MTHFR 677C → T genotype modulates the effect of a 5-year supplementation with B-vitamins on homocysteine concentration: The SU.FOL.OM3 randomized controlled trial

Aims To study how MTHFR 677C→T genotype modulates the effect of supplementation with B-vitamins on total homocysteine (tHcy) and B-vitamin concentrations. Methods 2381 patients with a personal history of cardiovascular disease were randomly assigned to one of four groups: 1) B-vitamins alone (560 μg of 5-methyl-THF, 3 mg of vitamin B6 and 20 μg of vitamin B12), 2) n-3 fatty acids alone (600 mg of EPA and DHA in a 2:1 ratio), 3) B-vitamins and n-3 fatty acids, and 4) placebo. Participants were followed up for 4.7 years. At baseline and annually thereafter, biological parameters were assessed. Multivariate and linear mixed models were fit to study the interaction between B-vitamins and MTHFR genotype. Results Among supplemented participants, concentrations of all three B-vitamins increased during the first year (all p<0.0001) across MTHFR genotype categories. tHcy decreased by 26.3% during the first year (p<0.0001), then steadily increased throughout the 5 years (ptrend<0.001). However, at the end of follow-up, that increase was smaller among TT than among CT or CC subjects (pinteraction<0.02). At baseline, the difference in tHcy concentrations between TT homozygous and CC homozygous subjects was 2.33 μmol/l (p<0.001). After 5 years, that difference was reduced to 1.06 μmol/l and remained statistically significant (p<0.001). Conclusion Participants with the TT genotype exhibited a lower 5-year decrease in tHcy concentrations following a B-vitamin supplementation than did participants with the CC or CT genotype. Clinical trial registration Current Controlled Trials # ISRCTN41926726.

CVD. Briefly, 2,501 participants (1,987 men and 514 women) with a history of CVD (acute coronary or cerebral ischemic event occurring within 1 to 12 months prior to randomization) were eligible to participate. Inclusion took place between 1 February 2003 and 1 June 2007, and followed up ended on 1 July 2009. Using a two-by-two factorial design, subjects were randomly assigned to one of four groups: Group A: B-vitamins (5-methyl-THF [560 μg], vitamin B 6 [pyridoxine hydrochloride, 3 mg] and vitamin B 12 [cyanocobalamin, 20 μg]); Group B: n-3 fatty acids (600 mg of eicosapentaenoic acid (EPA) and docosahexaenoic acid (DPA) in a ratio of 2:1); Group C: both B vitamins and n-3 fatty acids; Group D: placebo (gelatin). All supplements were given as two capsules to be taken once daily. The median treatment/follow-up duration was 4.7 years.
The trial protocol [31] was approved by the ethics committee of the Paris-Cochin Hospital ("Comité consultatif pour la protection des personnes se prêtant à la recherche biomédicale," CCPPRB no. 1933) and the national data protection board ("Comité National de l'Informatique et des Libertés," CNIL no. 901230). Participants provided written informed consent. All analyses and data interpretation were independent of the organizations that funded this study.

Measurements
Treatment compliance was assessed by biannual self-administered questionnaires (or via telephone calls by study physicians). Subjects were considered compliant if they took at least 80% of their allocated treatment. Body Mass Index (BMI in kg/m 2 ) was calculated at each annual examination using measurements of height and weight obtained by trained staff following standard protocols. Fasting blood samples were obtained at baseline and at each yearly followup examination (non-mandatory visits). All biochemical measurements were centralized. Regarding the clinic visits, the sample sizes at baseline and at years 1, 2, 3 and 5 were as follows: n = 2,381, 2,099, 2,001, 1,852 and 1,143, respectively (Fig 1). The number of participants lost to follow up had no impact on the allele distribution (the percentage of subjects with the TT variant at each of the five time points was 15.0%, 15.4%, 15.9%, 15.4% and 14.5%, respectively) or on the supplementation (the percentage of supplemented subjects at each of the five time points was: 50.1%, 49.1%, 48.9%, 49.2% and 49.1%, respectively).
