Homocysteine and coronary heart disease: the importance of a distinction between low and high risk subjects

Sirs—Recently an elegant meta-analysis of the relation between elevated concentrations of the total plasma homocysteine concentration (tHcy) and cardiovascular disease (CVD) was published in this journal.1 The authors conclude that prospective studies, which compared with case-control studies have a stronger methodological design, indicate only weak support for a causal relation between tHcy and coronary heart disease (CHD). While this is generally true, a distinction should be made among prospective studies that excluded and included subjects with CVD at baseline. Neglecting this difference may result in discarding tHcy as a risk factor for CHD in general, whereas we consider it potentially important in subjects with a high baseline risk of CVD. Figure 1 gives an overview of all prospective studies that provided a relative risk (RR) estimate for CHD for each 5 μmol/l increase in tHcy (or for a difference in tHcy quartiles2 or quintiles.3 Except for one study,4 the studied populations were not selected on the basis of their risk of CHD. The RR of studies that included (or did not report on exclusion of 5,6) subjects with baseline CVD were more often statistically significant as compared to studies that excluded subjects with pre-existing CVD. The observation of a more often statistically significantly positive, or stronger, RR is strengthened by findings of studies that performed separate analyses in high-risk subjects (e.g. with a history of myocardial infarction, stroke, angina pectoris, diabetes, or hypertension).3,7,8 Moreover, in elderly populations the number of subjects with silent, pre-clinical, CVD is probably larger; of the five prospective studies with elderly populations in Figure 1 (average age >60 years)6,8–11 only one8 did not show a significant association between the tHcy concentration and the risk of CHD. Finally, prospective studies performed in high-risk populations consistently show that the tHcy concentration is a strong predictor of cardiovascular mortality and morbidity (including CHD) in subjects with CHD,12 diabetes,13,14 renal insufficiency,15,16 peripheral artery disease,17 and systemic lupus erythematosus.18 Another argument that the tHcy concentration might especially be a risk factor for CHD in high-risk subjects, derives from the observation that cases occurring in the first few years of follow-up have the highest tHcy concentrations.8,11 Data of the Physicians’ Health Study illustrate this phenomenon: after 5 years of follow-up a significant increased risk of CHD was found in men with elevated tHcy levels,19 however, extending the follow-up to 7.5 years yielded a non-significant RR.20 Since subjects that die in the first few years of follow-up might already have had some form of pre-existing CVD, this further supports our argument. Thus, the available evidence seems to indicate that the tHcy concentration is a risk factor that provokes the acute event in subjects with a high risk of CVD, as suggested by some researchers.4,21 Intuitively it may seem unlikely to have a risk factor that operates only to predict disease in high-risk and not in low-risk subjects, however, elevated tHcy concentrations interact multiplicatively with conventional CVD risk factors that characterize high-risk subjects, like hypercholesterolaemia, hypertension and smoking.22 In addition, as elevated tHcy concentrations may interfere with normal coagulation and fibrinolysis,23 the stimulation of thrombosis in high-risk subjects that already have a certain degree of atherosclerosis, may be the crucial factor triggering vascular occlusion. This notion is supported by the results of a Norwegian prospective study in CHD patients.12 They found that the tHcy concentration was strongly associated with mortality, but not with the extent of coronary atherosclerosis. Furthermore, young patients with inborn errors of homocysteine metabolism that received effective treatment to lower their very high tHcy concentrations, still have a tHcy concentration well above the normal range (i.e. >30 μmol/l).24 The fact that these patients do not suffer from CVD events might be the results of the absence of other CVD risk factors in these young individuals. Thus, in subjects without conventional risk factors for CVD, an elevated tHcy concentration may be relatively harmless. The concomitant presence of CVD and a high tHcy concentration could also mean that elevations in the tHcy concentration are merely a marker of the degree of the underlying vascular disease. The latter seems, however, less likely given the recent finding, provided by a double-blind randomized placebo-controlled intervention trial, that lowering the tHcy concentration significantly reduced the rate of restenosis in coronary angioplasty patients.25 In addition, causality can also be inferred by the association between the 677 C > T variant of the MTHFR gene and the risk of CVD, as this genotype is present from birth onwards and will not change over the years. The 677TT variant of this genotype leads to an approximately 25% higher tHcy concentration compared with 677CC subjects.26 A meta-analysis performed in 1998, with almost 6000 cases and almost 7000 controls showed that the 677TT genotype is associated with an overall odds ratio (OR) for CVD of 1.12 (95% CI: 0.92–1.37).26 Although this OR is line with an increased risk of CVD for those with the 677TT genotype, the results was non-statistically significant, which is most likely attributable to a power problem. A more definite answer to the question whether this genotype is related to an increased risk of CHD will be provided by the largest meta-analysis performed to date (~11 000 CHD cases and ~13 000 controls),27 that will be published shortly. Ongoing intervention trials will be able to indicate whether a lower tHcy concentration, achieved through vitamin supplementation (folic acid, vitamin B6 and B12), prevents CVD in high-risk subjects.28 In conclusion, as tHcy may trigger an acute event in subjects with an unfavourable CVD risk profile, the