Dietary Intake of Folate and Riboflavin, MTHFR C677T Genotype, and Colorectal Adenoma Risk: A Dutch Case-Control Study

Colorectal adenomas are highly prevalent in the Western world. Among asymptomatic, average-risk patients, prevalence of adenomas is ∼25% in colonoscopy studies (1). As certain colorectal adenomas are considered precursors of colorectal cancer (2, 3), prevention of colorectal adenomas may decrease the occurrence of colorectal cancer.

Folate is hypothesized to have a beneficial effect on the development of colorectal adenomas and carcinomas. Folate is essential in DNA metabolism; deficiency affects DNA methylation and purine and pyrimidine synthesis (4). Vitamin B2 (riboflavin) plays a prominent role in folate metabolism. Flavin adenine dinucleotide, a metabolite of vitamin B2, serves as a cofactor for methylenetetrahydrofolate reductase (MTHFR; refs. 5-7). MTHFR is an important enzyme in folate metabolism; it catalyzes the conversion of 5,10-methylenetetrahydrofolate to 5-methyltetrahydrofolate (8). Flavin adenine dinucleotide was found to modify MTHFR activity in healthy subjects (9).

Most epidemiologic studies examining the association between folate intake or status and colorectal adenoma risk observed an inverse association (10-20), which was statistically significant in some studies (10-14), whereas two studies did not find an association (21, 22). We know of only two observational studies on vitamin B2 intake and colorectal adenoma risk: in one study no association was found (11), whereas the other study found a nonsignificant inverse association [odds ratio (OR) for highest versus lowest tertile of intake is 0.67; 95% confidence interval (95% CI) is 0.39-1.17; ref. 20].

A common C-to-T substitution in the MTHFR gene at nucleotide 677 converts an alanine to valine and is associated with decreased enzyme activity (23). Studies investigating the association of MTHFR C677T genotype and colorectal adenoma risk show nonsignificant relative risks ranging from 0.35 to 2.41 (13, 17, 19, 20, 22, 24-26). However, MTHFR C677T genotype may modify the association between intake of B-vitamins and colorectal adenomas; several studies indicate that the MTHFR TT genotype in combination with a low folate status may be a risk factor for colorectal adenomas (13, 18, 24, 26), although some studies do not show an interaction (17, 19). As far as we know, only one published study evaluated the interaction between vitamin B2 intake and MTHFR C677T genotype, in which there was no evidence of an interaction (20).

Most studies on folate and colorectal adenoma or cancer risk are conducted in the United States. Intake of folate and vitamin B2 in the United States is high compared with the Netherlands, where supplements are not regularly used and foods are not enriched with folate and only recently with vitamin B2. Therefore, we evaluated the associations between intake of folate and vitamin B2 and colorectal adenoma risk in a Dutch case-control study, taking into account potential confounding or effect-modifying variables, such as alcohol consumption, smoking, and intake of other B-vitamins. In addition, we examined whether these associations are modified by MTHFR C677T genotype.

The POLIEP study is a case-control study conducted in the Netherlands to investigate gene-environment interactions and risk of colorectal adenomas. Participants were recruited among those undergoing endoscopy in 10 outpatient clinics between June 1997 and October 2002. The study design has been previously described (27).

We defined cases as those with at least one histologically confirmed colorectal adenoma ever in their life. Controls had no history of any type of polyps, proven by complete visualization of the colon (i.e., full colonoscopy or sigmoidoscopy combined with X-ray). Eligibility criteria were Dutch speaking, of European origin, of ages 18 to 75 years at time of endoscopy, no hereditary colorectal cancer syndromes (i.e., familial adenomatous polyposis or hereditary nonpolyposis colorectal cancer), no chronic inflammatory bowel disease, no history of colorectal cancer, and no (partial) bowel resection. Response rates varied from 35% to 91% in different outpatient clinics; overall response was 54%.

