Abstract
Spinal ependymoma is a rare, often low-grade tumor, for which the clinical course and prognosis has been poorly defined. Classification has relied on histological criteria, as there are few defining molecular mutations, causing controversy in diagnosis and grading. DNA methylation analysis with the brain tumor classifier was used on tumor tissue from 37 patients identified from the Neuro-Oncology Branch Natural History Study and selected Collaborative Ependymoma Research Network (CERN) Tissue Repository. These were histologically diagnosed with non-myxopapillary spinal ependymoma (92% grade II and 8% grade III), with 12 designated “poor performers” (50% male, median age 30 years) defined by a progression free survival of less than two years (median 7 months) with at least one recurrence. Alternatively, the 25 “good performers” (56% male, median age 43 years) had no progression with a follow-up time greater than 5 years (median 84 months). Methylation classification matched to “spinal ependymoma” in 33% of the poor performer cases while 8% matched myxopapillary or RELA-fusion ependymoma and 50% were no-match. For the good performers, 52% agreed with histological diagnosis, 28% matched myxopapillary and 20% were no-match. Interestingly, we noted high-level MYCN amplification by copy number analysis in two (17%) cases, neither of which matched to a defined methylation class, and both of which were in the poor performers group. Overall, methylation results reveal complex biology in conventional spinal cord ependymomas, most evident in an increase of no-match cases in the poor performers group. Of importance, the no-match cases contain evidence of a potentially new diagnostic entity characterized by both a poor prognosis and MYCN amplification. For further elucidation of this entity, a larger cohort of patients with expanded clinical criteria has been identified for further testing.