Higher folate intake 1 and blood folate concentrations 2 4

Transcription

1 GASTROENTEROLOGY 2006;131: Dietary Factors and Biomarkers Involved in the Methylenetetrahydrofolate Reductase Genotype Colorectal Adenoma Pathway MARÍA ELENA MARTÍNEZ,*, PATRICIA THOMPSON*, ELIZABETH T. JACOBS,*, EDWARD GIOVANNUCCI,,,# RUIYUN JIANG,* WALT KLIMECKI,** and DAVID S. ALBERTS*,, *Arizona Cancer Center, Mel and Enid Zuckerman Arizona College of Public Health, Department of Pathology, **Arizona Respiratory Center, and the Department of Medicine, University of Arizona, Tucson, Arizona; Department of Nutrition and Department of Epidemiology, Harvard School of Public Health, Boston, Massachusetts; and # Channing Laboratories, Harvard Medical School, Boston, Massachusetts Background & Aims: Methylenetetrahydrofolate reductase (MTHFR) is involved in intracellular folate homeostasis and metabolism. We assessed 2 polymorphisms in the MTHFR gene (C677T and A1298C) in relation to colorectal adenoma recurrence and conducted analyses to investigate their joint effects with plasma and dietary markers of folate status. Methods: We prospectively analyzed data from 1598 individuals genotyped for the C677T polymorphism and 1583 with data on A1298C. Results: Among nonusers of multivitamin supplements, compared with wild-type carriage, higher odds of recurrence were observed for those with the 677 TT variant (odds ratio [OR], 1.66; 95% confidence interval [CI], ) and a nonsignificant increase was observed among those with the 1298 CC variant (OR, 1.50; 95% CI, ). Diplotype analyses among nonusers of multivitamins showed that individuals who carry the MTHFR 677TT_1298AA or 677CC_1298CC combination were significantly more likely to have a recurrence compared with those with the double wild-type (OR, 2.05 for TT_AA and 1.85 for CC_CC). Higher odds of recurrence were observed among participants with low folate intake or plasma folate and the 677 TT or 1298 CC variants compared with those with lower levels and the wild-type or heterozygous genotypes. Stronger associations were shown for the combination of high homocysteine and the 677 TT variant (OR, 2.29; 95% CI, ) but not the 1298 CC variant (OR, 1.09; 95% CI, ). Conclusions: We propose that the effect of the MTHFR genotypes on increasing risk of adenoma recurrence in the presence of a low folate status is through their increase in homocysteine concentrations, which in turn could result in DNA hypomethylation via pathways involving S-adenosylhomocysteine. Higher folate intake 1 and blood folate concentrations 2 4 have been shown to be associated inversely with risk of colorectal neoplasia. Methylenetetrahydrofolate reductase (MTHFR) is a crucial enzyme involved in folate homeostasis and metabolism. MTHFR converts 5,10 methylenetetrahydrofolate, required for conversion of deoxyuridylate to thymidylate, to 5-methyltetrahydrofolate, which provides a methyl group required to convert homocysteine to methionine. Methionine is the precursor of S-adenosylmethionine (SAM), the universal methyl donor necessary for biological methylation reactions. Depletion of thymidylate results in uracil misincorporation into DNA, leading to single-strand and double-strand breaks. 5,6 A common functional polymorphism at nucleotide 677 (C T) in the MTHFR gene in the catalytic domain that results in a substitution of alanine into valine has been associated with reduced MTHFR enzyme activity 7 and increased homocysteine concentrations Most studies show that individuals homozygous for the MTHFR TT variant are at lower risk of colon or colorectal cancer compared with those with the CC wild type. A recent meta-analysis 12 showed that compared with individuals with the CC wild-type, the odds ratio (OR) for colorectal cancer for carriers of the TT variant was 0.82 (95% confidence interval [CI], ). In contrast, no inverse association is evident for the presence of the MTHFR TT variant and adenoma end points, where a combined OR of 1.02 (95% CI, ) was shown in the meta-analysis. 12 A second polymorphism in the MTHFR gene, an A C transition at nucleotide 1298, which results in a substitution of glutamate by alanine, located in the presumed regulatory domain has been identified 13 and has been shown to be in strong linkage disequilibrium with the C677T site. The MTHFR A1298C polymorphism has been shown to result in lower enzyme activity 14 but the effect of reduced activity on risk of cancer is currently unknown. Carriage of the A1298C polymorphism does not uniformly lead to increased homocysteine levels 13,15 17 and its functional significance is controversial. Epidemiologic studies of the A1298C MTHFR variant also support an inverse association between the presence of the CC variant and colorectal cancer; however, only 1 study to date has been published on the association of this variant with adenoma prevalence, 18 which showed a weak nonsignificant inverse association. Data from the previously noted meta-analysis 12 showed an OR of 0.83 (95% CI, ) for colorectal cancer for the presence of the CC variant compared with the AA wild-type. A number of epidemiologic studies have provided data on the interaction between folate intake or blood levels and MTHFR C677T in relation to colorectal cancer or adenoma. Results of most published studies support the notion that an adequate folate status results in a lower risk of colorectal neoplasia in carriers of the MTHFR TT genotype, whereas a folate-deficient status increases risk. 21,25 In addition, given that Abbreviations used in this paper: CI, confidence interval; MTHFR, methylenetetrahydrofolate reductase; OR, odds ratio; SAH, S-adenosyl homocysteine; SAM, S-adenosylmethionine by the AGA Institute /06/$32.00 doi: /j.gastro

