Skip to main content
Download PDF

Associations of single nucleotide polymorphisms with mucinous colorectal cancer: genome-wide common variant and gene-based rare variant analyses

Biomarker Research volume 6, Article number: 17 (2018) Cite this article

Abstract

Background

Colorectal cancer has significant impact on individuals and healthcare systems. Many genes have been identified to influence its pathogenesis. However, the genetic basis of mucinous tumor histology, an aggressive subtype of colorectal cancer, is currently not well-known. This study aimed to identify common and rare genetic variations that are associated with the mucinous tumor phenotype.

Methods

Genome-wide single nucleotide polymorphism (SNP) data was investigated in a colorectal cancer patient cohort (n = 505). Association analyses were performed for 729,373 common SNPs and 275,645 rare SNPs. Common SNP association analysis was performed using univariable and multivariable logistic regression under different genetic models. Rare-variant association analysis was performed using a multi-marker test.

Results

No associations reached the traditional genome-wide significance. However, promising genetic associations were identified. The identified common SNPs significantly improved the discriminatory accuracy of the model for mucinous tumor phenotype. Specifically, the area under the receiver operating characteristic curve increased from 0.703 (95% CI: 0.634–0.773) to 0.916 (95% CI: 0.873–0.960) when considering the most significant SNPs. Additionally, the rare variant analysis identified a number of genetic regions that potentially contain causal rare variants associated with the mucinous tumor phenotype.

Conclusions

This is the first study applying both common and rare variant analyses to identify genetic associations with mucinous tumor phenotype using a genome-wide genotype data. Our results suggested novel associations with mucinous tumors. Once confirmed, these results will not only help us understand the biological basis of mucinous histology, but may also help develop targeted treatment options for mucinous tumors.

Background

Colorectal cancer is a global health problem and contributes substantially to worldwide cancer mortality [1]. In 2012, this disease was the 3rd most common cancer worldwide with higher rates occurring in developed countries [1]. In Canada, colorectal cancer is expected to cause 26,800 new cases and 9400 deaths in 2017. Newfoundland and Labrador, in particular, have the highest age-standardized rates of incidence and mortality in the country [2].

Mucins are a family of high-molecular-weight glycoproteins that are widely expressed by epithelial tissues [3]. According to the HGNC database [4], there are 22 members in this family that can be expressed in various tissues. They have been identified in two forms: cell surface (transmembrane), such as MUC1 and MUC4, and fully released (gel-forming) [3, 5, 6]. The gel-forming mucin-encoding genes are clustered at chromosome 11p15.5 [5, 7, 8]. These mucins, including MUC2, MUC5AC, MUC5B, and MUC6, constitute the major macromolecular components of mucus [5, 7, 9]. Among them, MUC2 is the most highly expressed one in the colorectum and is the predominant component of colorectal mucus [10,11,12]. MUC5B and MUC6 are highly expressed in the upper gastrointestinal (GI) tract, but low levels of both have been reported in the normal colon [12, 13]. MUC5AC is highly expressed in the upper GI tract and is not expressed in the normal colon, however, abnormal expression is observed in colorectal cancer [14,15,16].

Mucinous adenocarcinoma is a distinct form of colorectal cancer with the defining characteristic of a high mucin component (more than 50% of the tumor volume). This subtype accounts for 5–15% of colorectal cancer cases. Compared to non-mucinous colorectal cancer, mucinous adenocarcinoma patients are typically younger and are often at an advanced stage at diagnosis [17,18,19,20,21,22,23]. Mucinous tumors are more likely to occur in the proximal colon [20, 21, 24, 25] and tend to have an inferior response to systemic therapies [25, 26].

Specific molecular distinctions are also seen in mucinous compared to non-mucinous colorectal tumors, for example, increased rates of BRAF mutations and CpG island methylator phenotype (CIMP) [27]. In addition, overexpression of MUC2, strong ectopic expression of gastric MUC5AC, and decreased p53 expression in mucinous tumors are reported in the literature [28, 29]. Mucinous and non-mucinous tumors also appear to have differences in genome-wide gene expression patterns [23]. Some of the upregulated genes in mucinous tumors are involved in cellular differentiation and mucin metabolism, which are characteristics biologically relevant to the phenotype [23]. While the differences between mucinous and non-mucinous colorectal cancers are well recognized, the prognostic importance of a high mucin component has been controversial [19,20,21, 25, 26, 30,31,32,33,34,35].

Most studies investigating characteristics of mucinous colorectal tumors examined single or a limited number of candidate genes [10, 36, 37]. This study aimed to comprehensively identify common and rare genetic polymorphisms that may be influencing the production of mucin or formation of the mucinous tumor phenotype. To do so, we applied a genome-wide approach to identify genes and genetic regions that are associated with the risk of developing the mucinous tumor phenotype.

Methods

Patient cohort

The study cohort was a subgroup of the Newfoundland Colorectal Cancer Registry (NFCCR) and consisted of 505 Caucasian patients. Both the NFCCR and the study cohort were described in detail in other publications [38, 39]. In short, the NFCCR recruited 750 colorectal cancer patients in Newfoundland and Labrador collected between 1999 and 2003. All diagnoses were confirmed by pathological examination. Out of 750 patients, 505 patients constituted the study cohort as explained below.

