Unlike medulloblastomas, which occur predominantly in children, ependymomas arise at all ages. Pediatric and adult ependymomas can be distinguished by a predisposition for emergence of tumor in the posterior fossa in patients under the age of 10 years, while spinal ependymomas occur primarily in adults. Furthermore, the ependymoma location is associated with distinct clinical behavior (98). Molecular studies of ependymoma have revealed marked differences in gene expression profiles and have provided insights into the potential cell of origin. Further molecular characterization of ependymoma is expected to lead to a greater understanding of pathogenesis and open avenues to new treatment strategies.
Neuronal markers are associated with supratentorial pediatric ependymomas, but not infratentorial ependymomas, and expression of these markers is correlated with prolonged progression-free survival (9), a known feature of supratentorial ependymomas (100). Distinct gene profiles are found in ependymomas arising in the supratentorial, infratentorial, and spinal compartments. These gene expression profiles are characteristic of radial glia, which are neuronal precursor cells, localized to distinct areas of the CNS. Radial glia also share the same cell surface markers (CD133/Nestin/RC2/BLBP) as ependymoma cancer stem cells, lending further support to the theory that these are in fact the tumor cell of origin (102).
The link between radial glia and myxopapillary ependymoma has not yet been established. However, myxopapillary ependymomas demonstrate a gene expression profile enriched for HOX genes that resembles the gene expression profile of other spinal ependymomas studied, suggesting a common molecular background. Gene expression analysis in myxopapillary ependymoma shows that this tumor is distinct from intracranial ependymoma and that it is uniquely different from ependymoma, glioblastoma, medulloblastoma, pilocytic astrocytoma, and ATRT in its overexpression of the HOXB13 gene (103). HOX genes are transcription factors that control embryonic development and stem cell differentiation (94). The expression of these genes is regulated by chromatin structure and modification. Dysregulated expression of these genes has been demonstrated in multiple tumors, suggesting a potential role in oncogenesis (106). While not unique to myxopapillary ependymoma, overexpression of platelet-derived growth factor receptor alpha (PDGFR-α) confirmed by immunohistochemistry suggests the rational use of receptor tyrosine kinase inhibitors that are currently in clinical use, such as imatinib mesylate, sorafenib, and sunitinib.
Hereditary syndromes associated with ependymomas are multiple endocrine neoplasia type I, Turcot’s syndrome b, and neurofibromatosis type 2 (NF2) (95). Although specific genes have been implicated in these syndromes, the contribution of these genes to ependymoma formation is unknown. A patient with ependymoma demonstrating malignant progression on recurrence had acquired a loss of the wild-type MEN-1 gene located on 11q13, which suggests that loss of MEN-1 could play a role in tumor progression but not necessarily in tumor formation (107). Somatic NF2 mutations have been found predominantly in spinal ependymomas and are also found in other types of cancer such as thyroid cancer, mesothelioma, and melanoma. The NF2 gene is located on chromosome 22q12 and encodes Moesin-Ezrin-Radixin-Like Protein (Merlin), which functions as a scaffolding protein linking cell membrane proteins with the cytoskeleton (109). Loss of function of Merlin is the cause of NF2, with the phenotype characterized by the development of schwannomas, meningiomas, and ependymomas. Merlin has been implicated as essential in embryonic development, and its role as a tumor suppressor is highlighted by the observation that heterozygous Merlin knockout mice (NF2+/-) develop metastatic osteosarcomas, fibrosarcomas, and hepatocellular carcinomas (110). Other genes on chromosome 22 may be involved in ependymoma formation outside the spinal cord as one third of intracranial ependymomas demonstrate losses in the short arm of chromosome 22 (111).