Plasma tHcy, serum folate and serum vitamin B 12 were obtained by a competitive immunoassay with direct chemiluminescence detection; pyridoxal 5'-phosphate (circulating form of vitamin B 6 ) was determined in plasma by high-performance liquid chromatography. Genotyping for the genetic variants encoding MTHFR was carried out for the C677T mutation. For that purpose, genomic DNA was isolated from peripheral blood leucocytes using a Genomix kit (Talent-Euromedex, Souffelweyersheim, France) and following the manufacturer's instructions. The polymerase chain reaction for the 677C ! T mutation was performed according to the Frosst et al. method [32]; it generated a 198 bp fragment using the forward primer 50-TGA AGG AGA AGG TGT CTG CG and the reverse primer 50-AGG ACG GTG CGG TGA GAG TG. The 677C ! T mutation created a HinfI recognition sequence resulting in 175-and 23-bp products for the 677T allele. The enzyme-restricted bands for the MTHFR C677T genotypes were identified by a 10% polyacrylamide gel electrophoresis (PAGE) followed by silver staining. Each experimental batch of DNA was analysed in parallel with two control DNA samples with either MTHFR 677T or MTHFR 677C alleles, to avoid misinterpretation due to a lack of digestion.

Statistical analyses
Departure from the Hardy-Weinberg equilibrium was evaluated using a χ 2 test with one degree of freedom. Hcy, folate, vitamins B 6 and B 12 were not normally distributed and were log transformed to approach normality. Their geometric means are displayed in Table 1 and in Figs 2 and 3. Baseline characteristics of participants were compared between men and women and across MTHFR C677 genotype using the Student's t-test, analysis of variance, Wilcoxon-Mann-Whitney or Pearson's chi-squared tests, as appropriate.
Multivariate and linear mixed models (covariance structure: autoregressive) were used to study the baseline relationships among concentrations of tHcy, folate, vitamins B 6 and B 12 and  genotype and their trend over time. In these models, the outcome variables were the log-transformed plasma or serum tHcy, folate, vitamin B 6 and vitamin B 12 at baseline, respectively. The models were mutually adjusted for the remaining variables (i.e., those not modelled as the respective dependent variable), genotype and for baseline variables associated with the outcome of interest. The beta coefficients obtained from these models were further exponentiated. Beta coefficients above one are interpreted as a (β -1) Ã 100% increase in the outcome when the dependent variable increases by one unit (continuous variable), or compared to the reference category. Beta coefficients below 1 are interpreted as a (1β) Ã 100% decrease in the outcome when the dependent variable increased by one unit (continuous variable), or compared to the reference category. To compute the p-trends, time in years was considered as a continuous variable among the supplemented and the non-supplemented participants. Linear mixed models (covariance structure = autoregressive) with repeated measures were used to study whether the effect of supplementation with B-vitamins was different across genotype. The outcome variables were log-transformed values of plasma tHcy, folate, vitamin B 6 and vitamin B 12 , respectively, at baseline and at years 1, 2, 3 and 5 of follow-up. The models included the baseline value of the respective outcome variable, age, sex, intervention group, time point, and interaction terms for time point and each covariate.
In sensitivity analyses, the same statistical models were fit in the subsample of 1,143 participants with complete data on plasma and serum values of tHcy and B-vitamins over the entire follow-up. All the analyses were performed using SAS software (version 9.4; SAS institute Inc.).

Baseline characteristics of the subjects
Compliance with the supplementation regimen was high. In total, 86% of the subjects were considered compliant with the supplementation (reported taking > 80% of the assigned capsules) and compliance was similar (~86%) across the four treatment groups. In turn, the overall return rate of completed questionnaires was 99%, 96%, 94% and 95% at 6, 12 and 24 months and at the end of the trial, respectively.
