Clinical manifestations

Presentation and course

Infratentorial ependymomas arise from the roof, floor, or lateral recesses of the fourth ventricle (17; 14; 53; 34; 35). These tumors tend to be extremely infiltrative, and extension of the tumor caudally through the foramen magnum into the upper cervical cord and into the cerebellar pontine angles is common.

By the time of diagnosis, most posterior fossa ependymomas have resulted in blockage of the third or fourth ventricle and hydrocephalus. Children will most typically have headaches, nausea, and vomiting. However, clinical manifestations of ependymomas are notoriously variable. The duration of symptoms prior to diagnosis usually varies between 1 and 36 months, with the majority of patients having symptoms from 3 to 6 months. As the tumor extends along the floor of the fourth ventricle, it may cause multiple cranial nerve palsies, as well as cerebellar dysfunction. Extension into the cerebellar pontine angle may result in unilateral facial palsy, sensorineural hearing loss, and vestibular dysfunction.

At the time of diagnosis, the most common signs of posterior fossa ependymomas include papilledema and ataxia. Nystagmus is present in 40% to 50% of patients by the time of diagnosis, and other cranial nerve palsies occur in a somewhat lower percentage of patients.

The clinical manifestations of supratentorial tumors are related primarily to the location of the tumor and its degree of aggressiveness. Focal neurologic findings and seizures are common. By the time of diagnosis, the majority of patients have headaches and other signs and symptoms of increased intracranial pressure.

The clinical presentation of subependymomas is different (92). These tumors were initially described as incidental findings found on postmortem examination. A subgroup of patients with subependymomas will develop clinical symptomatology due to obstructive hydrocephalus. This primarily occurs with tumors arising in the septum pellucidum, the fourth ventricle floor, or the walls of the lateral ventriculars. In childhood, tumors may develop that have both classical ependymoma and subependymoma elements. In these cases, the tumor may act aggressively and be clinically indistinguishable from other forms of childhood ependymoma.

Prognosis and complications

Reported prognosis for childhood ependymomas is extremely variable (90; 107; 101; 74; 41; 12; 47; 83; 11; 68). Five-year, disease-free survival rates have ranged from 30% to 80%. Histology has been variably related to outcome, although several series and meta-analysis of published reports suggest that anaplastic histology is at least related to poorer event-free survival (25; 29; 43; 87; 100; 04). Infratentorial tumors carry a poorer prognosis and are often associated with greater anaplasia and less complete resections (04).

Other factors associated with outcome have included invasion of the tumor within the brainstem, age, extent of resection, and molecular subtype. Brainstem invasion in posterior fossa tumors has been associated with decreased survival, although it is unclear whether this is due to biological differences between tumors that invade the brainstem and those that do not, or is instead related to extent of resection (74). Older children are said to have improved survival compared to younger children; however, this finding is confounded by inclusion in some series of children with ependymoblastomas in the younger age group (41). Additional studies have questioned this relationship (15).

Molecular factors having impacts on ependymoma prognosis remain in active study (110; 50; 42; 78; 13; 76; 105). Transcriptional profiling has identified distinct molecular and clinical subgroups (67; 97; 87; 50; 73).

A comprehensive histologic and molecular genetic reassessment of pediatric supratentorial ependymomas demonstrated that the majority (70%+) were RELA-fused ependymomas (now ZFTA), and YAP1-fused tumors occurred but were much less common (76). Non-ZFTA, non-YAP1 tumors occurred more frequently than YAP1 tumors, and a fourth subgroup of ependymal/subependymal mixed tumors with an excellent prognosis was described. The YAP1-MAMLD1 tumors have a different biology than ZFTA fusions and usually excellent outcomes (27; 07). The ZFTA-fusion tumors have a variable prognosis, with a progression-free survival at 5-years of approximately 40%, but an overall survival of over 60%. Better survival is associated with gross total resection and high-dose (> 5400cGy) focal radiotherapy (03). Some investigators report survival rates of greater than 75% in other studies of those with ZFTA-fusion tumors (27; 07).

