LGG-10. EPIGENETIC/GENETIC/MORPHOLOGIC ANALYSES REVEAL CLINICAL/PROGNOSTIC INSIGHT OF PEDIATRIC LOW GRADE GLIOMAS

INTRODUCTION: Methylation analysis provides insight into the diagnosis and prognosis of pediatric brain tumors. However, the role of the methylome on pediatric low grade gliomas (PLGG) is still unclear. METHODS: We performed methylome analysis using the Illumina EPIC array combined with pathologic, molecular and outcome data on 153 well annotated PLGG from both St-Jude and SickKids. For BRAFV600E gliomas, high grade gliomas were also included. RESULTS: Hierarchical clustering and t-Distributed Stochastic Neighbor Embedding (tSNE) plots uncovered multiple factors that influence methylation-based clustering of PLGG. Importantly, tumor location and lymphocyte infiltration influence the cluster more than molecular status or pathology. Methylation data results in helpful information to change the clinical management in 2.2% of tumors but classified tumors incorrectly 4.3%. For BRAF-V600E gliomas (n=81), all tumors with CDKN2A deletion were included in a PXA cluster regardless of the pathology. Gene-ontology analysis shows that the genes with highly methylated promoter regions in the PXA cluster are relevant to central nervous system/cell/tissue differentiation/development. Tumors clustering as PXA (DKFZ classifier) had 78% 5-year overall survival (OS). However, this group could be further stratified into 100% OS for those with low grade histology versus 30% for HGG (p