Background: Hyperhomocysteinemia has been associated with a higher risk of cardiovascular disease (CVD) in epidemiological studies, but recent trials have failed to show a benefit of lowering homocysteine. To address this apparent paradox, we explored whether interaction between genetic and dietary factors related to homocysteine metabolism contributes to CVD risk.
Methods: We evaluated the associations of homocysteine, methylenetetrahydrofolate reductase (MTHFR) 677C>T genotype, and dietary intake of folate/B-vitamins with subsequent CVD events in 24 968 apparently healthy white American women followed for 10 years. Plasma homocysteine was measured using an enzymatic assay. MTHFR genotype was determined with a multiplex PCR using biotinylated primers.
Results: In unadjusted analyses, homocysteine showed moderately strong linear associations with CVD, with hazard ratios (95% CI) comparing top with bottom quintiles for total CVD of 1.92 (1.55–2.37), myocardial infarction 2.32 (1.52–3.54), and ischemic stroke 2.25 (1.45–3.50), all Ptrend <0.001. These ratios were markedly attenuated after adjusting for traditional risk factors and socioeconomic status to 1.08 (0.86–1.36), Ptrend = 0.12; 1.20 (0.76–1.87), Ptrend = 0.14; and 1.21 (0.75–1.94), Ptrend = 0.50, respectively. Homocysteine was associated with MTHFR genotype (1.4 μmol/L higher homocysteine for TT vs CC, P <0.001) and inversely with intake of folate, vitamin B2, B6, and B12, all Ptrend <0.001. However, there was no association of MTHFR genotype or dietary folate/B-vitamins with CVD. In addition, there were no gene–diet or gene–homocysteine interactions in relation to CVD.
Conclusions: In this large-scale prospective study, the association of homocysteine with CVD was markedly attenuated after adjusting for risk factors and was not modified by MTHFR 677C>T or intake of folate or B-vitamins.
Moderate increases of plasma homocysteine have been associated with a higher risk of cardiovascular disease (CVD)1 in observational studies, in particular case-control studies (1)(2). However, recent clinical trials in populations at high risk for CVD have failed to show treatment benefits with dietary interventions that lower homocysteine, such as folate and B-vitamins (3)(4)(5).
High concentrations of homocysteine may result from genetic or environmental and dietary factors that disrupt homocysteine metabolism (6)(7). Methylenetetrahydrofolate reductase (MTHFR) is the enzyme that catalyzes the conversion of 5,10-methylenetetrahydrofolate to 5-methyltetrahydrofolate, a cosubstrate for homocysteine remethylation to methionine (7)(8). The MTHFR 677C>T polymorphism, an alanine-to-valine substitution, results in the production of a thermolabile enzyme with decreased activity, with TT homozygotes having ∼50% reduction in enzyme activity (6)(7). The T allele has been associated with higher CVD risk in some studies but not others (9)(10)(11)(12). It has been proposed that predisposed individuals who have 1 or more copies of the T allele may become at increased CVD risk in the setting of a low dietary intake of folate or B-vitamins (8), the latter being important cofactors in homocysteine metabolism, and the resulting gene–diet interaction has been postulated to be a risk factor for CVD (13).
However, no large study to date has examined the associations of simultaneously measured plasma homocysteine, dietary nutrient intake, and MTHFR genotype in relation to incident CVD. Therefore, this study was conducted to explore whether an interaction between genetic and dietary factors related to homocysteine metabolism may contribute to CVD risk in a prospective study of apparently healthy women with 10-year follow-up.
Materials and Methods
Study participants were enrolled in the Women’s Health Study, a recently completed, randomized, double-blinded, placebo-controlled clinical trial of low-dose aspirin and vitamin E in the primary prevention of CVD and cancer in US female healthcare professionals (14)(15). Eligible participants were apparently healthy women, ages 45 years or older, who were free of self-reported CVD or cancer at study entry (1992–1995), with follow-up for incident CVD through February 2005. At the time of enrollment, participants gave written informed consent; completed questionnaires on race/ethnic status, demographics, medical history, medications, and dietary and lifestyle factors; and were asked to provide a blood sample. Because >95% of the study participants reported their race/ethnic status as white, we excluded nonwhite participants from this analysis to minimize potential confounding by population stratification. In total, 24 968 white women with both homocysteine and MTHFR genotype measurements made up the study population for this analysis. The study was approved by the institutional review boards of the Brigham and Women’s Hospital (Boston, Massachusetts).
