediatric brainstem tumors. Med Pediatr Oncol 17:116, 1989 44. Harisiadis L, Chang CH: Medulloblastoma in children: A correlation between staging and results of treatment. Int J Radiat Oncol BioI Phys 2:833, 1977 45. Hirsch JF, Reiner D, Czernichow P: Medulloblastoma in childhood. Survival and functional results. Acta Neurochir 48:1, 1979 46. Hood TW, McKeever PE: Stereotactic management of cystic gliomas of the brain stem. Neurosurgery 24:373, 1989 47. Hoppe-Hirsch E, Renier D, LeBouch-Tubiana A, et al: Medulloblastoma in childhood: Progressive intellectual deterioration. Child Nerv Syst 6:60, 1990 48. Horowitz ME, Mulhern RK, Kun LE, et al: Brain tumors in the very young child: Postoperative chemotherapy in combined modality treatment. Cancer 61:428, 1988 49. Hughes EN, Shillito J, Sallan SE: Medulloblastoma at the Joint Center for Radiation Therapy between 1968 and 1984. Cancer 61:1992, 1988 50. Jannoun L, Bloom HJG: Long-term psychological effects in children treated for intracranial tumors. Int J Radiat Oncol BioI Phys 18:747, 1990 51. Jooma R, Hayward RD, Grant DN: Intracranial neoplasms during the first year of life: Analysis of one hundred consecutive cases. Neurosurgery 14:31, 1984 52. Kanev PM, Lefebvre JF, Mauseth RS, et al: Growth hormone deficiency following radiation therapy of primary brain tumors in childhood. J Neurosurg 74:743, 1991 53. Kelly PJ: Volumetric stereotactic surgical resection of intra-axial brain mass lesions. Mayo Clin Proc 63:1186, 1988 54. Kovnar EH, Kellie SJ, Horowitz ME, et al: Preradiation cisplatin and etoposide in the treatment of high-risk medulloblastoma and other malignant embryonal tumors of the central nervous system: A phase II study. J Clin On col 8:330, 1990 55. Kovnar EH, Kun L, Burger P, et al: Patterns of dissemination and recurrence in childhood ependymoma: Preliminary results of Pediatric Oncology Group protocol # 8532. Ann Neurol 30:457, 1991 56. Lannering B, Albertsson-Wikland K: Improved growth response to GH treatment in irradiated children. Acta Paediatr Scand 78:562, 1989 57. Littman P, D' Angio GJ: Radiation therapy in the neonate. Am J Pediatr Hematol Oncol 3:279, 1981 58. Livesey EA, Brook CGD: Thyroid dysfunction after radiotherapy and chemotherapy of brain tumours. Arch Dis Child 64:593, 1989 59. Loeffler JS, Kretschmar CS, Sallan SE, et al: Preradiation chemotherapy for infants 876 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. DUFFNER & COHEN and poor prognosis children with medulloblastoma. Int J Radiat Oncol BioI Phys 15:177, 1988 Matson DD: Intracranial tumors of the first two years of life. West J Surg Obstet Gynecol 72:117, 1964 Meadows AT, Evans AE: Effects of chemotherapy on the central nervous system: A study of parenteral methotrexate in long-term survivors of leukemia and lymphoma in childhood. Cancer 37:1079, 1976 Milstein JM, Geyer JR, Berger MS, et al: Favorable prognosis for brains tern gliomas in neurofibromatosis: J Neuro Oncol 7:367, 1989 Mulhern RK, Horowitz ME, Kounar EH, et al. Neurodevelopmental status of infants and young children treated for brain tumors with preradiation chemotherapy. J Clin Oncol 7:1660, 1989 Nishio S, Fukui M, Tateshi J: Brainstem gliomas: A cliicopathological analysis of 23 histologically proven cases: J Neuro Oncol 6:245, 1988 Oberfield SE, Allen Je, Pollack J, et al: Long-term endocrine sequelae after treatment of medulloblastoma: Prospective study of growth and thyroid function. J Pediatr 108:219, 1986 Ochs JJ, Berger P, Brecher ML, et al: Computed tomography brain scans in children with acute lymphocytic leukemia receiving methotrexate alone as central nervous system prophylaxis. Cancer 45:2274, 1980 Oi S, Kokunai T, Matsumoto S: Congenital brain tumors in Japan (ISPN cooperative study): Specific clinical features in neonates. Childs Nerv Syst 6:86, 1990 Packer RJ, Allen JC, Goldwein JL, et al: Hyperfractionated radiotherapy for children with brainstem gliomas: A pilot study using 7200 cGy. Ann Neurol 27:167, 1990 Packer RJ, Siegel KR, Sutton LN, et al: Efficacy of adjuvant chemotherapy for patients with poor risk medulloblastoma: A preliminary report. Ann Neurol 24:503, 1988 Packer RJ, Sutton LN, Atkins TE, et al: A prospective study of cognitive function in children receiving whole-brain radiotherapy and chemotherapy:2-year results. J