Neutral Theory and the Somatic Evolution of Cancer

Abstract

Kimura's neutral theory argued that positive selection was not responsible for an appreciable fraction of molecular substitutions. Correspondingly, quantitative analysis reveals that the vast majority of substitutions in cancer genomes are not detectably under selection. Insights from the somatic evolution of cancer reveal that beneficial substitutions in cancer constitute a small but important fraction of the molecular variants. The molecular evolution of cancer community will benefit by incorporating the neutral theory of molecular evolution into their understanding and analysis of cancer evolution-and accepting the use of tractable, predictive models, even when there is some evidence that they are not perfect.

Fig. 1.

Frequency of a codon change versus the molecular distance of that change, in 23 cancers and in an organismal data set (lower right) analyzed by McLachlan (1972) and plotted by Kimura (1983). For the 23 cancer types, the frequency is normalized based on the relative frequency we would expect each amino acid substitution given the trinucleotide mutational signature (Rosenthal et al. 2016) of nonrecurrent variants in each tumor type, and the relative frequency of each amino acid in the exome. BLCA: bladder urothelial carcinoma; BRCA: breast invasive carcinoma; CESC: cervical squamous cell carcinoma and endocervical adenocarcinoma; COAD: colon adenocarcinoma; ESCA: esophageal carcinoma; GBM: glioblastoma multiforme; HNSC: head and neck squamous cell carcinoma, broken into HPV⁺ and HPV⁻ tumor samples using the criteria described within Cao et al. (2016); KIRC: kidney renal clear cell carcinoma; LAML: Acute Myeloid Leukemia; LGG: Brain Lower Grade Glioma; LIHC: liver hepatocellular carcinoma; LUAD: lung adenocarcinoma; LUSC: lung squamous cell carcinoma; OV: ovarian serous cystadenocarcinoma; PAAD: pancreatic adenocarcinoma; PRAD: prostate adenocarcinoma; READ: rectum adenocarcinoma; SKCM: skin cutaneous melanoma, broken into primary skin tumors and metastatic skin tumors; STAD: stomach adenocarcinoma; THCA: thyroid carcinoma; UCEC: uterine corpus endometrial carcinoma.

Fig. 2.

Density plots of the selection intensities of synonymous (dashed line) and amino acid replacement (solid line) somatic substitutions across 23 cancers (Cannataro, Gaffney, et al. 2017). The upward bias for nonrecurrent mutations is worse for tumors with lower rates of mutation and cancer types for which fewer tumors have been sequenced, because for those tumor types counts of zero substitutions among all tumors sequenced are not tallied and counts of one substitution among all tumors sequenced tend to be especially higher than the expected number of substitutions. The observed distributions are consistent with this bias (for instance, the UCEC distribution is based on the greatest number of substitutions and exhibits the least apparent bias; the LAML distribution is based on the least number of substitutions and exhibits the greatest apparent bias). This bias is presumably greater for nonsynonymous changes than synonymous changes, because synonymous mutations often can be produced by a higher number of potential nucleotide replacements in a codon, giving them a higher intrinsic mutation rate per codon.