Genotype data

The genotype data used in this study was explained in Xu et al. (2015) [39]. In short, DNA samples of 539 patients were subject to whole-genome single nucleotide polymorphism (SNP) genotyping using the Illumina Omni1-Quad human SNP genotyping platform (Centrillion Bioscience, USA). These patients were included into the genetic analysis because of the availability of their outcome and clinical data as well as the germline DNAs extracted from peripheral blood samples. The quality control analysis and filtering for this data included removing SNPs whose frequencies deviated from Hardy-Weinberg equilibrium, SNPs that had >5% missing values, and patients with discordant sex information, accidental duplicates, divergent or non-Caucasian ancestry, and first, second, or third degree relatives [39]. In Xu et al. (2015) [39], 505 patients were examined to investigate associations between overall and disease-free survival times after colorectal cancer diagnosis and genetic polymorphisms with a minor allele frequency (MAF) of at least 5%. In our study, there were 505 patients with 729,373 common SNPs (MAF ≥0.05) and 275,645 rare SNPs (MAF <0.05) that were included. No SNP was excluded due to high or perfect linkage disequilibrium (LD) with other SNPs. During this study, management and handling of these genotype data was done using PLINK v. 1.07 [40].

Statistical analysis

The response variable is a binary variable indicating existence of mucinous tumor histology or non-mucinous tumor histology.

Common SNP analysis

Univariable logistic regression analysis

Univariable logistic regression analysis was performed on each common SNP (MAF ≥5%) to determine if individual SNPs were significantly associated with mucinous tumor phenotype (i.e. mucinous versus non-mucinous tumor histology). For each SNP, the additive, co-dominant, dominant, and recessive genetic models were applied. Consequently, we report the 10 SNPs without excluding those in high LD with the highest level of significance in each genetic model (Additional file 1: Tables S1-S4).

Selection of baseline variables and multivariable logistic regression analysis

In order to select significant baseline factors to adjust for in the multivariable analyses, we first examined the variables shown in Table 1 using univariable logistic regression models. These variables were selected for inclusion into the selection process based on previous studies investigating mucinous colorectal tumors [27, 33]. Factors that had a p-value less than 0.1 were then included in a forward stepwise variable selection method. In addition, although there appeared to be a non-significant association between tumor histology and grade in the univariable analysis, tumor grade was still included in the multivariable model as has been shown to be linked to tumor histology [30, 41]. As a result, the baseline characteristics in the final models were sex, age at diagnosis, stage, and tumor location based on the 0.1 level of significance, and tumor grade (Additional file 1: Table S5). The 10 SNPs with the highest level of significance under each genetic model in the univariable logistic regression analysis were analyzed using the multivariable logistic regression model adjusting for the selected baseline characteristics (Additional file 1: Tables S1-S4).

Table 1 Baseline features of the study cohort and the results of univariable logistic regression analysis
Full size table

Plausibility of the genetic models

It is common in genetic association studies that only one genetic model is applied. In this study, we applied all four genetic models and assessed the plausibility of the genetic model under which the SNP was identified. To do this, we used the Akaike Information Criterion (AIC) calculations to compare the fit of four different genetic models per SNP under the multivariable logistic regression model. The genetic model with the smallest AIC estimate was considered to be the most plausible genetic model (i.e. the best fitting model). We first ranked the SNPs based on their p-value obtained in the multivariable model with the genetic model under which the SNP was identified (Additional file 1: Table S6). Then, we excluded those SNPs that were not identified in their plausible genetic model. Of note, we present in this manuscript only the 10 SNPs that have the highest association significance levels under the multivariable logistic regression models that were identified in their most plausible genetic model. We refer to these SNPs as “the top 10 SNPs”. The LD between SNPs was not taken into account when listing the top 10 SNPs.

Assessing the discriminatory accuracy of the estimated models

We aimed to check the ability of the multivariable models of the top 10 SNPs to discriminate between mucinous and non-mucinous phenotypes. A well-known method for assessing the discriminatory accuracy of a model is using a receiver operating characteristic (ROC) curve [42,43,44]. Calculating the area under the curve (AUC) of the ROC curve for the given models provides a single numeric representation for the performance of the model [43, 45, 46]. Comparing the AUC values and their corresponding confidence intervals provides a method for determining if one model is significantly superior to another in discriminatory accuracy [44, 47].

ROC curve analysis was performed by calculating the AUC using the pROC package in R [48]. The AUC estimates for (i) the model conditioning only on the baseline characteristics, (ii) the model conditioning on only the top SNPs, and (iii) the model conditioning on the baseline characteristics and the top SNPs. Comparing the AUC, specifically the 95% confidence intervals, between these three models can quantify the differences in the capacity of the models to distinguish mucinous and non-mucinous phenotypes.

Rare variant analysis

SKAT-O analysis

SKAT-O [49] test statistic was used to test the associations between the rare variants and the mucinous tumor phenotype. For this analysis, we prioritized gene-based regions including 5 kb long sequences before and after each gene. To do so, we first obtained genome location information for genome-wide gene-based regions (for the reference genome GRCh37.p13) using the biomaRt tool [50] in the Ensembl database [51]. The SNP information within these regions were then retrieved from the patient genome-wide data and used as the region-based SNP-sets in SKAT-O. During this analysis, each SNP was assigned to one gene-based region only. As a result, when a gene is located in close proximity to another gene, the second gene-based region does not include the SNPs that are analyzed in the first gene-based region. This limits redundancy since no SNP is analyzed more than once. For this analysis, only the additive genetic model was considered as using multiple genetic models is not a practical option for SKAT-O. The associations of gene regions were examined in multivariable models, adjusting for the significant baseline characteristics sex, age at diagnosis, stage, tumor location, and tumor grade.

All statistical analysis was performed using R v. 3.1.3 [52]. Correction for multiple testing was not applied to the results as this is an exploratory study and we did not want to increase false negative rate due to conservative corrections. While this increases the chances of obtaining false positives, we believe replication of these results in other studies will assist in reducing the potential false positive findings.

Bioinformatics analysis

Potential regulatory consequences of the identified SNPs were examined through RegulomeDB (http://www.regulomedb.org/) [53]. Ensembl [51] database was used to retrieve information related to the genes identified in the common and rare variant analysis.