2 December 2006 MTHFR AND COLORECTAL NEOPLASIA 1707 alcohol is a folate antagonist, 26 in some, 22,24,27 29 but not all, studies, high alcohol consumption either negates the protective effect of the TT genotype or leads to increased risk. We have reported that higher folate intake, blood levels, or lower homocysteine concentrations were associated with lower odds of recurrence of colorectal adenomas 4 and that this effect was most pronounced among nonusers of multivitamins. 34 We now investigate the role of the 2 common polymorphisms in the MTHFR gene (C677T and A1298C) in relation to recurrence of adenomatous polyps. Hypothesizing that genetic background has a strong modifying effect on folate status, we further conducted analyses to assess the joint effect between markers of folate (folate intake, plasma folate, and plasma homocysteine), alcohol consumption, and the MTHFR polymorphisms in relation to adenoma recurrence. We were particularly interested in assessing the role of homocysteine, given the paucity of published data on this important biomarker in the folate colon carcinogenesis pathway. Materials and Methods Study Populations Analyses were conducted in participants from the Wheat Bran Fiber and the Ursodeoxycholic Acid trials, the details of which have been reported elsewhere A brief description of each study is provided later. The studies were approved by the University of Arizona Human Subjects Committee and the Institutional Review Board. The Wheat Bran Fiber trial was a double-blind controlled trial conducted to compare the effect of a high-fiber vs a low-fiber cereal supplement on adenoma recurrence among individuals who had undergone colonoscopy and had 1 or more adenoma(s) removed. Participants were randomized between 1990 and 1995 and were followed-up for an average of 36.8 months. A total of 1429 participants were randomized into the trial and 1304 (91.3%) completed the study by undergoing 1 or more colonoscopies after randomization. The Ursodeoxycholic Acid trial was a double-blind controlled trial conducted to compare the effect of ursodeoxycholic acid on adenoma recurrence among patients who had undergone colonoscopy and had 1 or more adenoma(s) removed. Participants were randomized between 1996 and 2000 and were followed-up for an average of 31.0 months. A total of 1285 participants were randomized to either the treatment or the placebo group. Of these, 1192 participants (92.8%) completed the study. In both studies we excluded individuals with a personal history of inflammatory bowel disease or hereditary colon cancer syndromes. Adenoma recurrence was defined as detection of 1 or more colorectal adenomas(s) or a colorectal cancer occurrence after randomization. Neither intervention had a significant effect on adenoma recurrence. 36,37 Dietary Intake and Alcohol Consumption We assessed dietary intake and alcohol consumption using the Arizona Food Frequency Questionnaire, a computerscannable, semiquantitative food frequency questionnaire that asks about usual dietary intake over the previous year. Details regarding the development of the Arizona Food Frequency Questionnaire 38 and assessment of validity and reliability have been reported previously. 39 The Arizona Food Frequency Questionnaire also contains a vitamin/mineral supplement use section that includes data collected on brand, dose, and frequency of use. The corresponding vitamin/mineral supplement database contains more than 180 nutritional supplement preparations. Plasma Folate and Homocysteine Blood was drawn from fasting participants and placed in tubes containing heparin. Analyses of folate and homocysteine levels in plasma were conducted at the University of California Los Angeles using the Bio-Rad Quantaphase II Radioassay kit (Bio-Rad, Hercules, CA) and the IMMULITE Folic Acid analysis kit (Diagnostic Products Corporation, Los Angeles, CA) as previously described. 34 For analyses of homocysteine, the reversed-phase high-performance liquid chromatography method by Kuo et al, 40 and the IMMULITE 2000 solid-phase, competitive chemiluminescent enzyme immunoassay (Diagnostic Products Corporation) were used as previously described. 34 MTHFR Genotyping DNA samples were available for 697 participants in the Wheat Bran Fiber trial and 922 in the Ursodeoxycholic Acid trial for analyses of MTHFR genotypes. Genomic DNA was extracted from whole blood using a phenol-chloroform methodology. MTHFR genotyping reactions were performed using the 5= exonuclease reaction (TaqMan; Applied Biosystems, Foster City, CA). Real-time polymerase chain reaction allelic discrimination assays were designed using the Assay-by-DesignSM service offered by Applied Biosystems. Reaction mixtures were assembled robotically, consisting of 1 TaqMan Universal polymerase chain reaction Master Mix (Applied Biosystems), polymerase chain reaction primers (forward and reverse) at 900 nmol/l, dual-labeled probes (1 for each allele) at 200 nmol/l, and template (genomic DNA at 15 ng or water blank). Plates were sealed with optical sealing film (Applied Biosystems) and subjected to thermal cycling in polymerase chain reaction 9700 thermal cyclers (Applied Biosystems) according to the standard genotyping protocol of 95 C for 15 minutes, followed by 40 cycles at 92 C for 15 seconds, and at 60 C for 1 minute. After thermal cycling, plates were assayed for fluorescence using a 7900HT sequence detection system (Applied Biosystems). Assignment of genotypes was based on the ratio of reported fluorescence to passive dye standard, and was performed automatically using the autoclustering feature of the allelic discrimination software (Sequence Detection System Version 2.0; Applied Biosystems). Primer and probe combinations included Assay-on-DemandSM assay ID C_ _20 for the MTHFR C677T genotype (rs ). The complementary DNA (cdna) 677 C allele probe was labeled on the 5= end with the VIC reporter dye and contained the nucleotide sequence 5=- ATGAAATCGgCTCCCGC-3=A. The cdna 677 T allele probe was labeled on the 5= end with the FAM reporter dye and contained the nucleotide sequence 5=-ATGAAATCGaCTC- CCGC-3=. Forward and reverse primers were used to amplify the region surrounding the C677T polymorphism. The nucleotide sequence for the forward primer was 5=-GCACTTGAAG- GAGAAGGTGTCT-3=, and for the reverse primer was 5=-CCT- CAAAGAAAAGCTGCGTGATG-3=. For the MTHFR A1298C genotype, Assay-by DesignSM was used with the forward primer 5=-GGAGGAGCTGCTGAAGATGTG-3= and the reverse primer

