111MO - Extrachromosomal DNA drives oncogene spatial heterogeneity in IDH-wildtype glioblastomas

Background

Extrachromosomal DNA (ecDNA) is often observed in solid cancers and linked with poor prognosis. ecDNA tumours escape targeted therapies faster by dynamically changing ecDNA copy numbers. Here, we investigate the structure, contents and spatial-temporal evolution of ecDNA in human IDH-wildtype glioblastomas.

Methods

We analysed 59 newly-diagnosed adult IDH-wildtype glioblastomas prior to chemotherapy or radiation. We conducted whole genome sequencing for focal amplifications, and performed multi-region DNA FISH and nascent RNAscope to identify single cell ecDNA copy number. We developed a computational spatial-temporal model of tumour evolution to decipher patterns of tumour growth.

Results

In the primary cohort, 53% of tumours contained solely ecDNA compared with only 12% solely harbouring intrachromosomal amplifications. In the secondary cohort (PCAWG), ecDNA was identified in 86% of patients. The copy number of oncogenes on ecDNA was higher than that on intrachromosomal amplifications; there was increased diversity of oncogenes amplified on ecDNA compared to intrachromosomal amplifications, both in the variety of oncogenes and cell-to-cell variability in copy number. A subset of ecDNA harboured immunomodulatory genes. Multiple oncogenes were frequently co-amplified on the same ecDNAs, and 14% of tumours harboured multiple different ecDNA species with oncogene amplifications such as EGFR and CDK6. EGFR was the most common oncogene amplified on ecDNA, with the highest copy number (up to 250); amplified EGFR structural variants including vIII were exclusively found on ecDNA, typically combined with wild-type EGFR ecDNAs. Spatial modelling of single-cell ecDNA copy number across regions revealed oncogene specific evolutionary histories: EGFR amplified ecDNAs are under strong positive selection, while PDGFRA amplified ecDNAs are under weak positive or neutral selection. Computational modelling importantly also suggests that EGFR ecDNAs in some patients accumulate early in tumour evolution, prior to clonal expansion.

Conclusions

EcDNA dynamics drive substantial spatial heterogeneity and may provide a framework for interpreting responsiveness to therapy and for clinical trial molecular stratification in glioblasoma.

Editorial acknowledgement

Clinical trial identification

Legal entity responsible for the study

The authors.

Funding

CRUK; AMS; BRC.

Disclosure

All authors have declared no conflicts of interest.