Abstract 7504: Integrated multiomic analysis to infer longitudinal patterns of ecDNA in IDH-wildtype integrated multiomic analysis to infer longitudinal patterns of ecDNA in IDH wild-type glioblastoma

Abstract Extrachromosomal DNA (ecDNA), a prevalent mechanism of gene amplification, drives oncogene overexpression and is associated with poor patient survival. Over 60% of primary glioblastoma (GBM) cases harbor ecDNA-mediated oncogene amplifications. To better understand the functional consequences of ecDNA amplification, we performed a comprehensive analysis of ecDNA distribution in tumors using longitudinally sampled multiomic data from 35 GBM patient specimens. This dataset integrates 68 whole-genome sequencing (WGS) datasets, 243, 459 transcriptomic and chromatin accessibility profiles derived from the same nuclei. We developed a single-nucleus ATAC-seq-based pipeline to classify ecDNA status at single-cell resolution. To validate the robustness of this pipeline, we utilized patient-derived neurospheres (n = 2), generating multiomic datasets and over 100 fluorescent in situ hybridization (FISH) images per cell line for structural validation. This approach enabled the identification of ecDNA-containing regions in each patient sample through WGS data and the inference of single-cell-level copy number variations. In total, we classified the ecDNA status of over 86, 345 malignant cells. Our findings demonstrated that ecDNA copy numbers remain stable during tumor progression, with no significant changes in the proportion of ecDNA-containing cells over time. Notably, we observed that oncogenes amplified on ecDNA are associated with distinct tumor cell states. By applying non-negative matrix factorization and a random forest machine learning algorithm, we extracted gene programs linked to ecDNA that are enriched for cell-cycle regulation and chromatin regulating. These results provide new insights into the regulatory landscape of ecDNA-driven glioblastoma and highlight its potential role in shaping tumor cell identity and progression. Citation Format: Dacheng Zhao, Kevin Anderson, Amit Gujar, Megan Costa, Djamel Nehar-Belaid, Frederick Varn, Elise Courtois, Paul Gabrie, Paul Robson, Hyo-Eun Moon, Anna Golebiewska, Sun Ha Paek, Simone Niclou. Integrated multiomic analysis to infer longitudinal patterns of ecDNA in IDH-wildtype integrated multiomic analysis to infer longitudinal patterns of ecDNA in IDH wild-type glioblastoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 7504.