Loss of genomic DNA from chromosome 22 is the most common aberration observed in ependymoma (up to 55%) and appears to be a relatively isolated phenomenon compared to the second most common chromosome loss on chromosome 6, which is associated with multiple other abnormalities (112). In a population of children with ependymomas but no NF2, X-chromosome and chromosome 6 genomic losses were seen at slightly higher rates of 27% and 22%, respectively, compared to chromosome 22 losses (17%) (113). Partial monosomy of chromosome 22 has been observed in familial ependymoma, suggesting the presence of a tumor suppressor gene (114). Thus, further study of the regions of chromosome 22 loss and determination of gene functional consequences may prove fruitful in identifying unique molecular processes involved in the development of chromosome 22-deleted ependymoma. Four underexpressed genes have been mapped to the 12q12.3-q13.3 region: FBX7, which is part of the F-box family of proteins that bind substrates for ubiquitin-mediated proteolysis (115); CBX7, a polycomb protein that downregulates p16Ink4a by histone modification and acts as an oncogene in gastric cancer (117); SBF1, a pseudo-phosphatase that inhibits cell proliferation when expressed in transformed fibroblasts (118); and C22ORF2, discussed below (119). Real-time quantitative PCR, which can be used to determine the relative mRNA levels for specific genes, has been used to assess the deletion of 10 genes mapping to the 12q12.3-q13.33 region with the finding of loss of one of these genes occurring in 38 of 47 ependymomas. The most commonly deleted gene was C22ORF2 (CHIBBY), which is a nuclear protein that can compete with Tcf/Let for binding to beta-catenin, thereby inhibiting canonical Wnt signaling (120). This study also demonstrated that CHIBBY is frequently hypermethylated, suggesting that epigenetic regulation could also be important for its lack of expression and subsequent activated Wnt signaling, an important signal transduction pathway involved in the regulation of multiple cellular processes as already described for medulloblastoma. Similar to some medulloblastomas, activation of Notch signaling is also expected to play a role in aggressive tumor behavior in ependymomas. Chromosomal gain in the terminal part of the long arm of chromosome 9 is found in 54% of ependymomas and is associated with recurrence, patient age over 3 years, and posterior fossa location (121). Overexpression of the oncogene Notch-1 located in this region of chromosome 9 has been reported with associated overexpression of Notch ligands, receptors, and target genes (Hes-1, Hey2, c-Myc) and repression of the Notch repressor Fbxw7. Treatment of self-renewing pediatric ependymoma-derived neurospheres having surface CD133 stem cell marker with γ-secretase inhibitor resulted in reduction in neurosphere formation, bringing Notch pathway inhibition into the spotlight as a promising approach to targeted disruption of ependymoma growth.
Epigenetic gene regulation in ependymomas may play an important role in genome-wide gene expression profile changes promoting multiple tumor properties, and this could explain pathogenesis in tumors demonstrating balanced chromosomal complement as noted by Mack and Taylor (122). Genes known to be methylated in ependymoma are involved in cell cycle control (CDKN2A/B, p14ARF, RB1), apoptosis (DAPK1, CASP8, TSRSF10C and 10D), stress response (BLU, MCJ, TP73), differentiation (RARB), free-radical scavenging (GSTP1), purine metabolism (FHIT), and regulation of the extracellular matrix (TIMP3, THBS1) (115, 123). Also, the most commonly methylated tumor suppressor gene in ependymomas is RASSFA1 (83%), which has RAS-like protein structure functions in ATM-regulated DNA damage response (130), controls the cell cycle by limiting the accumulation of cyclin D1 and restricting exit from G1 (131), and promotes apoptosis by activating mammalian STE20-like kinases 1 and 2 (132).
Comparison of expressed gene set enrichment between recurrent and non-recurrent ependymomas points to the loss of adaptive immune response gene expression in recurrent ependymomas (133). This observation suggests that the ability of individual patients to establish an immune response to transformed ependymal cells may depend highly on the tumor cell expression of immune response genes. The mechanisms involved in host-tumor immune response have not been studied in ependymomas. Recurrent ependymomas have also been observed to frequently carry a greater ratio of partial to whole chromosome imbalances indicative of unbalanced structural rearrangements (134). This chromosomal configuration is expected to result in a higher proportion of gene loss or gain of function due to break-points within or near coding regions. The intrinsic biological behavior of ependymomas can be mapped to the pattern of chromosomal imbalance with lower 5-year survival rates in the group with a low number of imbalances but frequent 1q chromosome gain, compared to tumors with multiple whole chromosome imbalances or balanced chromosomes (134). Interestingly, ependymomas arising in children younger than 3 years of age were all found to harbor a balanced chromosome profile, raising the possibility of a distinct subclass of ependymomas restricted to this age category.
Chromosome 1q25 gain and overexpression of epidermal growth factor receptor gene (located on 7p11.2) are independent markers of poor prognosis in patients with sporadic ependymomas. While 1q gains are associated with the pediatric group of ependymomas, epidermal growth factor receptor (EGFR) overexpression can be found in all age groups (135). Chromosome 1q gain has been observed in both primary and recurrent ependymomas with translocation break points or amplifications mapped to a region spanning 1q11 to 31 (122). It is not known which genes in this chromosome region are involved in ependymoma formation or progression. However, one candidate gene located at 1q25 is the interferon-induced RNase (RNS4) gene, which functions in the antiviral and antiproliferative effects of interferons and may play important roles in cell cycle and apoptosis regulation as well as neuronal differentiation. The mRNA and protein levels of this gene or its methylation status have not been explored in ependymomas to date.
The epidermal growth factor family of receptors includes ErbB1 (EGFR), ErbB2, ErbB3, and ErbB4. ERB2/4 are co-expressed in over 70% of ependymomas and are correlated with enhanced proliferation and poor prognosis. Furthermore, ependymoma growth can be blocked in vitro with an inhibitor of ErbB2 (138). These findings are important because a specific antibody against the ErbB2 receptors (Herceptin) is currently used for the treatment of ErbB2-positive breast cancer. The development of specific epidermal growth factor inhibitors will be of interest not only in the treatment of ependymomas, but also in the treatment of medulloblastomas with ERB3/4 co-expression.