Chr20q Amplification Defines a Distinct Molecular Subtype of Microsatellite Stable Colorectal Cancer

Abstract

Colorectal cancer is the third leading cause of cancer-related death in the United States. About 15% of colorectal cancers are associated with microsatellite instability (MSI) due to loss of function in the DNA mismatch repair pathway. This subgroup of patients has better survival rates and is more sensitive to immunotherapy. However, it remains unclear whether microsatellite stable (MSS) patients with colorectal cancer can be further stratified into subgroups with differential clinical characteristics. In this study, we analyzed The Cancer Genome Atlas data and found that Chr20q amplification is the most frequent copy number alteration that occurs specifically in colon (46%) and rectum (61%) cancer and is mutually exclusive with MSI. Importantly, MSS patients with Chr20q amplification (MSS-A) were associated with better recurrence-free survival compared with MSS patients without Chr20q amplification (MSS-N; P = 0.03). MSS-A tumors were associated with high level of chromosome instability and low immune infiltrations. In addition, MSS-A and MSS-N tumors were associated with somatic mutations in different driver genes, with high frequencies of mutated TP53 in MSS-A and mutated KRAS and BRAF in MSS-N. Our results suggest that MSS-A and MSS-N represent two subtypes of MSS colorectal cancer, and such stratification may be used to improve therapeutic treatment in an individualized manner. SIGNIFICANCE: This study shows that chromosome 20q amplification occurs predominately in microsatellite-stable colorectal cancer and defines a distinct subtype with good prognosis, high chromosomal instability, distinct mutation profiles, and low immune infiltrations.

Figure 1.. Chr20q is frequently amplified in colon and rectum cancers.

A: Amplification/deletion frequencies of chromosomal bands across 33 cancer types. Red and blue denote amplification and deletion frequency, respectively. B: Chr20q is most frequently amplified in colon and rectum cancers. C: Chr20q amplification rates in different stages of colon and rectum cancer. D: Chr20q amplification rates in MSI and MSS samples. E: A combination of MSI and Chr20q amplification status stratifies colon cancer samples into three subtypes. F: The majority of Chr20q genes are up-regulated in MSS-A versus MSS-N samples. G: Distribution of correlation coefficients between expression level and copy number for Chr20q genes. Genes with correlation coefficients over 0.6 were selected to form a signature. H: The t-scores calculated by comparing Chr20q signature genes with non-Chr20q genes discriminate MSS-A from MSS-N samples. P value is calculated by using two-sided Wilcoxon rank-sum test. I: Receiver Operating Characteristic (ROC) for classifying MSS-A versus MSS-N samples based on inferred Chr20q amplification scores. Plots A-C are based on the TCGA data for 33 cancer types. All the other plots are based on the TCGA COAD data unless otherwise specified.

Figure 2.. Association of Chr20q amplification status with prognosis in colon cancer.

A: Kaplan-Meier plots for MSI, MSS-A and MSS-N patient groups. B: Forest plot for visualizing results from the multivariate Cox regression model. C-F: Survival difference between the MSS-A and MSS-N groups in patient subsets with mutated TP53, (C) wild type TP53, (D) distal colon cancer (E) and proximal colon cancer (F), respectively.

Figure 3.. Association of Chr20q amplification with genomic aberrations.

A-B: MSI, MSS-A and MSS-N subtypes have differential CIN levels represented as fraction of altered genomic regions (A) and aneuploidy score (B). C: Deficiency of the homologous recombination (HR) DNA repair pathway, represented as HR deficiency (HRD) scores, in MSI, MSS-A and MSS-N subtypes. D: MSI, MSS-A and MSS-N subtypes vary in tumor mutation burdens, represented as the total number of nonsynonymous somatic mutations. E: Top 25 most differentially mutated genes between MSS-A and MSS-N. F-H: MSS-A and MSS-N subtypes have significant differential mutation frequencies in TP53 (F), KRAS (G) and BRAF (H). Plots A-E are based on the TCGA COAD data, and F-H are based on the GSE39582 data. P values are calculated from two-sided Wilcoxon rank-sum test (A-D) or Fisher’s exact test (E-H).

Figure 4.. Association of Chr20q amplification with immune infiltration.

A-B: Differential infiltration levels of leukocytes (A) and lymphocytes (B) among MSI, MSS-A and MSS-N subtypes. B-C: Differential infiltration levels of naïve B cell, memory B cell, CD8+ T cell and monocyte among the three subtypes in the TCGA COAD data (C) and the GSE39582 data (D). E: Relative expression levels of immune related genes among the three subtypes. Plots A, B, C and E are based on the TCGA COAD data. P values are calculated by two-sided Wilcoxon rank-sum test.

Figure 5.. Chr20q amplification status is associated with prognosis in rectum cancer.

A: Chr20q-A patients have significantly longer survival than Chr20q-N patients. B: Forest plot for results from multivariate Cox regression model. Note that Chr20q amplification status is significant after adjusting clinical variables. C-E: CIN characterized by aneuploidy score (C), fraction altered (D) and HRD score (E) between Chr20q-A and Chr20q-N patients in TCGA READ dataset. F-G: Immune infiltration levels of leukocytes (F) and lymphocytes (G) between Chr20q-A and Chr20q-N patients. H: Memory B cell, CD8+ T cell and nature killer cell are found to be infiltrated in Chr20q-A and Chr20q-N differently at statistical significance. Plots A and B are based on the GSE87211 data and plots C-H are based on TCGA READ data. All p values for box plots are calculated from two-sided Wilcoxon rank-sum test.

Figure 6.. Drug efficacy difference between MSS-A and MSS-N cell lines.

A: Top 3 drugs more effective in MSS-N cell line. B: Top 3 drugs more effective in MSS-A cell lines. Drug efficacies are represented by natural logarithm of the half maximal inhibitory concentration (ln(IC50)). All p values for box plots are calculated from one-sided Wilcoxon rank-sum test.

Figure 7.. Potential cancer driver genes on Chr20q arm.

Amplification frequencies of 299 genes on Chr20q arm of colon and rectum cancer are shown in top two curves, respectively. Genes are ordered by their positions on Chr20q arm along x axis. Probability of genes affecting cell viability indicated by dependency of selected 37 genes across 19 colorectal cancer cell lines from DepMap are visualized in middle heatmap. Comparison/association analyses of selected 37 genes are visualized in the bottom heatmap. All significant p values were divided into 3 categories as low, medium and high significance. Red colored gene names indicate possible candidate driver genes. These genes are selected based on both significant essentiality from DepMap result and significance in at least 4 of the 6 comparison/association analyses.