Extrachromosomal DNA Gives Cancer a New Evolutionary Pathway

Studying tumor evolution has traditionally focused on individual cells gaining fitness advantages to outcompete their neighbors. However, this perspective may miss a crucial aspect of tumor resilience, particularly in cancers driven by extrachromosomal DNA (ecDNA). Rather than evolving at the individual level, ecDNA-driven cancers may operate on a population scale, where graded variation in oncogene copy numbers across cells enhances the adaptability of the entire tumor cell population.

Leveraging single cell multi-omics sequencing and machine learning, we demonstrate that cell-to-cell variation in ecDNA copy number scales directly with functional phenotypes, including gene expression, chromatin accessibility, protein levels, cell cycle distributions and growth. For the first time, we show that when this graded variation is disrupted, cells rapidly restore it, far faster than random ecDNA segregation would allow. Time-course experiments and computational simulations suggest that the inheritance of ecDNAs into daughter cells is governed by active, deterministic processes.

Our findings position ecDNA as a key driver of population-level adaptability, challenging the view of ecDNA as a mere byproduct of genomic instability. Instead, ecDNA forms the foundation of selection processes that enhance the tumor’s ability to evolve and survive in dynamic environments. This population-based adaptability makes ecDNA-driven cancers exceptionally resilient, offering new insights into their persistence and therapeutic resistance.