3 1708 MARTÍNEZ ET AL GASTROENTEROLOGY Vol. 131, 6 5=-CCCGAGAGGTAAAGAACAAAGACTT-3=. The cdna 1298 A allele probe, labeled on the 5= end with the VIC reporter dye, contained the nucleotide sequence 5=- ACCAGTGAA- GaAAGTGT-3=; the cdna 1298 C allele probe was labeled on the 5= end with the FAM reporter dye and contained the nucleotide sequence 5=-CAGTGAAGcAAGTGT-3=. Overall, a high proportion of the samples were genotyped successfully, with a 98% and 97% call rate for C677T and A1298C, respectively, including all control samples. All blanks failed to generate genotypes and no Mendelian inconsistencies were observed in controls. Genotype frequencies for both MTHFR genotypes were compatible with Hardy Weinberg equilibrium (MTHFR C677T: P.11 among those who recurred and P.78 among those who did not recur; MTHFR A1298C: P.11 among those who recurred and P.62 among those who did not recur). The PHASE version 2.1 software (available at washington.edu/stephens/software.html) was used to impute haplotypes. Risk Factor and Covariate Data Self-administered questionnaires were used to obtain data on sociodemographic variables, family history of colorectal cancer in first-degree relatives, history of polyps before the baseline adenoma, aspirin use, cigarette smoking, and history of polyps before the baseline examination. In addition, we considered baseline adenoma characteristics as potential confounding variables because these have been shown to be significant predictors of adenoma recurrence in this study population. 41 Data on adenoma characteristics (ie, number, size, location, and histology) were obtained from the medical records and pathology reports, as previously reported. 41 Statistical Analysis Summary data for baseline characteristics were compared using t tests for continuous variables and 2 analysis for categoric variables. We assessed the main effects of the MTHFR genotypes, diplotypes, and haplotypes on adenoma recurrence by unconditional logistic regression. 42 For the analysis of the diplotypes, we calculated ORs (95% CI) using individuals who were homozygous wild-type at both loci as the reference group. Modeling of the joint effects between MTHFR and folate markers was conducted by dichotomizing the folate markers using the following cut-off points: 400 g for folate intake (Recommended Dietary Allowance for adults), 43 1 drink/day for alcohol consumption, 7 nmol/l for plasma folate (cut-off point that defines deficient level), 44 and 13 mol/l for homocysteine (as a cut-off point to define a mild-to-moderate increase) By using the dichotomized values, we generated dummy variables for genotype and folate marker combinations using a common reference group. Given that no main effects on adenoma recurrence were observed for heterozygotes compared with wild type, we combined these 2 groups and included the higher folate intake or plasma value for use as the reference group; for analysis of homocysteine and alcohol consumption we used the lower concentration in combination with the wildtype and heterozygous individuals as the reference group. Modeling of advanced vs nonadvanced adenoma end points was conducted by polytomous logistic regression using maximumlikelihood multinomial logistic models. We defined advanced recurrences as adenomas that were 1 cm in size or larger, those with a tubulovillous or villous histology, and colorectal cancers; we excluded 3 colorectal cancer cases from this analysis given that these occurred without an adenoma recurrence. A priori we considered potential confounding variables that are suspected or established risk factors for colorectal neoplasia (ie, age, sex, cigarette smoking, family history of colorectal cancer, aspirin use, and so forth). Variables that were associated independently with both adenoma recurrence and baseline plasma markers were considered initially. We then tested additional covariates, including number of colonoscopies conducted during the trial and baseline adenoma characteristics (ie, number, size, location, and histology) because these have been shown to be significant predictors of adenoma recurrence in our study populations. 41 Our final results present multivariate models that comprise the final list of covariates that were significant in the model or that resulted in a difference of 10% or more in the main independent variable. Interactions were tested by using the log likelihood ratio test, in which the model that includes the interaction term was compared with that without the term. Results Among the 1598 individuals genotyped for MTHFR C677T, the prevalence of the T allele was 32.9%; 45.8% were CC homozygotes, 42.7% were heterozygotes, and the remaining 11.5% were TT homozygotes. The corresponding values for the 1583 individuals with MTHFR A1298C data were as follows: 47.3% AA homozygotes, 42.4% heterozygotes, and 10.3% CC homozygotes, with a C allele prevalence of 31.5%. No significant difference in regard to risk factors of interest was observed between this subset of individuals with genotype data and that of the total Wheat Bran Fiber and Ursodeoxycholic Acid populations (data not shown). Baseline characteristics of participants according to MTHFR genotypes are presented in Table 1. Individuals with the MTHFR 677 TT variant had lower plasma folate concentrations compared with CC wild-type individuals, but this was not statistically significant. The presence of the TT variant was associated with a significantly higher homocysteine concentration compared with wild type. As supported by the existing literature, no differences in the plasma markers were observed for the plasma markers and the MTHFR 1298 CC variant; however, a significantly higher proportion of males and of current smokers was shown for this variant compared with wild type. As shown in Table 2, we found no significant main effects of the MTHFR genotypes and adenoma recurrence in the total population. Given that our published studies on the relation between plasma folate, homocysteine, and adenoma recurrence indicate that the associations vary by self-reported multivitamin use, 34 we conducted analyses stratified by this variable. Fiftytwo percent of participants reported use of multivitamin/mineral supplements. Our results show a significantly higher odds of recurrence for individuals with the TT variant compared with CC/CT homozygotes who did not report supplement use (OR, 1.66; 95% CI, ) and no association among supplement users (P for interaction for MTHFR and supplement use.11). Similar but weaker and nonsignificant associations were shown for MTHFR A1298C with a borderline significant interaction for MTHFR and supplement use (P.08). As previously noted, published data showed that the MTHFR 677 TT variant was associated inversely with colorectal cancer, whereas null or positive associations have been shown for adenoma end points.