At baseline, participants completed a 131-item semiquantitative food frequency questionnaire as previously described (16). Briefly, participants were asked to estimate their average consumption over the past year of each food item and allowed 9 responses, ranging from “never” to “6 or more times per day.” Nutrient intake was then calculated based on the content of the portion sizes multiplied by the frequency of consumption. In addition, participants were asked to report the type and amount of multivitamin supplement use, and intakes of folate and B-vitamins from supplements were calculated with a comprehensive multivitamin database. Nutrient intake assessments made by use of this food frequency questionnaire have been previously shown to be valid and reliable (17)(18), with correlation coefficients of 0.49–0.55 for estimated folate and B-vitamin intake with this questionnaire method compared with measured plasma concentrations in apparently healthy female professionals similar to our study participants (19)(20). Nutrient intake was adjusted for total energy in kilocalories by the residual method to reduce measurement error due to general over- or underreporting of food items (21).
homocysteine and laboratory measurements
EDTA blood samples were obtained at the time of enrollment and stored in vapor phase liquid nitrogen (−170 °C). The concentration of homocysteine was determined using an enzymatic assay on the Hitachi 917 analyzer (Roche Diagnostics) using reagents and calibrators from Catch, Inc. Total and HDL cholesterol were assayed directly. Creatinine was measured by a rate-blanked method that is based on the Jaffé reaction. High-sensitivity C-reactive protein was measured with an immunoturbidimetric assay.
mthfr genotype determination
Genotyping was performed in the context of a multimarker assay using an immobilized probe approach, as previously described (Roche Molecular Systems) (22). In brief, each DNA sample was amplified by PCR with biotinylated primers. Each PCR product pool was then hybridized to a panel of sequence-specific oligonucleotide probes immobilized in a linear array. The colorimetric detection method was based on the use of streptavidin–horseradish peroxidase conjugate with hydrogen peroxide and 3,3′,5,5′-tetramethylbenzidine as substrates. Linear array processing was facilitated by the use of the AutoRELI-Mark II (Dynal Biotech). To confirm genotype assignment, scoring was carried out by 2 independent observers. Discordant results (<1% of all scoring) were resolved by a joint reading, and where necessary, a repeat genotyping.
ascertainment of incident cardiovascular events
Participants were followed for the composite endpoint of incident CVD (nonfatal myocardial infarction, nonfatal ischemic stroke, coronary revascularization, or death from cardiovascular causes) and the individual endpoints of nonfatal myocardial infarction and ischemic stroke. Medical records were obtained and reviewed for confirmation of events as previously described (23). Deaths from cardiovascular causes were identified by reports from family members, postal authorities, and a search of the National Death Index and confirmed by autopsy reports, death certificates, and medical records.
Statistical analyses were performed using STATA version 8.2 (StataCorp). First, we calculated allele frequencies and performed a Hardy–Weinberg equilibrium test using the Fisher probability test statistic (24). Next, we examined baseline characteristics and plasma homocysteine concentrations according to MTHFR genotype groups. P values for statistical comparison across genotype groups were obtained from χ2 tests for categorical variables, ANOVA for continuous variables expressed as means, and Kruskal–Wallis or Cuzick tests for continuous variables expressed as medians.
Next, cumulative probabilities of total CVD, myocardial infarction, and ischemic stroke were separately calculated according to plasma homocysteine quintiles, MTHFR genotypes, and dietary nutrient intake of folate and the B-vitamins. MTHFR genotypes were examined with additive, dominant, and recessive models. Cox proportional hazard regression models were then used to calculate the hazard ratios (HRs) and 95% CIs according to plasma homocysteine quintiles, MTHFR genotypes, and dietary nutrient intake. We used the median value for each quintile to perform tests for trends across quintiles of homocysteine or dietary intake.