overall estimate of the association between the tHcy concentration and the risk of CHD may be underestimated when epidemiological studies that included and excluded subjects with pre-existing CVD are evaluated together. Relative risks (error bars indicate 95% CI) from prospective studies on the relation between plasma homocysteine concentrations (tHcy) and fatal and non-fatal coronary heart disease This work was financially supported by a grant of the Netherlands Heart Foundation (grant no. 96.147). HJ Blom is established investigator of the Netherlands Heart Foundation (grant no. D97.021). Sirs—We thank Dr de Bree and her colleagues for their interest in our work. Although much has been learned about homocysteine, its role as a risk factor for cardiovascular disease (CVD) continues to be debated.1–3 Dr de Bree and colleagues argued that homocysteine may predict CVD best in those with pre-existing CVD. In this view, homocysteine serves as a marker for atherosclerosis rather than as a causative agent of incident CVD.4 Since we submitted our manuscript,5 additional nested case-control studies4,6–9 and cohort studies10–13 that met our study criteria have been published. Two of the three nested case-control studies from Finland found that homocysteine was significantly associated only with an increased risk for ischaemic heart disease (IHD) among men or women with baseline heart disease but not among those free of baseline heart disease.4,8 An additional Finnish study and one from Sweden failed to find a significant association of homocysteine with IHD.6,9 Furthermore, a 10-year follow-up of the Caerphilly study using a nested case-control study design also did not report a significant association between homocysteine and coronary heart disease as had an earlier report of 5 years of follow-up using a cohort design.7 In contrast, three cohort studies reported significant associations of homocysteine with IHD.10,12,13 In the fourth cohort study, homocysteine was not associated with stroke among all participants but was associated with stroke among participants whose stroke occurred before age 65 years.11 In addition, another nested case-control study found a significant association between homocysteine and sudden cardiac death.14 Using studies that we included in our analyses and more recently published studies, Dr de Bree and colleagues separated prospective studies into those that excluded participants with baseline CVD and those that included CVD. A visual inspection of their Figure suggests that studies that included baseline CVD were more likely to produce significant associations between homocysteine and CVD than were studies that excluded participants with baseline CVD. Their presentation of the data helps to illustrate a potential source of heterogeneity in the data. A study of Finnish women that Dr de Bree and colleagues the authors did not consider also found that homocysteine predicted risk for major coronary events only among women with histories of heart disease at baseline.8 Also, Kark and colleagues repeated their analyses after eliminating participants with prevalent CVD and diabetes and reported that ‘the multivariate association persisted undiminished (data not shown)’.10 Because Dr de Bree and colleagues did not present summary odds ratios (OR) for the two sets of studies, a quantitative comparison of these two sets is more difficult to perform. However, all risk estimates per 5 μmol/l change of homocysteine for the six studies that included baseline CVD appear to be below 1.5, and a summary OR appears to be similar to the summary OR (1.23; 95% CI: 1.07–1.41) that we estimated for nested case-control studies.5 Whether homocysteine predicts CVD risk better in the short-term than in the long-term is unclear. Several authors have found that homocysteine better predicted risk for the events occurring during the earlier stages of their studies than during the later stages.4,15 However, others have not found this effect.16,17 Nevertheless, positive results have been reported for studies with reasonably long follow-up periods,18,19 and negative results have been reported for studies with much shorter follow-up periods.20 An additional point that Dr de Bree and her colleagues make is that studies of older populations are more likely to produce significant findings. However, a number of prospective studies have found significant associations in younger but not older age groups as well,15,16,21,22 or found no age-related difference.18,23 The magnitude and the relation (linear or non-linear) between risk factors and any disease are always important. Considerable uncertainty about the magnitude of the risk estimates continues to exist. One of the principal messages from our meta-analysis was that the magnitude of the risk estimates varied in function of study design. Such estimates are important because they are used to determine population-attributable fractions and thus the societal burden of a risk factor. In concert with other information, such as the ease and cost-effectiveness of reducing the impact of risk factors, policy makers can render more informed decisions regarding the allocation of resources to disease prevention. We concur with Dr de Bree and colleagues about the importance of looking for groups, such as people with pre-existing CVD, in which homocysteine may better predict risk for future adverse events. Additional studies may help to clarify the view that homocysteine may better predict the risk for future IHD among people at high risk for coronary heart disease than among those without pre-existing heart disease. Although determination of whether homocysteine is a risk factor for IHD and other adverse events in the general population rather than in specific subgroups is clearly important, the fortification of the US food supply with folic acid and the subsequent drop in homocysteine concentration24 may to some degree make this debate moot in the US. The discussion about the role of homocysteine in CVD most likely will continue until the results from trials of homocysteine lowering through vitamin supplementation clarify this issue.