As regards posterior fossa ependymomas, two molecular subgroups have been identified, and these have been termed posterior fossa ependymoma subtype-A and posterior fossa subtype-B. Group-A ependymomas of the posterior fossa in childhood have been considered to have high rates of disease recurrence, although progression-free and overall survival rates have varied greatly between series with 5-year progression-free survival ranging between 35% and nearly 70%. In most reports, those patients with tumor with associated 1q gain or CDKN2A loss had poorer prognosis, as did children after subtotal resections (68; 105; 63). Total resections and the use of postoperative focal radiotherapy were associated with the best survival rates (68). In contrast, posterior fossa subgroup-B, which occurs more commonly in older children and adolescents, has a 5-year progression-free survival of over 70% and an overall survival near 100% at 5 years (78).

Other forms of ependymoma tend to have better overall survival rates, including subependymomas and myxopapillary ependymomas.

The timing of disease relapse in children with ependymomas is also variable. Although, as in medulloblastomas, the majority of children who relapse will do so within the first 24 to 30 months following diagnosis and treatment, a considerable number of children will also relapse 3 to 5 years following diagnosis.

Clinical vignette

A 5-year-old boy had headaches associated with nausea and vomiting for 3 months. Approximately two weeks before evaluation, his parents noted that he was unsteady and tended to bump into things. One week prior to evaluation, his face had pulled to the left, and his right eye was noted to turn in.

On examination, the child was awake and alert. General physical examination was normal. The child’s cranial nerve examination disclosed bilateral papilledema with normal visual acuity and visual fields. He had a right sixth nerve palsy with nystagmus on right lateral gaze. He had a peripheral right seventh nerve paresis and his hearing was decreased out of the right ear. His other cranial nerves were normal. The child had no clear-cut weakness, but a mild degree of truncal unsteadiness. He had bilateral dysmetria that was greater with the right hand. His walking gait was wide-based and he fell to the right. Reflexes were diffusely increased.

CT scan showed a large mass filling the fourth ventricle and extending to the right. The mass was isodense to normal brain with scattered calcification. After contrast enhancement, there was heterogeneous uptake of dye. MRI better demonstrated the full extent of the lesion; the mass appeared to be plastered to the back of the fourth ventricle with obvious involvement of the right cerebellar peduncle. The fourth ventricle was distorted and there was mild to moderate hydrocephalus.

Biological basis

Etiology and pathogenesis

Ependymomas are believed to arise from the ependymal surface of the brain or the spinal cord. Factors associated with the development of such tumors are unknown. Believed to arise from radial glial cells, supratentorial and infratentorial ependymomas have different genomic, gene expression, and immunohistochemical signatures (99; 97; 98; 06). Neuronal differentiation is seen frequently in supratentorial, but not infratentorial, tumors (06).

Ependymomas are cellular and exhibit relatively low mitotic areas. Histological features include perivascular pseudorosettes and ependymal rosettes. Glial fibrillary acidic protein expression is variable and is usually restricted to the radiating cell processes of the tumor.

Anaplastic ependymomas display histologic evidence of anaplasia and have higher cellular density, variable nuclear atypia, marked mitotic activity, and often, prominent vascular proliferation. A clear cell variant with rounded nuclei, clear halos, and, at times, anaplastic features exists and may be associated with a poorer outcome (26).

Ependymomas may occur in any location within the central nervous system. In childhood, ependymal tumors are more likely to occur in the posterior fossa, whereas in adulthood, a supratentorial site of origin is more common (17; 53; 34; 35). Malignant ependymomas (or anaplastic tumors) are somewhat more common supratentorially than infratentorially. The tumors usually arise within or adjacent to the ependymal linings of the ventricular system or the central canal of the spinal cord. However, at times, especially in the supratentorial space, ependymal tumors are so large that it is difficult to determine their site of origin.

Myxopapillary ependymomas are considered a distinct variant of ependymoma, occurring almost exclusively in the region of the cauda equina.

These tumors are histologically characterized by elongated tumor cells arranged in a perivascular papillary pattern around central cores of mucinous or hyalinized perivascular stroma.

Subependymomas are nodular tumors composed of nests of ependymal cells in a dense glial fibrillary matrix. The majority of subependymomas are well delineated and arise as nodules in the fourth and lateral ventricles.

They are thought to have a different, more benign prognosis than other types of ependymomas that occur in the brain.

Ependymoblastoma was initially considered a subtype of the ependymoma. Histologically, ependymoblastomas are primarily composed of small, undifferentiated cells associated with regions of relatively well-formed ependymal blastic rosettes. The tumor is now classified as an embryonal tumor and is primarily considered a subtype of primitive neuroectodermal tumor of childhood. In the revised World Health Organization classification of CNS tumors, ependymoblastomas are given a separate designation under the general category of embryonal tumors.