Results

The demographic and clinicopathological information for the sample population is shown in Table 1. We observed a non-significant association of histology with age at diagnosis (>65 versus ≤60), grade, microsatellite instability (MSI) status, lymphatic invasion (LI), and BRAF V600E mutation; a moderately significant association with stage, sex, and age at diagnosis between 60 and 65 versus ≤60; and a strongly significant association with tumor location (Table 1). In this cohort, there was a trend for female sex having increased risk of mucinous tumors. As expected, the proportion of mucinous tumors was higher in colon cancer patients compared to rectum cancer patients and in stage II-IV patients compared to stage I patients (Table 1).

Common SNP analysis

None of the associations in this analysis reached the traditional genome-wide significance level (P < 5 × 10−8), but each genetic model identified promising associations.

After the univariable analysis, there were 33 SNPs that were nominally associated with the mucinous tumor phenotype (Additional file 1: Tables S1-S4). Associations of two SNPs (rs11216624 & rs17712784) were identified in both the dominant and co-dominant genetic models; one SNP (rs7314811) was detected in the additive, recessive, and co-dominant genetic models; and three SNPs (rs4843335, rs10511330, & rs16822593) were detected in both the additive and dominant genetic models. The estimates obtained in the univariable analysis did not change significantly when the models were adjusted for the baseline characteristics (Additional file 1: Tables S1-S4).

As explained in the Methods section, the AIC estimates (Additional file 1: Table S6) were used to determine the most plausible genetic models for each of 33 SNPs. Ten SNPs with the smallest p-value in the multivariable analysis under the most plausible genetic models were further prioritized (i.e., the top 10 SNPs). The results of the univariable and the multivariable logistic regression analyses for these top 10 SNPs are summarized in Table 2. Seven of these SNPs were located within gene sequences. These genes were quite diverse and belong to a variety of biological processes and pathways (Table 3).

Table 2 Top ten promising SNPs identified in univariable analysis and the subsequent multivariable analysis under their plausible genetic models
Full size table
Table 3 Genes identified in the common and rare analyses
Full size table

Before the ROC analysis, the LD among the top 10 SNPs were assessed using patient genotype data. These calculations indicated that rs13019215 and rs12471607 were in complete pairwise LD (r2 = 1). The SNPs rs4837345 and kgp10457679 were also in high LD with each other, as well as rs10511330 and rs16822593 (0.99 ≤ r2 ≤ 1.0). Therefore, we kept one SNP per SNP set in high LD, which left the following SNPs for the ROC analysis: rs9481067, rs10511330, rs13019215, rs716897, rs4843335, rs11968293, and kgp10457679.

Figure 1 shows the ROC curves comparing the accuracy of the models to discriminate mucinous and non-mucinous tumor phenotypes. The model (iii) including both the baseline characteristics and the SNPs (AUC = 0.916, CI: 0.873–0.960) had the most discriminatory accuracy followed by model (ii) including only the SNPs (AUC = 0.868, CI: 0.813–0.923) and model (i) including only the baseline characteristics (AUC = 0.703, 95% CI: 0.634–0.773). Since the confidence intervals of models (i) and (iii) do not overlap, we can confidently claim that there is a statistically significant improvement in the discriminating accuracy of the model containing the SNPs [44, 47]. This also suggests that these SNPs explain some of the variation between the mucinous and non-mucinous tumor phenotypes.

Fig. 1
figure 1

ROC curves and corresponding AUC values for multivariable models. Due to high LD among some of the top 10 SNPs, ROC analysis was performed on only the following SNPs: rs9481067, rs10511330, rs13019215, rs716897, rs4843335, rs11968293, and kgp10457679. AUC: area under the ROC curve, CI: confidence interval, LD: linkage disequilibrium, ROC: receiver operator characteristic

Full size image

Rare SNP analysis

In the gene region-based rare variant analysis, we investigated 29,966 regions in the patient cohort using the multivariable SKAT-O method. Table 3 and Table 4 summarize the most significant regions (P < 10−4) that potentially contain causal rare variants associated with the mucinous tumor phenotype. The number of variants aggregated in these gene-based regions varied from 5 to 10. While three of these regions (including the SEC24B, SEC24B-AS1, and CCDC109B regions) were located close to each other on chromosome 4, other regions come from different parts of the genome (Table 4).

Table 4 Most significant gene regions identified from SKAT-O multivariable analysis
Full size table

Discussion

Mucinous tumors are considered an aggressive type of colorectal tumors that are poorly understood [22, 24, 54]. While their role in prognosis is not well established, several studies suggested these tumors are associated with poorer prognosis when compared to non-mucinous tumors [25, 26, 32, 33, 35]. Identification of genes and genetic variations that can have a role in mucinous tumor development, therefore, has both scientific (e.g. dissecting the biology behind the mucinous tumor histology) as well as clinical value (e.g. biological information gained may assist with development of targeted treatment for this cancer subtype). Accordingly, for the first time with this study, we examined associations of both common and rare variants with the risk of developing the mucinous tumor phenotype using a genome-wide dataset.

While our results did not reach the conservative genome-wide significance level, promising associations were detected in both the common and rare variant analyses. In common SNP analysis, we identified seven unlinked polymorphisms that significantly increased our capacity to discriminate between mucinous and non-mucinous tumor phenotypes (Fig. 1, Table 2). Their effects on tumor histology were independent from the effects of the baseline variables (Fig. 1, Table 2). It is possible these polymorphisms (or others in high LD with them (Additional file 1: Table S7), including three additional SNPs shown in Table 2) are biologically linked to tumor histology or mucin production. Since there was no reported functional consequence of these SNPs in the literature, we searched the RegulomeDB database [53] for their potential biological characteristics. As of March 2018, the only SNP with a predicted/reported regulatory function in this database was kgp10457679 (rs10819474) (RegulomeDB score = 1f). This intergenic SNP is categorized as an expression quantitative trait locus (eQTL)/Transcription Factor (TF) binding/DNAse peak site, with a likely role of influencing the expression of target genes (Additional file 1: Table S8). Specifically, PPP2R4 is noted as the eQTL for this SNP. PPP2R4 is a tumor suppressor protein [55] which has been shown to have low activity in a large portion of a small cohort of colorectal tumors [56] and is associated with shorter survival times in metastatic colorectal cancer patients [57]. A potential link of PPP2R4 to mucinous tumor phenotype risk should be examined in further studies. Interestingly, one GWAS identified a SNP within the sequences of ZBTB20, other than the one reported in this study, that is significantly associated with the risk of non-cardia gastric cancer in the Han Chinese population [58]. Overall, all the novel loci identified by the common variant analysis are interesting candidates in examination of mucinous tumor development.

