Mismatch repair polymorphisms and the risk of colorectal cancer
Article first published online: 4 JAN 2007
Copyright © 2006 Wiley-Liss, Inc.
International Journal of Cancer
Volume 120, Issue 7, pages 1548–1554, 1 April 2007
How to Cite
Berndt, S. I., Platz, E. A., Fallin, M. D., Thuita, L. W., Hoffman, S. C. and Helzlsouer, K. J. (2007), Mismatch repair polymorphisms and the risk of colorectal cancer. Int. J. Cancer, 120: 1548–1554. doi: 10.1002/ijc.22510
- Issue published online: 30 JAN 2007
- Article first published online: 4 JAN 2007
- Manuscript Accepted: 6 NOV 2006
- Manuscript Received: 20 JUN 2006
- National Institute on Aging. Grant Number: IU01AG18033
- mismatch repair;
- colorectal cancer;
- DNA repair;
- meat intake
Rare germline variants in mismatch repair genes have been linked to hereditary nonpolyposis colorectal cancer; however, it is unknown whether common polymorphisms in these genes alter the risk of colorectal cancer. To examine the association between common variants in mismatch repair genes and colorectal cancer, we conducted a case-cohort study within the CLUE II cohort. Four single nucleotide polymorphisms in 3 mismatch repair genes (MSH3 R940Q, MSH3 T1036A, MSH6 G39E and MLH1 I219V) were genotyped in 237 colorectal cancer cases and a subcohort of 2,189 participants. Incidence rate ratios (RRs) and 95% confidence intervals (95% CIs) for each polymorphism were estimated. The MSH3 1036A variant was found to be associated with an increased risk of colorectal cancer (RR = 1.28, 95% CI: 0.94–1.74 and RR = 1.65, 95% CI: 1.01–2.70 for the AT and TT genotypes, respectively, with ptrend = 0.02), particularly proximal colon cancer. Although the MSH3 940Q variant was only weakly associated with colorectal cancer overall (ptrend = 0.07), it was associated with a significant increased risk of proximal colon cancer (RR = 1.69, 95% CI: 1.10–2.61 and RR = 2.68, 95% CI: 0.96–7.47 for the RQ and QQ genotypes, respectively with ptrend = 0.005). Processed meat intake appeared to modify the association between the MSH3 polymorphisms and colorectal cancer (pinteraction < 0.10 for both). No association was observed with the MSH6 and MLH1 polymorphisms overall. This study suggests that common polymorphisms in the mismatch repair gene, MSH3, may increase the risk of colorectal cancer, especially proximal colon cancer. © 2006 Wiley-Liss, Inc.
Alterations in mismatch repair are implicated in the etiology of colorectal cancer. Inherited mutations in mismatch repair genes cause hereditary nonpolyposis colorectal cancer (HNPCC), a familial cancer syndrome that accounts for ∼2–4% of colorectal cancers.1 Individuals with HNPCC inherit an inactivating mutation in a mismatch repair gene (e.g., MSH2, MLH1, MSH6 and PMS2). Although in most cases, the individual only carries one copy of the inactivating mutation and displays normal mismatch repair,2 the somatic loss of the wild-type allele in the colorectal tissue leads to a dramatic increase in small intragenic mutations, microsatellite instability and the development of colorectal cancer. Nearly all tumors from HNPCC patients display microsatellite instability3, 4; however, ∼10–16% of colorectal tumors from patients without HNPCC also exhibit microsatellite instability,3, 5 suggesting that defects in mismatch repair may also play a role in the etiology of some non-HNPCC colorectal cancers.
