Colorectal Tumors in Diverse Patient Populations Feature a Spectrum of Somatic Mutational Profiles

Abstract

Admixed populations, including the Hispanic/Latino/a community, are underrepresented in cancer genetic/genomic studies. Leveraging the Latino Colorectal Cancer Consortium (LC3) and other existing datasets, we analyzed whole-exome sequencing data on tumor/normal pairs from 718 individuals with colorectal cancer to map somatic mutational features by ethnicity and genetic similarity. Global proportions of African, Asian, European, and Native American genetic ancestries were estimated using ADMIXTURE. Associations between these proportions and somatic mutational features were examined using logistic regression. APC, TP53, and KRAS were the top three mutated genes across all participants and in the subset of Latino individuals in LC3. In analyses examining recurrently mutated genes, tumors from patients of Latino ethnicity had fewer KRAS and PIK3CA mutations compared with tumors from non-Latino patients. Genetic ancestry overall was associated with CDC27 mutation status, and African genetic ancestry was associated with SMAD2 mutation status. In exome-wide analyses, African genetic ancestry was significantly associated with higher odds of mutation in KNCN and TMEM184B. Native American genetic ancestry was associated with a lower frequency of microsatellite instability-high tumors. The SBS11 mutational signature was associated with Native American genetic ancestry as well as Latino ethnicity. In an independent replication dataset, MSK-IMPACT, estimates of association were largely consistent in direction but nonsignificant. A meta-analysis of LC3 and MSK-IMPACT showed that African genetic ancestry was significantly associated with KRAS mutation status and MSI status. This work facilitates precision medicine initiatives by providing insights into the contribution of genetic ancestry to molecular features of colorectal tumors. Significance: Analysis of tumors from various populations can broadly characterize genomic landscapes and enhance precision medicine strategies.

Figure 1.

Estimated population structure and genetic ancestry for 718 colorectal cancer patients stratified by ethnicity. This figure highlights that genetic variability occurs on a continuum and cannot be captured by the population descriptor of ethnicity, which is a sociopolitical construct. A. Principal component analysis (PCA) for Latino and non-Latino patients with the 1KGP and the PAGE Study reference populations inferred by Eigenstrat. PC1, PC2 and PC3 refer to the first three principal components. B. t-SNE plot for Latino and non-Latino patients with the 1KGP and the PAGE Study reference populations. Each patient is represented by a point, while color represents the continental reference population (African = dark green; Asian = orange; European = blue; Native American = red) and the study participants (black). C. Genetic ancestry proportions in Latino and non-Latino patients. Each patient’s genome is represented by a column partitioned into different colors corresponding to genetic ancestry group (AFR = dark green; ASN = orange; EUR = blue; NAT = red). Patients in each ethnic group are ordered by the most common genetic ancestry component in decreasing order. D. Boxplots show the distribution of each ancestral components in Latino and non-Latino patients. Median genetic ancestry value is represented as a solid line, interquartile range [IQR] as a box, and whiskers extend up to 1.5*IQR from the upper and lower quartiles. Outliers are depicted as solid points.

Figure 2.

Mutational landscape in 718 colorectal cancer patients. A. Percentage of mutations by total number of bases in each gene. Each point in the plot represents a gene. Genes at the top left of the plot (APC, KRAS, TP53) have a higher mutation rate relative to gene length. B. Genes with the highest standardized residual from the robust regression method. Each bar in the plot represents a gene. APC, TP53 and KRAS were the only genes with standardized residual >15. C. Boxplots show the mean proportion of each mutational signature in the overall study sample. Median genetic ancestry value is represented as a solid line, interquartile range [IQR] as a box, and whiskers extend up to 1.5*IQR from the upper and lower quartiles. Outliers are depicted as solid points. D. Mean proportion of the six substitution classes (C>A, C>G, C>T, T>A, T>C, T>G) across study samples are shown. Boxplot lines reflect lower quartile, median, and upper quartile of base change, while extended points represent outliers.