Of 1,526 eligible participants, we excluded 49 subjects with insufficient dietary data. Thus, the analyses included 1,477 participants: 768 cases and 709 controls. Of 24 participants, no DNA sample was available for MTHFR genotyping and the MTHFR gene could not be amplified in one DNA sample; therefore, analyses using data on MTHFR genotype included 1,452 participants: 751 cases and 701 controls.

Participants filled out self-administered questionnaires on diet, medical history, and several lifestyle factors, according to their habits in the year previous to their colonoscopy or complaints. Dietary intake was assessed with a standardized and validated semiquantitative food-frequency questionnaire that was originally developed for the Dutch cohort of the European Prospective Investigation into Cancer and Nutrition (EPIC; ref. 28). Subsequently, intake of energy and nutrients was calculated using the Dutch food composition table. Folate intake was calculated using recently updated information of folate content in Dutch foods from Konings et al. (29). Vegetables, bread, and meat contributed more than 50% of folate intake. The reproducibility of these products as assessed with the questionnaire was for men 0.76, 0.86, and 0.68, respectively, and for women 0.65, 0.78, and 0.80, whereas the relative validities compared with 24-hour recalls were 0.31, 0.76, and 0.47 for men and 0.38, 0.78, and 0.70 for women (28). The main sources of vitamin B2 were milk and milk products, accounting for about 50% of intake. The reproducibility for milk and milk products was 0.71 for men and 0.79 for women; the relative validity was 0.73 for men and 0.78 for women (28). The Dutch EPIC food-frequency questionnaire has not yet been validated for folate intake.

To determine the MTHFR C677T polymorphism, we used the PCR-RFLP method described in detail by Frosst et al. (23) in DNA isolated from whole blood. PCR was done with internal negative controls. Laboratory staff was blinded to case-control status. To study reproducibility, 20% of the samples were analyzed in duplicate and yielded the same result. In addition, we participated in an external quality control program. Results showed a 100% match with expected genotype.

To investigate the association between nutrient intake and colorectal adenomas, we used logistic regression models. Intakes of nutrients were adjusted for total energy intake using the linear residual regression method of Willett and Stampfer (30). Sex-specific tertiles of intake were calculated based on the distribution among controls. We calculated ORs and 95% CIs to estimate the relative risk of developing colorectal adenomas. Reference groups were those with the lowest nutrient intake. To examine the association between MTHFR C677T genotype and colorectal adenomas, we used individuals with the MTHFR CC genotype as reference.

We examined whether the associations with vitamin intake were modified by sex, alcohol intake, or smoking habits, but no differences were observed between strata. We examined if potential confounding factors (i.e., age, body mass index, physical activity, educational level, smoking, use of nonsteroid anti-inflammatory drugs, indication for colonoscopy, family history of colorectal cancer, use of contraceptives, hormone replacement therapy, outpatient clinic, and intake of fat, fiber, alcohol, folate, vitamin B2, vitamin B6, vitamin B12, calcium, total meat, organ meat, fruits, and vegetables) were associated both with colorectal adenomas and vitamin intake, and changed the crude estimates by more than 10% when added to the logistic regression models. The final logistic regression models included the covariate age; the vitamin B2 models also included dietary intake of folate and calcium.

Interaction between MTHFR C677T genotype and intake of vitamins was studied by stratification to genotype, using those categorized in the lowest tertile of intake and with the MTHFR CC genotype as reference, and by testing for different slopes associated with nutrient intake across genotype.

To test for linear trend, we modeled the tertile of nutrient intake as a continuous variable in the logistic regression model, in which each tertile was assigned its median value.

All tests of statistical significance were two sided and the significance level was set at 5%. We used Statistical Analysis Software (SAS version 8, SAS Institute, Gary, NC) for all analyses.

In Table 1, characteristics of the study population are shown. Compared with controls, cases were more likely to be male, older, have a slightly higher body mass index, and smoke more. Furthermore, cases usually had a colonoscopy for screening rather than because of complaints, used less contraceptives and hormone replacement therapy, and had a higher intake of total energy, vegetables, fat, alcohol, vitamin B6, and folate. Most of these differences remained after standardization for age and sex.