4 Table 1. Baseline Characteristics According to MTHFR Genotype MTHFR C677T (N 1598) MTHFR A1298C (N 1583) Characteristic CC (n 732) CT (n 682) TT (n 184) AA (n 749) AC (n 671) CC (n 163) Age, y a Male, n (%) 485 (66.3) 469 (68.8) 122 (66.3) 488 (65.2) 454 (67.7) 120 (73.6) b Family history c 0.22 ( ) 0.25 ( ) 0.22 ( ) 0.22 ( ) 0.25 ( ) 0.20 ( ) Aspirin use d 0.26 ( ) 0.30 ( ) 0.26 ( ) 0.27 ( ) 0.28 ( ) 0.27 ( ) Current smoker e 0.11 ( ) 0.10 ( ) 0.09 ( ) 0.10 ( ) 0.11 ( ) 0.15 ( ) b Previous polyp f 0.41 ( ) 0.39 ( ) 0.36 ( ) 0.38 ( ) 0.42 ( ) 0.39 ( ) colonoscopies g 1.70 ( ) 1.71 ( ) 1.66 ( ) 1.71 ( ) 1.69 ( ) 1.72 ( ) Supplement use 0.48 ( ) 0.48 ( ) 0.50 ( ) 0.48 ( ) 0.49 ( ) 0.46 ( ) Plasma biomarker Folate level, nmol/l ( ) ( ) ( ) ( ) ( ) ( ) Homocysteine level, 9.71 ( ) 9.66 ( ) ( ) b, h 9.87 ( ) 9.73 ( ) 9.86 ( ) mol/l Dietary intake Dietary folate, /day ( ) ( ) ( ) ( ) ( ) ( ) Total folate, /day ( ) ( ) ( ) ( ) ( ) ( ) Alcohol, g/day 8.18 ( ) 7.10 ( ) 9.88 ( ) 8.07 ( ) 7.52 ( ) 8.94 ( ) Total B 6, /day 8.08 ( ) 7.20 ( ) 9.50 ( ) 8.31 ( ) 7.10 ( ) 8.28 ( ) Total B 12, g/day ( ) ( ) ( ) ( ) ( ) ( ) Methionine, mg/day 1.43 ( ) 1.45 ( ) 1.38 ( ) 1.42 ( ) 1.45 ( ) 1.37 ( ) Adenoma characteristics Number i 1.71 ( ) 1.66 ( ) 1.57 ( ) 1.66 ( ) 1.69 ( ) 1.66 ( ) Size, mm j 9.23 ( ) 9.21 ( ) 9.29 ( ) 9.28 ( ) 9.18 ( ) 9.54 ( ) Proximal location k 0.29 ( ) 0.32 ( ) 0.32 ( ) 0.31 ( ) 0.30 ( ) 0.24 ( ) Villous histology l 0.21 ( ) 0.21 ( ) 0.20 ( ) 0.20 ( ) 0.21 ( ) 0.20 ( ) a Values are means SD; other values represent age- and sex-adjusted proportions for categoric variables and age- and sex-adjusted means for continuous variables (95% CI). b P.05 comparing variant with wild type. c History of colorectal cancer in parent or sibling. d Regular use of aspirin in the previous month. e Current smoker; excludes 21 participants with missing data for MTHFR677 and 20 for MTHFR1298. f History of previous polyps before qualifying colonoscopy; excludes 115 individuals with missing data for MTHFR677 and 114 for MTHFR1298. g Number of colonoscopies after qualifying procedure. h P.05 for analysis of variance 3-way comparison with Bonferroni correction. i Total number of adenomas; excludes 5 participants with missing data for both MTHFR677 and MTHFR1298. j Size of largest adenoma; excludes 6 participants with missing data for both MTHFR 677 and MTHFR k Proximal location; excludes 7 participants with missing data for both MTHFR 677 and MTHFR l Includes tubulovillous or villous adenomas; excludes 7 participants with missing data for MTHFR677 and 6 for MTHFR1298. December 2006 MTHFR AND COLORECTAL NEOPLASIA 1709

5 1710 MARTÍNEZ ET AL GASTROENTEROLOGY Vol. 131, 6 Table 2. Association Between MTHFR Genotypes and Adenoma Recurrence in the Total Population, by Supplement Use, and According to Recurrence Type Total population Supplement use Nonsupplement use Recurrence type OR a (95% CI) OR b (95% CI) OR b (95% CI) MTHFR C677T All recurrences CC 331 (45.2) (43.9) (46.5) 1.00 CT 302 (44.3) 0.92 ( ) 145 (44.8) 0.98 ( ) 157 (43.9) 0.86 ( ) TT 90 (48.9) 1.21 ( ) 39 (42.4) 0.96 ( ) 51 (55.4) 1.54 ( ) TT vs CC/CT 1.26 ( ) 0.97 ( ) 1.66 ( ) P for interaction c.11 Nonadvanced CC/CT 429 (30.5) (31.7) (29.4) 1.00 TT 56 (30.4) 1.15 ( ) 28 (30.4) 0.97 ( ) 28 (30.4) 1.40 ( ) Advanced d,e,f CC/CT 197 (14.0) (12.4) (15.5) 1.00 TT 34 (18.5) 1.59 ( ) 11 (12.0) 0.98 ( ) 23 (25.0) 2.28 ( ) P for interaction c.19 MTHFR A1298C All recurrences AA 348 (46.5) (45.1) (47.7) 1.00 AC 287 (42.8) 0.85 ( ) 144 (43.9) 0.88 ( ) 143 (41.7) 0.83 ( ) CC 78 (47.9) 1.04 ( ) 29 (39.2) 0.76 ( ) 49 (55.1) 1.37 ( ) CC vs AA/ AC 1.12 ( ) 0.81 ( ) 1.50 ( ) P for interaction c.08 Nonadvanced AA/AC 420 (29.7) (32.0) (27.5) 1.00 CC 56 (34.4) 1.21 ( ) 18 (24.3) 0.70 ( ) 38 (42.7) 1.85 ( ) Advanced d,e,f AA/AC 211 (14.7) (12.4) (17.2) 1.00 CC 22 (13.5) 0.94 ( ) 11 (14.9) 1.16 ( ) 11 (12.4) 0.85 ( ) P for interaction c.01 a Total population ORs were adjusted for age, sex, number of colonoscopies, previous polyps, study, and supplement use. b Stratified ORs were adjusted for age, sex, number of colonoscopies, previous polyps, and study. c Interaction between MTHFR and supplement use. d Advanced recurrences defined as the presence of a large (1 cm) adenoma or one with tubulovillous/villous histology and any colorectal cancers. e P values for difference between advanced and nonadvanced recurrence and MTHFR C677T associations are as follows:.17 for the total population,.98 for supplement users, and.11 for nonsupplement users; corresponding values for MTHFR A1298C are:.35,.22, and.03, respectively. f Excludes 4 participants with missing data on advanced adenoma recurrence and 3 participants who developed colorectal cancer without an adenoma recurrence. Therefore, we conducted analyses of MTHFR genotypes according to type of recurrence, for which we hypothesized that the presence of an advanced recurrence would be more comparable with that of a carcinoma, thereby showing no positive associations. However, our results show stronger effects of the 677 TT variant and advanced recurrences in the total population and among nonsupplement users. In the latter, compared with CC wild-type or CT heterozygous state, the presence of the TT variant was associated with a 2.28 odds of recurrence (95% CI, ); the corresponding OR for nonadvanced recurrences was 1.40 (95% CI, ). Analyses of the C1298A genotype did not follow the same pattern; in fact, results suggested stronger effects in nonadvanced recurrences. We also derived haplotypes and diplotypes from the 2 MTHFR variants. Estimated haplotype frequencies were: 35.7% at codon 677C codon 1298A, 31.4% at codon 