The specific molecular genetic changes of pediatric ependymomas are increasingly delineated, although comparative genomic hybridization studies and other molecular genetics techniques such as all karyotyping with microsatellite marker have most often implicated the 6q, 17q, and 22q regions; findings have also been found on chromosomes 13, 16, 19, and 20 (86; 102). Radial glia have been identified as the candidate stem cells of ependymoma (99). A retrospective analysis, in 75 patients, found that elevated human telomere reverse transcriptase expression was related to poorer outcome (97; 84). Gain of 1q25 and overexpression of EGFR have also been related to poorer outcome (67; 87; 84).

Based on transcriptional profiling studies, it is now widely accepted that despite histologic similarities, ependymomas rising from different regions of the brain have distinct gene profiles and signature cellular lineages (62). Supratentorial ependymomas, which are in great part from radial glial lineage, have been demonstrated by mRNA profiling, cytogenetic analysis, and epigenetic analysis to be molecularly different from infratentorial ependymomas (110; 32). The largest subgrouping of pediatric supratentorial ependymomas are characterized by gene fusions involving ZFTA. ZFTA is a transcriptional factor important in NF-kB pathway activation; such tumors comprise approximately 70% of supratentorial ependymomas. The ZFTA fusions result from chromothripsis of chromosome 11q. Such tumors have evidence of NF-kB pathway activation and low rates of mutation other than the fusion. Gain of chromosome 1q occurs in approximately one quarter of cases, and 1q25 copy number gain is a poor prognostic factor (05).

The second supratentorial molecular subtype is the YAP1 subtype. These tumors tend to occur at a younger age and demonstrate fusion of YAP1 with MAMLD1. The tumor has a relatively stable genome and a favorable prognosis.

YAP tumors are uncommon, and non-ZFTA fusions/non-YAP fusions are more common. Mixed ependymomas/subependymomas have an excellent prognosis (76).

With regards to posterior fossa ependymomas in childhood, the most common type is posterior fossa subtype-A. Presenting predominantly in young children, at a median age of 3 years, this tumor has a CpG hypermethylated phenotype (CIMP+). It tends to have low rates of mutations that affect protein structure without recurring mutations and a balanced chromosomal profile. Approximately 25% will have a mutation of chromosome 1. Also, H3K27M mutations have been found in posterior fossa subtype-A tumors and may connote an even more dire outcome (30; 79), and they have been used as a surrogate to designate methylation-based PFA tumors. At the time of recurrence of post-fossa group A tumors, chromosome gains and losses often occur; 1q gain and/or 6 q loss present in approximately one-quarter of children at time of diagnosis increases to over 60% at relapse (18).

Posterior fossa subtype-B is less common in childhood. It does not have the CpG island methylator phenotype (CIMP-, differentiating it from the type A tumor). It has low rates of mutations that affect protein structure and no recurring mutations. Numerous cytogenetic abnormalities are seen, and as noted previously, it has a more favorable prognosis (110; 89; 42; 82; 55; 78).

Using a mouse model of ependymoma, high throughput screening, kinome-wide binding assays, and in vivo efficacy studies identified potential treatments against neural stem cells (08). This work elucidated possible novel approaches and targets for therapy.

Epidemiology

Ependymomas comprise 5% to 10% of all childhood primary CNS tumors (23). Approximately 90% of childhood ependymomas are intracranial, with the remainder arising in the spinal cord or along the conus medullaris. These latter two types of tumors are almost always myxopapillary ependymomas. Two thirds to three quarters of childhood ependymomas arise in the posterior fossa. The majority of posterior fossa ependymomas are histologically benign (grade II). The mean age of diagnosis is 5 to 6 years, although over one half of patients will be less than 5 years of age at diagnosis, and a significant number (up to 25% in some series) occur in children under age two years. As noted, specific molecular subtypes tend to arise at different ages. There is no clear-cut sexual predominance.

Prevention

No means of prevention is known for the development of an ependymoma. There is no clear-cut genetic predilection, although, as in many other types of CNS tumors, there is a suggestion that children with a neurofibromatosis may be at slightly higher risk for the development of ependymomas.

