February 21, 2012 (Oxford, United Kingdom) — By examining both published and unpublished genetic studies, an international group of distinguished researchers has reported what they say is final proof that levels of homocysteine have no meaningful effect on the risk of developing coronary heart disease (CHD) [1]. Their meta-analysis is published online February 21, 2012 in PLoS Medicine.
"This has settled the homocysteine issue and illustrates how the published literature has been systematically wrong," lead author Dr Robert Clarke (CTSU, University of Oxford, UK) told heartwire .
After two decades, this report is a definitive answer and has closed the door on the use of folic acid for prevention of heart disease.
"This paper has two messages that we'd like to get across. One is a medical message: after two decades of research, this report is a definitive answer and has closed the door on the use of folic acid for prevention of heart disease." The second message "is methodological," he says, flagging the importance of dealing with publication bias. "It is recognizing that meta-analyses and studies--particularly if they are small--are more likely to be prioritized for publication according to the positivity of the results. The net effect of this is that you can get, systematically, the wrong answer, even for a very well-studied question."
But another researcher who has worked in this field strongly disagrees with these conclusions. Dr David Wald (Wolfson Institute of Preventive Medicine, London, UK) says: "The analysis is flawed because it does not use all the available data, only parts of it." And Wald added that while it is recognized that publication bias "may affect the precision of the estimated risk, it has also previously been concluded that publication bias is insufficient to turn a positive result into a negative one."
Genetics Supported Role of Homocysteine Lowering in Primary Prevention
Clarke explained that numerous trials and meta-analyses have shown that lowering homocysteine levels with folic acid does not prevent CHD events in those with existing cardiovascular disease. But previously published genetic studies have shown a link between the homocysteine gene and heart disease, leading to assertions that homocysteine lowering might work in the primary prevention of heart disease.
The analysis is flawed because it does not use all the available data, only parts of it.
"Trials assess the effects of short-term differences in homocysteine levels, of about five years," Clarke notes, "whereas genetic studies assess the effects of lifelong differences in homocysteine on the risk of CHD and also of stroke."
He and his colleagues began by looking at all the published genetic data sets that were available but then went on to also consider unpublished studies. Most of the latter are genomewide association studies (GWAS) that have been conducted within the past three years, says Clarke.
Genetic studies assess the effects of lifelong differences in homocysteine on the risk of CHD and also of stroke.
The data sets contained information on the methylenetetrahydrofolate reductase (MTHFR) gene, coding for an enzyme involved in the disposal of homocysteine. Individuals with the TT variant of a common polymorphism of this gene--(MTHFR C677T)--have a dysfunctional enzyme, so homocysteine levels build up. In the previously published genetic studies, those with the TT variant have had a significantly increased risk of ischemic heart disease (approximately 25% higher) compared with those with the CC variant (the one associated with lower homocysteine).
But Unpublished Data Refute the Findings, or Do They?
In the new analysis, Clarke and colleagues examined 19 unpublished data sets--including a total of 48 175 CHD cases and 67 961 controls--in which multiple genetic variants had been measured, including MTHFR C677T. "We didn't have folate on everybody, but we knew the country and the year of blood collection, and we could map out the effect of the homocysteine difference by category using genotypes," Clarke notes.
In meta-analyses of these unpublished data sets, the case-control CHD odds ratio (OR) comparing TT vs CC homozygotes was 1.02 overall (95% CI 0.98–1.07; p=0.28) and 1.01 (95% CI 0.95–1.07) in unsupplemented low-folate populations.
The overall results from large unpublished data sets show lifelong moderate homocysteine elevation has little or no effect on CHD.
In stark contrast, in 86 published studies--including 28 617 CHD cases and 41 857 controls--the same odds ratio comparing TT vs CC homozygotes was 1.15 (95% CI 1.09–1.21), "significantly discrepant (p=0.001) with the OR in the unpublished data sets," the researchers observe.
"The confidence interval for the overall results from large unpublished data sets shows lifelong moderate homocysteine elevation has little or no effect on CHD," they state. "The discrepant overall result from previously published studies reflects publication bias or methodological problems."
But Wald says "a serious omission" is the lack of information on homocysteine levels in the unpublished studies. "This is a critical area of uncertainty. If the folate levels among participants in these studies were satisfactory, there would be no difference in risk of ischemic heart disease between people with and without the MTHFR mutation, because folate rectifies the harm done by the mutation. It would be like studying a population of occasional and nonsmokers and concluding that smoking does not cause lung cancer."
Clarke contends, however, that they did have "very precise indirect estimates of folate status of the study populations," calculated from a variety of data sources, such as food companies and WHO reports. "We know precisely when the blood was collected and where the population was studied and can then assign a folate status to the study population."
A Combined Analysis Would Have Been the Appropriate One to Perform
But Wald also argues that the "appropriate analysis" to perform would be to combine the published data from the 86 studies and the new data from the 19 unpublished studies, "not present them separately and then ignore the published ones.
"A combined analysis, not presented by the authors, supports a causal association that is consistent with previously published estimates, with about a 10% increased risk of CHD rather than the 15% reported," he adds.
The authors report no conflicts of interest.
Heartwire from Medscape © 2012 Medscape, LLC
Cite this: Is New Gene Analysis the Final Chapter in the Homocysteine Saga? - Medscape - Feb 21, 2012.
Comments