None of the baseline characteristics was statistically different according to supplementation group. Among the 2,381 subjects genotyped, 14.9% were homozygous for the MTHFR 677 TT variant, 47.4% were heterozygous (CT) and 37.7% were CC homozygous ( Table 1). The observed SNP genotype frequencies conformed to the Hardy-Weinberg proportions (p = 0.57). The mean baseline age was 61.3 y (SD = 9 y). Myocardial infarction was the most prevalent CVD (45.9%), followed by unstable angina (28.2%). The mean tHcy concentration was 13.2 umol/l at baseline. Men were younger than were women (mean age = 60.8 y in men, and 63.1 y in women). Also, men had higher tHcy concentrations and lower folate and vitamin B 12 concentrations than did women. TT homozygous participants had higher plasma tHcy and lower plasma folate concentrations (p<0.0001) than did individuals with the CC or CT genotype.

Factors associated with tHcy, folate, vitamins B 6 and B 12 at baseline
At baseline, multivariate regression analyses ( Table 2) showed that plasma tHcy concentration was associated with age (p = 0.001), creatinine concentration (p = 0.001), prevalent CVD (3% lower in subjects with a history of unstable angina (p = 0.03) or 6% lower in subjects with a history of myocardial infarction (p = 0.001) compared with subjects with a history of ischemic stroke), and MTHFR genotype (19% higher in subjects with TT genotype compared with those with CC genotype, p = 0.001). It was negatively associated with serum concentrations of folate and vitamins B 6 and B 12 (all p < 0.001).
Plasma vitamin B 6 concentrations were 6% higher in men compared with women, were negatively associated with age (p = 0.001) and positively associated with concentrations of creatinine (p = 0.001) and folate (p = 0.001). Serum folate concentrations were 9% higher in women compared with men, positively associated with age (p = 0.001), plasma vitamin B 6 (p = 0.001) and serum vitamin B 12 (p = 0.001). TT homozygous individuals had a 12% lower serum folate concentrations compared to CC homozygous individuals. Finally, serum cobalamin concentration was positively associated with creatinine (p = 0.01) and folate (p = 0.001) concentrations and negatively correlated with tHcy, age, type of prevalent CVD and male sex (all p<0.04). At baseline, MTHFR genotype was independently associated with plasma tHcy and folate concentrations. MTHFR genotype accounted for approximately 4% of the explained variance in folate concentration and 7% in homocysteine concentration.

Effect of supplementation with B vitamins on tHcy concentration
Owing to the trial's factorial design, we first tested for interaction between the two types of supplements (n-3 fatty acids and B-vitamins) regarding the evolution of plasma tHcy concentration. As the interaction term was not significant (data not shown), subjects in the two groups receiving B-vitamin supplements (with or without n-3 fatty acids) were pooled. Fig 2 shows the evolution of adjusted mean values of vitamin B 6 , folate, vitamin B 12 and tHcy by Bvitamin supplementation group (yes/no). In multivariate analyses, the statistical adjustment included variables significantly associated with baseline B-vitamin and tHcy values ( Table 2). At baseline, the values were not significantly different (all p values > 0.27) between the two supplementation groups. All the values were significantly different (p < 0.0001) at 1, 2, 3 and 5 years post-baseline between the two supplementation groups. Among the non-supplemented subjects, concentrations of vitamin B 6 (p trend = 0.88), folate (p trend = 0.65) and vitamin B 12 (p trend = 0.55) did not change over time, while concentrations of tHcy (p trend <0.001) significantly increased. Among supplemented subjects, plasma values of all vitamins tended to plateau after the first year of supplementation while concentrations of tHcy significantly increased (ptrend<0.001) Among subjects not supplemented with B vitamins, concentrations of vitamin B 6 (p trend = 0.88), folate (p trend = 0.65) and vitamin B 12 (p trend = 0.55) did not vary over time, while concentrations of tHcy (percent change per year = +3%, p trend <0.001) significantly increased over time. The trends regarding tHcy concentration were not significantly different by MTHFR genotype (tHcy: p for interaction = 0.29, Fig 3). Plasma values of tHcy of TT homozygous subjects were systematically 19% higher compared with those of CC subjects (p<0.001).
