MTHFR 677C→T Polymorphism and Risk of Coronary Heart Disease:  A Meta-analysis FREE

Homocysteine is a sulfur-containing amino acid that plays a pivotal role in methionine metabolism. Genetic defects of the enzymes or dietary deficiency of B-vitamin cofactors involved in this metabolism result in elevated homocysteine levels. Elevated homocysteine levels have been associated with increased risk of coronary heart disease (CHD),¹ but whether this association is causal is uncertain.² Observational studies have shown that individuals with low folate levels or intake have a higher risk of CHD,^3- 6 and it is possible that these associations may be independent of homocysteine.⁷

A common polymorphism exists for the gene that encodes the methylene tetrahydrofolate reductase (MTHFR) enzyme, which converts 5,10-methylene tetrahydrofolate to 5-methyltetrahydrofolate, required for the conversion of homocysteine to methionine. Individuals who have a C-to-T substitution at base 677 of the gene (amino acid change A222V) have reduced enzyme activity and higher homocysteine⁸ and lower folate levels than those without this substitution.^9- 13 Elucidation of an association, if any, between this polymorphism and CHD risk might be informative regarding the hypothesis that impaired folate metabolism, resulting in high homocysteine concentrations, plays a causal role in the occurrence of CHD.

Individual studies and previous meta-analyses of such studies^8,14 included too few subjects to provide conclusive evidence for or against an association of this polymorphism and CHD risk.¹⁵ The aim of this study was to assess the relation of the MTHFR 677C→T polymorphism with risk of CHD by conducting a meta-analysis of individual participant data from all case-control observational studies that had data on this polymorphism and risk of CHD.

Eligible studies were identified by searching the electronic literature (MEDLINE and Current Contents) for relevant reports published before June 2001 (using the search terms MTHFR and coronary heart disease), by hand searching reference lists of original studies and review articles (including meta-analyses) on this topic, and by personal contact with investigators in the field. Studies were included if they had data on the MTHFR 677C→T genotype and a case-control design (retrospective or nested case-control) and involved CHD as an end point.

Among a total of 53 published studies that examined the relation between the MTHFR 677C→T polymorphism and CHD risk, 6 studies were not included because they did not have a proper case-control design^16- 20 or they studied cardiovascular mortality²¹ only. Data on 13 further studies were unavailable because the investigators were unable or unwilling to collaborate.^22- 34 Data from 6 unpublished studies that fulfilled the eligibility criteria were included after personal contact with the investigators. Among the 6 unpublished studies, 4 had previously reported on the relationship between homocysteine and CHD,^35- 38 whereas no data had been previously reported in 2 studies. Hence, data were available for these analyses from 40 studies (34 published^{4,11- 12,14,39- 67} and 6 unpublished^35- 38) involving 11 162 cases and 12 758 controls (Table 1).

Data were collected on MTHFR 677C→T genotype, case-control status, and plasma levels of homocysteine, folate, and other cardiovascular risk factors, if available. Data were checked for consistency with the published article or with information provided by the investigators and converted into a standard format for incorporation into a central database. In the majority of the studies that included cases with myocardial infarction, diagnosis was defined using World Health Organization criteria.⁶⁸ In most studies that included cases with coronary artery disease, diagnosis was based on angiographic confirmation of significant stenosis (≥50%) in at least 1 of the 3 major coronary arteries. However, 1 study also included cases with silent myocardial infarction and coronary revascularization.⁴ If studies included both population-based controls and hospital-based controls, only data on population-based controls were included. All studies used a standardized method to determine MTHFR 677C→T genotype,⁶⁹ with 2 exceptions in which the method had been validated elsewhere.^70- 71

Assuming that a prolonged increase in plasma homocysteine of 1 µmol/L (0.14 mg/L) is associated with a 5% increase in CHD risk^1,72 and that the average homocysteine concentration is 2.5 µmol/L (0.34 mg/L) higher in TT-genotype patients than CC-genotype patients,⁸ the expected odds ratio (OR) of CHD for the TT compared with the CC genotype would be about 1.13. With an average prevalence of the TT genotype of 12%,⁸ more than 9526 cases and an equal number of controls were required to have sufficient statistical power to estimate an OR in the expected range, using a 2-sided α of .05 and 80% power.⁷²