677C codon 1298C, and 32.9% at codon T codon 1298A. Although additional haplotypes were derived, their prevalence was too low ( 10%) to conduct meaningful analyses. Results for diplotype and haplotype associations with adenoma recurrence also were generated (Table 3). Our diplotype analyses showed the effect of modification by supplement use. Individuals who reported no supplement use and had either the 677TT_1298AA or 677CC_1298CC diplotypes had significantly higher odds of recurrence compared with the double wild-type CC_AA combination (ORs, 2.05 and 1.85, respectively). No main effects or effect modification by supplement use was shown for any of the haplotypes. We assessed whether the differences in risk of adenoma recurrence in the diplotype analyses might be caused by their effect on plasma folate and/or homocysteine concentrations (Table 4). Results of these analyses showed that the highest homocysteine concentrations were observed for the 677TT_1298AA diplotype: compared with the double CC_AA wild-type, the differences were significant for homocysteine in the total population (10.61 vs 9.53 mol/l; P.05) and more pronounced among nonsupplement users (12.01 vs mol/l; P.05). Individuals with the 677CC_1298CC diplotype had the next highest homocysteine concentration, but the differences were not statistically significant (9.91 vs 9.53 mol/l in the total population and vs mol/l among nonsupplement users). Plasma folate concentrations were lowest in the same diplotype group for the total population (11.86 vs nmol/l) and nonsupplement users (8.62 vs

6 December 2006 MTHFR AND COLORECTAL NEOPLASIA 1711 Table 3. Association Between MTHFR Diplotypes and Haplotypes and Adenoma Recurrence in the Total Population and by Supplement Use Total population Supplement use Nonsupplement use OR a (95% CI) OR b (95% CI) OR b (95% CI) Diplotype c (677_1298) CC_AA 88 (42.11) (44.7) (39.6) 1.00 CT_AA 174 (49.3) 1.28 ( ) 78 (48.5) 1.15 ( ) 96 (50.0) 1.43 ( ) TT_AA 86 (47.8) 1.32 ( ) 37 (41.1) 0.85 ( ) 49 (54.4) 2.05 ( ) CC_AC 160 (45.9) 1.18 ( ) 76 (45.2) 0.99 ( ) 84 (46.4) 1.41 ( ) CT_AC 123 (39.3) 0.85 ( ) 65 (41.9) 0.78 ( ) 58 (36.7) 0.93 ( ) CC_CC 76 (48.1) 1.23 ( ) 29 (40.3) 0.80 ( ) 47 (54.7) 1.85 ( ) P for interaction d.21 Haplotype C_A 510 (45.5) (46.0) (45.1) 1.00 C_C 439 (44.4) 0.95 ( ) 199 (42.3) 0.81 ( ) 240 (46.4) 1.10 ( ) T_A 477 (46.1) 1.02 ( ) 221 (44.2) 0.90 ( ) 256 (47.9) 1.16 ( ) P for interaction d.20 a Total population ORs were adjusted for age, sex, number of colonoscopies, previous polyps, study, and supplement use. b Stratified ORs were adjusted for age, sex, number of colonoscopies, previous polyps, and study. c There were no observations for CT_CC, TT_AC, or TT_CC diplotypes. d Interaction between MTHFR and supplement use nmol/l), although only the latter was statistically significant (P.05). Results of the joint effects of folate intake, plasma folate, homocysteine, and alcohol consumption and MTHFR genotypes in relation to adenoma recurrence are presented in Table 5. For total folate intake, a higher odds of recurrence was shown for individuals who were TT homozygotes who had intake of less than 400 g/day (OR, 1.71; 95% CI, ) as compared with CC or CT variants with higher folate intake. A similar trend was shown for plasma folate, although the association was not statistically significant. We observed a stronger association for the presence of high homocysteine concentrations and the TT variant as compared with lower concentrations and the presence of wild-type or heterozygous variant (OR, 2.29; 95% CI, ). Nonsignificant associations were shown for folate intake and plasma folate and the MTHFR 1298 CC variant; however, no effect was shown for homocysteine and this variant. No effect modification was evident for alcohol consumption for either MTHFR genotype. When we conducted analyses stratified at the median for all variables, associations were slightly weaker but generally similar to those presented in our tables. No significant associations were observed when we assessed interactions among the 3 markers of folate status and MTHFR haplotypes (data not shown). Discussion The availability of methyl groups for DNA methylation reactions is regulated primarily by the activity of MTHFR through its role in 1-carbon metabolism. The totality of the published data on the association of MTHFR C677T polymorphism and colorectal cancer indicate that individuals with the TT variant are at lower risk of developing colorectal cancer 12 ; however, the opposite has been shown for other gastrointestinal cancers As well, the carriage of the 677 TT genotype does not confer protection for colorectal adenoma. 12 In the present study, individuals with the MTHFR TT genotype who were nonsupplement users were at significantly higher odds of adenoma recurrence compared with those with the CC or CT genotype. Reasons for the lack of effect among heterozygotes are unknown and might be related to the TT variant s effect on biochemical markers or Table 4. Plasma Folate and Homocysteine According to MTHFR C677T and A1298C Diplotypes Folate, nmol/l Homocysteine, mol/l Diplotype Total population Supplement use Nonsupplement use Total population Supplement use Nonsupplement use CC_AA ( ) a ( ) ( ) 9.53 ( ) 8.88 ( ) ( ) CT_AA ( ) ( ) 9.93 ( ) 9.67 ( ) 9.05 ( ) ( ) TT_AA ( ) ( ) 8.62 b ( ) b ( ) 9.22 ( ) b ( ) CC_AC ( ) ( ) ( ) 9.74 ( ) 9.02 ( ) ( ) CT_AC ( ) ( ) 9.91 ( ) 9.73 ( ) 8.84 ( ) ( ) CC_CC ( ) ( ) ( ) 9.91 ( ) 8.78 ( ) ( ) P for interaction c.39 P for interaction c 09 a Adjusted mean for age and sex (95% CI). b P.05 compared with CC_AA. c Interaction between diplotype and supplement use.