Typical association studies, such as the common variant analysis, focus on a variant-by-variant approach, which is underpowered for rare variants. It has been suggested that gene/region-based approaches can be useful in increasing the power under these circumstances where the direct effects of multiple variants on a phenotype can be examined [59]. Hence, in this study, we performed the first rare variant analysis to explore gene regions that may have a role in mucinous tumor formation using SKAT-O [49]. SKAT-O is a multi-marker association test which has reasonable type I error rate and is a powerful test under many scenarios [49]. In our study, this method identified a number of gene-based regions that may harbor rare variants associated with mucinous phenotype (Tables 3 and 4). Interestingly, three of the gene-based regions in Table 4 (SEC24B, SEC24B-AS1, and CCDC109B-based regions) were located in a 341,243 bp long genomic region on chromosome 4q. Since we assigned each SNP to only one gene region, these results suggest that these three gene regions are associated with the mucinous phenotype independent of each other. A search on the RegulomeDB database [53] indicated that one of the SNPs in LINC00596 (rs8005541) could have a strong regulatory function (RegulomeDB score = 1f). This variant is located in an eQTL and seems to affect the expression of two nearby genes; DHRS4 and DHRS4L2. These two genes are a part of a gene cluster on chromosome 14 that code for dehydrogenases/reductases [60] and have not been previously linked to mucinous tumors. Similarly, none of the genes in Table 4 had a previously identified connection to the risk of developing mucinous tumors. In conclusion, these regions, genes, or SNPs, alone or in combination, may be influential on the mucinous tumor phenotype and should be explored further.

Several strengths and limitations of this study should be mentioned. Studying the mucinous tumor phenotype is inherently challenging since it is not frequently detected. Despite this and the large number of SNPs/gene-based regions investigated, this study identified promising genetic variants and genomic regions that may have a biological connection to the mucinous tumor phenotype. We are aware that our results need to be replicated in independent cohorts and remain to be verified. Of note, SNPs and genetic regions we report are different than the MUC genes, which are the typical candidate genes for mucin production and mucinous phenotype. In the common variant analysis, the recessive and co-dominant models yielded some high odds ratio estimates but also wide confidence intervals (as expected, as these are the models with relatively low power). Consequently, the interpretation of these results should be made with caution. SKAT-O is a robust test and an attractive choice for rare variant analysis, however, it cannot determine which SNPs or how many SNPs within a SNP-set are truly associated with the phenotype. Also, in the rare variant analysis, due to the assignment of one SNP to one gene region, there could be some genes whose associations may have been missed. In addition, in contrast to previous studies, we used a comprehensive genome-wide SNP genotype data, however, analysis of a more comprehensive data (such as those obtained by whole genome sequencing) would be desirable. This is particularly true for rare variants as most genotyping technologies target primarily common SNPs.

Conclusions

In this study, we performed the first genome-wide association study on common and rare SNPs in colorectal cancer patients to identify novel genetic associations with the mucinous tumor phenotype. We identified novel, promising, and independent associations of specific SNP genotypes with the risk of developing mucinous tumors. In the common and rare variant analysis, we reported SNPs within the sequences of genes encoding transporter proteins, such as SLC22A16 and SLC35F1, which may have a role in transporting molecules related to excessive mucin production. In addition, the rare variant analysis reported associations with several regulating RNA molecules, which may influence the expression of genes related to mucin production. Finally, the common SNP analysis reports genes whose protein products are involved in DNA replication (CCDC141) and transcription (ZBTB20) that could have downstream effects on the mucin genes. Furthermore, the common SNPs reported in this study significantly improved the discriminatory accuracy of the multivariable model to distinguish between mucinous and non-mucinous tumors. In addition, we detected novel promising associations between gene-based sets of rare SNPs and mucinous tumors. The results of this study, once replicated in other cohorts, can contribute further information to the molecular characteristics of this under-studied but clinically important colorectal cancer subtype.

Abbreviations

AIC:

Akaike information criterion

AUC:

Area under the curve

CI:

Confidence interval

DNA:

Deoxyribonucleic acid

eQTL:

Expression quantitative trait loci

GI:

Gastrointestinal

LD:

Linkage disequilibrium

LI:

Lymphatic invasion

MAF:

Minor allele frequency

MSI:

Microsatellite instability

NFCCR:

Newfoundland Colorectal Cancer Registry

ROC:

Receiver operating characteristic

SNP:

Single nucleotide polymorphism

TF:

Transcription factor

USA:

United States of America

References

  1. Ferlay J, Soerjomataram I, Dikshit R, Eser S, Mathers C, Rebelo M, et al. Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012. Int J Cancer. 2015;136(5):E359–86.

    Article  PubMed  CAS  Google Scholar 

  2. Canadian Cancer Society’s Advisory Committee on Cancer Statistics. Canadian Cancer Statistics 2017. Toronto, ON: Canadian Cancer Society; 2017.

    Google Scholar 

  3. Moniaux N, Escande F, Porchet N, Aubert JP, Batra SK. Structural organization and classification of the human mucin genes. Front Biosci. 2001;6:d1192–206.