We first examined the unadjusted HRs and then adjusted the models for age (years), smoking status, systolic blood pressure, total and HDL cholesterol (mg/dL), diabetes mellitus, and hormone use. Additional models for homocysteine quintiles were also adjusted for MTHFR genotype or dietary intake of folate/B-vitamins to examine whether these might explain the association of homocysteine with CVD. We also added indicators of socioeconomic status (education and income) to the risk factor-adjusted models. Subsequently, we adjusted for body mass index, creatinine (mg/dL), high-sensitivity C-reactive protein (CRP), alcohol use, physical activity, parental history of premature coronary disease, and treatment status (aspirin/vitamin E) to determine whether these factors may account for part of the association of homocysteine with CVD.
Finally, we examined gene–environment interactions among dietary intake of folate or B-vitamins and homocysteine concentrations with the MTHFR genotypes in relation to CVD using likelihood ratio tests. Cut-points for folate, B-vitamins, and homocysteine were based on median concentrations. All P values were 2-tailed.
MTHFR genotypes were found to be in Hardy–Weinberg equilibrium (P = 0.21), with allele frequencies of 66.9% and 33.1% for the C and T alleles, respectively. Baseline characteristics of the study participants were similar across MTHFR genotypes (Table 1⇓ ) except for hypertension. Homocysteine concentrations were significantly associated with MTHFR genotype (1.4 μmol/L higher mean homocysteine for TT vs CC, P <0.001) and inversely associated with calorie-adjusted intake of folate, vitamins B2 (riboflavin), B6 (pyridoxine), and B12 (cyanocobolamin), all Ptrend <0.001 (see Fig. 1 in the Data Supplement that accompanies the online version of this article athttp://www.clinchem.org/content/vol53/issue4).
During a mean (SD) follow-up period of 9.9 (1.3) years (246 852 person-years), there were 812 total incident CVD events (including 205 myocardial infarctions and 216 ischemic strokes). Cumulative probability curves for incident CVD with quintiles of homocysteine showed a significant association (Plog-rank <0.001) with higher homocysteine concentrations. MTHFR genotype showed no significant association with incident CVD (Plog-rank = 0.50), nor did baseline dietary intake of folate or of vitamins B2 or B6. Higher vitamin B12 intakes were positively associated (Plog-rank = 0.03) with incident CVD.
In unadjusted analyses, homocysteine showed moderately strong associations with CVD in a linear manner (Table 2⇓ ). The crude HRs (95% CIs) comparing top with bottom homocysteine quintiles were 1.92 (1.55–2.37) for CVD, 2.32 (1.52–3.54) for myocardial infarction, and 2.25 (1.45–3.50) for ischemic stroke, all Ptrend <0.001. These HRs markedly attenuated after adjusting for traditional risk factors to 1.17 (0.94–1.46), Ptrend = 0.02 for CVD; 1.32 (0.85–2.04), Ptrend = 0.06 for myocardial infarction; and 1.27 (0.80–2.00), Ptrend = 0.32 for ischemic stroke. Additional adjustment for MTHFR genotype resulted in hardly any change in the adjusted HRs (1.19 for CVD comparing top with bottom quintile of homocysteine (95% CI 0.95–1.48), nor did adjustment for folate and B-vitamin intake (1.24, 95% CI 0.99–1.56). Further adjustment for creatinine, high-sensitivity CRP, body mass index, alcohol intake, physical activity, parental history of premature coronary disease, and treatment status also did not change the results (adjusted HR 1.22, 95% CI 0.94–1.57). When we adjusted for socioeconomic status indicators (education and income), the HRs diminished further and the linear trend became nonsignificant (Table 2⇓ ).
When we examined intake of folate and B-vitamins in relation to incident CVD, we found no association with folate or vitamins B2 or B6, with or without adjusting for CVD risk factors (Table 3⇓ ). Vitamin B12 was positively associated with CVD in crude analysis, but the association attenuated and became nonsignificant after we adjusted for risk factors.
The use of additive (Table 4⇓ ), dominant, or recessive gene models (data not shown) showed no association between MTHFR genotype and CVD, with or without adjusting for risk factors. Similarly, no associations were found with myocardial infarction or ischemic stroke. Additional adjustment for homocysteine did not alter the results. Finally, the results were not altered by high or low homocysteine concentrations, and we found no gene–environment interactions when we stratified by folate or B-vitamin intake above or below the median (data not shown).