Neurosurg 70:707, 1989 Packer RJ, Sutton LN, Rorke LB, et al: Prognostic importance of cellular differentiation in medulloblastoma of childhood. J Neurosurg 61:296, 1984 Packer RJ, Zimmerman RA, Bilaniuk LT: Magnetic resonance imaging in the evaluation of treatment-related central nervous system damage. Cancer 58:635, 1986 Pasqualini T, Diez B, Domene H, et al: Long-term endocrine sequelae after surgery, radiotherapy, and chemotherapy in children with medulloblastoma. Cancer 59:801, 1987 Patterson R, Farr RF: Cerebellar medulloblastoma: Treatment by irradiation of the whole central nervous system. Acta Radiol 39:323, 1953 Peylan-Ramu N, Poplack DG, Pizzo PA, et al: Abnormal CT scans of the brain in asymptomatic children with acute lymphocytic leukemia after prophylactic treatment of the central nervous system with radiation and intrathecal chemotherapy. N Engl J Med 298:815, 1978 Raffel C, McComb JG, Bodner S, et al: Benign brain stem lesions in pediatric patients with neurofibromatosis: Case reports. Neurosurgery 25:959, 1989 Raimondi AJ, Tomita T: Advantages of "total" resection of medulloblastoma and disadvantages of full head post-operative radiation therapy. Child Brain 5:550, 1979 Raimondi AJ, Tomita T: Medulloblastoma in childhood: Comparative results of partial and total resection. Child Brain 5:310-328, 1979 Rajukulasingam K, Cerullo LJ, Raimondi AJ: Childhood moyamoya syndrome: Postradiation pathogenesis. Child Brain 5:467, 1979 Rauhut F, Reinhardt V, Budach V, et al: Intramedullary pilocytic astrocytomas-a clinical and morphological study after combined surgical and photon or neutron therapy. Neurosurg Rev 12: 309, 1989 Riva D, Pantaleoni C, Milani N, et al: Impairment of neuropsychological functions in children with medulloblastomas and astrocytomas in the posterior fossa. Childs Nerv Syst 5:107, 1989 Rowland JH, Glidewell OJ, Sibley RF, et al: Effects of different forms of central nervous system prophylaxis on neuropsychologic function in childhood leukemia. J Clin Oncol 2:1327, 1984 r I I CHANGES IN THE APPROACH TO CNS TUMORS IN CHILDHOOD 877 83. Sanford RA, Freeman CR, Burger P, et al: Prognostic criteria for experimental protocols in pediatric brainstem gliomas. Surg Neurol 30:276, 1988 84. Shalet SM, Beardwell CG, Morris-Jones P, et al: Growth hormone deficiency in children with brain tumors. Cancer 37:1144, 1976 85. Shalet SM, Beardwell CG, Pearson D, et al: The effect of varying doses of cerebral irradiation on growth hormone production in childhood. Clin Endocrinol 5:287, 1976 86. Shalet SM, Clayton PE, Price DA: Growth impairment following treatment for childhood brain tumors: Acta Paediatr Scand (Suppl)343:137, 1988 87. Shalet SM, Gibson B, Swindell R, et al: Effect of spinal irradiation on growth. Arch Dis Child 62:461, 1987 88. Shibamoto Y, Takahashi M, Dokoh S, et al: Radiation therapy for ·brain stem tumor with special reference to CT feature and prognosis correlations: Int J Radiat Oncol Bioi Phys 17:71, 1989 89. Simpson DA, Carter RF, Ducrou W: Intracranial tumours in infancy. Dev Med Child Neurol 10:190, 1968 90. Spun berg JJ, Chang CH, Goldman M, et al: Quality of long-term survival following irradiation for intracranial tumors in children under the age of two. Int J Radiat Oneal Bioi Phys 7:727, 1981 91. Squires RH: Intracranial tumors: Vomiting as a presenting sign. Clin Pediatr 28:351, 1989 92. Suc E, Kalifa C, Brauner R, et al: Brain tumors under the age of three. The price of survival. Acta Neurochir 106:93, 1990 93. Tomita T, McLone DG: Brain tumors during the first twenty-four months of life. Neurosurgery 17:913, 1985 94. Tomita T, McLone DG, Naidich TP: Brain stem gliomas in childhood. J Neuro Oncol 2:117,1984 95. Van Eys J, Cangir A, Coody, et al: MOPP regimen as primary chemotherapy for brain tumors in infants. J Neuro Oncol 3:237, 1985 96. Voorhess M, Brecher M, Glickman AF, et al: Hypothalamic pituitary function in children with acute lymphocytic leukemia following CNS prophylaxis: A retrospective study. Cancer 57:1287, 1986 97. Zimmerman RD, Fleming CA, Lee BCP, et al: Periventricular hyperintensity as seen by magnetic resonance: Prevalence and significance. AJNR 7:13, 1986 Address reprint requests to Patricia K. Duffner, MD Department of Neurology Children's Hospital of Buffalo 219 Bryant Street Buffalo, NY 14222