Epidemiology

Ependymomas are rare neoplasms in the pediatric population (incidence of 0.23 per 100,000 person-years), accounting for only 10% of primary CNS tumors. Ninety percent of pediatric ependymomas are intracranial, and only 10% are located within the spine.4 Ependymomas have been reported to comprise 16 to 30% of pediatric intramedullary spinal cord tumors, whereas astrocytomas and gangliogliomas account for the majority of histological diagnoses.5–7 However, a recent epidemiological study of 1,485 pediatric nervous system tumor cases in Beijing, China, from 2001 to 2005 found ependymomas to be the most common spinal tumor, comprising nearly 20% of all spinal tumors.8 The highest incidence of ependymomas is in the 10- to 15-year age group.9 Although pediatric intramedullary ependymomas can be found throughout the spinal axis, they demonstrate a rostral preference, with cervical cord involvement reported to range from 30 to 45%.10 Myxopapillary ependymomas are a distinct histological subtype that is typically confined to the filum terminale. Although myxopapillary ependymomas comprise only 8 to 12% of pediatric spinal ependymomas, they account for over 90% of lesions in this region.11,12

Historical Context

Important milestones in the identification and treatment of ependymomas are listed in Table 41.1.

Pathology

Ependymomas arise throughout the neuraxis in association with the ependyma or its remnants. These tumors can appear cellular, glial, or epithelial, but all manifest ependymal differentiation, often in the form of perivascular pseudorosettes.13 Occasionally, ependymomas can exhibit true rosettes. Spinal ependymomas usually affect adults, but they are more common in the pediatric population when localized to the filum terminale. Rarely, ependymomas may be found outside the CNS, and such extra-spinal ependymomas are typically found in the sacrococcygeal region in children.14

Pathogenesis

The pathogenesis of ependymomas is unclear; however, Simian viral DNA sequences have previously been found in high frequency (9/16, 56%) within intracranial ependymomas.15 Whether the presence of Simian viral DNA causes malignant transformation or if ependymomas merely provide a favorable environment for viral replication in patients with latent infection has been a hotly contested topic. Currently there is inadequate evidence to support a direct role of Simian virus in human cancers.16 Intramedullary spinal cord ependymomas have previously been found in patients with neurofibromatosis type 2,17 and one study detected mutations of the neurofibromatosis type 2 gene transcript in five of seven sporadic ependymomas.10 The mutations occurred in the region of the transcript that is homologous to previously identified cytoskeletal proteins, resulting in significant truncation of the predicted protein product. This inactivation of the NF2 gene on chromosome 22 appears to be limited to ependymomas specifically found within the spinal cord.18,19

Clinical Presentation

The clinical presentation of pediatric spinal ependymomas is typically a slow and indolent course. On average, patients can have symptoms for several months and even years prior to diagnosis. Typically, patients report dysesthesia correlating to the level of the tumor. Common signs and symptoms in the pediatric population include localized or radicular pain, muscle weakness, sensory changes, gait disturbances, torticollis, hydrocephalus, spastic paraparesis, hyperreflexia, scoliosis, and sphincter disturbance20,21 ( Table 41.3 ).

Imaging

Plain radiographs are able to detect bone abnormalities resulting from ependymoma growth in over 60% of cases.6 Typical findings on radiographs and computed tomography (CT) include scoliosis, canal widening, vertebral body scalloping, and pedicle erosion. Although these imaging modalities do not play a diagnostic role since the advent of magnetic resonance imaging (MRI), they are often a part of the initial imaging evaluation and are still useful for assessing spinal alignment and integrity following surgery.

Computed tomography myelography can reveal an enhancing mass or a blocked spinal canal, and can be of use in localizing a spinal mass to the intradural intramedullary, intradural extramedullary, or extradural compartments. CT myelography typically reveals a centrally located, regularly fusiform cord in the case of intramedullary ependymomas. Ependymomas of the filum typically cause fusiform swelling in the cauda equina and can block contrast enhancement.

Magnetic resonance imaging is the primary imaging modality utilized in assessing spinal ependymomas because of its superior delineation of tumor margins, multiplanar imaging capability, ability to obtain complementary information with T1- and T2-weighted sequences, and avoidance of ionizing radiation.11 Despite the advancement in preoperative evaluation of spinal cord tumors with MRI, a tissue biopsy is still required to definitively establish a histological diagnosis.3 The standard MRI protocol for pediatric spinal ependymomas generally consists of sagittal T1-, T2-, and postgadolinium T1-weighted sequences with axial T2- and postgadolinium T1-weighted images through the lesion.

Spinal ependymomas on MRI are typically contrast-enhancing, homogeneous (occasionally heterogeneous), sausage-shaped masses located centrally in the cord and demonstrate high T2 and low T1 signal intensity. Remote episodes of hemorrhage may cause T2 hypointensity along the borders of the lesion forming a “hemosiderin cap.” Spinal ependymomas typically span three to four vertebral bodies.22 Rostral, caudal, and intratumoral cysts can often be seen accompanying the tumors ( Fig. 41.1 ).

Leptomeningeal dissemination of spinal ependymomas is relatively uncommon in comparison to other primary CNS neoplasms; however, evidence of spread into the cerebrospinal fluid (CSF) is a critical factor in staging, prognosis, and treatment.23 Imaging of leptomeningeal spread within the spine is variable and includes enhancement along the surface of the spinal cord, enhancing foci in the extramedullary intradural space, and nerve root or thecal sac abnormalities ( Table 41.4 ).

Magnetic resonance imaging is obtained postoperatively to determine the surgical approach and extent of tumor involvement, and is repeated periodically to monitor for recurrence. A thin enhancing rim along the resection margin (granulation tissue) begins to form 24 hours after surgery; thus, postoperative imaging should not be delayed. This enhancement along the resection cavity begins to progressively decrease 5 weeks after surgery and resolves within 1 year of surgery. Thus, new or increasing enhancement seen in follow-up MRIs 5 weeks or longer after the surgery merit further investigation, as this finding may indicate tumor recurrence.11