Methionine Synthase D919G Polymorphism, Folate Metabolism, and Colorectal Adenoma Risk

Substantial evidence suggests that folate-mediated one-carbon metabolism plays a role in colorectal carcinogenesis (1, 2). Low-folate diets have been found to increase the risk of colorectal cancer and adenomatous polyps (a cancer precursor; Refs. 3, 4). Increased risks observed with low intake of vitamin B₁₂, vitamin B₆, and methionine and with high intake of alcohol, which acts as a folate antagonist, may be attributable to their involvement in this metabolic pathway [for reviews, see Choi and Mason (1) and Giovannucci (5)]. In addition, polymorphisms in the folate-metabolizing enzymes methylenetetrahydrofolate reductase (MTHFR) and thymidylate synthase (TS or TYMS) have been associated with colorectal cancer and adenoma risk in several studies, possibly via interaction with relevant dietary factors (6, 7, 8, 9, 10). Potential mechanisms for these associations involve both DNA methylation and nucleotide synthesis (1, 11).

Methionine synthase [5-methyltetrahydrofolate-homocysteine S-methyltransferase (MTR)], on chromosome 1q43, encodes one of several key enzymes involved in the folate-mediated one-carbon metabolism. It catalyzes the methylation of homocysteine to methionine with simultaneous conversion of 5-methyl-tetrahydrofolate (5-methyl-THF) to tetrahydrofolate (THF). MTR is essential for the provision of S-adenosylmethionine, the universal donor of methyl groups, as well as the provision of THF for use in nucleotide synthesis.

A common MTR variant consists of an A-to-G transition at base-pair 2756 and leads to a change from aspartic acid to glycine at codon 919 (D919G; Ref. 12). Although the direct functional impact of this polymorphism has not been established, there is some evidence that this may be an activating polymorphism; in some studies, individuals with GG genotype have lower homocysteine concentrations (13, 14, 15, 16) and higher serum folate concentrations (14). Other studies, however, suggest no functional differences (17, 18, 19, 20). In a clinic-based case-control study, we sought to assess associations between the MTR D919G polymorphism and risk of colorectal adenoma. We also examined possible interactions with dietary intakes of vitamin B₁₂, vitamin B₆, methionine, folate, and alcohol and with polymorphisms in MTHFR (C677T) and in TS [TS enhancer region (TSER) 28-bp tandem repeats and 1494del6].

Subject recruitment for the Minnesota Cancer Prevention Research Unit case-control study has been described previously (21). Briefly, cases and controls were recruited through a multiclinic private gastroenterology practice in metropolitan Minneapolis, Digestive Healthcare. Patients who were scheduled for colonoscopy between April 1991 and April 1994 were screened for eligibility and recruited before colonoscopy. Recruitment at Digestive Healthcare sites was initiated at the time of scheduling so that both patient and recruiter would be blind to the final diagnosis. The study protocol was approved by the internal review boards of the University of Minnesota and each Digestive Healthcare site, and written informed consent was given by each study participant.

Eligibility criteria for both cases and controls were (a) resident of the Twin Cities metropolitan area; (b) age 30–74 years; (c) English speaking; (d) no known genetic syndrome associated with colonic neoplasia; and (e) no history of cancer (except nonmelanoma skin cancer), adenomatous polyps, or inflammatory bowel disease. Indications for colonoscopy have been published previously (9); 68% of all colonoscoped patients participated.

Colonoscopy findings were recorded at the colonoscopy visit, and only participants with a complete colonoscopy reaching the cecum were eligible. All polyps were removed and evaluated histologically. Patients with polyps showing invasive carcinoma were not included. Adenomatous polyp cases were those found to have at least one adenomatous polyp; controls were polyp free at colonoscopy. To reduce the possibility of population admixture, this analysis was restricted to Caucasians (97.7% of the study sample).