Plasma homocysteine and folate values were log-transformed to improve normality, and geometric means are shown. Differences between cases and controls and between MTHFR 677C→T genotypes were assessed using analysis of variance for continuous data and χ² tests for categorical data. We assessed whether the frequencies of CC, CT, and TT genotypes among controls in individual studies were consistent with the expected distribution (ie, in Hardy-Weinberg equilibrium) using the Pearson χ² test.

The OR and 95% confidence interval (CI) of CHD for the TT genotype or for the CT genotype compared with the CC genotype were assessed in each individual study using logistic regression. The analyses ignored matching of cases and controls on age and sex, which had been applied in some studies. The study-specific ORs were then pooled with adjustment for study. Possible heterogeneity between the results of individual studies or in groups defined by continent of origin or by study design was assessed using χ² tests.

To explore interaction between the MTHFR 677C→T genotype and folate status, 6 subgroups were created whereby folate status was defined as below or above the median serum/plasma folate level. Odds ratios were calculated for all subgroups, with the subgroup with CC genotype and high folate as the reference group.

Complete data on age, sex, smoking, hypertension, and hypercholesterolemia were only available in a subset of studies, and the possible effects of confounding by these risk factors on the relationship between MTHFR and CHD risk were assessed using multivariable logistic regression in this subset.

A funnel plot was created by plotting the OR of CHD for TT vs CC genotype against the number of individuals in each study. A pattern resembling a symmetrical inverted funnel implied absence of significant selection or publication bias. All analyses were performed using SAS, version 6.12 (SAS Institute Inc, Cary, NC).

Table 1 shows the number of cases and controls and selected characteristics for the controls of included studies. About half the data came from studies involving European populations and about a quarter from those of North American populations. The age distribution was similar in all studies. The prevalence of TT genotype among controls varied considerably among studies, ranging from 3.2 (in UK Indians)⁵² to 30.2 (in an Italian population).⁴² The MTHFR 677C→T genotype frequencies in controls were in Hardy-Weinberg equilibrium in all but 3 studies.^53- 54,60

Table 2 shows the geometric mean plasma concentrations of homocysteine and folate and the presence of established cardiovascular risk factors for cases and controls and within MTHFR 677C→T genotypes. Cases had a higher mean homocysteine concentration and a more adverse cardiovascular risk profile. There were no significant differences in plasma folate concentrations between cases and controls. Among both cases and controls, individuals with the TT and CT genotypes had higher plasma homocysteine concentrations and lower folate concentrations than individuals with the CC genotype. Among controls, individuals with the CT genotype had a lower body mass index and individuals with the TT genotype had lower creatinine concentrations compared with individuals with the CC genotype. Among cases, there were significant differences in the prevalence of male sex, hypercholesterolemia, and smoking among genotypes.

Figure 1 shows the OR of CHD for the TT genotype compared with the CC genotype in individual studies and a summary estimate for the combined analysis of all studies with adjustment for study. Overall, individuals with the TT genotype had a significantly higher odds of CHD compared with individuals with the CC genotype (OR, 1.16; 95% CI, 1.05-1.28). There was a trend toward an increased risk for the CT genotype compared with the CC genotype (OR, 1.04; 95% CI, 0.98-1.10). There was significant heterogeneity among the results of individual studies (χ²₃₉ = 63.8; P<.01). The continent of origin appeared to account for most of this heterogeneity. Continent-specific ORs showed that CHD risk was significantly increased for individuals with the TT genotype compared with those with the CC genotype in Europe (OR, 1.14; 95% CI, 1.01-1.28) but not in North America (OR, 0.87; 95% CI, 0.73-1.05). There was no heterogeneity within European studies (χ²₂₁ = 27.1; P = .17) or North American studies (χ²₉ = 4.2; P = .90), but there was significant heterogeneity between the pooled estimates for Europe and North America (χ²₁ = 6.6; P = .01). Data on studies from other continents were too sparse to assess a continent-specific OR.

