MDB-21. Uncovering Two New Ependymoma Subtypes and Novel Medulloblastoma Pathways Through Transcriptomic Profiling

Abstract Medulloblastoma and ependymoma are common pediatric brain tumors, characterized by considerable molecular and clinical diversity. To create a comprehensive reference for these tumors, we gathered bulk RNA sequencing data from 888 medulloblastoma and 370 ependymoma samples. After correcting for batch effects, normalizing the data, and performing dimensionality reduction, we developed a unified landscape to analyze gene expression, signaling pathways, gene fusions, and copy number alterations. Our findings revealed distinct clustering patterns, such as two major ependymoma subgroups, EPN-E1 and EPN-E2, each with unique molecular features and gene fusions. In medulloblastoma, we achieved precise categorization of Group 3/4 tumors by subtype, and of SHH tumors based on patient age. Both known gene expression patterns and novel pathways identified by our bulk RNA-seq landscape were validated using single-cell transcriptomics in medulloblastoma patients. This landscape offers a critical resource for discovering biomarkers, enhancing diagnoses, and allows the mapping of new patient RNA-seq data onto the reference framework to ensure accurate subtype classification. The landscape is available on Oncoscape, an interactive platform, which supports global access and exploration.