7 1712 MARTÍNEZ ET AL GASTROENTEROLOGY Vol. 131, 6 Table 5. Interactions Between Markers of Folate Status MTHFR C677T and A1298C Genotypes and Adenoma Recurrence MTHFR 677 recurrence (%) OR a (95% CI) MTHFR 1298 recurrence (%) OR a (95% CI) Folate intake, g/day 400 CC/CT 379 (42.9) 1.00 AA/AC 383 (43.0) CC/CT 254 (47.9) 1.17 ( ) AA/AC 252 (47.6) 1.16 ( ) 400 TT 54 (45.8) 1.17 ( ) CC 44 (44.0) 1.05 ( ) 400 TT 36 (54.6) 1.71 ( ) CC 34 (54.0) 1.46 ( ) Plasma folate level, nmol/l 7.0 CC/CT 437 (43.8) 1.00 AA/AC 432 (43.5) CC/CT 93 (47.0) 1.05 ( ) AA/AC 93 (45.8) 0.99 ( ) 7.0 TT 52 (43.0) 1.08 ( ) CC 50 (45.1) 1.02 ( ) 7.0 TT 20 (60.6) 1.69 ( ) CC 17 (63.0) 1.99 ( ) Homocysteine level, mol/l 13.0 CC/CT 466 (44.6) 1.00 AA/AC 452 (43.6) CC/CT 64 (43.0) 0.83 ( ) AA/AC 73 (45.9) 1.00 ( ) 13.0 TT 54 (42.9) 0.99 ( ) CC 58 (47.9) 1.17 ( ) 13.0 TT 18 (64.3) 2.29 ( ) CC 9 (52.9) 1.09 ( ) Alcohol, drinks/day 1 CC/CT 429 (44.6) 1.00 AA/AC 433 (44.3) CC/CT 204 (45.2) 0.96 ( ) AA/AC 202 (45.7) 1.00 ( ) 1 TT 62 (50.8) 1.40 ( ) CC 50 (50.5) 1.34 ( ) 1 TT 28 (45.2) 0.99 ( ) CC 28 (43.8) 0.84 ( ) a Adjusted for age, sex, number of colonoscopies, previous polyps, and study. other mediating factors, which are not affected by the presence of the CT genotype. We conducted analyses of advanced adenoma recurrence to assess whether a protective effect of the TT genotype would be shown, given that these lesions are markers of more advanced disease and a similar finding to the carcinoma studies might be expected. However, we found stronger associations between the TT variant and advanced recurrences, particularly among nonusers of multivitamins, whose intake of folate and other B vitamins involved in the folate metabolic pathway are lower. Reasons for the divergence between studies of colorectal adenoma and carcinoma are unknown but early vs late stage effects of folate in colon tumorigenesis are speculated as primary determinants of the differences between studies. Disruptions in the folate metabolism pathway can result in either alterations in DNA synthesis or methylation, 2 outcomes that are important in carcinogenesis and presumably are independent. Reduced activity of the MTHFR enzyme has been associated with an increase in the pool of 5,10-methylenetetrahydrofolate, the primary donor for methylating deoxyuridine monophosphate to deoxythymidine monophosphate in DNA synthesis. This occurs at the expense of the pool of methyltetrahydrofolate, the primary methyl donor for methionine synthesis. This shift in donor pools in relation to MTHFR activity in favor of methylenetetrahydrofolate has been shown to sensitize mice to folate deficiency, causing a gene environment associated disturbance in the balance between SAM and S- adenosyl homocysteine (SAH) (lower SAM:SAH ratio). 54 It has been proposed 29 that the role of MTHFR C677T per se in lowering the risk of colorectal cancer is mediated in part by the shift that occurs in favor of DNA synthesis and repair at the expense of methionine synthesis. Chen et al 55 proposed that DNA synthesis may play a more important role in later stages of the colon carcinogenesis sequence during which rapid cell growth is a critical determinant of tumor progression in contrast to methylation status, which is thought to be an early acting event in tumorigenesis, leading to genomic instability. Animal studies have been highly informative for showing a role for folate deficiency as an early acting risk factor for colon carcinogenesis. 56 These studies also provide support for the apparent paradoxic findings in which high supplemental folate after the establishment of microscopic neoplastic foci in the colorectal mucosa enhances the development of colorectal carcinogenesis. 57 Based on the accumulated evidence, it is reasonable to suggest that individuals with the TT variant may be at higher risk for the development of colorectal adenomas through an early acting disruption in DNA methylation, driven in part by increased SAH levels, which are especially significant under conditions of methyl deficiency (ie, low folate, high alcohol). 55 Fewer studies of the role of the MTHFR A1298C variant in colorectal neoplasia have been published. As noted earlier, unlike the C677T variant, this second polymorphism has not been shown uniformly to alter circulating levels of homocysteine concentrations 13,15 17, and has not been shown consistently to be associated with risk of colorectal cancer. 12 Our results show no effect of the A1298C MTHFR variant on homocysteine concentrations, and positive but not significant associations with adenoma recurrence. Furthermore, contrary to our findings of the C677T variant, a stronger effect of the A1298C variant and advanced lesions was not present. Therefore, an important role for the main effect of the A1298C genotype remains unclear. In our diplotype analyses, odds of recurrence were higher among nonsupplement users with the 677TT_1298AA and the 677CC_1298CC combinations as compared with the 677CC_1298AA double wild-types. Reasons for this increase are unknown, although they suggest that carriage of a homozygous mutant allele combination is important. In a recent publication, 17 the effect of the MTHFR genotype combinations was assessed in relation to genomic DNA methylation, an important epigenetic feature involved in regulation of gene expression and regulation. 17,58,59 Results of this study 17 show that under conditions of low folate status, DNA methylation was reduced significantly among individuals with the 677TT_1298AA com-