Mismatch repair primarily corrects single base-pair mismatches and small insertion-deletion loops that arise during replication. In mammals, single base-pair mismatches and single base insertion-deletion loops are recognized by the MSH2–MSH6 heterodimer (MutSα), whereas larger insertion-deletion loops are detected by the MSH2–MSH3 heterodimer (MutSβ).6, 7 A second heterodimer, such as MLH1-PMS2 (MutLα), MLH1-PMS1 (MutLβ) or MLH1–MLH3 (MutLγ), is then recruited to the damage site and the repair process is initiated with the help of other enzymes, such as exonuclease I, polymerase δ and DNA ligase.8
Although rare germline variants in mismatch repair genes are known to cause HNPCC, it is unknown whether common polymorphisms in mismatch repair genes are associated with the risk of colorectal cancer. In this case-cohort study, we examined the association between 4 common polymorphisms in 3 mismatch repair genes (MSH3 R940Q, MSH3 T1036A, MSH6 G39E and MLH1 I219V) and the risk of colorectal cancer. The MSH3 R940Q and T1036A polymorphisms encode nonconservative amino acid changes and are located in the C-terminal region, which contains the putative nucleotide binding domain and helix-turn-helix DNA-binding domain.9 The MSH6 G39E polymorphism resides in exon 1 and encodes a nonconservative amino acid change. The MLH1 I219V polymorphism is located within a conserved residue in exon 8. Although the polymorphism is unlikely to affect mismatch repair capacity,10 it may characterize an ancestral haplotype with an increased mutation frequency.11, 12 We hypothesized that these polymorphisms may be associated with colorectal cancer susceptibility.
Subject and methods
Our study was nested within the CLUE II cohort, which has been described elsewhere.13, 14 Briefly, CLUE II is a community-based cohort consisting primarily of Caucasians (94%) from Washington County, Maryland, that was started in 1989. A total of 22,887 adult residents were enrolled in the cohort, completed a brief questionnaire and provided a blood specimen at baseline. Demographic characteristics and information on risk factors (e.g., smoking status, education and body mass index) were ascertained through the baseline questionnaire. Additional information regarding medical history and other risk factors (e.g., family history) were obtained through follow-up questionnaires. All participants provided written consent at study entry, and the study was approved by the Institutional Review Board at the Johns Hopkins Bloomberg School of Public Health.
Identification of subcohort
A case-cohort study design was used to examine the association between mismatch repair polymorphisms and colorectal cancer risk. Among adult CLUE II participants who were residents of Washington County, Maryland, and donated blood specimens at baseline, we selected a 10% age-stratified, random sample as our subcohort (n = 2,289). Individuals diagnosed with anal cancer (n = 1) or colorectal cancer prior to 1989 (n = 21) and subjects without sufficient DNA for genotyping (n = 78) were excluded, leaving 2,189 persons available for analysis. Baseline characteristics (e.g., age, sex and smoking status) of the subcohort participants were similar to those of the full cohort (p > 0.05).
Identification of cases
Incident cases of colorectal cancer were ascertained through the Washington County Cancer Registry (starting in 1989) and the Maryland Cancer Registry (starting in 1992). A total of 272 pathology-confirmed cases of colorectal cancer were identified among CLUE II residents of Washington County, Maryland, from the date of blood draw through December 2003. Since the etiology of nonadenocarcinoma cancers may be different from that of adenocarcinoma tumors, all cancers that were not adenocarcinoma (n = 11) were excluded. Twenty-four cases with insufficient DNA for genotyping were excluded, leaving 237 colorectal adenocarcinoma cases for the analysis. Of these cases, 191 individuals were diagnosed with colon cancer (International Classification of Diseases, Ninth Revision [ICD-9] codes 153.0–153.9 and International Classification of Diseases, Tenth Revision [ICD-10] codes C18.0–C18.9) and 46 were diagnosed with rectal cancer (ICD-9 codes 154.0, 154.1 and 154.8 and ICD-10 codes C19 and C20). Tumors from the cecum through the splenic flexure were classified as proximal colon cancers, whereas tumors from the descending and sigmoid colon were considered to be distal colon cancers. Of the 191 colon cancer cases, 103 cases had proximal colon tumors, 76 cases had distal colon tumors and 12 cases had tumors in an unspecified location of the colon or had tumors in both the proximal and distal colon.
Using an alkaline lysis method,15 DNA was extracted from the buffy coat fraction of the blood specimens, and aliquots of DNA were shipped to Applied Biosystems (Foster City, CA) for genotyping. Four single nucleotide polymorphisms (SNPs) in 3 mismatch repair genes [MSH3 Arg940Gln (rs184967), MSH3 Thr1036Ala (rs26279), MSH6 Gly39Glu (rs1042821) and MLH1 Ile219Val (rs1799977)] were genotyped using TaqMan® assays. Laboratory personnel were blinded to case–control status, and replicate samples showed 100% concordance for all genotypes. Genotyping was successfully completed for 90–99% of subjects, depending upon the assay.
