ABSTRACT/PROJECT SUMMARY Extrachromosomal circular DNA (ecDNA) is a structural rearrangement of the genome whereby 1 or more genomic regions breaks from the chromosomes and is circularized, forming an acentric circular DNA sequence 100kbp-10Mbp in length and generally containing multiple genes and regulatory regions. These bodies are found exclusively in tumor cells and are associated with exceptionally malignant cancers. Although ecDNA has been observed in tumor tissue since the 1960s, recent work has highlighted the central role of ecDNA in oncogene amplification and as a contributor to intra- tumoral heterogeneity. Intra-tumoral heterogeneity is one of the leading determinants of therapeutic resistance and treatment failure and one of the main reasons for poor overall survival in cancer patients. However, the functional relevance of ecDNA as a driver of tumor heterogeneity and drug resistance in pediatric cancers has lagged behind advances in adult tumors. We hypothesize that ecDNA is a frequent molecular driver of tumorigenesis, metastasis and resistance to treatment in pediatric cancer. We propose to test this hypothesis by the following specific aims: Aim 1: Establish the genomic, clinical, and molecular properties of ecDNA in medulloblastoma. We will characterize ecDNA structures across a multi-institutional cohort of medulloblastoma patients and patient-derived xenograft models. To better understand molecular mechanisms of ecDNA-driven malignancy, we will perform comprehensive multi-omic sequence analysis of ecDNAs in MB. Aim 2: Characterize intratumoral copy-number heterogeneity conferred by ecDNA at single-cell resolution. Because ecDNAs lack centromeres, a tumor may acquire intratumoral copy-number heterogeneity with every cell division. We will develop methods to detect ecDNA in single cells, and apply them to characterize ecDNA heterogeneity in a paired patient tumor and patient derived xenograft model. Aim 3: Survey the genomic, clinical and molecular properties of ecDNA across pediatric cancer types. We will screen for ecDNA across thousands of patients from 22 pediatric cancer types by deploying computational tools for ecDNA detection to large pediatric cancer genomic data platforms. We will also leverage the high-quality mutation annotations already available for these tumors to find genomic correlates of ecDNA. Upon completion of this project, we will have established the prognostic relevance of ecDNA status across pediatric cancer types, and developed open-source software to decompose ecDNA heterogeneity at the single-cell level.