Tobacco smoke alters the trajectory of lung adenocarcinoma evolution via effects on somatic selection and epistasis

Tobacco smoke is a known mutagen. However, its physiological effects on the lung may also influence the somatic selective pressures acting on mutations, further shaping cancer evolution. The relative contributions of these mutagenic and physiological effects to oncogenesis have not been quantified, despite their importance in predicting the differential therapeutic effect of targeting variants in lung cancers of smokers and nonsmokers.

We classified 1,722 lung adenocarcinoma (LUAD) sample genomes from The Cancer Genome Atlas and other projects as ever-smoker (ES) or never-smoker (NS) LUAD based on smoking-associated mutational signature attribution or clinical annotation. We then independently calculated background oncogenic mutation rates in the ES- and NS-LUAD groups. Comparing these background rates to observed variant prevalences enabled us to estimate and compare the selective advantages conferred by each mutation in ES- and NS-LUAD. Finally, we quantified pairwise and higher-order epistatic effects by estimating selection for each mutation in specific somatic genotypes.

As expected, background oncogenic mutation rates were gene-specifically elevated in ES-LUAD. However, differences in oncogenic mutation rates between ES- and NS- LUAD were insufficient to explain differences in the prevalence of some mutated genes, implying that such mutations must have further experienced differential somatic selection. In particular, KRAS, KEAP1, and STK11 mutations experienced substantially stronger selection in ES-LUAD, whereas mutations of EGFR, PIK3CA, SMAD4, and other genes were more strongly selected in NS-LUAD. Mechanistically, EGFR mutations were associated with upregulation of genes involved in the epithelial-mesenchymal transition in NS-LUAD, but not in ES-LUAD. Epistasis was pervasive and distinct between the subtypes: ES-LUAD featured more frequent synergy and substantially less antagonism. These patterns entail divergent evolutionary trajectories, with NS-LUAD constrained to fewer, narrower paths and ES-LUAD exploring a broader, more permissive adaptive landscape. Furthermore, we identified higher-order epistasis-systematically examined here for the first time in cancer-manifesting as sub-additive and emergent synergistic interactions that selectively promote trajectories of oncogenesis.

This disambiguation of the mutagenic and selective effects of tobacco smoke reveals how environmental insult can reshape the evolutionary trajectories of LUAD and enables quantitative prediction of treatment vulnerabilities based on smoking status and tumor somatic genotype.