At the colonoscopy visit, questionnaires were administered and blood was drawn. Data collected included information on dietary intake, physical activity, smoking habits, anthropometric measurements, medical information, demographic information, and reproductive history. When data were incomplete, study staff followed up by telephone. The dietary questionnaire used was an adaptation of the Willett food-frequency questionnaire, which has been studied previously for validity and repeatability (22). Folate values derived from this questionnaire correlated (r = 0.56) with RBC folate levels (an indicator of long-term folate status; Ref. 4).

Genomic DNA was extracted from peripheral WBCs with a Puregene kit (Gentra Systems, Minneapolis, MN). The MTR D919G polymorphism was genotyped on a 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA) with a fluorescent allelic discrimination assay. Amplification was performed in 20-μl reactions using the TaqMan core reagent kit (Applied Biosystems) with 200 nm each of the amplification primers 5′-TGTTCCCAGCTGTTAGATGAAAATC-3′ and 5′-AATCTGTTTCTACCACTTACCTTGAGAGA-3′, 100 nm each of the MGB probes (Applied Biosystems) 5′VIC-AGACAGGACCATTATG3′-NFQ (D allele) and 5′-6-FAM-AGACAGGGCCATTAT3′-NFQ (G allele), and 10 ng of genomic DNA. Cycling conditions were 50°C for 2 min, 95°C for 10 min, and 40 cycles of 95°C for 15 s and 60°C for 1 min. Each plate contained negative and positive controls for all genotypes. For quality control purposes, genotyping was repeated for 94 randomly selected samples, and there were no discrepancies between results. Genotype frequencies did not deviate from those expected under Hardy–Weinberg equilibrium. Genotyping methods for MTHFR C677T, the TSER 28-bp repeat polymorphism, and TS 1494del6 [deletion of TTAAAG in the 3′-untranslated region (UTR)] are described elsewhere (8, 9).

Logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for associations between MTR genotypes and risk of colorectal adenoma. The following confounders or adjustment variables were identified as those whose inclusion modified MTR ORs: age, sex, hormone therapy (women only; ever/never), body mass index, pack-years of smoking, regular use of nonsteroidal anti-inflammatory drugs (≥1/week), daily energy intake (kilocalories), dietary fiber, percentage of energy from fat, and daily dietary intakes of folate, vitamin B₁₂, vitamin B₆, methionine, and alcohol. In general, the impact of confounding was minimal; however, for consistency of reporting, multivariable-adjusted ORs are presented. Regular use of aspirin (≥1/week), hours of physical activity, daily dietary intake of protein, and percentage of energy from protein did not confound any association studied.

Examination of gene-diet interactions (MTR D919G with daily dietary intakes of folate, vitamin B₁₂, vitamin B₆, methionine, and alcohol) consisted of OR estimation using a common referent group of individuals with DD genotype and the highest nutrient intake level (or lowest alcohol intake level). In this manner, patterns of genotype-specific risks within dietary intakes and dietary-specific risks within genotype were observed. Tests of trends for risks associated with dietary intakes within genotype groups consisted of likelihood ratio comparisons of grouped linear versus quantitative parameterizations of dietary variables. Because trends by dietary intakes were expected, significance testing of gene-diet interactions consisted of comparisons across MTR genotype of the within-genotype trends in risk associated with dietary intakes.

Examination of gene-gene interactions (MTR D919G with MTHFR C677T, the TSER repeat polymorphism, and TS 1494del6) consisted of OR estimation using a common referent group of individuals with wild-type genotypes. In this manner, patterns of MTR genotype-specific risk within other genotype groups and other genotype-specific risks within the MTR genotype were observed. Significance of gene-gene interactions was assessed with likelihood ratio testing of models with relevant multiplicative interaction terms.