At baseline, tHcy was not significantly different between individuals with CC and CT genotype (p = 0.072 for non-supplemented subjects and 0.93 for supplemented subjects), and was higher in individuals with TT compared with CT genotype (p = 0.002 for non-supplemented subjects and 0.001 for supplemented subjects). During follow-up, tHcy concentrations significantly increased among non-supplemented subjects (p trend <0.001, with no interaction between time and MTHFR genotype (p for interaction = 0.12). Among supplemented participants, after a more marked decrease in tHcy concentrations in TT subjects (p for interaction < 0.02) over the first year, the same trend was found as that in non-supplemented participants.
Among the subjects supplemented with B-vitamins, concentrations of all studied vitamins increased over the first year (all p values<0.0001), and then tended to plateau (Fig 2). The same trend was found among MTHFR categories (data not shown). Also, plasma values of tHcy significantly decreased by 26.3% after the first year of supplementation (p<0.0001, Fig 2), then gradually increased over the subsequent years (linear trend from year 1 to year 5: +5.0% per year, p<0.001). The same trend was found among MTHFR categories (Fig 3). However, the reduction in tHcy after the first year of supplementation was higher in TT homozygous subjects than in subjects with the other two genotypes (p interaction <0.02). At baseline, the difference in tHcy between TT homozygous and CC homozygous subjects was = 2.33 μmol/l (p<0.001). At the end of the supplementation period, that difference was reduced to 1.06 μmol/l while remaining statistically significant (p<0.001).
Results of the sensitivity analyses (n = 1,143 subjects with complete biomarker data) were consistent with those obtained in the main analysis.

Discussion
To our knowledge, this is the first study reporting findings on effect modification by MTHFR genotype of the association between a long-term B-vitamin supplementation (560 μg of 5-methyl-THF, 3 mg vitamin B 6 and 20 μg of vitamin B 12 ) and plasma values of tHcy in adults with CVD history. We found that baseline concentrations of folate and tHcy were associated with different MTHFR 677 genotypes. Throughout the supplementation, the MTHFR 677 TT genotype was associated with lower serum folate concentrations and higher plasma tHcy concentrations than were the CT or CC genotype. MTHFR 677 genotype was an independent predictor of response to supplementation, and it interacted with it by lowering plasma tHcy values after year 1. This study, therefore, could help inform the debate regarding 5-MTHF fortification for public health reasons and regarding folic acid supplementation for the prevention of neural tube defects or congenital CVD [33][34][35][36]. Another implication of our results concerns migraine treatment. Menon et al. [14,15] have demonstrated that supplementation with B vitamins to relieve headaches caused by migraine was more efficient in carriers of the C677T variant. Thus, MTHFR genotype may be a consideration if B-vitamin supplementation is to be used as a migraine treatment. Our data also support suggestions from other observational and experimental studies [25,26,37,38] that TT homozygotic individuals have greater folate requirements than do their CC or CT counterparts. However, active B-vitamin supplementation produces tHcy concentrations in TT homozygous individuals that are closer to the concentrations found in CC homozygotous individuals.
An interesting observation of the present study was the fact that folate status improved moderately even in the non-supplemented group, perhaps due to dietary changes in this cohort of patients. As we did not assess dietary intake of B-vitamins over time, this hypothesis could not be confirmed. However, there are no reasons to suspect that this change (if indeed present) could differ according to the B-vitamin supplementation group. We collected information on dietary supplement use (including folate) outside of that provided by the supplementation under study. Such use was not significantly different between the placebo and the supplemented group, and could also partly explain the above-mentioned folate status improvement. Nevertheless, a more notable impact on folate levels was seen after B vitamin supplementation, such that mean serum folate levels more than doubled after the first year of supplementation. As expected, tHcy decreased as folate levels increased.
