Abstract
Ependymomas in children occur most commonly in the posterior fossa of the brain and cause significant morbidity and mortality. The pathogenesis of these tumors remains obscure as recent whole genome and whole exome sequencings have not yielded recurrent genetic alterations. These tumors are classified into PFA and PFB groups based on abnormal CpG island methylation suggesting that epigenetic alterations may be a significant driver. To gain insights into the epigenetics of childhood ependymomas, we performed mass spectroscopy for histone modifications to discover that H3K27me3 was globally reduced in PFA tumors, similar to that observed in H3K27M diffuse intrinsic pontine gliomas (DIPGs). Despite global reduction in H3K27me3, ChIP-sequencing for H3K27me3 revealed genomic enrichment at several loci important for neurodevelopment. Comparison of DNA methylation and genome-wide H3K27me3 enrichment from both H3K27M mutant DIPGs and PFAs showed many similarities and pointed to factors that regulate radial glial (neuronal stem cells) development. Moreover, radial glia in the developing human posterior fossa were negative for H3K27me3. To assess the clinical utility of this finding, H3K27me3 immunohistochemistry was assessed as positive or negative in a blinded manner in a cohort (n=112) of previously defined childhood PFA (n=72) and PFB tumors (n=40). Next, H3K27me3 immunohistochemistry status in two independent, non-overlapping cohorts of childhood EPN (n=133 and n=98) was determined and risk stratification was assessed retrospectively using univariate and multivariate Cox proportional hazards analyses. H3K27me3 immunohistochemistry showed 99% sensitivity and 100% specificity in delineating PFA from PFB tumors and was sufficient to delineate patients with worse prognosis (P<0.0001) in both cohorts. Together these data suggest that reduction in H3K27me3 is central to the pathogenesis of PFA ependymomas and immunohistochemical evaluation of H3K27me3 is an economic, easily available and readily adaptable molecular surrogate for PFA tumors to accurately classify childhood ependymomas to inform prognosis, treatment and future clinical trials.