8 December 2006 MTHFR AND COLORECTAL NEOPLASIA 1713 bination compared with those with the 677CC_1298AA diplotype; individuals with the 677CC_1298CC diplotype also had slightly lower, nonsignificantly different methylation levels. Therefore, it is possible that in the presence of low folate status (ie, nonusers of multivitamins), these specific genotype combinations result in lower global methylation, thereby increasing the risk for adenoma recurrence. Our own data suggest that the increased risk of adenoma recurrence associated with the presence of the 677TT_1298AA diplotype also might be owing to higher plasma homocysteine levels, which was increased markedly and significantly among individuals with this diplotype, an association that has been reported in smaller studies. 60,61 Although these results are opposite of those reported for childhood leukemia, in which the 1298AA_677TT combination was associated with a significantly lower risk, 62 in childhood leukemia the presence of the MTHFR 677 TT genotype was associated with an overall lower risk of this malignancy. Our data on the interaction between folate intake and plasma folate and the C677T MTHFR polymorphism in relation to adenoma recurrence are consistent with reports in the literature that show a higher risk of colorectal neoplasia among individuals with the TT variant and low folate status. 19,25,63 65 Our results related to the MTHFR A1298C genotype show similar patterns for these interactions. Contrary to some reports in the literature, 18,22,66 we did not observe any effect modification in either the C677T or A1298C variants by alcohol consumption. One possible reason for the lack of effect of alcohol might be that consumption of alcohol in our populations was low (median, 1.7 g/day). Therefore, we were limited to using a cut-off level of 1 drink/day because higher levels of alcohol consumption, for which associations have been detected, 18 were relatively low in our populations. In general, studies of colorectal cancer or adenoma have found an association with alcohol only at levels exceeding g/day. 18,67 Lacking in the literature is a comprehensive understanding of the role of homocysteine in relation to MTHFR and colorectal neoplasia, despite its key role in the folate metabolism pathway. Metabolism of homocysteine occurs via 2 pathways: it can be catabolized via the trans-sulphuration pathway or it can be remethylated to methionine via the remethylation pathway. Its metabolism involves regulation of enzymes involved in these pathways, one of which is MTHFR. Variants of the MTHFR gene that decrease enzymatic activity result in high homocysteine levels, given that remethylation of homocysteine into methionine is diminished. Thus, high homocysteine levels in the fasting state primarily reflect defects in homocysteine remethylation. 68 Our results, which show more than a 2-fold increase in odds of adenoma recurrence for mild to moderate hyperhomocysteinemia and the MTHFR TT variant, further support the importance of a low folate status in increasing risk of colorectal neoplasia. Reason(s) for the stronger interactive effect of homocysteine compared with plasma or dietary folate are unknown. Given that several studies, 9,69 72 including our own, 4,34 have shown that the presence of the TT variant results in higher homocysteine concentrations compared with the CC wild type, this underscores the importance of this marker in pathways involving MTHFR. It is plausible that individuals with the TT variant require higher intake of folate to reduce homocysteine concentrations or that factors other than folate are required for further reduction It also is plausible that high homocysteine levels caused by low folate and the presence of the TT variant increases risk whereas high levels owing to other factors (eg, deficiencies in the trans-sulphuration pathway) do not confer a higher risk. Giovannucci et al 28 proposed that MTHFR 677 TT individuals are hyper-responders to folate status, whereby risk of colorectal neoplasia is decreased in the presence of higher folate status (including low alcohol consumption) but increased when folate status is low. The investigators further proposed that in the presence of adequate folate status, MTHFR 677 TT individuals accumulate the form of intracellular folate that is beneficial for DNA methylation whereas under conditions of low folate, low levels of 5,10-methylenetetrahydrofolate can lead to impairment in DNA synthesis. Our data on the adverse effect of high homocysteine levels in TT homozygous individuals further indicate that homocysteine is a sensitive marker because it is influenced not only by folate intake but also by intake of other B vitamins and alcohol, as well as the integrity of the pathways responsible for its metabolism. It is unclear whether homocysteine is related causally to colorectal neoplasia or whether it is a marker of a more complex mechanism(s). Genetic and nutritional alterations that inhibit the removal of homocysteine can result in accumulation of SAH. Because methyltransferases bind to SAH, the efficiency of methyltransferase reactions is dependent on the efficient removal of SAH. Published data show that moderate increases in homocysteine concentrations ( mol/l) are associated with higher SAH but not SAM levels in human and animal studies. 76,77 Thus, homocysteine may be a marker of high SAH intracellular concentrations, which could alter methylation. In fact, higher homocysteine concentrations were shown to be correlated positively with DNA hypomethylation in both healthy individuals 76 and in those with vascular disease. 78 Furthermore, published studies show that the MTHFR genotypes are associated with lower DNA methylation, particularly in the presence of low folate status, 17,61,79 which is consistent with our data. Yi et al 76 proposed an indirect mechanism for the effect of homocysteine on increasing risk of disease in which an SAHmediated inhibition of DNA methyltransferases occurs. Taking these observations into account, we propose a pathway for the effect of the MTHFR genotypes on risk of colorectal adenoma, particularly in the presence of low folate status (Figure 1). In this pathway, carriage of low activity MTHFR genotypes can result in higher homocysteine concentrations, which in turn can either increase levels of SAH (and increase SAH:SAM ratio) or directly suppress DNA methylation. Higher levels of SAH and its corresponding SAH:SAM ratio also can result in DNA hypomethylation. This proposed pathway highlights the fact that genomic instability and methylation are early events in carcinogenesis. Strengths of our studies include the relatively large study population of close to 1600 individuals with plasma biomarker and genotype data to assess combination genotypes and effect modification between the genotypes and markers of folate status in relation to adenoma recurrence, including recurrence of advanced lesions. Further strengths include the prospective nature of our analyses, which allow the investigation of adenoma formation. Limitations related to our biochemical markers include the single, one-time measure and the inability to assess red blood cell folate, a more stable biochemical marker of folate status. Furthermore, it also is plausible that 1 or more factors related to multivitamin use other than folate are respon-

