ecDNA: unfortunate accident or active player in glioblastoma development?

To address this, we curated and studied a cohort of human glioblastoma samples. Glioblastoma is an aggressive brain cancer with few successful treatments. By sequencing and reconstructing the ecDNA molecules in each sample, we mapped out the ecDNA landscape in these tumours revealing that, overwhelmingly, ecDNA is the dominant mode of focal oncogene amplification in glioblastoma, as opposed to linear focal amplifications on the chromosomes. ecDNA enables these tumours to attain higher, and more diverse, expression levels of important oncogenes such as epidermal growth factor receptor (EGFR). This drove home the challenges in treating glioblastoma, why these are particularly deadly diseases, and possibly why targeted therapies like EGFR inhibitors have so far failed to make meaningful improvements to survival for glioblastoma in clinical trials. 

We wanted to dig further into the nature of ecDNA in glioblastoma, pinpointing the degree to which ecDNA is simply an unfortunate accident, generated during the transition to cancer, or an active player driving cancer development. When Imran presented the rest of the eDyNAmiC team with some impressive multi-region glioblastoma tissue samples, we spotted an opportunity to answer these exact questions. 

We developed a spatial simulation platform called SPECIES (SPatial ECdna Intratumor Evolution Simulation) which enabled us to investigate how underlying evolutionary properties of ecDNA manifest in their spatial patterns of abundance. Here our partnership really came into its own, with Imran’s clinical expertise and Magnus’ mathematical knowledge combining to extract as much information as possible from the data. Using this tool, we reverse-engineered the evolutionary trajectories of each patient tumour, tuning SPECIES to find the evolutionary parameters which produced the best matching spatial ecDNA patterns from the multi-region measurements. 

This endeavour led to some unexpected results. It revealed that glioblastomas with different ecDNA-borne oncogenes evolve along unique paths. One of the most fascinating implications of this was that EGFR-ecDNA, unlike other oncogenic ecDNAs, appear to emerge before the transition to cancer – something never before suggested in glioblastoma. This was a profound prediction: amplifications of EGFR, previously assumed to be chromosomal in nature, have long been speculated to be early events in the genesis of glioblastoma, however our analysis implicates ecDNA as crucial players in the process. These predictions add glioblastoma to the list of other human tumour types, along with oesophageal and breast cancers, in which ecDNA accumulate prior to malignant transformation. To investigate this further, we then demonstrated that ecDNA can accumulate in healthy brain cells of mice, without the need for a tumour, lending further weight to our claims that this phenomenon is possible in the human brain. 

What then triggered the transition to a fully-fledged cancer in these cases? Perhaps by rising to high copy numbers via its unique random mode of inheritance, ecDNA primes cell populations to acquire further driver mutations, promoting the transition to cancer – like fire awaiting a spark. To test this idea, we studied mutations on extrachromosomally amplified EGFR across our cohort, revealing a landscape of EGFR-ecDNA harbouring structural variants (SVs) and point mutations, some of which, like the EGFRvIII SV, are known to be oncogenic. These mutated EGFR-ecDNA never dominated the tumours, instead existing in a mixture with the wild-type EGFR-ecDNAs from which they originated. Further SPECIES analysis supported our “spark” hypothesis, suggesting that these mutations occurred during, or shortly before, the transition to cancer, therefore representing the potential catalyst which drove the cascade towards cancer.  

These findings shed further light on the processes which drive the transformation to glioblastoma and its evolutionary path and may help inform clinical practice. Heterogeneity of EGFRvIII underlies the failure of targeted therapies, and our subtle but crucial observation, that tumours hedge their bets with a mixture of wild-type and mutant EGFR, amplified on agile ecDNA circles, could possibly help to explain the plastic response of glioblastomas to anti-EGFR and other targeted therapies. It will be important to determine whether tumours with and without pre-cancerous ecDNA accumulation respond differently to targeted therapies in clinical trials, and if ecDNA characterisation could be used for molecular stratification. Given the critical role for ecDNA in initiating glioblastoma formation, we speculate that there is potential for further research to enable early intervention and make meaningful improvements to these patients’ outlook. 

After training in theoretical physics, I received my PhD in mathematics from Queen Mary University of London in 2022, in the group of Weini Huang. During this time, I was particularly interested in spatial biology, and its intersection with evolution, and investigated how somatic evolutionary dynamics can be measured from spatial mutation patterns in colorectal cancer and healthy human liver. 

I joined team eDyNAmiC in 2022 as a postdoctoral researcher in the group of Benjamin Werner, where I extended the spatial analysis techniques to study spatial patterns of extrachromosomal DNA in glioblastoma. I am interested in deciphering the patterns of ecDNA evolution, building mathematical descriptions of how ecDNA shapes the evolution of tumours, and how they themselves evolve over time. 

Read the paper in Cancer Discovery

This work was co-led by Benjamin Werner at Queen Mary University of London and Paul Mischel at Stanford University, both part of Cancer Grand Challenges’ team eDyNAmiC, as well as Charles Swanton at The Francis Crick Institute. 

Through Cancer Grand Challenges team eDyNAmiC is funded by Cancer Research UK and the National Cancer Institute in the US with generous support to Cancer Research UK from Emerson Collective and The Kamini and Vindi Banga Family Trust.