Mutation Analysis of TMB-High Colorectal Cancer: Insights Into Molecular Pathways and Clinical Implications

Abstract

Colorectal cancer (CRC) is well characterized in terms of genetic mutations and the mechanisms by which they contribute to carcinogenesis. Mutations in APC, TP53, and KRAS are common in CRC, indicating key roles for these genes in tumor development and progression. However, for certain tumors with low frequencies of these mutations that are defined by tumor location and molecular phenotypes, a carcinogenic mechanism dependent on BRAF mutations has been proposed. We here analyzed targeted sequence data linked to clinical information for CRC, focusing on tumors with a high tumor mutation burden (TMB) in order to identify the characteristics of associated mutations, their relations to clinical features, and the mechanisms of carcinogenesis in tumors lacking the major driver oncogenes. Analysis of overall mutation frequencies confirmed that APC, TP53, and KRAS mutations were the most prevalent in our cohort. Compared with other tumors, TMB-high tumors were more frequent on the right side of the colon, had lower KRAS and higher BRAF mutation frequencies as well as a higher microsatellite instability (MSI) score, and showed a greater contribution of a mutational signature associated with MSI. Ranking of variant allele frequencies to identify genes that play a role early in carcinogenesis suggested that mutations in genes related to the DNA damage response (such as ATM and POLE) and to MSI (such as MSH2 and MSH6) may precede BRAF mutations associated with activation of the serrated pathway in TMB-high tumors. Our results thus indicate that TMB-high tumors suggest that mutations of genes related to mismatch repair and the DNA damage response may contribute to activation of the serrated pathway in CRC.

Keywords: CpG island methylator phenotype (CIMP); colorectal cancer; molecular phenotype; precision medicine; tumor mutation burden (TMB).

FIGURE 1

Oncoplot of somatic mutations and CNAs for genes related to the cell cycle or WNT, RTK‐RAS, PI3K, NOTCH, or MMR pathways as well as for frequently mutated genes (HighFreqGenes) in 153 tumor samples. Each cell of the plot is color coded according to the type of mutation. The stacked bar plots above and to the right show the frequency of mutation types for each sample and gene, respectively. In addition, TMB, MSI, and tumor location data are shown for each sample at the bottom. ACC, ascending colon cancer; DCC, descending colon cancer; ICC, ileocolic cancer; RC, rectal cancer; SCC, sigmoid colon cancer; TCC, transverse colon cancer.

FIGURE 2

Distribution, type, and frequency of mutations within KRAS (A), BRAF (B), PIK3CA (C), and PTEN (D) for the 14 TMB‐high samples compared with the 139 TMB‐low samples.

FIGURE 3

Analysis of VAF ranking for mutations according to TMB status in colorectal cancer. (A) Heat map generated by clustering of genes and samples on the basis of VAF ranking. Red or blue coloring of cells in the map indicates genes with a high or low VAF ranking within a sample, respectively. TMB, MSI, and tumor location data for each sample are also shown at the top. With the use of all 153 samples, the highest VAF of somatic mutations that does not exceed tumor purity was assigned as the VAF for each gene in each sample. Genes were ranked by VAF within each sample. (B, C) Plots of the average VAF ranking of each gene versus the number of samples with mutations both for all 153 samples (B) and the 14 TMB‐high samples (C). Gene names are displayed for genes with a high VAF ranking and high mutation count as well as for other genes of interest.

FIGURE 4

Somatic mutation patterns for synonymous and nonsynonymous substitutions in all 153 samples reconstructed with SBS6 (a signature of defective MMR) and SBS5 (a clock‐like signature). (A) Heat map generated by clustering on the basis of the proportions of each signature. Red or blue coloring of each cell in the map indicates higher or lower proportions, respectively. TMB, MSI, and tumor location data for each sample are also shown on the left. (B, C) Box plots for the contributions of SBS6 (B) and SBS5 (C) according to tumor location. *p < 0.01 versus SCC, †p < 0.01 versus RC (Benjamini‐Hochberg–corrected t test).