B Vitamins for the Prevention of Vascular Disease: Title and subTitle BreakInsufficient Evidence to Justify Treatment

Individuals with homocystinuria, a rare genetic disorder associated with markedly elevated plasma homocysteine levels (ie, 100-300 μmol/L [to convert to mg/dL, divide by 7.397]—at least 10 times higher than the general population), experience rapidly progressive atherosclerosis and associated thromboembolic events in early adulthood. This observation prompted the “homocysteine hypothesis” that moderately elevated homocysteine levels might be of causal relevance to cardiovascular disease in the general population.¹

Many observational epidemiological studies have reported that cases with coronary heart disease (CHD) or stroke have higher homocysteine levels compared with age and sex-matched controls. In 1995, a meta-analysis of observational studies, involving a total of 2297 CHD cases, indicated that a 5 μmol/L higher homocysteine level was associated with about a 70% increase in the risk of CHD.² These studies were mainly retrospective in design, however, and were unable to exclude the possibility that the atherosclerotic disease process might itself have increased homocysteine levels (reverse causality bias). In 1998, an updated meta-analysis of both retrospective and prospective (or nested case-control) studies reported weaker associations in prospective studies than in retrospective studies; a 5 μmol/L higher homocysteine level was associated with only a 30% increased risk of CHD within the prospective studies.³

Because homocysteine levels are closely correlated with established risk factors for vascular disease, meta-analyses of published results of observational studies are unable to adjust fully for the effects of confounding due to such risk factors. In 2002, the Homocysteine Studies Collaboration sought to remedy this weakness by performing a meta-analysis based on individual participant data from observational studies. The results of the latter meta-analysis indicated that, after adjustment for known cardiovascular risk factors (age, sex, blood pressure, cigarette smoking, and total cholesterol) and for the effects of measurement error, each 25% lowering of homocysteine (ie, about 3) was associated with only an 11% (odds ratio [OR], 0.89; 95% confidence interval [CI], 0.83-0.96) lower risk of CHD and a 19% (OR, 0.81; 95% CI, 0.69-0.95) lower risk of stroke.⁴

In addition to the prospective studies assessing the relationships of homocysteine levels with vascular risk, a separate group of epidemiological studies has examined the effects of lifelong differences in homocysteine levels resulting from genetic variants of the 5, 10-methylene tetrahydrofolate reductase (MTHFR) enzyme. Individuals with a C to T substitution at base 677 of the gene that encodes MTHFR have reduced enzyme activity, and homocysteine levels are typically 20% higher in those with the TT genotype compared with the CC genotype in nonfortified populations. Because parental genes are allocated at random during gamete formation, individuals with the TT genotype differ only randomly from those allocated the CC genotype, so case-control studies of vascular disease risk comparing individuals with these 2 genotypes should not be influenced by confounding or other biases. Meta-analyses of such case-control studies indicate that a 3 μmol/L difference in homocysteine level among individuals with the TT and CC genotypes is associated with about a 10% to 15% difference in CHD risk^{5 - 7} and a 20% to 25% difference in stroke risk.⁸ The concordance of the results in prospective and genetic studies provides strong support for a causal relationship between elevated plasma homocysteine levels and vascular disease risk.⁹

The homocysteine hypothesis has attracted considerable interest because plasma homocysteine can be reduced by folic acid and vitamin B₁₂, raising the prospect that dietary supplementation with B vitamins could reduce the risk of CHD and stroke. A number of large-scale trials (ie, with >1000 participants) of B vitamin supplementation designed to test this hypothesis were initiated in the mid-1990s. All 4 large trials that have reported their results to date (Second Cambridge Heart Antioxidant Study [CHAOS-2],¹⁰ Vitamin Intervention for Stroke Prevention [VISP],¹¹ Norwegian Vitamin Trial [NORVIT],¹² and Heart Outcomes Prevention Evaluation-2 [HOPE-2]¹³ ), have so far failed to demonstrate significant reductions in their primary outcomes (although the HOPE-2 study did report a significant reduction in stroke, which was a secondary outcome). It is against this backdrop of generally disappointing findings that the results of the Homocysteinemia in Kidney and End Stage Renal Disease (HOST), sponsored by the Veterans Affairs Cooperative Study Program, are reported by Jamison and colleagues¹⁴ in this issue of JAMA.

The HOST trial was designed to assess the effects of lowering homocysteine levels on mortality and vascular outcomes in 2056 patients with renal disease. Participants with advanced chronic kidney disease (estimated creatinine clearance ≤30 mL/min [0.50 mL/s]) or end-stage renal disease, and with homocysteine levels of 15 μmol/L or higher, were randomized to receive 40 mg of folic acid, 2 mg of cyanocobalamin (vitamin B₁₂), and 100 mg of pyridoxine hydrochloride (vitamin B₆) daily or placebo. Despite a 26% reduction in homocysteine levels (achieved with a dose of folic acid substantially in excess of that required to achieve a maximal reduction in homocysteine), the HOST study reported no significant reduction in the primary end point of all-cause mortality, and no significant reductions in any of the secondary outcomes, which included myocardial infarction, stroke, and amputations.

With the publication of yet another apparently negative study, it is reasonable to review the overall results of randomized controlled trials in the context of the evidence from the prospective and genetic studies of vascular risk. In a recent meta-analysis of randomized controlled trials, which included the 4 previously reported large trials and 8 smaller trials, the summary relative risk (RR) for CHD was 1.04 (95% CI, 0.92-1.17) and for stroke was 0.86 (95% CI, 0.71-1.04).¹⁵ If these results are updated with the results of the HOST trial, then the overall inverse-variance weighted OR per each 3 μmol/L reduction in homocysteine levels for a mean duration of treatment of 3.1 years is 1.00 (95% CI, 0.92-1.09) for CHD events (in 12 trials) and 0.88 (95% CI, 0.78-1.00) for stroke (in 9 trials). The reductions in RR achieved in the trials are somewhat smaller than those predicted by prospective studies (0.89 [95% CI, 0.83-0.96] for CHD and 0.81 [95% CI, 0.69-0.95] for stroke⁴ ) and genetic studies (0.84 [95% CI, 0.75-0.94] for CHD and 0.74 [95% CI, 0.43-1.28] for stroke),^{6 ,8} but the 95% CIs for the trial results remain compatible with the associations observed in the epidemiological studies.

Possible reasons for the failure of the 5 completed trials to demonstrate statistically definite effects on vascular risk include an inadequate number of recorded events or insufficient duration of treatment; an attenuation of the benefit owing to folic acid fortification in North America, where most of the trials have been conducted to date; or a true failure of treatment to reduce vascular risk.

To resolve the relevance, if any, of B vitamin supplements for prevention of vascular disease, a meta-analysis of combined individual participant data from all available randomized controlled trials of B vitamin supplementation for prevention of vascular disease is planned, once the ongoing trials are completed and published.¹⁶ Among the 12 randomized homocysteine-lowering trials for prevention of cardiovascular disease that have recruited more than 1000 participants, data should be available on about 52 000 participants (32 000 with prior cardiovascular disease in unfortified populations, 14 000 with prior cardiovascular disease in fortified populations, and 6000 with renal disease in fortified populations); thus, the meta-analysis should be sufficiently powered to detect a 10% reduction in rates of major vascular events, major coronary events, and stroke.

In the meantime, based on existing data, including the findings of the HOST trial by Jamison et al, there is insufficient evidence to justify routine use of homocysteine-lowering vitamin supplements for the prevention of vascular events among individuals at high risk for vascular disease.