Differential diagnosis

Confusing conditions

The primary clinical distinction from other tumors that arise in a similar region of brain in childhood is separation of posterior fossa ependymomas, including cerebellar astrocytomas, brainstem gliomas, and medulloblastomas. Rarely, other conditions may be confused with an ependymoma of the posterior fossa, including postviral cerebellar ataxia (cerebellitis), drug ingestion, or posterior fossa abscess. Cerebellar astrocytomas are three to four times more common than childhood ependymomas. They tend to present with lateralized cerebellar deficits, followed by truncal unsteadiness, and signs and symptoms of increased intracranial pressure. A subvariety of cerebellar astrocytoma of childhood, the so-called "diffuse" or midline cerebellar astrocytoma, is more difficult to clinically separate from ependymoma because the tumor tends to infiltrate the fourth ventricle floor and is more likely to cause cranial nerve palsies and nystagmus. Medulloblastomas tend to arise in the roof of the fourth ventricle and result in early signs of midline unsteadiness and increased intracranial pressure. Medulloblastomas are also three to four times more common than ependymomas. Brainstem gliomas are more likely to present with multiple cranial nerve deficits, unsteadiness, and long tract signs. When ependymomas infiltrate the brainstem, they may be more difficult to separate from exophytic brainstem gliomas. Neuroradiographic studies can usually separate brainstem gliomas and cerebellar astrocytomas from ependymomas. Distinction between ependymomas and medulloblastomas is often somewhat more difficult.

Supratentorial ependymomas are difficult to distinguish clinically from other tumors that arise in the cerebral cortex. Cerebral gliomas, whether high grade or low grade, are essentially indistinguishable from childhood supratentorial ependymomas.

Diagnostic workup

If a posterior fossa tumor is suspected, initial diagnostic workup usually consists of either a CT scan or an MRI, performed with or without contrast enhancement (53; 94). Both procedures are 95% to 100% reliable in demonstrating the presence of ependymoma. Neuroradiographically, posterior fossa ependymomas seem to arise primarily within the fourth ventricle. Approximately 15% of tumors arise in the cerebellar pontine angle, and 5% to 10% in the cerebellar hemisphere. One half of tumors will involve one or both cerebellar pontine angles at the time of diagnosis. On noncontrast CT, ependymomas usually appear as an isodense, at times heterogeneous, mass with ill-defined borders. They fill an expanded fourth ventricle and extend to 1, or both, foramina of Luschka. Approximately one half of lesions will be calcified. Contrast enhancement is present in 80% to 90% of cases. The contrast enhancement is equally as likely to be homogeneous as it is heterogeneous. Hemorrhage will be present in up to 10% of cases. Small cysts may also be present, but large cysts are uncommon. Occasionally, ependymomas may present as homogeneous hyperdense masses and are indistinguishable from medulloblastomas on CT.

On MR, the infiltrative features of the ependymoma are more clearly delineated.

The tumor is often noted to insinuate itself along the fourth ventricle and extend through ventricular outlets (37). Cisternal and cervical cord involvement is frequently greater. Ependymomas are usually hypointense to isointense on T1-weighted images and hyperintense, as compared to gray matter, on T2-weighted images. As on CT, a variable degree of contrast enhancement is usually present. Nearly one half of patients will have heterogeneous contrast enhancement. On both CT and MRI, hydrocephalus is usually present at the time of diagnosis.

Supratentorial ependymomas tend to share the same CT and MR characteristics as cerebral gliomas (21). These tumors often extend into the ventricles, but may be entirely extraventricular, or only connected to the ventricles at the margin of the ventricular surface. Cysts are common in supratentorial ependymomas and calcification occurs in approximately 40% of cases. Malignant supratentorial ependymomas are more likely to invade surrounding brain parenchyma and cause marked edema.

Following initial diagnosis, staging studies for the extent of disease are usually recommended, primarily for detection of leptomeningeal seeding (51). Although ependymomas are well documented to metastasize along the leptomeninges, the timing and incidence of such spread is poorly documented (53; 96; 34; 35). In other series, less than 10% of children with ependymomas had leptomeningeal disease at the time of diagnosis. Some evidence indicates that children with higher grade lesions are more likely to disseminate the neuraxis early in the course of illness, although this is not well substantiated (71). Children with posterior fossa tumors are somewhat more likely to develop leptomeningeal dissemination than those with supratentorial lesions. However, in both groups of patients, leptomeningeal tumor spread is uncommon until after recurrence, often initially occurring only after multiple recurrences. Spinal magnetic resonance imaging, performed with and without contrast enhancement and either prior to or after surgery, is the imaging procedure of choice for the determination of leptomeningeal disease. Cerebrospinal fluid cytological examination is also usually employed for determination of the extent of disease; however, its yield is low, and evidence for its continued use is lacking (22). Dissemination outside of the CNS is extremely rare, and evaluation for extraneural dissemination at diagnosis is not indicated.

