Integrated Multiregional Analysis Proposing a New Model of Colorectal Cancer Evolution

Abstract

Understanding intratumor heterogeneity is clinically important because it could cause therapeutic failure by fostering evolutionary adaptation. To this end, we profiled the genome and epigenome in multiple regions within each of nine colorectal tumors. Extensive intertumor heterogeneity is observed, from which we inferred the evolutionary history of the tumors. First, clonally shared alterations appeared, in which C>T transitions at CpG site and CpG island hypermethylation were relatively enriched. Correlation between mutation counts and patients' ages suggests that the early-acquired alterations resulted from aging. In the late phase, a parental clone was branched into numerous subclones. Known driver alterations were observed frequently in the early-acquired alterations, but rarely in the late-acquired alterations. Consistently, our computational simulation of the branching evolution suggests that extensive intratumor heterogeneity could be generated by neutral evolution. Collectively, we propose a new model of colorectal cancer evolution, which is useful for understanding and confronting this heterogeneous disease.

Fig 1. An integrated view of ITH in the 9 colorectal tumors.

Multiregional profiles of mutations, CN and methylation alterations were visualized as heat maps. Orange and green bars indicated founder and progressor alterations, respectively. Colored labels for each sample were prepared so that color similarity represents similarity between mutation profiles. For case3, activities of expression signatures were also provided.

Fig 2. Evolutionary trees of the 9 colorectal tumors.

Evolutionary trees inferred from the multiregional mutation profiles have orange trunks, green branches and variously colored leaves, which correspond to founder, progressor mutations and samples, respectively. The leaves were colored based on the color-coding scheme used in Fig 1. Mutation timings of reported driver genes in colorectal cancer were indicated along the trees, and schemas or photos of multiregionally sampled tumors were also provided. Red and blue scales measure tumor size and tree size based on the number of mutations, respectively.

Fig 3. Analysis of genetic ITH.

(A) The number of samples having mutation (black letters), CN gains (red letters) and losses (blues letters) were counted for each of the founder and progressor categories. The APC item counts one sample subjected to focal deletion. (B) Correlation between founder mutation counts and patients’ ages in our 8 cases. Case 9 was excluded due to a hypermutation phenotype. ρ is Spearman’s correlation coefficients. (C) Mutational signatures were calculated from founder (F) and progressor (P) mutations in our 9 cases, and also from clonal and subclonal mutations in non-hypermutated TCGA samples. P-values were calculated by Wilcoxon signed-rank test on the 9 cases. (D) Distribution of cancer cell fraction in which founder, shared and unique mutations occur. P-values were calculated by The Wilcoxon rank-sum test. (E) Correlation between clonal mutation rates and patients’ ages in TCGA non-hypermutated samples. P-values were calculated by The Wilcoxon rank-sum test. (F) Proportion of arm-level gain, loss, focal amplification and deletion was calculated for founder and progressor CN alterations in the 8 cases subjected to CN profiling.

Fig 4. Analysis of epigenetic ITH.

(A) A heat map of multiregional methylation profiles of the 8 cases. (B) Differential contribution of different categories of probe to intratumor and intertumor variance. According to intratumor and intertumor variance, Probes were ranked to obtain each indicated number of top-ranked probes. Probes were categorized based on their genomic positions and enrichment of each category in the top-ranked probes was measured. (C) Proportion of indicated types of methylation alterations was calculated for founder (F) and progressor (P) methylation alterations in the 8 cases. P-values were calculated by Wilcoxon signed-rank test on the 8 cases.

Fig 5. Simulation of cancer evolution.

(A) A simulated tumor. Different colors represent different clones. White rectangles labeled with alphabets indicate regions subjected to multiregional sampling. (B) A simulated single-cell mutation profile matrix. Columns represent 500 cells sampled from the simulated tumor, and the top colored bars label each clone. Rows represent mutated genes and driver genes are indicated by left blue bars. (C) A simulated multiregional mutation profile matrix. VAFs of each gene were calculated for cell subpopulations from the 8 regions indicated in (A). (D and E) Distribution of VAFs (D) and Proportion of driver genes (E) in different categories of mutations. The mutations were obtained from 20 multiregional mutation profile matrices generated by independent simulation trials. In (E), the width of each bar is proportional to the count of each category of mutations. Therefore, the area of each bar is proportional to the count of driver genes that belong to each category of mutations.

Fig 6. Our model of colorectal cancer evolution.

First, founder alterations containing a set of drive alterations are accumulated in the genome and epigenome as a result of aging. After establishment of a parental clone, extensive ITH is generated by neutral evolution, although a few driver alterations are acquired as progressor alterations. Note that this illustration is based on the evolutionary tree of case 3 (Fig 2). However, an actual tumor should harbor numerous subclones, as suggested by the local ITH data (Fig 5D) and simulated single-cell mutation profile (Fig 5B).