The genotype frequencies in the subcohort were similar to those reported elsewhere16 and on the National Cancer Institute SNP500 Cancer Project website (http://snp500cancer.nci.nih.gov) for Caucasians. When stratified by race, the subcohort genotype frequencies were consistent with Hardy-Weinberg proportions (p > 0.05) for all SNPs using a goodness-of-fit χ2 test (or exact test if cell counts were small).
Information on diet and alcohol consumption was obtained at baseline using a 60-item Block food frequency questionnaire (FFQ),17 which was returned by ∼80% of the participants included in this study. Participants were considered to have an invalid FFQ if they reported eating fewer than 3 foods per day, answered less than half the questions or had values for total energy intake outside the range of 500–3,500 kcal per day for women or 800–5,000 kcal per day for men. After excluding individuals who did not complete the FFQ (n = 479) or had an invalid FFQ (n = 192), a total of 194 cases and 1,556 subcohort participants remained for the dietary analyses.
Usual meat intake was estimated using the data from the 13 FFQ questions pertaining to meat consumption. Intake of each meat item was estimated by multiplying the number of grams per medium serving size by the average frequency per day. Total meat, red meat and processed meat intake was assessed by summing the relevant meat items. Processed meat included bacon, sausage, hot dogs, ham, bologna, salami and other lunch meats. Medians and thirds of dietary factors were estimated using the distribution of intake in the subcohort.
The study was analyzed as a case-cohort study. Person-years at risk for the entire cohort were estimated from the subcohort, and cases outside the subcohort only entered the analysis at the time of their diagnosis. Follow-up time for the subcohort participants was determined from the date of blood draw to the date of colorectal cancer diagnosis (for cases within the subcohort), death or the end of follow-up (December 30, 2003), whichever came first, and colorectal cancer cases were assigned a negligible amount of follow-time at the time of their diagnosis.
Incidence rate ratios (RRs) and 95% confidence intervals (CIs) for the association between individual polymorphisms and colorectal cancer were estimated using a modified Cox regression model18 with a robust variance estimate19 and adjusted for age (continuous) and race. The inclusion of age as a categorical variable (<60, 60–69 and 70+ years) instead of a continuous variable yielded similar results, and additional adjustment for body mass index, sex and education did not substantially alter the results. The results were similar when the analysis was limited to Caucasians or restricted to individuals who were ≥35 years of age at baseline. To test for trend, a single variable coded as the number of variant alleles was included in the regression model, and the p-value from the Wald test was used to assess statistical significance.
To examine whether lifestyle or dietary factors modified the association between each polymorphism and colorectal cancer, stratified analyses were conducted. The statistical significance of the interaction was assessed by comparing nested models with and without the cross product terms using a likelihood ratio test. Heterogeneity between cancer locations (e.g., colon vs. rectal) was assessed by treating one location as the “case” and the other as the “control” in a logistic regression model, and comparing the models with and without the genotype of interest using a likelihood ratio test.
Measures of linkage disequilibrium and haplotype analyses were conducted for MSH3 among Caucasians. Pairwise linkage disequilibrium measures (D′ and r2) were examined using Haploview (http://www.broad.mit.edu/personal/jcbarret/haploview/), and haplotype analyses were conducted using HaploStats (http://www.mayo.edu/hsr/people/schaid.html). Haplotypes were estimated using an expectation-maximization (EM) algorithm,20 overall differences between the cases and noncases were assessed using a global score test,21 and risks for individual haplotypes were estimated using a generalized linear model.21, 22
Unless otherwise specified, statistical analyses were performed using SAS version 8.1 (SAS Institute, Cary, NC).
