Rare disruptive mutations and their contribution to the heritable risk of colorectal cancer

Abstract

Colorectal cancer (CRC) displays a complex pattern of inheritance. It is postulated that much of the missing heritability of CRC is enshrined in high-impact rare alleles, which are mechanistically and clinically important. In this study, we assay the impact of rare germline mutations on CRC, analysing high-coverage exome sequencing data on 1,006 early-onset familial CRC cases and 1,609 healthy controls, with additional sequencing and array data on up to 5,552 cases and 6,792 controls. We identify highly penetrant rare mutations in 16% of familial CRC. Although the majority of these reside in known genes, we identify POT1, POLE2 and MRE11 as candidate CRC genes. We did not identify any coding low-frequency alleles (1-5%) with moderate effect. Our study clarifies the genetic architecture of CRC and probably discounts the existence of further major high-penetrance susceptibility genes, which individually account for >1% of the familial risk. Our results inform future study design and provide a resource for contextualizing the impact of new CRC genes.

Figure 1. Study design to investigate the contribution of rare disruptive mutations to the heritable risk of CRC.

WES was performed on a highly enriched subset of CRC patients recruited to the NSCCG and a series of UK population controls from the 1958 Birth Cohort. After QC, this Discovery set comprised 1,006 early-onset (diagnosed ≤55 years) familial cases (≥1 first-degree relative with CRC) and 1,609 controls. To test the hypothesis that low-frequency variants confer risk for CRC, we performed a meta-analysis alongside 12,344 UK samples genotyped on the Illumina HumanExome-12v1_A Beadchip. To test the hypothesis that a burden of rare mutations in a gene confers risk for CRC, we performed a burden test on rare (<1% frequency) coding variants and subsequent meta-analysis alongside additional sequencing data from 3,770 UK samples (WGSET). Finally, we performed GSEA to investigate whether the burden of rare variants are overrepresented in a specific biological pathway.

Figure 2. Quantile–quantile plot of the T1 burden test results applying three different variant classifications.

(a) Class 1, disruptive; (b) Class 2, predicted damaging; (c) Class 3, all non-synonymous variants. MLH1, MSH2 and APC genes annotated. Dotted line corresponds to P_T1-value of 8.0 × 10⁻⁷.

Figure 3. Contribution of rare mutations in known predisposition genes to familial CRC applying three different variant classifications.

(1) Class 1, disruptive; (2) Class 2, predicted damaging; (3) Class 3, all non-synonymous variants, and in addition, splice region variants catalogued as pathogenic by InSight or Clinvar.