A total of 513 individuals with adenomatous polyps (190 women and 323 men) and 609 controls (373 women and 236 men) were genotyped at the D919G polymorphism of MTR. The G allele frequency was 17% among women controls and 19% among men controls. MTR genotype frequencies and selected characteristics of study participants are shown in Table 1; other characteristics have been described previously (9, 21). Adjusted ORs for individuals with DG (n = 344) and GG (n = 42) genotypes were 1.1 (95% CI, 0.8–1.5) and 1.6 (95% CI, 0.8–3.2), respectively, compared with individuals with DD genotype (n = 736), suggesting only a modest, statistically nonsignificant increase in adenoma risk with D919G. These results contradict the inverse association predicted by some functional studies (13, 14, 15), but are consistent with a previous study of colorectal cancer (23). No statistically significant interaction between genotype and sex was seen; we observed adjusted ORs of 1.4 (DG versus DD; 95% CI, 0.9–2.1) and 1.7 (GG versus DD; 95% CI, 0.7–4.3) for women and 1.0 (DG versus DD; 95% CI, 0.6–1.4) and 1.6 (GG versus DD; 95% CI, 0.5–4.6) for men. These estimates did not vary when stratified by polyp size, as another study of colorectal adenoma in women suggested (24), or age, number of polyps, or location of largest polyp (data not shown).

Results of analyses stratified by daily dietary intake of folate, vitamin B₁₂, vitamin B₆, methionine, and alcohol are shown in Table 2. We hypothesized that associations between MTR D919G genotype and adenoma may be modified by the amount of substrate (folate), product and inhibitor (methionine), relevant cofactors (vitamins B₁₂ and B₆), or factors affecting folate absorption (alcohol); however, no interactions with intake of these nutrients were seen (Table 2).

Among women, an interaction was suggested between daily methionine intake and MTR D919G (P = 0.05), such that the expected increased adenoma risk with lower methionine intake was seen only among women with a MTR DG or GG genotype (Table 2). No association with methionine intake was found among women with the DD genotype. When women with DG or GG genotypes were analyzed as one group, increasing ORs of 0.6 (95% CI, 0.3–1.4), 1.2 (95% CI, 0.5–2.7), and 1.6 (95% CI, 0.7–3.7) were seen with high (>2.18 g), medium (1.54–2.18 g), and low (<1.54 g) daily methionine intake, respectively, compared with women with DD genotype and in the high tertile of methionine intake (P for interaction = 0.01). Among men, patterns of risk associated with methionine intake did not seem to vary by MTR genotype (Table 2). Similar results were seen when sex-specific methionine tertile cut-points were used (women, 1.44 and 2.01 g/day; men, 1.68 and 2.37 g/day).

A statistically significant interaction between MTR D919G genotype and current alcohol intake was observed among women (P = 0.03). No association with alcohol intake was observed among women with the DD genotype; however, an increased risk of adenoma was seen with intake of alcohol among women with DG or GG genotypes (Table 2). When women with DG or GG genotypes were analyzed jointly, increasing ORs of 0.8 (95% CI, 0.4–1.4), 1.9 (95% CI, 1.0–3.6), and 3.3 (95% CI, 1.4–8.1) were seen with intake of no alcohol, less than median intake among alcohol consumers (≤7 g/day), and more than median intake (>7 g/day), respectively, compared with women with DD genotype and no alcohol intake (P for interaction = 0.01). When the same referent group was used, women with DG or GG genotypes who drank >11 g/day had a somewhat greater increase in risk (OR = 4.8; 95% CI, 1.5–15.7). A similar pattern of interaction was suggested among men (Table 2). Similar trends were seen when sex-specific median alcohol intake cut-points were used (women, 3.9 g/day; men, 12.0 g/day).

We hypothesized that associations between MTR D919G genotype and colorectal adenoma may be modified by variation in genes encoding other folate-metabolizing enzymes. MTHFR C677T and the TSER 28-bp repeat polymorphism have been shown to be functionally relevant (giving decreased MTHFR activity and TS expression, respectively). The MTHFR and TSER genotypes did not appear to interact with the MTR genotype in relation to adenoma risk among women or men (Table 3).