Baseline characteristics of the CLUE II subcohort participants by mismatch repair genotype are given in Table I. The mean age at blood draw was 48.4 ± 16.2 years, and the average follow-up time was 13.5 ± 2.6 years. The majority of participants were Caucasian (98%), most had at least a high school education (67%) and slightly more women than men participated (58 vs. 42%). Baseline characteristics of the participants were similar across the mismatch repair genotypes. The MLH1 219V allele was more common among Caucasians than other races, which is consistent with other studies.16
|Characteristic||Mismatch repair genotype|
|MSH3 R940Q||MSH3 T1036A||MSH6 G39E||MLH1 I219V|
|No. of participants||1551||565||49||1063||851||167||1352||561||58||968||896||226|
|Age at blood draw (mean ± SD)||48 ± 16||48 ± 17||49 ± 17||48 ± 16||48 ± 16||49 ± 17||48 ± 16||48 ± 16||51 ± 18||48 ± 16||48 ± 16||48 ± 16|
|Less than 12 years||21.4||20.7||22.5||21.0||21.1||22.8||20.3||20.9||31.0||19.9||21.0||27.4|
|High school graduate||45.2||46.2||55.1||45.7||44.9||46.1||46.6||46.2||37.9||45.2||46.6||42.9|
|Beyond high school||33.4||33.1||22.5||33.3||34.0||31.1||33.1||33.0||31.0||34.9||32.4||29.7|
|Smoking status (%)|
|Former cigarette smoker||28.0||32.6||22.5||27.8||29.6||32.3||28.2||29.4||29.3||30.5||27.5||29.2|
|Current cigarette smoker||20.6||19.8||26.5||20.8||20.7||19.2||20.9||20.5||19.0||20.4||20.4||22.1|
|Ever cigar or pipe smoker but never cigarette smoker||1.9||2.3||4.1||1.7||2.5||2.4||2.4||1.6||0.0||1.9||2.7||0.4|
|Body mass index in kg/m2 (mean ± SD)||26.2 ± 5.0||26.0 ± 4.9||25.8 ± 5.4||26.1 ± 5.0||26.3 ± 5.1||25.9 ± 4.9||26.2 ± 5.0||26.3 ± 5.3||25.3 ± 4.7||26.4 ± 5.1||26.1 ± 5.0||25.8 ± 4.6|
The majority of the 237 colorectal cancer cases were Caucasian (98%), and the average age at diagnosis was 70.8 ± 10.8 years. Older age was associated with a significant increased risk of colorectal cancer (RR = 1.45/5 year increase, 95% CI: 1.41–1.48). However, only weak associations were observed between colorectal cancer and sex (male vs. female: RR = 1.21, 95% CI: 0.91–1.60), tobacco smoking (ever vs. never: RR = 1.23, 95% CI: 0.93–1.63), body mass index (BMI ≥25 kg/m2: RR = 1.18, 95% CI: 0.89–1.58) and education (≥12 years: RR = 0.78, 95% CI: 0.58–1.07) after adjustment for age.
In examining the mismatch repair polymorphisms, the MSH3 1036A variant was associated with an increased risk of colorectal cancer (Table II). The association appeared stronger for tumors in the proximal colon (RR = 1.57, 95% CI: 1.03–2.39) compared to the distal colon (RR = 1.06, 95% CI: 0.66–1.70) or rectum (RR = 1.11, 95% CI: 0.61–2.01). Although not statistically significant, the MSH3 940Q variant was also associated with slight increased risk of colorectal cancer (ptrend = 0.07). When the results were stratified by tumor location, the MSH3 940Q variant was associated with a significant increased risk of proximal colon cancer (RR = 1.69, 95% CI: 1.10–2.61 and RR = 2.68, 95% CI: 0.96–7.47 for the RQ and QQ genotypes, respectively, with ptrend = 0.005). No association was observed with distal tumors, and the p for heterogeneity between proximal and distal tumors for the MSH3 940Q variant was 0.004.
