Abstract Molecular variants drive tumorigenesis. The degree by which specific molecular variants drive cell proliferation varies between tissues, is dependent upon mutational background within tissues, and also varies along the progression from normal tissue development through cancer growth. Knowledge regarding the magnitude by which variants drive cell proliferation, quantifiable as the scaled selection coefficient or cancer effect size of variants, throughout the trajectory from healthy tissue into cancer informs early cancer evolution, detection, and diagnosis—and also has direct relevance to clinical trials, targeted therapeutics, and basic research prioritization. However, in order to quantify these scaled selection coefficients one must disentangle the prevalence of fixed substitutions from the rate at which they occur within cells—the mutation rate. These mutation rates also vary between tissues, throughout genomes, and over stages of tissue development and carcinogenesis within the same tissue lineage. Here, we present a novel methodology to quantify scaled selection coefficients in a stepwise manner over several discrete steps in tissue evolution and tumorigenesis. We highlight results from ongoing work in two tissue types: the esophagus and the endometrium. Within the normal esophagus and esophageal squamous cell carcinoma we find that mutation rates vary substantially along the trajectory from embryogenesis through normal tissue and from normal tissue through tumorigenesis. Accordingly, we also find that selective advantages vary between these steps of evolution, with variants of NOTCH1, NOTCH2, and FAT1 having measurable proliferative advantage within normal tissues but no detectable proliferative advantage during tumorigenesis, and variants of PIK3CA, TP53, NFE2L2, and FBXW7 being selected to fix within both normal and tumorigenic tissues. Intriguingly, we complement this analysis with a quantification of pairwise epistasis and find that mutation of NOTCH1 decreases the selective advantage of variants within TP53 and other drivers, informing recent research suggesting NOTCH1 substitutions may actively inhibit tumorigenesis. We similarly deconvolute mutation and selection between embryogenesis through atypical hyperplasia of the endometrium and atypical hyperplasia through Stage 1 endometrial carcinoma and find that variants within CTNNB1 and PTEN have higher selective advantage within hyperplasia, whereas variants within ARID1A and PIK3R1 are more highly selected through Stage 1 tumorigenesis. Overall, we demonstrate how accounting for differences in mutational processes and rates among discrete steps in tumorigenesis and epistatic interactions allows one to determine which variants are contributing to clonal expansion within normal tissues, which variants are contributing to tumorigenesis, which variants are important to both, and in certain instances which variants may impede the progression of a tumor. Citation Format: Vincent L. Cannataro, Kira A. Glasmacher, Mary Summers, Jeffrey D. Mandell, J. Nic Fisk, Mia Jackson, Sem Asmelash, Jeffrey P. Townsend. Unraveling mutation and selection to better understand early cancer evolution [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Cancer Evolution and Data Science: The Next Frontier; 2023 Dec 3-6; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_2):Abstract nr IA010.
Read full abstract