A statistically significant interaction (P = 0.007) was seen between MTR D919G and TS 1494del6 among women, suggesting that they may have a joint role in adenoma risk. Women who were either homozygous variant at one polymorphism and homozygous wild type at the other or were doubly heterozygous appeared to have an increased risk compared with those with wild-type genotypes at both polymorphisms [ORs (95% CIs) versus MTR DD and TS 3′ UTR wild type (wt)/wt were as follows: DD and deletion (del)/del, 2.8 (1.2–6.5); GG and wt/wt, 5.0 (1.3–19.5); DG and wt/del, 2.6 (1.4–4.9)]. The functional significance of the TS 3′-UTR polymorphism has not yet been reported. We found only 22 women with the GG genotype, and Bonferroni adjustment for the number of strata and gene-gene interactions assessed reduced the significance of the interaction to P = 0.04.

In summary, our analysis of colorectal adenoma cases and colonoscopy-screened controls suggests that the MTR D919G polymorphism may have a modest role in colorectal adenoma development, particularly among women. Our data additionally suggested that the role of MTR D919G in women may become more apparent in the presence of higher alcohol intake or lower methionine intake. We observed a statistically significant interaction between this MTR polymorphism and the TS 3′-UTR polymorphism 1494del6 among women. A strength of this study is the use of fully colonoscoped subjects, which led to a clear distinction between cases and polyp-free controls.

Our observation of a modest increase in colorectal adenoma risk with the MTR D919G variant genotypes, particularly among women [for DG/GG versus DD, OR = 1.4 (95% CI, 0.9–2.1)] contributes to a varied literature. Previous case-control analyses of MTR D919G have suggested increased risk for adenoma (25) and colorectal cancer (23) and no reported sex differences, and one analysis found no suggestion of colorectal cancer association (25). Some functional studies have indicated that the variant G allele is associated with lower homocysteine (13, 14, 15, 16) and higher folate concentrations (14); because these levels have been consistently associated with decreased risk of adenoma, we hypothesized that the G allele would be inversely associated with adenoma risk. In addition, nested case-control analyses from the Nurses Health Study and the Physician’s Health and Health Professional Follow-up Studies found decreased adenoma risks for women (24) and decreased colorectal cancer risks for men (26), respectively.

We examined whether differences in results among studies may be attributable to differing baseline folate-related nutrient levels and found that women in one study showing an inverse association had slightly higher folate intake than the women in the present study [tertiles of 310 and 508 μg/day (24) versus 267 and 434 μg/day]. Exclusion of the participants with lower folate levels (<280 or <300 μg/day) in the current analysis did not change the results, and an inverse association was not suggested. It is important to note that, in this and other studies, 95% CIs for variant genotype OR estimates included 1.0, and that the true relationship between MTR D919G and colorectal cancer or adenoma may thus be no association. Our observation that MTR D919G may play a more important role in colorectal carcinogenesis among women than men was not reported in other colorectal cancer or adenoma studies.

One difficulty in interpreting the observed associations arises from the lack of data regarding the functional impact of the MTR D919G polymorphism on protein function. Studies of homocysteine levels that suggested this might be an activating polymorphism examined only male study participants (13, 14, 15) or consisted of more than 75% men (16). Other studies of both men and women have not observed this association (17, 18, 19, 27), and the authors of one study reported higher homocysteine levels only among men with the variant allele and no difference among women (28). It is possible that the impact of the variant allele differs between men and women and that additional studies stratified by sex may help discern any differences. It is also possible that discrepancies in the results from functional studies may be attributable to differences in the studied populations in folate or vitamin B₁₂ intakes, which were often not reported.

We had hypothesized that any association between MTR D919G genotype and risk of adenoma could be modified by intake of folate (which interacts with other folate-related polymorphisms and adenoma risk), vitamin B₁₂ (a required MTR cofactor), vitamin B₆ (a required cofactor for conversion of THF to 5,10-methylene-THF), methionine (the product of and inhibitor of MTR), or alcohol (an inhibitor of folate absorption). No suggestions of interaction with folate, vitamin B₁₂, or vitamin B₆ were seen among women or men, consistent with other published studies of colorectal cancer and adenoma (23, 24, 26). MTR D919G may interact with a wider variation in intake levels of these nutrient than existed in the present study; however, no evidence for this was seen in the current clinic-based sample or other population-based samples.