|Polymorphism||No. of PY||Location of cancer|
|Colorectal||Colon||Proximal colon||Distal colon||Rectum|
|No. of cases||RR1||No. of cases||RR1||No. of cases||RR1||No. of cases||RR1||No. of cases||RR1|
|RQ||7592||72||1.24 (0.91–1.71)||57||1.21 (0.85–1.71)||38||1.69 (1.10–2.61)||13||0.58 (0.31–1.08)||15||1.41 (0.73–2.70)|
|657||9||1.85 (0.83–4.14)||7||1.77 (0.73–4.31)||5||2.68 (0.96–7.47)||2||1.06 (0.24–4.66)||2||2.21 (0.50–9.78)|
|RQ or QQ||1.29 (0.95–1.75)||1.25 (0.89–1.75)||1.77 (1.16–2.69)||0.62 (0.34–1.11)||1.47 (0.78–2.75)|
|AT||11405||101||1.28 (0.94–1.74)||82||1.33 (0.95–1.87)||48||1.54 (0.99–2.39)||27||0.93 (0.56–1.56)||19||1.10 (0.59–2.06)|
|AA||2212||27||1.65 (1.01–2.70)||23||1.80 (1.06–3.05)||11||1.72 (0.84–3.51)||10||1.65 (0.79–3.45)||4||1.14 (0.38–3.40)|
|AT or AA||1.34 (1.01–1.80)||1.41 (1.03–1.94)||1.57 (1.03–2.39)||1.06 (0.66–1.70)||1.11 (0.61–2.01)|
|GE||7600||60||0.92 (0.66–1.29)||46||0.84 (0.58–1.22)||22||0.71 (0.43–1.18)||21||0.99 (0.58–1.70)||14||1.33 (0.68–2.61)|
|EE||750||7||0.90 (0.39–2.06)||3||0.45 (0.14–1.49)||1||0.27 (0.04–1.99)||1||0.39 (0.05–2.95)||4||3.25 (1.08–9.83)|
|GE or EE||0.92 (0.66–1.27)||0.80 (0.55–1.15)||0.66 (0.40–1.09)||0.93 (0.55–1.58)||1.53 (0.82–2.86)|
|IV||12071||94||1.03 (0.75–1.41)||75||1.05 (0.74–1.48)||41||1.10 (0.70–1.74)||30||0.98 (0.59–1.64)||19||0.97 (0.52–1.83)|
|VV||3033||17||0.74 (0.42–1.30)||16||0.89 (0.50–1.61)||9||0.97 (0.46–2.07)||7||0.91 (0.38–2.15)||1||0.20 (0.03–1.51)|
|IV or VV||0.97 (0.72–1.32)||1.02 (0.73–1.42)||1.08 (0.70–1.67)||0.97 (0.59–1.58)||0.81 (0.44–1.52)|
|VV vs. IV or II||0.73 (0.43–1.25)||0.87 (0.50–1.53)||0.93 (0.45–1.90)||0.92 (0.41–2.08)||0.20 (0.03–1.50)|
The two MSH3 polymorphisms were in strong linkage disequilibrium (D′ = 1.0) and moderately correlated (r2 = 0.45) among Caucasians. When both polymorphisms were entered into the same model, the association between MSH3 1036A and proximal colon cancer was substantially attenuated (ptrend = 0.93), but the MSH3 940Q variant remained associated with an increased risk of proximal colon cancer (ptrend = 0.02), suggesting that the increased risk observed previously with the MSH3 1036A variant may have been driven by its correlation with the MSH3 940Q allele. Similarly, when haplotype analyses were conducted, only the haplotype contain the variants at both MSH3 loci was significantly associated with an increased risk of proximal colon cancer among Caucasians (Table III). No association was observed for the haplotype containing the 940R wild type and 1036A variant.
|Location of cancer|
|Colorectal||Colon||Proximal colon||Distal colon||Rectum|
|MSH3||Haplotype||Noncases (%)||Cases (%)||OR1||Cases (%)||OR1||Cases (%)||OR1||Cases (%)||OR1||Cases (%)||OR1|
|R||A||13.3||14.3||1.10 (0.82–1.47)||15.6||1.22 (0.891.67)||11.9||0.96 (0.60–1.52)||19.4||1.41 (0.92–2.17)||9.1||0.69 (0.33–1.43)|
|Q||A||15.4||18.8||1.26 (0.96–1.64)||18.4||1.25 (0.93–1.68)||22.8||1.56 (1.09–2.24)||11.5||0.74 (0.44–1.26)||20.0||1.27 (0.74–2.18)|
No association was observed for the MSH6 G39E and MLH1 I219V polymorphisms and colorectal cancer risk overall. However, MSH6 39E homozygotes displayed an increased risk of rectal cancer compared to the wild type (RR = 3.25, 95% CI: 1.08–9.83, p = 0.04). No association was observed for the MLH1 I219V polymorphism with tumors in any location.
