MB-21 * INTEGRATED GENOMICS ELUCIDATES RELATIVE SPATIAL HOMOGENEITY OF EMBRYONAL BRAIN TUMORS

Abstract

INTRODUCTION: Comprehensive, genome-wide profiling and next-generation based sequencing studies have dramatically improved our understanding of pediatric brain tumor biology in the recent years. However, the vast majority of theses studies are based on the assumption that single biopsies are representative for the entire primary tumor. Intratumor heterogeneity constitutes a common phenomenon previously described in renal cell carcinoma (RCC), breast cancer, and high-grade glioma (HGG). Highly disparate molecular profiles of spatially separated tumor areas within the same tumor may preclude development of personalized, molecularly targeted therapies based on single tumor biopsies. MATERIAL AND METHODS: To address this issue, we conducted multiregion whole exome sequencing, high-resolution DNA copy number analysis (Cytoscan HD), and transcriptional profiling on 39 distinct pediatric and adult tumors with a median of six spatially distant biopsies per tumor (range 4-11). Histological entities included AT/RT (n = 2), HGG (n = 17), medulloblastoma (n = 9), medulloepithelioma (n = 1), and RCC (n = 10). We assessed the degree of intratumor heterogeneity and subgroup affiliation using integrated genomics and unsupervised hierarchical clustering algorithms. RESULTS: Embryonal brain tumors demonstrated highly spatially homogenous transcriptomes. In contrast to adult glioblastoma, we showed that subgroup affiliation was stable in multiregion biopsies from the same medulloblastoma patient. Furthermore, embryonal brain tumors displayed highly similar focal and broad DNA copy number alterations compared to HGG and RCC. Multiregion sequencing further reinforced the relatively higher degree of intratumor homogeneity in embryonal brain tumors. Compared to HGG or RCC, somatic mutations in embryonal tumors were much more likely to be ubiquitous throughout the tumor. CONCLUSIONS: The relative spatial homogeneity of embryonal brain tumors suggests that limited biopsies are representative of the tumor genomics landscape, which has important implications for biological classification and development of targeted therapies for these tumors.