Management

At the present time, management for children with ependymomas remains somewhat controversial. Complete resection is recommended for children with myxopapillary spinal cord ependymomas. In those patients with a complete resection, further therapy is usually not necessary.

For children with posterior fossa ependymoma, present recommendations include initial attempts at gross total resections (101; 96; 83). Even within specific molecular subtypes, total or near total resections are associated with better outcomes, especially in subtype-A (85). This is problematic, as posterior fossa ependymomas are often entwined with multiple cranial nerves and “gross” total resections may result in significant transient and permanent neurologic sequelae.

Although there have been anecdotal reports of long-term disease control after gross total resection alone, most series suggest that radiotherapy is indicated following surgery, independent of the histological grade of the tumors for posterior fossa tumors (71; 68).

Given the low incidence of leptomeningeal disease at the time of diagnosis, treatment with conformal local radiotherapy is indicated for patients who are completely staged by spinal MRI and cerebrospinal fluid cytology and found to have no evidence of leptomeningeal disease spread (33; 72; 71; 68; 105; 44). Disease-free survival rates of 60% to 80% have been reported after conformal radiotherapy in patients with non-aplastic lesions (72; 71; 31). However, staging for evidence of leptomeningeal spread is required prior to using local therapy (83). Doses ranging from 5000 to 6000 cGy have been recommended. As important as the overall dose of radiotherapy is the volume of radiotherapy. Given the locally infiltrative nature of ependymomas, local radiotherapy fields often have to be extensive (including multiple segments of the cervical cord) to encompass the entire tumor volume. Proton beam irradiation has been extensively used (60; 91; 111; 81). One study suggested that children may have cognitive and other neurologic deficits after conventional radiotherapy (111). In a series of 70 children treated with proton beam radiotherapy with both posterior fossa and supratentorial ependymomas, overall survival compared favorably to that reported in the literature in a series of children treated with photon irradiation (60; 61). In addition, those treated with proton radiotherapy apparently maintained intellectual and endocrinologic function (61; 80). However, there has been concern that proton beam irradiation may cause greater focal brain injury than standard radiation, especially in the brain stem (40).

Disease control is poorer in patients with subtotally resected tumors of the posterior fossa. In the one randomized prospective trial that used adjuvant chemotherapy for children with ependymomas, the addition of lomustine and vincristine to local and craniospinal irradiation therapy was not shown to demonstrate a survival advantage when compared to treatment with craniospinal and local irradiation therapy alone (10). A second small randomized trial demonstrated that the chemotherapy treatment regimen used did not affect outcome (88). One study suggested benefit from preirradiation combination chemotherapy in subtotally resected patients (28). Another suggested that adjuvant chemotherapy using an ifosfamide regimen may be of some benefit (75). A potential use of pre-irradiation chemotherapy is to allow a window for re-resection prior to radiotherapy. This has been found to be possible in one study (66).

Optimal management for children with myxopapillary ependymomas remains unclear, although “complete” resection confers the best chance for long-term control. In a meta-analysis, the addition of radiotherapy resulted in better survival than treatment with surgery alone, especially in those patients younger than 20 years of age who underwent a subtotal resection (52). There is no consensus concerning the need for postsurgical radiotherapy for children after total resections. In addition, given the proclivity of myxopapillary ependymomas to disseminate to the leptomeninges at relapse, there is no agreement on the most effective volume of radiotherapy for long-term disease control (52). In one series, the use of radiation therapy seemed to be of benefit in children with multifocal myxopapillary ependymomas (56).

The new molecular understanding of ependymomas is leading to a reassessment of the significance of extent of resection and histology in prognosis and development of management plans (110; 108; 78). Also, novel molecular targets are being identified (110; 95; 55). Integrated in vitro and in vivo high-through screening in mouse models are also being used to identify novel treatment approaches (08).

Given the relatively small numbers of children with supratentorial ependymomas, management is unclear. For totally resected tumors, there have been anecdotal reports that patients can be safely observed without further treatment. Other recommendations have primarily been to use localized radiotherapy for these patients, as long as there is no evidence of leptomeningeal disease at the time of diagnosis (45; 54). Cognitive decline may be seen after focal radiotherapy, especially in those with uncontrolled seizures (54).

