Inferring modes of evolution from colorectal cancer with residual polyp of origin

// Minsoo Kim 1, 2 , Brooke R. Druliner 3 , Nikolaos Vasmatzis 2 , Taejeong Bae 2 , Nicholas Chia 2 , Alexej Abyzov 1, 2 and Lisa A. Boardman 3 1 Program in Bioinformatics and Computational Biology, University of Minnesota, Minneapolis, MN 55455, USA 2 Department of Health Sciences Research, Center for Individualized Medicine, Mayo Clinic, Rochester, MN 55905, USA 3 Division of Gastroenterology and Hepatology, Mayo Clinic, Rochester, MN 55905, USA Correspondence to: Lisa A. Boardman, email: Boardman.Lisa@mayo.edu Alexej Abyzov, email: Abyzov.Alexej@mayo.edu Keywords: colorectal cancer; adenomatous polyp; cancer evolution; carcinogenesis; mutation Received: July 12, 2017      Accepted: December 20, 2017      Published: December 26, 2017 ABSTRACT Besides the classical evolutionary model of colorectal cancer (CRC) defined by the stepwise accumulation of mutations in which normal epithelium transforms through an intermediary polyp stage to cancer, a few studies have proposed alternative modes of evolution (MOE): early eruptive subclonal expansion, branching of the subclones in parallel evolution, and neutral evolution. However, frequencies of MOEs and their connection to mutational characteristics of cancer remain elusive. In this study, we analyzed patterns of somatic single nucleotide variations (SNVs) and copy number aberrations (CNAs) in CRC with residual polyp of origin from 13 patients in order to determine this relationship. For each MOE we defined an expected pattern with characteristic features of allele frequency distributions for SNVs in cancers and their matching adenomas. From these distinct patterns, we then assigned an MOE to each CRC case and found that stepwise progression was the most common (70% of cases). We found that CRC with the same MOE may exhibit different mutational spectra, suggesting that different mutational mechanisms can result in the same MOE. Inversely, cancers with different MOEs can have the same mutational spectrum, suggesting that the same mutational mechanism can lead to different MOEs. The types of somatic substitutions, distribution of CNAs across genome, and mutated pathways did not correlate with MOEs. As this could be due to small sample size, these relations warrant further investigation. Our study paves the way to connect MOE with clinical and mutational characteristics not only in CRC but also to neoplastic transformation in other cancers.