Since defects in mismatch repair confer tolerance to DNA damage from alkylating agents23 (e.g., N-nitroso compounds), we hypothesized that meat intake may modify the association between mismatch repair genotypes and colorectal cancer risk. Although no significant interactions were observed with total or red meat intake (p > 0.05), processed meat intake was found to modify the association between the MSH3 polymorphisms and colorectal cancer risk (p < 0.10 for both, Table IV). Among participants with processed meat intakes above the median (≥10.1 g/day), individuals with at least one copy of the MSH3 940Q or MSH3 1036A allele displayed an increased risk of colorectal cancer, while no association was observed among participants with processed meat intakes below the median (Table IV). Similar results were observed in the haplotype analysis with both haplotypes containing the MSH3 1036A variant conveying an increased risk among participants with processed meat intakes above the median (Table V).
|Processed meat intake (g/day)||pinteraction|
|No. of cases||No. of PY||RR1||No. of cases||No. of PY||RR1|
|RQ or QQ||31||2990||0.99 (0.62–1.59)||34||2878||1.69 (1.03–2.76)||0.08|
|AT or AA||47||4726||0.88 (0.57–1.36)||55||4755||2.25 (1.34–3.76)||0.002|
|GE or EE||34||2996||1.10 (0.69–1.76)||20||2979||0.74 (0.42–1.31)||0.29|
|IV or VV||46||5160||0.88 (0.56–1.38)||45||5612||1.11 (0.67–1.84)||0.40|
|Haplotype||Processed meat intake (g/day)||pinteraction|
|R||A||13.0||9.4||0.66 (0.39–1.09)||12.6||20.9||1.93 (1.27–2.95)|
|Q||A||15.7||17.3||1.06 (0.71–1.57)||14.8||19.9||1.68 (1.07–2.65)||0.002|
No significant differences were observed between the mismatch repair polymorphisms and colorectal cancer risk when the results were stratified by sex, smoking status, folate intake, alcohol consumption or family history.
In this study, we found that genetic variants in MSH3 were associated with an increased risk of colorectal cancer. MSH3 is located at chromosome 5q11–q12 and encodes the mismatch repair protein, MSH3. MSH3 forms a heterodimer with MSH2 (MutSβ) that detects insertion-deletion loops and targets them for mismatch repair.8 No germline mutations in MSH3 have been associated with HNPCC, but mutations in MSH3 lead to a partial defect in mismatch repair and some microsatellite instability.24 Cells from MSH3-deficient mice are defective in the repair of insertion/deletion mismatches but not single base mismatches and show some microsatellite instability,25 suggesting a role for MSH3 in preventing carcinogenesis.
In our study, the MSH3 variants were more strongly associated with proximal colon cancer, which is consistent with what is known regarding the pathophysiology of colorectal tumors. Defects in mismatch repair lead to microsatellite instability,26 which is found in ∼10–16% of colorectal tumors.3, 5 The majority colorectal tumors with microsatellite instability are located in the proximal colon,27, 28, 29, 30 suggesting that defects in mismatch repair have a greater role in the etiology of proximal tumors compared to distal tumors. Consistent with this hypothesis, we found that the MSH3 940Q and 1036A variants were more strongly associated with proximal colon cancer.
Although both MSH3 variants were associated with the risk of proximal colon cancer, our analysis suggested the association observed with the 1036A variant may be because of its correlation with the 940Q allele. When both variants were examined in the same proportional hazards model, only the MSH3 940Q allele remained significantly associated with risk. Similarly, only the haplotype containing the MSH3 940Q allele was associated with proximal colon cancer; no association was observed with the haplotype containing the 1036A variant but not the 940Q allele. Both MSH3 variants encode nonconservative amino acid changes, which could potentially alter the function of the protein. The 1036T allele is not conserved across either the mouse or chimpanzee species; however, the 940R allele is conserved across both species, suggesting that the residue at codon 940 may be more important for the protein. Although it is possible that both associations observed in this study were because of the linkage disequilibrium with another unknown variant, the MSH3 R940Q polymorphism is a strong candidate for having biological importance, and additional studies examining the functional consequences of the variant are warranted.