Our results suggest that the MTR DG and GG genotypes are associated with increased risk of colorectal adenoma in the presence of low methionine intake among women; no interaction was seen among men. Because of limited information on methionine in most nutrient databases, methionine measurements from food frequency questionnaires may be error prone, leading to a bias of ORs toward the null. However, this would not be expected to occur preferentially among women with DD genotypes, in whom decreased risk with lower methionine intake was in fact suggested. Methionine synthase is subject to product inhibition through methionine (29), although to our knowledge, a potential impact of the MTR polymorphism on the product inhibition through methionine has not yet been investigated. Other colorectal cancer and adenoma studies have not reported any interaction between MTR genotype and methionine intake.

We also report that MTR DG and GG genotypes were associated with increased risk of colorectal adenoma in the presence of higher alcohol consumption (>7 g/day) among women and that additional increased risk was observed with higher intake (>11 g/day). A similar pattern was also seen among men, although tests of interaction did not meet statistical significance at the α = 0.05 level. These observations are consistent with a previous study of colorectal cancer in men in which those with GG genotype who had higher alcohol intake (≥1 drink/day) had an increase in risk for colorectal cancer (OR = 2.6; 95% CI, 0.7–10.8) that was not seen in other genotype groups (P = 0.04; Ref. 26). Unfortunately, measurement of alcohol consumption may be unreliable, particularly among women, who underreport usage more often than men. High alcohol intake has been shown to decrease folate absorption (30, 31), and acetaldehyde can inhibit MTR activity (32). although the effects of moderate alcohol consumption are less well described and results based on alcoholic subjects, animals, or in vitro analyses may be less relevant to the population in the present study.

We observed a statistically significant interaction between MTR D919G and the TS 3′-UTR polymorphism 1494del6 among women (P = 0.007). Women who were either homozygous wild type at one polymorphism and homozygous variant at the other or doubly heterozygous appeared to have an increased risk compared with those with wild-type genotypes at both polymorphisms. Although the functional relevance of the TS deletion polymorphism is unknown at present, these results are consistent with the two polymorphisms having opposite associations with adenoma risk that become obvious for each only in the presence of the other. The relationship between these two polymorphisms and adenoma risk may be attributable to differential provision of metabolites to various “products” of one-carbon metabolism, particularly S-adenosylmethionine production or pyrimidine synthesis. No interactions were suggested with MTHFR C677T or the TSER 28-bp repeat polymorphism. Both of these polymorphisms have previously been shown to yield a functional change and to be associated with colorectal cancer or adenoma risk in the presence of low folate levels (6, 8, 9, 23, 33). It is possible that additional higher-order interactions exist between intakes of relevant nutrients and polymorphisms in MTR, MTHFR, and TS; however, a dataset the size of our study population lacks the power to detect three-way interactions between these exposures and MTR genotypes with the stratification approach used at present.

Our results should be interpreted with caution because the functional impact of the MTR D919G polymorphism is not well established and small numbers of individuals in some strata increase the likelihood that variation in risk estimates by MTR D919G genotype may be attributable to chance. This study is also limited in not having data on homocysteine levels, which may be considered an intermediate phenotype. Considering the complexity of folate-mediated one-carbon metabolism and genetic variability in multiple enzymes, even relatively large molecular epidemiological studies may not capture these interactions adequately, particularly as the number of possible comparisons continues to increase. Although the present study provides hypotheses for future research, a continued interdisciplinary approach integrating nutrition science, biochemistry, epidemiology, and mathematical modeling is needed to understand this increasingly important biological pathway.

We thank Roberd Bostick and Lisa Fosdick for contributions to the original study design and data collection; Marian Neuhouser, Yutaka Yasui, Kim Robien, and Jacob Selhub for helpful discussions; Juanita Leija for genotyping; and David Rider for programming.