Cancer evolution in its native host: oncogenic selection and epistasis revealed by human tumor genomes

Abstract Somatic mutations contribute to tumorigenesis through mutation, selection, and clonal expansion within the native host. Quantifying oncogenic fitness—the extent to which specific somatic mutations promote tumorigenesis—is vital to understanding cancer initiation, progression, and therapeutic response. Perturbation of tumorigenesis by experimental alteration of oncogenes and tumor suppressors in genetically engineered mouse models has long served to inform human cancer research and such experimental systems have been argued to be necessary for definition of tumorigenic effects and fitness landscapes. However, such experimentally constrained systems do not recapitulate the evolutionary processes through which human cancers arise and are highly affected by interspecific differences in genetic background, physiology, and the environment. Here, we argue that large-scale human tumor genomic datasets can be viewed as repeated natural evolutionary experiments that enable inference of oncogenic potential and selective epistatic interactions within our own species. Using lung adenocarcinoma as an illustrative case, we show that inferred oncogenic potential and selective epistatic interactions vary across both somatic genetic and ecological contexts, including tobacco exposure. These observations highlight the value of evolutionary modeling of human tumor genomic data for accurate and precise characterization of oncogenic fitness landscapes, while situating such analyses within a broader context. Computational analyses can leverage human cancer genomic data as a primary resource for accessing oncogenic fitness landscapes, guiding experimental research, and refining therapeutic strategies. Comparative analyses across species can extend this framework by examining conserved and lineage-specific features of somatic selection in each cancer’s native species, germline genetics, and environment with phylogenetic tools, thereby revealing both conserved and lineage-specific features of mutation, somatic selection, and epistasis.