    Article  PubMed  CAS  Google Scholar 

  4. Gray KA, Yates B, Seal RL, Wright MW, Bruford EA. Genenames.org: the HGNC resources in 2015. Nucleic Acids Res. 2014;43:D1079–85.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  5. Desseyn J, Aubert J, Porchet N, Laine A. Evolution of the large secreted gel-forming mucins. Mol Biol Evol. 2000;17(8):1175–84.

    Article  PubMed  CAS  Google Scholar 

  6. Dhanisha SS, Guruvayoorappan C, Drishya S, Abeesh P. Mucins: structural diversity, biosynthesis, its role in pathogenesis and as possible therapeutic targets. Crit Rev Oncol Hematol. 2018;122:98–122.

    Article  PubMed  Google Scholar 

  7. Desseyn J, Buisine M, Porchet N, Aubert J, Degand P, Laine A. Evolutionary history of the 11p15 human mucin gene family. J Mol Evol. 1998;46(1):102–6.

    Article  PubMed  CAS  Google Scholar 

  8. Gosalia N, Leir S, Harris A. Coordinate regulation of the gel-forming mucin genes at chromosome 11p15.5. J Biol Chem. 2012;288(9):6717–25.

    Article  CAS  Google Scholar 

  9. Corfield AP. Mucins: a biologically relevant glycan barrier in mucosal protection. Biochim Biophys Acta Gen Sub. 2015;1850(1):236–52.

    Article  CAS  Google Scholar 

  10. Okudaira K, Kakar S, Cun L, Choi E, Wu Decamillis R, Miura S, et al. MUC2 gene promoter methylation in mucinous and non-mucinous colorectal cancer tissues. Int J Oncol. 2010;36(4):765–75.

    PubMed  CAS  Google Scholar 

  11. Johansson MEV, Larsson JMH, Hansson GC. The two mucus layers of colon are organized by the MUC2 mucin, whereas the outer layer is a legislator of host-microbial interactions. Proc Natl Acad Sci U S A. 2010;108:4659–65.

    Article  PubMed  PubMed Central  Google Scholar 

  12. Ho SB, Niehans GA, Lyftogt C, Yan PS, Cherwitz DL, Gum ET, et al. Heterogeneity of mucin gene expression in normal and neoplastic tissues. Cancer Res. 1993;53(3):641–51.

    PubMed  CAS  Google Scholar 

  13. Toribara NW, Roberton AM, Ho SB, Kuo WL, Gum E, Hicks JW, et al. Human gastric mucin. Identification of a unique species by expression cloning. J Biol Chem. 1993;268(8):5879–85.

    PubMed  CAS  Google Scholar 

  14. Biemer-Hüttmann A, Walsh MD, McGuckin MA, Ajioka Y, Watanabe H, Leggett BA, et al. Immunohistochemical staining patterns of MUC1, MUC2, MUC4, and MUC5AC mucins in hyperplastic polyps, serrated adenomas, and traditional adenomas of the colorectum. J Histochem Cytochem. 1999;47(8):1039–48.

    Article  PubMed  Google Scholar 

  15. Bartman AE, Serson SJ, Ewing SL, Niehans GA, Wiehr CL, Evans MK, et al. Aberrant expression of MUC5AC and MUC6 gastric mucin genes in colorectal polyps. Int J Cancer. 1999;80(2):210–8.

    Article  PubMed  CAS  Google Scholar 

  16. Amini A, Masoumi-Moghaddam S, Ehteda A, Liauw W, Morris DL. Depletion of mucin in mucin-producing human gastrointestinal carcinoma: results from in vitro and in vivo studies with bromelain and N-acetylcysteine. Oncotarget. 2015;6(32):33329–44.

    Article  PubMed  PubMed Central  Google Scholar 

  17. Wu C, Tung S, Chen P, Kuo Y. Clinicopathological study of colorectal mucinous carcinoma in Taiwan: a multivariate analysis. J Gastroenterol Hepatol. 1996;11(1):77–81.

    Article  PubMed  CAS  Google Scholar 

  18. Odone V, Chang L, Caces J, George SL, Pratt CB. The natural history of colorectal carcinoma in adolescents. Cancer. 1982;49(8):1716–20.

    Article  PubMed  CAS  Google Scholar 

  19. Chew M, Yeo SE, Ng Z, Lim K, Koh P, Ng K, et al. Critical analysis of mucin and signet ring cell as prognostic factors in an Asian population of 2,764 sporadic colorectal cancers. Int J Color Dis. 2010;25(10):1221–9.

    Article  Google Scholar 

  20. Papadopoulos VN, Michalopoulos A, Netta S, Basdanis G, Paramythiotis D, Zatagias A, et al. Prognostic significance of mucinous component in colorectal carcinoma. Tech Coloproctol. 2004;8(1):s123–5.

    Article  PubMed  Google Scholar 

  21. Kang H, O'Connell BJ, Maggard AM, Sack J, Ko YC. A 10-year outcomes evaluation of mucinous and signet-ring cell carcinoma of the colon and rectum. Dis Colon Rectum. 2005;48(6):1161–8.

    Article  PubMed  Google Scholar 

  22. Consorti F, Lorenzotti A, Midiri G, Di Paola M. Prognostic significance of mucinous carcinoma of colon and rectum: a prospective case-control study. J Surg Oncol. 2000;73(2):70–4.

    Article  PubMed  CAS  Google Scholar 

  23. Melis M, Hernandez J, Siegel EM, McLoughlin JM, Ly QP, Nair RM, et al. Gene expression profiling of colorectal mucinous adenocarcinomas. Dis Colon Rectum. 2010;53(6):936–43.

    Article  PubMed  Google Scholar 

  24. Nozoe T, Anai H, Nasu S, Sugimachi K. Clinicopathological characteristics of mucinous carcinoma of the colon and rectum. J Surg Oncol. 2000;75(2):103–7.