Table 2 shows the associations between dietary intake of folate and vitamin B2 and colorectal adenoma risk for the whole study population and stratified by MTHFR C677T genotype. A high dietary folate intake was positively associated with risk of colorectal adenomas, with a borderline significant test for trend (P = 0.054). A high intake of vitamin B2 was inversely associated with colorectal adenoma risk, with a statistically significant test for trend. No association was observed between intake of vitamins B6 and B12 and colorectal adenomas (data not shown). ORs did not change substantially when we excluded those using multivitamin supplements (data not shown).

Compared with individuals with the MTHFR CC genotype, MTHFR CT and TT genotypes were not related to colorectal adenoma risk. The age- and sex-adjusted ORs (95% CIs) were 1.06 (0.84-1.33) and 0.96 (0.66-1.39), respectively (data not shown). For intake of folate (Table 2), vitamin B6, and vitamin B12 (data not shown), no effect modification by MTHFR C677T genotype was observed. The inverse association for vitamin B2 intake was present in all MTHFR genotypes, but seemed more pronounced among those with the MTHFR TT genotype. The interaction term did not reach statistical significance (Table 2).

Folate seemed to be a risk factor with low or medium vitamin B2 intake (Table 3). Moreover, the inverse association between vitamin B2 and colorectal adenomas was particularly seen when folate intake was relatively high. Statistically, the interaction was not significant (P = 0.64).

In this endoscopy-based case-control study, we observed a slightly positive association between dietary folate intake and colorectal adenoma risk, which was especially apparent for those with low vitamin B2 intakes. MTHFR genotype did not modify this association. An inverse association between dietary intake of vitamin B2 and colorectal adenoma risk was found, which may especially be important among those with relatively high folate intake or those with the MTHFR TT genotype.

As far as we know, one other publication exists on the interaction between vitamin B2 and folate in colorectal carcinogenesis, in which there was a pattern of decreased adenoma risk among those with high intakes of folate and vitamin B2 compared with those with low intakes of folate and vitamin B2. However, in that study the intake of vitamin B2 was substantially higher than in our study (20).

In a rat model, it was shown that MTHFR was affected by riboflavin deficiency (5). However, this does not explain why folate intake increases colorectal adenoma risk in those with a low vitamin B2 status. We speculate that folate will become protective in colorectal carcinogenesis only when vitamin B2 intake is high enough. One in vitro study showed that at low riboflavin level, nuclear buds decreased significantly with increasing folic acid level, whereas at high riboflavin level, nuclear buds decreased even more with increasing folic acid level. In that study, the interaction between folic acid and riboflavin was statistically significant (31). These observations might explain why we cannot reproduce most American results, suggesting a protective role of folate intake in colorectal carcinogenesis; cereals in the United States have been enriched with vitamin B2 since 1943 (32), and mean vitamin B2 intake is about 28% higher than in the Netherlands (33, 34).

The positive, but statistically nonsignificant, association between dietary folate intake and colorectal adenoma risk conflicts with results from other studies showing an inverse association (10-20) or no association at all (21, 22). Our study population had a low folate intake and the range of intake was narrow compared with that in other studies. Supplement use is not common in the Netherlands; in the Dutch Food Consumption Survey 1998, using a 2-day dietary record, 9.5% of participants reported having used multivitamin supplements, including B-vitamin supplements, on one or both days (34). Furthermore, as vitamin intake from supplements was not calculated in our study, we focused on dietary intake. In some studies, the inverse association between folate intake and colorectal adenoma risk weakened when the analyses were restricted to dietary folate (10, 24). A limitation of the present study and many other epidemiologic studies to date, is the use of a food-frequency questionnaire for the assessment of dietary folate intake. This EPIC questionnaire was not validated for folate intake. However, most food-frequency questionnaires that were validated for folate intake show poor correlation between erythrocyte folate and dietary folate intake (15, 35, 36). Validations for total folate intake show higher correlations (10).

