Abstract 4889: Integrative multi-omic analysis reveals novel prognostic biological entities in pediatric medulloblastoma

Abstract

Abstract Medulloblastoma is a cerebellar tumor for which relapses are associated with poor survival rates of less than 5%. Molecular heterogeneity and dose-limiting toxicities that occur with standard of care approaches, which include surgical resection, craniospinal irradiation, and chemotherapy, complicate the treatment of both primary and recurrent medulloblastoma due to adverse long-term sequela. While the integration of molecular analyses into the histopathology of pediatric medulloblastomas has changed the way these diseases are diagnosed, classified, and treated, there remains an unmet need to further understand the underpinnings of the disease to improve clinical outcomes and minimize neuronal, cognitive, and hormonal toxicities that occur with standard-of-care therapies. We therefore aimed to investigate novel subgroups of pediatric medulloblastoma as a basis for further defining biological properties and targets for the disease through multi-omic characterization. We derived single-nucleotide variant, copy number variant, expression, alternative splicing, and methylation array data from pediatric medulloblastomas profiled as part of the Children’s Brain Tumor Network (CBTN) and the Open Pediatric Cancer (OpenPedCan) projects. Using non-negative matrix factorization across the five datasets derived from genomic, transcriptomic, and epigenomic profiling, we identified 14 clusters found to be prognostically significant by Kaplan-Meier analysis of overall survival (p < 0.0001). Our model further subdivided sonic hedgehog (SHH) and Group 3/4 tumors, but not Wnt-driven tumors, suggesting that the latter forms a homogeneous molecular subgroup. Among our subgroups, we recapitulated the four classically defined medulloblastoma subtypes known to be associated with mutations in KDM6A, KMT2C, CTNNB1, PTCH1, TP53, KBTBD4, MYC, and MYCN. Cascading biological relationships across the copy number, methylation, and expression domains were observed among novel subgroups, potentially driving migratory, immuno-regulatory, GPCR-related, growth factor-related, and histone methylation biological programs. A comparison of subgroup expression profiles to transcriptional signatures from the Library of Integrated Network-Based Cellular Signatures (LINCS) program identified canonical oncogenic pathways unique to each subgroup as potentially targetable, including, but not limited to, MAPK, VEGFA, inflammatory, fatty acid metabolic, and PI3K/AKT signaling. Our work uncovers novel biological entities in pediatric medulloblastoma potentially driven by higher order mechanisms across the genome, transcriptome, and epigenome, with the possibility of informing novel precision medicine approaches for the disease. Citation Format: Komal Rathi, Varun Kesherwani, Ammar S. Naqvi, Yuankun Zhu, Alex Sickler, Xiaoyan Huang, Bo Zhang, Brian Rood, Adam C. Resnick, Adam A. Kraya. Integrative multi-omic analysis reveals novel prognostic biological entities in pediatric medulloblastoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4889.