A Comprehensive Study of the Effect on Colorectal Cancer Survival of Common Germline Genetic Variation Previously Linked with Cancer Prognosis

Colorectal cancer is the second leading cause of cancer-related deaths worldwide (1). However, current knowledge on germline genetic influences over colorectal cancer prognosis is sparse. There is evidence that shared germline genetic basis exists across multiple cancer types in several key regulatory pathways of cancer pathogenesis (2) and progression (3). Previous genome-wide association studies (GWAS) have identified a number of genetic loci that might be associated with prognostic outcomes for various cancers. These genetic variants may also influence survival outcomes of patients with colorectal cancer. Here, we report a large population-based study investigating the effects of published GWAS-identified variants associated with cancer prognosis on colorectal cancer survival.

We searched the NHGRI-EBI GWAS Catalog (https://www.ebi.ac.uk/gwas/ accessed in December 2018) to retrieve GWAS-identified variants (P < 5 × 10⁻⁵) associated with survival-related traits for patients of any types of cancer. Patients with colorectal cancer with available information on age at diagnosis, sex, American Joint Committee on Cancer (AJCC) stage, and GWAS data were included from the Study of Colorectal Cancer in Scotland (SOCCS). The MultiCentre Research Ethics committee for Scotland and other committees approved the study and written informed consent was obtained from all participants. Additional details on the study cohort and quality control measures on genotyping have been reported previously (4, 5). Patients with colorectal cancer were prospectively followed up until death or censored on July 1, 2017. We evaluated overall survival (OS) and colorectal cancer–specific survival (CSS) as outcomes. A Cox proportional hazards model was adopted to estimate the effect of each variant (under an additive genetic model) on survival outcomes adjusting for age, sex, and AJCC stage. We also performed stratified analyses by sex, AJCC stage, and tumor site. With the type I error at α < 5 × 10⁻⁴ (a Bonferroni corrected threshold), we estimated the study power for variants of various minor allele frequencies (MAF) and effect sizes using the method proposed by Owzar and colleagues (6).

A total of 5,675 colorectal cancer cases were included in this analysis and their basic characteristics are summarized in Table 1. A total of 128 genetic variants (linkage disequilibrium r² < 0.2) were identified from GWAS Catalog (details presented in Supplementary Table S1) and were included in the analysis. Power calculation indicated a power of at least 0.80 to detect a hazard ratio (HR) of 1.25 for 75% of the included variants (MAF > 0.1). Power estimates with various parameters are presented in Supplementary Table S2. In the overall analysis of all 5,675 patients with colorectal cancer, none of the included variants were significantly associated with either OS or CSS (at P < 0.0005). We observed eight variants (rs17026425, rs17057166, rs6854845, rs1728400, rs17693104, rs202280, rs6797464, and rs823920) with P < 0.05 in the same direction of effects with previous findings (Table 2); of them, three variants (rs17026425, rs17057166, and rs6854845) were previously reported to be associated with rectal cancer survival. In stratified analysis, the variant rs17026425 was statistically significantly associated with OS for male patients with colorectal cancer (HR = 1.37; 95% confidence interval, 1.15–1.62; P = 3.3 × 10⁻⁴). In addition, we observed two variants to be associated at P < 0.05 (rs17026425 and rs6854845) with OS in patients with rectal cancer. No statistically significant associations were found in other stratified analyses (Supplementary Tables S3–S5).

Here, we studied all common variants previously reported to be associated with prognosis in different cancer types. Overall, our results do not support any associations between these variants and survival outcomes for patients with colorectal cancer. There are some suggestive signals that may merit further investigation in even larger datasets. For instance, we report a suggestive effect of rs17026425 in both overall and stratified analysis of patients with rectal cancer, which concords with a previous GWAS (7). Of note, neither our study nor the previous GWAS detected association of this variant with colon cancer survival, indicating that this potential effect may be specific to rectal cancer. The variant is an intron variant of IQ motif containing M (IQCM) gene and is located in the binding region of JUN/JUND transcription factors, which manifest higher expression in colorectal cancer (8). The IQCM gene itself is highly expressed in testis only, making results restricted to males only in our study even more intriguing.

Presented here, the study has sufficient power to detect 75% of previously reported survival variants, but failed to do so. Notably, 90% (19/21) of identified studies (Supplementary Table S1) have sample size below 5,675, which is required to detect effect of genetic variants with MAF of 10% and HR of 1.25, thus suggesting potential false-positive association as well as overestimation of real effects in original studies (winner's curse). Lack of pleiotropic and common effects across different cancers could also be behind the observed results, given the fact that variants reportedly associated with prognosis of other cancers except colorectal cancer showed mostly null effects in SOCCS. Our findings show poor reproducibility of results in the field and a pressing need for collaborative efforts, so as to aggregate larger colorectal cancer cohorts with genotype data to unravel the genetic architecture of colorectal cancer survival.

M.G. Dunlop is a scientific advisory board member at Oxford Cancer Biomarkers. No potential conflicts of interest were disclosed by the other authors.

Conception and design: Y. He, S.M. Farrington, M.G. Dunlop, E. Theodoratou

Development of methodology: M.G. Dunlop

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. He, F.V.N. Din, J.P. Blackmur, P. Vaughan-Shaw, S.M. Farrington, H. Campbell, M.G. Dunlop, E. Theodoratou

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. He, M. Timofeeva, X. Li, J.P. Blackmur, V. Svinti, M.G. Dunlop

Writing, review, and/or revision of the manuscript: Y. He, M. Timofeeva, X. Li, J.P. Blackmur, P. Vaughan-Shaw, S.M. Farrington, H. Campbell, M.G. Dunlop, E. Theodoratou

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S.M. Farrington

Study supervision: H. Campbell, M.G. Dunlop, E. Theodoratou

Other (supervision of statistical analysis): M. Timofeeva

The authors acknowledge the excellent technical support from Stuart Reid and Marion Walker. The authors are grateful to Donna Markie and all those who continue to contribute to recruitment, data collection, and data curation for the Study of Colorectal Cancer in Scotland studies. The authors acknowledge that these studies would not be possible without the patients and surgeons who take part. The authors acknowledge the expert support on sample preparation from the Genetics Core of the Edinburgh Wellcome Trust Clinical Research Facility. This work was supported by CRUK programme grant C348/A18927 (to M.G. Dunlop). It was also supported by funding for the infrastructure and staffing of the Edinburgh CRUK Cancer Research Centre. E. Theodoratou is supported by a CRUK Career Development Fellowship (C31250/A22804). J.P. Blackmur is supported by an Edinburgh Clinical Academic Track-linked CRUK PhD Fellowship. P. Vaughan-Shaw was supported by MRC Clinical Research Training Fellowship (MR/M004007/1), a research fellowship from the Harold Bridges Bequest, and by the Melville Trust for the Care and Cure of Cancer. Y. He and X. Li were supported by the China Scholarship Council. This work received support from COST Action BM1206. This work was also funded by a grant to M.G. Dunlop as Project Leader with the MRC Human Genetics Unit Centre Grant (U127527202 and U127527198 from 1/4/18).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.