We also found that processed meat intake modified the association between the MSH3 polymorphisms and colorectal cancer. Among the participants with processed meat intake above the median, the MSH3 940Q and 1036A variants were associated with an increased risk of colorectal cancer. Most prospective studies have shown a positive association between processed meat intake and colorectal cancer risk [reviewed by Sandhu et al.31]. Meat, particularly processed meat, contains N-nitroso compounds as the result of the curing process.32 Nitrosamines cause alkylation damage, predominantly N7- and O6-alkyl-guanine DNA adducts.33 Although mismatch repair does not repair O6-alkyl-guanine adducts, defects in mismatch repair lead to increased tolerance to O6-alkyl-guanine adducts and enhanced mutagenesis.23, 34 However, the mechanism underlying this tolerance is not well understood, and it unclear how MSH3 may play a role. Alternatively, the observed association could be because of linkage disequilibrium with a variant in the neighboring dihydrofolate reductase (DHFR) gene, which is important for the de novo synthesis of purines.
No association was observed between the MSH6 G39E polymorphism and colorectal cancer overall, but a positive association was observed between the variant and rectal cancer. MSH6 forms a heterodimer with MSH2 (MutSα) that recognizes single base insertion-deletions and single base–base mismatches.8 Like MSH3, mutations in MSH6 result in a partial loss of mismatch repair.35, 36 However, unlike MSH3-deficient cells, MSH6-deficient cells are defective in the repair of single nucleotide mismatches but not larger mismatches.35, 36, 37 The MSH6 G39E polymorphism encodes a nonconservative amino acid change, and the 39G allele is conserved across the mouse genome. It is unknown whether the variant affects the function of the protein. In our study, homozygotes for the MSH6 39E variant displayed an increased risk of rectal cancer. However, since no association was found between the MSH6 variant and colorectal cancer overall and the number of rectal cases in this study was small, the observed association with rectal cancer should be interpreted with caution.
Our study had several advantages and limitations. Since the study was nested in a fairly homogenous, community-based prospective cohort, it is unlikely to suffer from significant selection bias or population stratification. In addition, information on environmental factors, such as diet, was collected prospectively, curbing the potential for recall bias. However, since the majority of participants in our study were Caucasian Americans, the results of this study may not be generalizable to other racial or ethnic populations. As with any genetic association study, it is possible that some of our findings are false-positives, and additional studies are needed to replicate these results. In addition, tumor specimens were unavailable for this study, so we were unable to explore the relationship between genetic variants in mismatch repair genes and tumor microsatellite instability, which could have lead to additional insight.
In conclusion, our study suggests that common polymorphisms in mismatch repair genes may be associated with the risk of colorectal cancer. Genetic variants in MSH3 were associated with an increased risk of proximal colon cancer in our study, which is consistent with the hypothesis that defects in mismatch repair contribute to the etiology of proximal tumors. Additional studies are needed to confirm these results and to evaluate the functional effects of these MSH3 variants on mismatch repair capacity. Future studies may benefit from genotyping additional polymorphisms in MSH3 and conducting more comprehensive haplotype analysis.
We thank Ms. Judy Hoffman-Bolton and Ms. Alyce E. Burke for managing the data and Dr. Kenneth W. Kinzler for providing his expertise. We also thank the participants in the CLUE II cohort for making this study possible.
- 1Colorectal tumors. In: Vogelstein B,Kinzler KW, eds. The genetic basis of human cancer. New York: McGraw-Hill, 2002. 583–612., .
- 2Hereditary nonpolyposis colorectal cancer (HNPCC). In: Vogelstein B,Kinzler KW, eds. The genetic basis of human cancer. New York: McGraw-Hill, 2002. 307–21..