    Article  PubMed  CAS  Google Scholar 

  25. Catalano V, Loupakis F, Graziano F, Torresi U, Bisonni R, Mari D, et al. Mucinous histology predicts for poor response rate and overall survival of patients with colorectal cancer and treated with first-line oxaliplatin- and/or irinotecan-based chemotherapy. Br J Cancer. 2009;100(6):881–7.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  26. Negri FV, Wotherspoon A, Cunningham D, Norman AR, Chong G, Ross PJ. Mucinous histology predicts for reduced fluorouracil responsiveness and survival in advanced colorectal cancer. Ann Oncol. 2005;16(8):1305–10.

    Article  PubMed  CAS  Google Scholar 

  27. Tanaka H, Deng G, Matsuzaki K, Kakar S, Kim GE, Miura S, et al. BRAF mutation, CpG island methylator phenotype and microsatellite instability occur more frequently and concordantly in mucinous than non-mucinous colorectal cancer. Int J Cancer. 2006;118(11):2765–71.

    Article  PubMed  CAS  Google Scholar 

  28. Hanski C, Tiecke F, Hummel M, Hanski M, Ogorek D, Rolfs A, et al. Low frequency of p53 gene mutation and protein expression in mucinous colorectal carcinomas. Cancer Lett. 1996;103(2):163–70.

    Article  PubMed  CAS  Google Scholar 

  29. Park SY, Lee HS, Choe G, Chung JH, Kim WH. Clinicopathological characteristics, microsatellite instability, and expression of mucin core proteins and p53 in colorectal mucinous adenocarcinomas in relation to location. Virchows Arch. 2006;449(1):40–7.

    Article  PubMed  CAS  Google Scholar 

  30. Farhat MH, Barada KA, Tawil AN, Itani DM, Hatoum HA, Shamseddine AI. Effect of mucin production on survival in colorectal cancer: a case-control study. World J Gastroenterol. 2008;14(45):6981–5.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  31. Nitsche U, Zimmermann A, Späth C, Müller T, Maak M, Schuster T, et al. Mucinous and signet-ring cell colorectal cancers differ from classical adenocarcinomas in tumor biology and prognosis. Ann Surg. 2013;258(5):775–83.

    Article  PubMed  PubMed Central  Google Scholar 

  32. Numata M, Shiozawa M, Watanabe T, Tamagawa H, Yamamoto N, Morinaga S, et al. The clinicopathological features of colorectal mucinous adenocarcinoma and a therapeutic strategy for the disease. World J Surg Oncol. 2012;10:109.

    Article  PubMed  PubMed Central  Google Scholar 

  33. Verhulst J, Ferdinande L, Demetter P, Ceelen W. Mucinous subtype as prognostic factor in colorectal cancer: a systematic review and meta-analysis. J Clin Pathol. 2012;65(5):381–8.

    Article  PubMed  CAS  Google Scholar 

  34. Nitsche U, Friess H, Agha A, Angele M, Eckel R, Heitland W, et al. Prognosis of mucinous and signet-ring cell colorectal cancer in a population-based cohort. J Cancer Res Clin Oncol. 2016;142(11):2357–66.

    Article  PubMed  Google Scholar 

  35. Park JS, Huh JW, Park YA, Cho YB, Yun SH, Kim HC, et al. Prognostic comparison between mucinous and nonmucinous adenocarcinoma in colorectal cancer. Medicine. 2015;94(15):e658.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  36. Hanski C. Is mucinous carcinoma of the colorectum a distinct genetic entity? Br J Cancer. 1995;72(6):1350–6.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  37. Kim DH, Kim JW, Cho JH, Baek SH, Kakar S, Kim GE, et al. Expression of mucin core proteins, trefoil factors, APC and p21 in subsets of colorectal polyps and cancers suggests a distinct pathway of pathogenesis of mucinous carcinoma of the colorectum. Int J Oncol. 2005;27:957–64.

    PubMed  CAS  Google Scholar 

  38. Woods MO, Hyde AJ, Curtis FK, Stuckless S, Green JS, Pollett AF, et al. High frequency of hereditary colorectal cancer in Newfoundland likely involves novel susceptibility genes. Clin Cancer Res. 2005;11(19):6853.

    Article  PubMed  CAS  Google Scholar 

  39. Xu W, Xu J, Shestopaloff K, Dicks E, Green J, Parfrey P, et al. A genome wide association study on Newfoundland colorectal cancer patients' survival outcomes. Biomarker Res. 2015;3(1):6.

    Article  Google Scholar 

  40. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira M, Bender D, et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet. 2007;81(3):559–75.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  41. Leopoldo S, Lorena B, Cinzia A, Gabriella D, Angela Luciana B, Renato C, et al. Two subtypes of mucinous adenocarcinoma of the colorectum: clinicopathological and genetic features. Ann Surg Oncol. 2008;15(5):1429–39.

    Article  PubMed  Google Scholar 

  42. Lasko TA, Bhagwat JG, Zou KH, Ohno-Machado L. The use of receiver operating characteristic curves in biomedical informatics. J Biomed Inform. 2005;38(5):404–15.

    Article  PubMed  Google Scholar 

  43. Zhou X, Obuchowski NA, McClish DK. Chapter 2. Measures of diagnostic accuracy. In: Statistical methods in diagnostic medicine. 2nd ed. Hoboken: Wiley; 2011. p. 13–57.

    Chapter  Google Scholar 

  44. Søreide K. Receiver-operating characteristic curve analysis in diagnostic, prognostic and predictive biomarker research. J Clin Pathol. 2008;62(1):1.

    Article  PubMed  Google Scholar 

  45. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology. 1982;143(1):29–36.

    Article  PubMed  CAS  Google Scholar 

  46. Kumar R, Indrayan A. Receiver operating characteristic (ROC) curve for medical researchers. Indian Pediatr. 2011;48(4):277–87.