In this prospective study of 24 968 initially healthy women with 246 852 person-years of follow-up, we found no significant associations of the MTHFR 677C>T polymorphism with incident CVD, myocardial infarction, or ischemic stroke, despite higher homocysteine in TT homozygotes. In unadjusted analyses, homocysteine showed moderately strong associations with CVD, myocardial infarction, and ischemic stroke, with nearly 2-fold increased risk for top vs bottom quintiles. After adjustment for cardiovascular risk factors and socioeconomic indicators, however, the association of homocysteine with events was markedly attenuated and became nonsignificant. Adjustment for MTHFR genotype or folate/B-vitamin intake did not alter the association. Higher dietary intake of folate and B-vitamins was associated with lower homocysteine but not with lower incident CVD. Finally, there was no evidence of gene–diet or gene–homocysteine interaction in relation to CVD. Taken together, these findings do not support a strong role for either homocysteine or the MTHFR 677C>T polymorphism as etiologic cardiovascular risk factors in healthy women.
The magnitude of the association between homocysteine and incident CVD in this study was consistent with a small case-control analysis in these women (25) and a recent metaanalysis of observational studies that included predominantly men (2), although the findings differ from those of a 24-year prospective study in healthy women that found significant associations with myocardial infarction (26). Although case-control studies have tended to report strong associations between homocysteine and CVD, many prospective studies have shown weaker or no associations (1)(2).
Several small case-control studies have included simultaneous measurements of MTHFR genotype and plasma homocysteine concentrations in relation to CVD (10), but only 1 study, the Copenhagen City Heart Study, was conducted in a prospective manner, although data on gene–diet interactions were not reported (27). In accordance with those studies, we found that the T allele was associated with higher plasma homocysteine concentrations. Whereas the case-control studies found mixed results regarding the association of MTHFR homozygosity and CVD risk (10), the Copenhagen City Heart Study found no association between MTHFR genotype with incident CVD or venous thromboembolic events (27). Another prospective study with 18 years of follow-up in women also found no increased CVD risk associated with the T allele, although homocysteine concentrations were not studied (28).
It has been suggested that the MTHFR T allele carries higher CVD risk in the setting of low dietary intake of folate or B-vitamins (13). Folate and B-vitamins are important cofactors in the metabolism of homocysteine. In this study, despite the lower concentrations of homocysteine in women with higher intake of folate or B-vitamins, there was no significant association between diet and incident CVD. Our finding of a significant association between vitamin B12 and CVD was likely due to confounding, because it was no longer significant after we controlled for CVD risk factors, but we cannot exclude the possibility of potential harm with the use of vitamin B12 (5)(29).
In these data, the associations between dietary intakes of folate or B-vitamins and incident CVD did not differ according to MTHFR genotype. This finding contrasts with results from a recent metaanalysis that found a significant interaction by folate status, with a higher risk associated with the TT genotype in studies with low folate (13). The studies with low folate, however, were also predominantly case-control studies conducted in Europe, in comparison with the high-folate studies that were predominantly North American and prospective. It is possible that our results differed from the case-control studies at least in part because of the prospective nature of our study or its geographic location, given that our results were similar to those of the North American prospective studies that showed no gene–folate intake interaction (13).
Our study findings in the primary prevention setting are consistent with several recent clinical trials that have failed to show benefit with folate and B-vitamin treatment for the secondary prevention of CVD, despite the increased homocysteine concentrations in these trials, which predicted outcomes and were substantially lowered by study treatment (3)(4)(5)(29)(30). Clinical trials for low-risk primary prevention would require extremely large study populations and extended durations of follow-up, both major barriers for conducting such trials.
Our study has potential limitations. The study population was limited to white female healthcare professionals. The lack of association of the MTHFR genotype with CVD should not be viewed as suggesting no role for the MTHFR genotype in CVD, but when coupled with the finding that adjustment for MTHFR genotype also did not alter the association of homocysteine with CVD, these null findings argue against a major role for MTHFR in CVD. It is unlikely that this result represents a type II error, because these null findings for MTHFR are consistent with results from 2 recent comprehensive genetic metaanalyses that found substantial geographical heterogeneity and reported null associations for MTHFR and CVD in North American populations (10)(13), consistent with our study findings. Mandatory folate fortification in the US may have lowered plasma homocysteine concentrations of our study participants later in the follow-up period, but the baseline homocysteine measurements, and more than half of the follow-up in our study, took place after the mandatory fortification of grain products in 1998 (31). Finally, the data on dietary intake of folate/B-vitamins were obtained from food frequency questionnaires, which may be subject to measurement error.
