Abstract
BACKGROUND: Somatic mutations commonly accumulate in histologically normal tissues and contribute to cancer development. However, many somatic variants found at high frequencies in normal tissues are also found at high frequencies in cancers arising from the same tissue types. Consequently, the roles of these variants in cancer development remain poorly understood. To address this, we assessed the selective pressures across steps of tumorigenesis, as well as epistasis between driver mutations. METHODS: We employed evolutionary modeling to quantify the strength of selection on somatic variants across two key steps of esophageal development: from organogenesis to clonal histologically normal epithelium, and from there to esophageal squamous-cell carcinoma. Analyzing sequence data from 2171 samples, our first-of-its kind model revealed the stepwise contributions of somatic mutations to increased cellular division and survival and their pairwise epistatic effects on driver mutation selection. RESULTS: NOTCH1 substitutions were strongly selected along the step from organogenesis to clonal histologically normal esophageal tissue, explaining their high prevalence in samples of this tissue. However, for the first time, we show that there is little to no positive selection for NOTCH1, NOTCH2, and FAT1 mutations during progression from clonal histologically normal esophageal tissue to esophageal squamous-cell carcinoma in humans, leading to a conclusion that these substitutions promote clonal growth in normal tissue, but do not drive tumorigenesis from extant clonal tissue growths. Moreover, we provide a somatic genetic basis for this differential role: we demonstrate for the first time that mutations in NOTCH1 exhibit antagonistic epistasis with mutations of well-known tumor suppressor genes TP53 and RB1, reducing selection for these progressive mutations in tumorigenesis. CONCLUSIONS: Our findings demonstrate that early positively selected mutations in NOTCH1 and other genes can shape evolutionary trajectories in ways that ultimately constrain malignant progression. Quantification of step-specific selection and identification of antagonistic epistatic interactions with key tumor suppressors reveals that somatic evolution is often context-dependent-what promotes clonal expansion in normal tissue may later impede growth or survival in tumors. These insights underscore the need for precision strategies that account for the shifting fitness landscape across premalignant and malignant stages, informing early detection, prevention, and therapeutic prioritization.