The heterogeneity between European and North American studies may be explained by an interaction between MTHFR 677C→T polymorphism and folate status. Table 3 shows ORs of CHD within strata of the MTHFR 677C→T genotype and folate status for a subset of studies for which data on folate status was available. The results show that the TT genotype is associated with increased CHD risk only when folate status is low, which indicates an interaction between the MTHFR 677C→T polymorphism and folate status.

To explore potential differences in the association between prospective and retrospective studies, we assessed pooled ORs for each study design. There was significant heterogeneity between the pooled estimates of prospective and retrospective studies (χ²₁ = 16.1; P<.001). The pooled OR of CHD for the TT genotype compared with the CC genotype was 0.86 (95% CI, 0.67-1.10) for prospective studies (5 studies involving 1288 cases and 1749 controls) and 1.21 (95% CI, 1.10-1.33) for retrospective studies (35 studies involving 9874 cases and 11 009 controls). However, since 3 of the 5 prospective studies were North American studies, it is likely that this subgroup analysis reflects a continent effect rather than an effect of the prospective study design.

The effect of confounding was explored in a subgroup of studies with available data on age, sex, smoking, hypertension, and hypercholesterolemia. Complete data on these cardiovascular risk factors were available for 5343 cases and 7308 controls. In this subgroup, the crude OR of CHD for the TT genotype vs the CC genotype was 1.15 (95% CI, 1.02-1.30). After adjustment for these confounding factors, the OR of CHD was 1.21 (95% CI, 1.06-1.38), thereby indicating that confounding is of little relevance to the overall results.

Figure 2 shows a funnel plot in which the OR of CHD for TT vs CC genotype was plotted against the number of individuals in each study. The figure includes data from published and unpublished studies and from studies for which data were not available. The shape of the funnel plot suggests that a few small studies finding an inverse association may not have been published. In addition, we calculated the average OR of CHD associated with TT compared with CC genotype among the 12 studies in which individual data were not provided for these analyses. The average OR for these studies was 1.15 (using the inverse of the variance as a weighting factor), suggesting that the present findings were probably not materially altered by the exclusion of studies for which data were unavailable.

This study involving 11 162 CHD cases and 12 758 controls from 40 studies demonstrated that individuals with the MTHFR 677 TT genotype have a 16% higher odds of CHD compared with individuals with the CC genotype. The results support the hypothesis that impaired folate metabolism, resulting in high homocysteine concentrations, plays a causal role in the occurrence of CHD. This meta-analysis illustrates the need to study a very large number of cases and controls to provide conclusive evidence for an association between genotype and disease in a setting in which the disease risk associated with a genotype is moderate.

The MTHFR 677 TT genotype was significantly associated with a 14% increase in CHD risk in European populations but not in North American populations. Previous studies had shown that the MTHFR 677C→T polymorphism is only associated with high homocysteine levels or increased CHD risk in a setting of low folate status.^{11- 12,43- 44,67,73- 74} Hence, at higher dietary intakes of folate, the effect of the MTHFR 677C→T genotype has no adverse effect on plasma homocysteine levels or on subsequent risk of CHD. Our results confirm that a positive association between the MTHFR 677 TT genotype and CHD risk is mainly present when folate levels are low. However, we think that these results should be interpreted with caution since they are based on only part of the data and there might be misclassification of folate status because of the different assays used. Therefore, the absolute estimates might not be completely valid.