9 1714 MARTÍNEZ ET AL GASTROENTEROLOGY Vol. 131, 6 Figure 1. Proposed pathway for the effect of MTHFR genotypes on risk of colorectal adenoma. Polymorphisms in the MTHFR gene result in high homocysteine concentrations. If homocysteine is not removed, SAH accumulation will occur, resulting in a higher SAH:SAM ratio; higher concentrations of these 2 products can lead to DNA hypomethylation. This pathway is especially important in the presence of low folate status. sible for the observed differences between users and nonusers of these. Additional limitations related to studies of adenoma recurrence also have to be considered, including a short length of follow-up evaluation, and inability to assess late events in neoplastic progression. However, as was noted earlier, given that findings for MTHFR are not uniform between colorectal adenoma and carcinoma studies, we had the ability to investigate the role of this genotype specifically to earlier events in the carcinogenesis sequence. Results of the present study underscore the importance of assessing MTHFR variants independently and in combination and the complexity of the mechanisms involved in understanding the 1-carbon metabolism pathway in colorectal neoplasia. Our findings also highlight the significance of taking into account the folate status or the use of multivitamins in the population when evaluating the associations. Lastly, our data support a critical role of homocysteine, a largely understudied marker in the 1-carbon metabolism pathway in colorectal neoplasia. References 1. Sanjoaquin MA, Allen N, Couto E, Roddam AW, Key TJ. Folate intake and colorectal cancer risk: a meta-analytical approach. Int J Cancer 2005;113: Glynn SA, Albanes D, Pietinen P, Brown CC, Rautalahti M, Tangrea JA, Gunter EW, Barrett MJ, Virtamo J, Taylor PR. Colorectal cancer and folate status: a nested case-control study among male smokers. Cancer Epidemiol Biomarkers Prev 1996;5: Kato I, Dnistrian AM, Schwartz M, Toniolo P, Koenig K, Shore RE, Akhmedkhanov A, Zeleniuch-Jacquotte A, Riboli E. Serum folate, homocysteine and colorectal cancer risk in women: a nested case-control study. Br J Cancer 1999;79: Martinez ME, Henning SM, Alberts DS. Folate and colorectal neoplasia: relationship between plasma and dietary markers of folate and adenoma recurrence. Am J Clin Nutr 2004;79: Blount BC, Mack MM, Wehr CM, MacGregor JT, Hiatt RA, Wang G, Wickramasinghe SN, Everson RB, Ames BN. Folate deficiency causes uracil misincorporation into human DNA and chromosome breakage: implications for cancer and neuronal damage. Proc Natl Acad Sci USA1997;94: Duthie SJ. Folic acid deficiency and cancer: mechanisms of DNA instability. Br Med Bull 1999;55: Goyette P, Sumner J, Milos R, Duncan A, Rosenblatt D, Matthews R, Rozen R. Human methylenetetrahydrofolate reductase: isolation of cdna, mapping and mutation identification. Nat Genet 1994;7: van den Put NMJ, Steegers-Theunissen RPM, Frosst P, Trijbels FJM, Eskes TKAB, van den Heuvel LP, Mariman ECM, den Heyer M, Rosen R, Blom HJ. Mutated methylenetetrahydrofolate reductase as a risk factor for spina bifida. Lancet 1995;346: Jacques PF, Bostom AG, Williams RR, Ellison RC, Eckfeldt JH, Rosenberg IH, Selhub J, Rozen R. Relation between folate status, a common mutation in methylenetetrahydrofolate reductase, and plasma homocysteine concentrations. Circulation 1996;93: Blom HJ. Genetic determinants of hyperhomocysteinaemia: the roles of cystathionine beta-synthase and 5,10-methylenetetrahydrofolate reductase. Eur J Pediatr 2000;159(Suppl3):S208 S Rozen R. Genetic modulation of homocysteinemia. Semin Thromb Hemost 2000;26: Kono S, Chen K. Genetic polymorphisms of methylenetetrahydrofolate reductase and colorectal cancer and adenoma. Cancer Sci 2005;96: van der Put NM, Gabreels F, Stevens EM, Smeitink JA, Trijbels FJ, Eskes TK, van den Heuvel LP, Blom HJ. A second common mutation in the methylenetetrahydrofolate reductase gene: an additional risk factor for neural-tube defects? Am J Hum Genet 1998;62: Weisberg I, Tran P, Christensen B, Sibani S, Rozen R. A second genetic polymorphism in methylenetetrahydrofolate reductase (MTHFR) associated with decreased enzyme activity. Mol Genet Metab 1998;64: Chango A, Potier De Courcy G, Boisson F, Guilland JC, Barbe F, Perrin MO, Christides JP, Rabhi K, Pfister M, Galan P, Hercberg S, Nicolas JP. 5,10-methylenetetrahydrofolate reductase common mutations, folate status and plasma homocysteine in healthy French adults of the Supplementation en Vitamines et Mineraux Antioxydants (SU.VI.MAX) cohort. Br J Nutr 2000;84: Friedman G, Goldschmidt N, Friedlander Y, Ben-Yehuda A, Selhub J, Babaey S, Mendel M, Kidron M, Bar-On H. A common mutation A1298C in human methylenetetrahydrofolate reductase gene: association with plasma total homocysteine and folate concentrations. J Nutr 1999;129: Friso S, Girelli D, Trabetti E, Stranieri C, Olivieri O, Tinazzi E, Martinelli N, Faccini G, Pignatti PF, Corrocher R. A1298C methylenetetrahydrofolate reductase mutation and coronary artery disease: relationships with C677T polymorphism and homocysteine/folate metabolism. Clin Exp Med 2002;2: Giovannucci E, Chen J, Smith-Warner SA, Rimm EB, Fuchs CS, Palomeque C, Willett WC, Hunter DJ. Methylenetetrahydrofolate reductase, alcohol dehydrogenase, diet, and risk of colorectal adenomas. Cancer Epidemiol Biomarkers Prev 2003;12: Ma J, Stampfer MJ, Giovannucci E, Artigas C, Hunter DJ, Fuchs C, Willett WC, Selhub J, Hennekens CH, Rozen R. Methylenetetrahydrofolate reductase polymorphism, dietary interactions and risk of colorectal cancer. Cancer Res 1997;57: Slattery ML, Potter JD, Samowitz W, Schaffer D, Leppert M. Methylenetetrahydrofolate reductase, diet, and risk of colon cancer. Cancer Epidemiol Biomarkers Prev 1999;8: Levine AJ, Siegmund KD, Ervin CM, Diep A, Lee ER, Frankl HD, Haile RW. The methylenetetrahydrofolate reductase 677C T polymorphism and distal colorectal adenoma risk. Cancer Epidemiol Biomarkers Prev 2000;9: Le Marchand L, Wilkens LR, Kolonel LN, Henderson BE. The MTHFR C677T polymorphism and colorectal cancer: the multieth-