Glioblastoma (GBM) is the most common and aggressive primary brain tumor in adults. It has a poor prognosis with a median survival of about 14 months, although with maximal treatment. Genetic heterogeneity caused by extrachromosomal DNA (ecDNA) is the main barrier to therapeutic advancement. EcDNA is separated during cell division and frequently contains oncogenes like MYC, CDK4, PDGFRA, and EGFR. Although the presence of ecDNA in cancer is well established, the extent to which distinct ecDNA species influence tumor progression in GBM remains unclear due to the complex dynamics of tumor expansion and ecDNA inheritance.
This study aimed to comprehensively characterize ecDNA-driven intratumor heterogeneity in IDH-wildtype (IDHwt) GBM using multi-model experimental methods combined with computational modeling. The researchers sought to find oncogene-containing ecDNA structures, evaluate their transcriptional activity and spatial distribution, and determine how ecDNA dynamics affect heterogeneity and tumor growth. They developed a spatial simulation framework, spatial ecDNA intratumor evolution simulation (SPECIES), which combines computational and experimental analyses to predict ecDNA selection, inheritance, and expansion dynamics in human tumors.
The study analyzed a cohort of 59 consecutive adult patients with surgically resected IDHwt GBM treated at a single UK center from 2017 to 2020. Formalin-fixed paraffin-embedded (FFPE) samples were collected before adjuvant therapy. Tissue microarrays were constructed from 3 distinct tumor regions: the core, infiltrating zone, and leading edge. Whole-genome sequencing (WGS) was performed on available samples, with DNA copy number changes identified using CNVkit, and candidate ecDNA structures reconstructed and visualized with the AmpliconSuite pipeline, AmpliconArchitect, and CycleViz.
Nascent transcriptional activity of ecDNA-harbored oncogenes was detected using an intron-targeted RNAscope method. EGFRvIII protein expression was assessed by immunohistochemistry. DNA FISH was applied to quantify ecDNA copy number per nucleus and evaluate spatial heterogeneity. In parallel, genetically engineered mouse models carrying Myc-ecDNA were studied to validate the oncogenic role of ecDNA. Computationally, the SPECIES framework used a stochastic cellular automaton model to simulate ecDNA inheritance, tumor growth, and spatial heterogeneity. Approximate Bayesian computation (ABC) with rejection sampling was applied to fit simulated distributions to patient-derived single-cell ecDNA FISH data.
The results revealed that ecDNA was highly prevalent in IDHwt GBMs frequently carrying EGFR, PDGFRA, or CDK4 oncogenes. These ecDNA structures were confirmed to be circular, and their presence correlated with highly variable gene expression at the single-cell level, consistent with random segregation during mitosis. DNA FISH demonstrated spatial heterogeneity in tumors, with tumor core enriched for ecDNA-containing cells, while infiltrating margins exhibited more variable ecDNA copy-number distributions.
Engineered Myc-ecDNA in mice validated the functional oncogenicity of these structures, showing that ecDNA contributed to aggressive tumor growth. Computational modeling indicated that random ecDNA inheritance alone could explain much of the observed heterogeneity, but selective pressures further influenced ecDNA expansion. Bayesian inference suggested that selection coefficients varied across oncogenes, with EGFR-ecDNAs often displaying stronger selective advantages compared to PDGFRA. The presence of EGFRvIII-mutated ecDNA provided additional growth advantages, while EGFR wild-type ecDNAs exhibited variable effects. Tumors differed in whether ecDNA emerged early in initiation or accumulated later during progression, suggesting distinct evolutionary trajectories.
This study highlights the role of ecDNA in driving genetic heterogeneity and shaping the dynamic evolution of GBM under selective pressures. It shows that ecDNA-driven oncogenesis is context-dependent, influenced by oncogene variants, mutation, and tumor microenvironmental factors. EcDNA dynamics contribute to the spatial tumor differences between tumor cases and margins, explaining the treatment resistance and recurrence.
The study uses multi-region sequencing, functional validation, imaging, and computational modeling to show that ecDNA contributes to the intratumor diversity by a selective process. These findings highlight ecDNA biology as a promising therapeutic target and establish the SPECIES framework as a powerful tool for modeling ecDNA dynamics.
References: Noorani I, Haughey M, Luebeck J, et al. Extrachromosomal DNA–driven oncogene spatial heterogeneity and evolution in glioblastoma. Cancer Discov. 2025. doi:10.1158/2159-8290.CD-24-1555