The average use of vitamin supplements has been consistently higher for several years in North America (25%-40%)^57,75- 78 than in Europe (5%-15%).^35,79 While the North American studies were carried out before the enhancement of folate fortification in 1998, fortification of breakfast cereals had been introduced several years before this. Hence, it is very likely that effect modification by dietary intake of folate may account for at least some of the difference in the ORs of CHD obtained for the European and North American populations.^44,72,80 In the present study, combined data from both cases and controls for each study showed that the mean homocysteine concentration was higher in European studies (10.9 µmol/L [1.47 mg/L]) than in North American studies (10.5 µmol/L [1.42 mg/L]). Moreover, the differences between MTHFR TT and CC genotypes were greater in European studies compared with North American studies for both homocysteine (2.1 vs 1.3 µmol/L [0.28 vs 0.18 mg/L]) and folate (2.5 vs 1.7 nmol/L [1.1 vs 0.75 ng/mL]) concentrations, respectively.

Additional sources of heterogeneity between Europe and North America may include effect modification by other cardiovascular risk factors^65,80- 83 or linkage disequilibrium with other polymorphisms, such as the MTHFR 1298A→C polymorphism.^39,46,84 While the prevalence of hypercholesterolemia, smoking, and alcohol use was higher in European compared with North American studies (data not shown), these data were too sparse to examine possible effect modification by these factors.

Studies on the association between the MTHFR 677C→T polymorphism and mortality or longevity have shown inconsistent results.^20,85- 89 However, if individuals with TT have a higher case-fatality rate, then one might expect that the association in retrospective studies would be attenuated compared with that observed in prospective studies because retrospective studies are restricted to survivors, whereas prospective studies can include fatal and nonfatal outcomes. The present study showed that the TT genotype was associated with increased CHD risk in retrospective studies, but not in prospective studies, but this is likely to reflect differences in populations rather than an effect of prospective studies, considering that 3 of 5 prospective studies were North American studies.

Although confounding is generally not anticipated in analyses of an association of a genotype with disease, there may be some imbalance in the distribution of cardiovascular risk factors by the MTHFR genotypes. Adjustment for the possible confounders in a subset of studies with available data did not attenuate the OR of CHD for the TT compared with CC genotype for MTHFR. However, the possibility of residual confounding cannot be completely excluded.

Another potential source of bias might be the inclusion of individuals from heterogeneous ethnic backgrounds. For example, the prevalence of the TT genotype is much lower in blacks (˜1%) than in whites.⁹⁰ If the distribution of individuals with a specific ethnic background is unequal between cases and controls (so-called population stratification), this may bias an association between a genotype and risk of CHD. In a recent study, however, bias from population stratification in case-control studies was quantified and it was concluded that its impact is likely to be small, even if ethnicity is ignored.⁹¹ Furthermore, the risk of population stratification in this meta-analysis is small since adjustment for study ensured that cases from each study were compared with their own controls.

It is unlikely that publication bias accounted for the results obtained; the funnel plot shows that only a few small negative studies may have been missed. Furthermore, selection bias is unlikely to have influenced the results, since the average OR of CHD associated with the TT genotype compared with the CC genotype of 12 studies that were unavailable for inclusion in these analyses was similar to our pooled OR.

An accompanying article in this issue (see p 2015) describes a meta-analysis of 30 studies involving 5000 cases with ischemic heart disease, which showed that among prospective studies, a 25% lower usual homocysteine was associated with 11% (OR, 1.11; 95% CI, 1.04-1.17) lower risk of ischemic heart disease.⁹² The concordance between the risk estimates obtained in these studies provides support for a causal association between homocysteine and CHD. Several large trials are currently under way to assess if homocysteine lowering by supplementation with folic acid and other B vitamins can reduce the risk of CHD.⁹³ Neither the meta-analyses nor these trials can solve the issue of whether high homocysteine levels per se or the accompanying low folate levels, which may operate via other mechanisms, are the cause of CHD. However, the present study provides some indirect evidence of the likely benefits of increasing population mean levels of folate, as the MTHFR genotype has no adverse effect on cardiovascular risk in the setting of normal folate status. Hence, provided that folate status is adequate, there is little clinical value of screening for MTHFR 677C→T genotype in the general population for prediction of CHD risk.