Management of infants with ependymomas still remains problematic. Some studies have suggested that infants with ependymomas have a poorer prognosis and are more likely to have leptomeningeal disease at the time of diagnosis (or to develop it early in the course of illness). Because of the reluctance of physicians to use extensive irradiation in young children due to the potential long-term sequelae of such therapy, there have been attempts to use chemotherapy alone, following surgery. There are data to suggest that the combination of cyclophosphamide, cisplatin, VP-16, and vincristine can delay, if not obviate, the need for radiotherapy in some children with ependymomas (19). Different multi-agent postoperative chemotherapy regimens controlled disease, without radiotherapy, in approximately 40% of children younger than three years of age, primarily in those who had undergone a gross total resection (38; 39; 64). It is unclear whether children who respond completely to chemotherapy (as determined by neuroradiographic studies), or those who have no residual disease following surgery and remain disease-free during treatment, require radiotherapy at the end of chemotherapeutic treatment (38).

However, studies have suggested that children as young as one year of age can be “safely” treated with conformal radiotherapy (72; 71; 68; 105). Others have found intellectual deterioration over time even in older patients treated with conformal radiotherapy (106; 111).

At the time of disease recurrence, it has been shown that after first recurrence some patients will experience long-term disease control after treatment with a second resection radiation therapy and possibly chemotherapy (34; 35; 104; 16). For this reason, early detection of recurrence is important, and routine surveillance studies should be part of long-term management (36). Studies suggest that the best chance for long-term survival is after re-irradiation, primarily in patients with locally recurrent disease who undergo a complete reresection (69; 02). Proton beam re-irradiation has been shown to be effective (20). There is some evidence that stereotactic radiosurgery can be effective in some cases (49). In seeming contradiction, there is also retrospective evidence that the use of craniospinal radiotherapy may result in a survival benefit compared to treatment with local radiotherapy (103; 104).

Studies evaluating the efficacy of chemotherapy in children with recurrent ependymomas have demonstrated activity for some chemotherapeutic agents, such as cisplatinum, ifosfamide, cyclophosphamide, VP 16, and vincristine (used singularly or typically in combinations); however, no clear survival advantage has been seen (01).

Outcomes

Studies, both retrospective and prospective, have strongly suggested that for children with ependymomas the extent of resection is the most important predictor of outcome, independent of histologic grade of tumor (101; 96; 12; 47; 83; 88; 43; 72; 69; 68; 104). As noted previously, molecular subtype impacts outcome to a greater extent than histological findings (15; 68; 76; 105). Patients with totally resected tumors, primarily of the posterior fossa, had an overall 5-year, progression-free survival rate of nearly 70%, as compared to 30% to 40% for those patients with partially resected tumors.

The dose and volume of radiotherapy has also been related to outcome (90). In one series, the employment of craniospinal irradiation, along with local radiotherapy, was shown to improve survival for children with posterior fossa ependymomas. However, because the majority of treatment failures occurred at the primary tumor site, it is unclear why craniospinal irradiation improves outcome (33; 70). Patients who received craniospinal irradiation in this series were also somewhat more likely to receive larger local radiation volumes and higher total doses of radiotherapy. Given the tendency of ependymomas to widely infiltrate contiguous areas of brain, especially the upper cervical cord, it is conceivable that much of this improved disease control may be related to better local control by the use of larger volumes of local radiation (33; 93; 71; 68). However, in attempts to decrease toxicity, more focused radiotherapy techniques, such as proton beam, IMRT, and stereotactic fractionated radiotherapy, have been evaluated (109; 20; 46). Given the potential long-term sequelae of craniospinal irradiation, there has been significant reluctance to use extensive whole-brain irradiation, unless there is clear evidence that such treatment improves long-term survival. Hyperfractionated local radiotherapy, with radiotherapy being delivered in 110 cGy fraction, twice daily, to a total dose of 7040 cGy, did not result in better survival rates than lower doses of once-per-day radiotherapy (65).

The prognosis for children with myxopapillary spinal cord (primarily caudae equina) ependymomas is excellent after treatment with surgery alone. Occasionally, patients will develop recurrent disease and, even less frequently, leptomeningeal disease, after resection of myxopapillary tumors (24).