    Article  PubMed  Google Scholar 

  47. Zweig MH, Campbell G. Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine. Clin Chem. 1993;39(4):561.

    PubMed  CAS  Google Scholar 

  48. Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez J, et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics. 2011;12(1):77.

    Article  PubMed  PubMed Central  Google Scholar 

  49. Lee S, Wu MC, Lin X. Optimal tests for rare variant effects in sequencing association studies. Biostatist. 2012;13(4):762–75.

    Article  Google Scholar 

  50. Kinsella RJ, Kähäri A, Haider S, Zamora J, Proctor G, Spudich G, et al. Ensembl BioMarts: a hub for data retrieval across taxonomic space. Database. 2011:bar030.

  51. Flicek P, Amode MR, Barrell D, Beal K, Billis K, Brent S, et al. Ensembl 2014. Nucleic Acids Res. 2014;42(D1):D749–55.

    Article  PubMed  CAS  Google Scholar 

  52. Core Team R. R: a language and environment for statistical computing. R Foundation for Statistical Computing 2013.

    Google Scholar 

  53. Boyle AP, Hong EL, Hariharan M, Cheng Y, Schaub MA, Kasowski M, et al. Annotation of functional variation in personal genomes using RegulomeDB. Genome Res. 2012;22(9):1790–7.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  54. Yamamoto S, Mochizuki H, Hase K, Yamamoto T, Ohkusa Y, Yokoyama S, et al. Assessment of clinicopathologic features of colorectal mucinous adenocarcinoma. Am J Surg. 1993;166(3):257–61.

    Article  PubMed  CAS  Google Scholar 

  55. Janssens V, Goris J, Van Hoof C. PP2A: the expected tumor suppressor. Curr Opin Genet Dev. 2005;15(1):34–41.

    Article  PubMed  CAS  Google Scholar 

  56. Cristóbal I, Rincón R, Manso R, Madoz-Gúrpide J, Caramés C, del Puerto-Nevado L, et al. Hyperphosphorylation of PP2A in colorectal cancer and the potential therapeutic value showed by its forskolin-induced dephosphorylation and activation. Biochim Biophys Acta Mol Basis Dis. 2014;1842(9):1823–9.

    Article  CAS  Google Scholar 

  57. Cristóbal I, Manso R, Rincón R, Caramés C, Zazo S, del Pulgar TG, et al. Phosphorylated protein phosphatase 2A determines poor outcome in patients with metastatic colorectal cancer. Br J Cancer. 2014;111(4):756–62.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  58. Shi Y, Hu Z, Wu C, Dai J, Li H, Dong J, et al. A genome-wide association study identifies new susceptibility loci for non-cardia gastric cancer at 3q13.31 and 5p13.1. Nat Genet. 2011;43:1215.

    Article  PubMed  CAS  Google Scholar 

  59. Li B, Leal SM. Methods for detecting associations with rare variants for common diseases: application to analysis of sequence data. Am J Hum Genet. 2008;83(3):311–21.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  60. Gabrielli F, Tofanelli S. Molecular and functional evolution of human DHRS2 and DHRS4 duplicated genes. Gene. 2012;511(2):461–9.

    Article  PubMed  CAS  Google Scholar 

  61. Sherry ST, Ward MH, Kholodov M, Baker J, Phan L, Smigielski EM, et al. dbSNP: the NCBI database of genetic variation. Nucleic Acids Res. 2001;29(1):308–11.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  62. Kent WJ, Sugnet C,W., Furey TS, Roskin KM, Pringle TH, Zahler AM, et al. The human genome browser at UCSC. Genome Res 2002;12(6):996–1006.

  63. Enomoto A, Wempe MF, Tsuchida H, Shin HJ, Cha SH, Anzai N, et al. Molecular identification of a novel carnitine transporter specific to human testis: insights into the mechanism of carnitine recognition. J Biol Chem. 2002;277(39):36262–71.

    Article  PubMed  CAS  Google Scholar 

  64. Aouida M, Poulin R, Ramotar D. The human carnitine transporter SLC22A16 mediates high affinity uptake of the anticancer polyamine analogue bleomycin-A5. J Biol Chem. 2009;285(9):6275–84.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  65. Fukuda T, Sugita S, Inatome R, Yanagi S. CAMDI, a novel disrupted in schizophrenia 1 (DISC1)-binding protein, is required for radial migration. J Biol Chem. 2010;285(52):40554–61.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  66. Kayser G, Gerlach U, Walch A, Nitschke R, Haxelmans S, Kayser K, et al. Numerical and structural centrosome aberrations are an early and stable event in the adenoma-carcinoma sequence of colorectal carcinomas. Virchows Arch. 2005;447(1):61–5.

    Article  PubMed  CAS  Google Scholar 

  67. Ishida N, Kawakita M. Molecular physiology and pathology of the nucleotide sugar transporter family (SLC35). Pflugers Arch. 2004;447(5):768–75.

    Article  PubMed  CAS  Google Scholar 

  68. Xie Z, Zhang H, Tsai W, Zhang Y, Du Y, Zhong J, et al. Zinc finger protein ZBTB20 is a key repressor of alpha-fetoprotein gene transcription in liver. Proc Natl Acad Sci U S A. 2008;105(31):10859–64.

    Article  PubMed  PubMed Central  Google Scholar 

  69. Zhao J, Ren K, Tang J. Zinc finger protein ZBTB20 promotes cell proliferation in non-small cell lung cancer through repression of FoxO1. FEBS Lett. 2014;588(24):4536–42.

    Article  PubMed  CAS  Google Scholar 

  70. Wang Q, Tan Y, Ren Y, Dong L, Xie Z, Tang L, et al. Zinc finger protein ZBTB20 expression is increased in hepatocellular carcinoma and associated with poor prognosis. BMC Cancer. 2011;11(1):271.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  71. Fan W, Koch CA, de Hoog CL, Fam NP, Moran MF. The exchange factor Ras-GRF2 activates Ras-dependent and Rac-dependent mitogen-activated protein kinase pathways. Curr Biol. 1998;8(16):935–9.