Furthermore, the methods of assessment of folate contents from foods may differ between studies. In our study, we used data of folate contents in foods assessed using a high-performance liquid chromatography–based method, which overall leads to about 25% lower estimates of folate contents than data based on microbiological assays (29). However, in the Netherlands Cohort Study on Diet and Cancer, which used the same method of folate measurement, an inverse, although not statistically significant, association between folate and colon cancer in men and women and between folate and rectal cancer in men was suggested. The intake of folate was about 210 μg/d, which is somewhat higher than in our study (37).

An alternative explanation for the positive association between folate intake and colorectal adenomas as found in our study might have been consumption of liver. Liver contains not only a high level of folate but also of carcinogens and thus may drive up the ORs for folate. Consumption of liver was not specifically assessed in our study. However, total organ meat was assessed and did not confound the results. Therefore, we assume that the influence of liver consumption, if present at all, will be small.

We found an inverse relationship between vitamin B2 intake and colorectal adenoma risk. An explanation may be that vitamin B2 deficiency reduces MTHFR activity (5). An inverse association was also found in one other study (20), whereas another study did not find an association (11).

We did not find an indication that MTHFR C677T genotype modifies the association between intake of folate and colorectal adenoma risk, which is in line with some studies (17, 19), but not with others (13, 18, 24, 26), not depending on study size. Our data suggest a nonsignificant interaction between vitamin B2 intake and MTHFR C677T genotype in colorectal adenoma occurrence; the inverse association between vitamin B2 intake and colorectal adenomas seemed more pronounced among MTHFR TT individuals. In a study examining the structure and properties of MTHFR from Escherichia coli, it was found that the E. coli MTHFR Ala177Val polymorphism, corresponding to the human C677T polymorphism, increases the tendency for MTHFR to lose its flavin adenine dinucleotide cofactor (6). This finding was also observed in recombinant human MTHFR, although more subtle (7), which may explain our finding of an interaction between vitamin B2 intake and MTHFR genotype. In human studies, it was observed that the homocysteine-lowering effect of vitamin B2 was essentially confined to subjects carrying the MTHFR TT genotype (9, 38), or even to subjects who carry the MTHFR TT genotype and have low folate status (39).

Case-control studies may be limited by selection and information bias. As screening for colorectal cancer is not common in the Netherlands, most endoscopies are conducted for bowel complaints, which may influence dietary patterns and introduce information bias. However, when we excluded people who had changed their dietary habits because of these complaints (166 cases and 232 controls), it did not affect the results. Information bias might also be caused by the fact that we included prevalent cases, but when we excluded prevalent cases (n = 363), results were essentially the same.

In summary, the results from this study do not provide evidence for a decreased colorectal adenoma risk for subjects with high levels of folate intake. Folate seems to be a risk factor for colorectal adenomas, especially when vitamin B2 intake is low. This may particularly be relevant for populations where products are not (yet) or only recently enriched with B-vitamins, such as the Netherlands.

Additionally, the study indicates an inverse association between vitamin B2 intake and colorectal adenoma risk, especially with higher folate intake. This study indicates that there may be an interplay between MTHFR C677T genotype and intake of vitamin B2, but not of folate, in association with colorectal adenoma risk. Although the study does not provide enough evidence to draw firm conclusions, it brings an interesting speculation about the interactive roles of folate and vitamin B2 in colorectal carcinogenesis, which should be further elucidated.

We are grateful to all the people who kindly participated in this study. The participants were recruited with help from the endoscopy staff of the following hospitals in the Netherlands: Slingeland Ziekenhuis (Doetinchem), Ziekenhuis Gelderse Vallei (Ede), Radboud University Nijmegen Medical Centre (Nijmegen), Antonius Ziekenhuis (Nieuwegein), Meander Medisch Centrum (Amersfoort), Ziekenhuis Rijnstate (Arnhem), Ziekenhuis Rivierenland (Tiel), Slotervaart Ziekenhuis (Amsterdam), Jeroen Bosch Ziekenhuis ('s Hertogenbosch), and Canisius Wilhelmina Ziekenhuis (Nijmegen). We thank Maria van Vugt and Elly Monster (Division of Human Nutrition, Wageningen University, the Netherlands) for collecting blood samples and assistance with the conduct of this study.