The strength of the current study is the comprehensive and simultaneous assessment of risk factors, dietary intake, plasma homocysteine, and MTHFR genotype status in the same study population, which is the largest study to date that has examined these simultaneously measured risk factors in relation to incident CVD. The detailed information on risk factors allowed for the control for potential confounding, in particular for socioeconomic status. The prospective study design allowed us to minimize biases due to recall, selection, or population stratification, biases that may limit case-control studies. Case-control studies are also limited by temporal bias and reverse causality, making it difficult to establish whether increased concentrations of homocysteine precede clinical CVD or vice versa. Finally, potential interactions of gene–diet and gene–homocysteine concentration in relation to incident CVD were examined, because dietary factors have been postulated to explain the differing associations of MTHFR 677C>T genotype status with CVD by geographical regions (10).
In summary, baseline plasma homocysteine was associated with incident CVD, but this association was weakened and became nonsignificant after adjustment for cardiovascular risk factors and socioeconomic status. In addition, the association of homocysteine with CVD did not appear to be mediated or modified by the MTHFR 677C>T genotype or dietary intake of folate and B-vitamins, despite the variation seen in homocysteine concentrations in accordance with the MTHFR genotype and dietary nutrient intake. Taken together, the findings of this study indicate limited roles for either plasma homocysteine or the MTHFR 677C>T polymorphism as cardiovascular risk factors in healthy women and are consistent with current guidelines that do not recommend population-wide screening for plasma homocysteine.
Grant/funding support: The research for this article was supported by grants from the Donald W. Reynolds Foundation, the Leducq Foundation, and the Doris Duke Charitable Foundation. The Women’s Health Study is supported by Grants HL-43851 and CA-47988 from the National Heart, Lung, and Blood Institute and the National Cancer Institute. S.M. is supported by Grant 0670007N from the American Heart Association.
Financial disclosures: S.C., H.A.E., and K.L. are employees of Roche, which provided the genotyping reagents at no cost under a research collaboration. The other authors have no conflicts of interest related to this work.
Acknowledgments: We thank the investigators, staff, and participants of the Women’s Health Study for their valuable contributions. We also thank Hillary H. Hegener, Jessica L. Gould, and Kirsti A. Diehl for performing the genotyping; Calvin Mano and Nang Tan from Roche Molecular Systems for facilitating the genotyping reagents and technical support; and Jeff Post from Roche Molecular Systems for developing the image processing software. Clinical trial registration: http://clinicaltrials.gov/ct/show; unique identifier, NCT00000479.
1 Values shown for continuous variables are mean (SD) unless otherwise indicated. IQR is interquartile range (25th to 75th percentile). HDL cholesterol is high-density cholesterol. Hs-CRP is high-sensitivity C-reactive protein. P values were obtained for continuous variables from ANOVA for variables expressed as means and from Kruskal-Wallis tests for variables expressed as medians. Pvalues for categorical variables were obtained from chi-square tests. Abbreviations: IQR, interquartile range; Hs-CRP, high-sensitivity CRP.
1 Obtained from Cox proportional hazard regression models that adjusted for age, smoking status, systolic blood pressure, total cholesterol, HDL cholesterol, diabetes mellitus, and hormone use. The reference category for HRs was the lowest quintile of biomarker.
2 Additionally adjusted for the MTHFR genotype.
3 Additionally adjusted for the socioeconomic status (SES) indicators education and income.
1 Obtained from Cox proportional hazard regression models that adjusted for age, smoking status, systolic blood pressure, total cholesterol, HDL cholesterol, diabetes mellitus, and hormone use. The reference category for HRs was the lowest quintile of nutrient intake.
1 Obtained from Cox–proportional hazard regression models that adjusted for age, smoking status, systolic blood pressure, total cholesterol, HDL cholesterol, diabetes mellitus, and hormone use.
2 Adjusted for the risk factors above plus homocysteine (μmol/L).
↵2 These authors contributed equally to this work.
↵1 Nonstandard abbreviations: CVD, cardiovascular disease; MTHFR, methylenetetrahydrofolate reductase; HR, hazard ratio; CRP, C-reactive protein.
- © 2007 The American Association for Clinical Chemistry