    Article  PubMed  CAS  Google Scholar 

  72. Crespo P, Calvo F, Sanz-Moreno V. Ras and rho GTPases on the move: the RasGRF connection. BioArchitecture. 2011;1(4):200–4.

    Article  PubMed  PubMed Central  Google Scholar 

  73. Wendeler MW, Paccaud J, Hauri H. Role of Sec24 isoforms in selective export of membrane proteins from the endoplasmic reticulum. EMBO Rep. 2006;8(3):258–64.

    Article  CAS  Google Scholar 

  74. Goldenring JR. A central role for vesicle trafficking in epithelial neoplasia: intracellular highways to carcinogenesis. Nat Rev Cancer. 2013;13(11):813–20.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  75. Raffaello A, De Stefani D, Sabbadin D, Teardo E, Merli G, Picard A, et al. The mitochondrial calcium uniporter is a multimer that can include a dominant-negative pore-forming subunit. EMBO J. 2013;32(17):2362–76.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  76. Duchen MR. Mitochondria and calcium: from cell signalling to cell death. J Physiol. 2000;529:57–68.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  77. Brown GR, Hem V, Katz KS, Ovetsky M, Wallin C, Ermolaeva O, et al. Gene: a gene-centered information resource at NCBI. Nucleic Acids Res. 2014;43:D36–42.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  78. Green RC, Green JS, Buehler SK, Robb JD, Daftary D, Gallinger S, et al. Very high incidence of familial colorectal cancer in Newfoundland: a comparison with Ontario and 13 other population-based studies. Familial Cancer. 2007;6(1):53–62.

    Article  PubMed  CAS  Google Scholar 

Download references

Acknowledgements

We thank the patients and families that participated in NFCCR and all the NFCCR personnel and investigators who contributed to NFCCR.

Funding

This work was funded by the Research and Development Corporation (RDC) Newfoundland and Labrador (NL) [5404.1723.101] and the Faculty of Medicine of Memorial University of Newfoundland (awarded to Y.E. Yilmaz). M.E. Penney is partly supported by a Translational and Personalized Medicine Initiative (TPMI)/NP SUPPORT fellowship. The funding sources were not involved in the study design; in the collection, analysis or interpretation of data; in the writing of the report; or in the decision to submit the paper for publication.

Availability of data and materials

The data that support the findings of this study are available from NFCCR but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Permission to obtain the data can be requested from NFCCR, Faculty of Medicine, Memorial University of Newfoundland, St. John’s, NL, Canada. Ethics approval shall be obtained from the Health Research Ethics Board (HREB), Ethics Office, Health Research Ethics Authority, Suite 200, 95 Bonaventure Avenue, St. John’s, NL, A1B 2X5, Canada (e-mail: info@hrea.ca).

Author information

Authors and Affiliations

  1. Discipline of Genetics, Faculty of Medicine, Memorial University of Newfoundland, St. John’s, Canada

    Michelle E. Penney, Sevtap Savas & Yildiz E. Yilmaz

  2. Discipline of Medicine, Faculty of Medicine, Memorial University of Newfoundland, St. John’s, Canada

    Patrick S. Parfrey & Yildiz E. Yilmaz

  3. Discipline of Oncology, Faculty of Medicine, Memorial University of Newfoundland, St. John’s, Canada

    Sevtap Savas

  4. Department of Mathematics and Statistics, Faculty of Science, Memorial University of Newfoundland, St. John’s, Canada

    Yildiz E. Yilmaz

Authors
  1. Michelle E. Penney
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Patrick S. Parfrey
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Sevtap Savas
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Yildiz E. Yilmaz
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

MP, SS, and YY designed the study and revised the manuscript. PP provided patient characteristics and disease outcome data. SS provided the genome-wide SNP genotype data and the patient cohort investigated. MP conducted the statistical analysis, interpreted the results, wrote the first draft of the manuscript, and prepared the figure and Tables. SS and YY reviewed the results and their interpretation, and supervised the study. All authors reviewed the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Yildiz E. Yilmaz.

Ethics declarations

Ethics approval and consent to participate

Patient consents were obtained by the Newfoundland Colorectal Cancer Registry (NFCCR) at the time of recruitment. If the patient was deceased, consent was sought from a close relative [78]. Ethics approval for this study was obtained from the Health Research Ethics Board (HREB; #15.043).

Competing interests

The authors declare that they have no competing interests.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Additional file

Additional file 1:

Table S1. Top ten most significant common SNPs identified based on the univariable analyses and the subsequent multivariable analyses under the additive genetic models. Table S2. Top ten most significant common SNPs identified based on the univariable analyses and the subsequent multivariable analysis under the dominant genetic models. Table S3. Top ten most significant common SNPs identified based on the univariable analyses and the subsequent multivariable analyses under the recessive genetic models. Table S4. Top ten most significant common SNPs identified under the univariable analyses and the subsequent multivariable analyses under the co-dominant genetic models. Table S5. Baseline characteristics selected through a stepwise variable selection method under the multivariable model. Table S6. AIC estimates under the multivariable models of common SNPs identified in the univariable analysis. Table S7. Haploreg results for the top 10 SNPs in the common variant analysis. Table S8. Proteins which have reported evidence of binding to the genomic region in which kgp10457679 resides (extracted from RegulomeDB). (PDF 360 kb)

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Penney, M.E., Parfrey, P.S., Savas, S. et al. Associations of single nucleotide polymorphisms with mucinous colorectal cancer: genome-wide common variant and gene-based rare variant analyses. Biomark Res 6, 17 (2018). https://doi.org/10.1186/s40364-018-0133-z

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s40364-018-0133-z

Keywords

Biomarker Research

ISSN: 2050-7771

Contact us