We found that subjects homozygous for the 677T allele had higher tHcy concentrations compared with the CC/CT genotype before and after B vitamin supplementation, even though the magnitude of the difference significantly decreased over the course of the supplementation. At the end of the 4.7 years of follow-up, supplemented subjects still had lower tHcy concentrations than did non-supplemented ones, and also had lower tHcy concentration compared with those measured at the start of the study (Fig 2). Indeed, many investigators have reported lower folate concentrations in TT subjects than in CC or CT subjects [29,39,40]. These results were confirmed by two meta-analyses of intervention studies performed recently [41,42]. Colson et al. demonstrated that a short-term supplementation with folate affected plasma homocysteine and serum folate differently according to MTHFR status [41]. Tsang et al. studied healthy women aged 12-49 y and reported that low blood folate concentrations were associated with the MTHFR C677T mutation [42]. As B vitamin supplementation reduced the difference in tHcy concentration between TT and CT (or CC) subjects (Fig 2), our results support the fact that B vitamins provided as low-dose supplements may help in lowering tHcy concentrations in subjects with 2 T alleles. A major objective of the present study was to examine how the effect of B vitamin supplementation on tHcy concentrations was moderated by the MTHFR 677C!T mutation. After the first year of supplementation, TT homozygous participants experienced a significantly lower absolute decrease in tHcy concentrations than did CC or CT participants. In fact, the tHcy concentration was lower in TT homozygous participants supplemented with Bvitamins than in CC or CT participants not supplemented with these vitamins (Fig 2). This gene-nutrient interaction has been described by some authors [25,29,37,38,43,44]. Tsaï et al. [38] investigated the effect of the MTHFR 677 C !T variant on tHcy concentrations before and after fortification of grain products with folic acid in a sample of 884 Caucasian and 587 African American subjects. They found a gene-nutrient interaction and recommended the measurement of tHcy concentration rather than genotyping MTHFR 677 TT as the primary assay for diagnosis and monitoring of moderate hyperhomocysteinemia. Forh et al. [37] conducted a randomized controlled trial during which 160 women were supplemented with folate over 8 weeks. The authors found that women with the TT genotype exhibited a greater decrease in tHcy concentrations following vitamin supplementation. For the studies without gene-nutrient interaction findings, lower values of folate and higher baseline values of tHcy concentrations were given as explanations. Selhub et al. have suggested that the tHcy-lowering effect of folate reaches a plateau at intakes of around 400 ug/day, with higher doses of folate producing little additional benefit [45].
Important strengths of this study include its design, the large sample size, long treatment duration, and the low-dose supplementation with three different B-vitamins known to impact reduction of tHcy concentration. Most of the studies in the literature have used either the three B-vitamins separately [25,29,37,38,43,44], or a combination of two among the three vitamins [46,47]. In fact, plasma B-vitamins are not the preferred markers of B-vitamin status. Next, this study used 5-methy-THF, which can be put into supplements, but not used in fortification, because it is not heat-stable. The daily dosage of 5-methyl-THF used as vitamin supplementation is about 3 times higher than the usual daily dosage of folic acid obtained from mandatory fortification (mean < 150 micrograms/day) in the United States [48]. Therefore, the same magnitude of effect cannot be expected. Different assay methods for measuring folate provide somewhat different results [49][50][51]. Therefore, caution should be used when comparing absolute values obtained in our population with other populations. However, as the same assay method was used all along the trial, it did not influence the time trends of folate values. Also, as a dynamic measure reflecting recent nutritional intake, plasma B-vitamin values fluctuate; thus, they are normally used as a marker of short-term B-vitamin status. The participants in this study had a personal history of CVD. This may impede the extrapolation of our results to a general population. However, the magnitude of this bias is limited by the fact that prevalent CVD does not have an impact on plasma values of either tHcy or B-vitamins. Red blood cell folate and other vitamins are regarded as markers of long-term B-vitamin status because they reflect B-vitamin status during the erythropoiesis. Shorter and more frequent blood sampling intervals, especially early in the intervention, would have provided more data on the timing and the trajectories of genotype-dependent responses to B-vitamin supplementation. There was a reduction over time of the number of participants for whom blood samples were available. This led to a reduction of the statistical power when comparing the prevalence of hyperhomocysteinemia at the end of the study.
In summary, this study confirmed that reduction in plasma tHcy concentration by B-vitamin supplementation could be influenced by MTHFR genotype. Subjects with the TT genotype may be able to compensate for the effects of the 677C!T mutation on tHcy and folate metabolism if their B-vitamin status is adequate. These aspects must be considered in food fortification and supplementation policy development.