Ependymoma accounts for 8% to 10% of all childhood central nervous system (CNS) tumors, with fewer than 170 new cases diagnosed annually in the United States in children and adults younger than 25 years old. The mean age at the time of diagnosis ranges from 51 to 71 months,[2-5] and 25% to 40% of those diagnosed are younger than age 3. Survival statistics for ependymoma are generally disappointing: The historical 5-year survival estimate is 50% to 64%, and the historical progression-free survival estimate is 23% to 45%.[2,4,7-9] Recurrences are typically local, and the median time to recurrence is 13 to 25 months.[2-4,7,9,10] Approximately 20% of failures involve distant recurrence, and late recurrences are not uncommon.
Ependymoma develops from the neuroepithelial lining of the ventricles of the brain and the central canal of the spinal cord; 90% of tumors are located intracranially, with 30% occurring above the tentorium and 60% below it (Figure 1). Supratentorial ependymoma arises either from the lateral or third ventricle (60%) or from the cerebral hemisphere (40%).[1,11] Infratentorial ependymoma arises from one of three specific sites within the fourth ventricle: the floor (60%), the lateral aspect (30%), or the roof (10%).[12,13] Tumors that arise from the floor of the fourth ventricle may extend through the foramen of Magendie and over the dorsal surface of the spinal cord. Those that arise from the lateral aspect of the fourth ventricle can extend out of the foramen of Luschka and into the cerebellopontine angle and along the anterior aspect of the pons and medulla (Figure 2).
Complete surgical removal of posterior fossa ependymoma arising from the floor or lateral aspect of the fourth ventricle is difficult because these tumors are typically close to the surface of the brainstem and cranial nerves. Fortunately, neuraxis dissemination at the time of diagnosis is rare and occurs in fewer than 7% of patients (Figure 3).
Numerous studies have sought to identify prognostic factors for intracranial ependymoma; most have been single-institution, retrospective reports that span several decades and consequently include numerous advances in neuroimaging, neurosurgery, radiation oncology, chemotherapy, and supportive measures. Surgical resection appears to be the most important prognostic factor.[2-5,7-11,14] In patients with completely resected tumors, the 5-year survival estimate is 67% to 80% and the 5-year progression-free survival estimate is 51% to 75%. Among patients with incompletely resected tumors, the 5-year survival estimate is 22% to 47%, and the 5-year progression-free survival estimate is 0% to 26% (Figure 4).
Age at Diagnosis
Age, at the time of diagnosis, may also be an important prognostic factor. Very young children typically have a poorer outcome.[4,7,15] For children younger than 3 years old at diagnosis, Pollack et al reported a 5-year survival estimate of 22% and a progression-free survival estimate of 12%. In older children, the 5-year survival estimate is 75%, and the progression-free survival estimate is 60%.
Duffner et al reported the experience of the Pediatric Oncology Group (POG) in very young children. For children who were younger than age 3 and had undergone gross total resection, they calculated a 5-year survival rate of 61%, whereas for those who had undergone subtotal resection, the estimate was 30%. The POG study also showed a 63% 5-year survival for young children (aged 24 to 35 months) in whom radiation therapy was delayed for 1 year, but a 26% 5-year survival for infants and very young children (aged 0 to 23 months) in whom radiation therapy was delayed for 2 years. The POG findings suggest that the poor survival estimates frequently reported for very young children are most likely related to the higher incidence of infratentorial tumors, the lower rate of complete resection, and the delay in the administration of radiation therapy.
Historic studies have shown that patients with ependymoma who receive radiation therapy experience a better outcome than those who are not treated with irradiation.[17,18] In addition, one study suggested that the improvement in outcome may be radiation dose-dependent (ie, higher doses may improve outcome). However, no randomized trials have unequivocally demonstrated that improved outcome is caused solely by radiation therapy; other factors such as extent of resection and age at the time of diagnosis also contribute to the outcome.
Histologic Grade of the Tumor
One of the most controversial prognostic factors in childhood ependymoma is the histologic grade of the tumor. Although numerous reports have suggested that patients with differentiated ependymoma achieve a better outcome than do those with anaplastic ependymoma,[11,17,20-25] some investigators believe that histologic grade has no prognostic significance.
We recently reported histologic characterization of tumors and outcome in a contemporary series of 50 patients. In a blinded review of pathology, we determined that histologic grade was significantly related to progression-free survival after irradiation (P < .001). The 2-year event-free survival estimate (± SE) was 32% ± 14% for patients with anaplastic ependymoma and 84% ± 7% for patients with differentiated ependymoma. Statistical significance was maintained when the analysis was adjusted for age (< 3 years), chemotherapy with or without tumor progression before radiation therapy, and extent of resection.
The finding of a poor progression-free survival estimate after irradiation for patients with anaplastic ependymoma parallels the findings from other contemporary series.[19,24] Data from the blinded pathology review at St. Jude also showed that anaplastic ependymoma was more likely to occur in the supratentorial brain (P = .002). Of 12 patients with supratentorial tumor, 6 experienced a recurrence despite gross total resection and irradiation.
Histologic evaluation plays an important role in the design and interpretation of prospective trials and in determining the significance of prognostic factors in the current treatment era (Figure 5). Cooperative multi-institutional protocols will enable us to determine the significance of the various factors that will be used to estimate prognosis and stratify individual treatments.