Plasma homocysteine levels and mortality in patients with coronary artery disease 

Authors Nygard O, Nordrehaug J, Refsum H, Ueland P, Farstad M, Vollset S. 
Source New England Journal of Medicine. 337:230-6. July 24, 1997. 
Institutions University of Bergen and Haukeland University Hospital, Bergen, Norway. 
Support Norwegian Council on Cardiovascular Diseases and the Norwegian Research Council. 



Homocystinuria, a disease characterized by very high levels of circulating and urinary homocysteine, is associated with premature vascular disease. There has been increasing interest in the role of moderately elevated homocysteine levels in the pathogenesis of vascular and thrombotic diseases, in particular coronary disease. There is evidence that homocysteine may be involved in the pathogenesis of thrombotic events (which complicate atherosclerotic disease), rather than in the development of atherosclerosis per se. In order to investigate this hypothesis further, the authors looked at five-year mortality in a cohort of patients with angiographically proven coronary disease who had documented homocysteine levels at the time of angiography. 



587 patients with angiographically documented coronary disease, studied between February 1991 and June 1992.

Angiogram interpretation

Angiograms were interpreted blindly, and stenoses of at least 50% were considered significant. CAD was scored as single, double or three vessel disease, (left mainstem without RCA disease was scored as double vessel disease). LV ejection fraction was also assessed.

History data collected

History of angina, hypertension, diabetes, family history of premature coronary disease, other medical problems, medications, lipid-lowering diets, smoking history and history of vascular events (myocardial infarction, cerebrovascular disease, peripheral vascular disease).

Biochemical measurements

At the time of angiography, blood was collected and analyzed for: plasma total homocysteine, serum total cholesterol, HDL cholesterol, triglycerides, serum apolipoproteins A-I and B and serum lipoprotein Lp(a). LDL cholesterol was calculated.

Follow-up and causes of death

Deaths occurring between the time of angiography and April 30, 1996 were determined from the National Population Register, and causes of death were obtained from death certificates. 


    Demographics: 478 men (81%), 109 women. Median age 62 years.

    Clinical CAD: 22% had unstable angina; 57% had previous MI; 11% had history of CABG.

    Risk factors: Diabetes in 7% of patients; 27% had hypertension; 27% were smokers, 48% former smokers.

    Medications: Aspirin 45%; beta-blocker 73%; calcium blocker 44%; ACEI 9%; lipid-lowering medication 6%.

    Angiographically documented CAD: single vessel disease in 16% of patients; double vessel in 29%; three vessel disease in 55%. Ejection fraction was below 50 in 13% of patients.

    After angiography, 20% of patients were referred for PTCA, 54% for CABG, 13% were not accepted for revascularization, mainly because of diffuse peripheral CAD and 12% had no indication for revascularization.

Predictors of homocysteine level

A number of clinical and biochemical factors were related to total homocysteine level. 

  • Sex: mean level was 11.4 umol/l in men, 10.5 umol/l in women
  • Age: increase with age (1.3 umol/l with each 20 years of age)
  • Prior MI: 1.0 umol/l higher
  • EF below 50%: 1.2 umol/l higher
  • Hypertensive therapy: 0.7 umol/l higher



    After adjustment for age and sex, the strongest predictors were:

  • serum folate level (r=-0.36)
  • serum creatinine (r=0.30)
  • serum uric acid (r=0.17)
  • serum B12 (r=-0.15)
  • LV ejection fraction (r=0.13)



Homocysteine levels and mortality

After median follow-up of 4.6 years, overall mortality was 11.1% among men (53/478) and 10.1% among women (11/109), with a strong, graded dose-response relationship to homocysteine levels. Kaplan-Meier estimate of 4-year mortality was 3.8% for patients with homocysteine levels under 9 umol /l, 8.6% for those with levels between 9 and 14.9, and 24.7% for those with levels of 15 or greater. These results were not adjusted for age, sex or other potential confounders, however.

In a multivariate model, after adjustment for age, sex, homocysteine level, LV ejection fraction, serum creatinine, total cholesterol, extent of CAD, treatment for hypertension, diabetes, smoking status, platelet count and aspirin use, the three most significant predictors of mortality were LV ejection fraction, serum creatinine and serum homocysteine. The adjusted mortality ratio for these three predictors were (adapted from Table 1 in the article):

Variable Number of patients Adjusted mortality ratio

Left Ventricular ejection fraction
>69 205 1.00
55-69 243 1.51
40-54 108 4.68
<40 30 4.69

Serum Creatinine (mg/dl)
<0.90 79 1.00
0.90-1.35 426 0.64
1.35-1.69 46 1.81
>1.69 27 2.55

Plama homocysteine (umol/l)
<9.0 130 1.00
9.0-14.9 372 1.92
15.0-19.9 59 2.78
>19.9 26 4.51

This relationship between homocysteine and mortality held up in multiple subgroup analyses.

The dose-response relationship between homocysteine level and mortality was nearly linear, with a steeper slope at higher levels, however. At a level of 15 umol/l, the mortality ratio for an increase in homocysteine of 5 umol/l was roughly 2.

Lipid factors were only weakly or not at all related to mortality, in this study. Serum folate was weakly related to mortality, but this association disappeared after adjustment for homocysteine level. B12 level was not associated. The relationship between homocysteine and mortality was not affected by adjustment for either serum folate or B12 levels.

When only cardiovascular causes of mortality were considered (50 out of the 78 deaths), the relationship was strengthened slightly. 

Predictors of coronary disease and previous MI

The extent of coronary disease at the time of angiography was only weakly related to the homocysteine level, but strongly related to serum lipids. In contrast, a history of previous MI was strongly related to homocysteine level, but not to lipid levels. Serum folate and B12 levels were related to neither of these. 

Author's discussion

The authors make a number of points in their discussion, including the following:

The relationship between homocysteine levels and mortality was apparent early (within a few months) and the dose-response effect was seen across a broad range of homocysteine levels.

Although the results of this study suggest that homocysteine is associated with the risk of thrombotic events, while lipid levels are associated with the development of atherosclerosis, these processes are not mutually exclusive and other studies have found an association between homocysteine and extent of vascular disease.

Although there is always the potential for confounding in studies such as these (homocysteine being just a marker for another, causal factor), the multiple adjustments that were made here, and the consistency across various subgroups makes this less likely.

The authors conclude that their results should provide an additional incentive to performing intervention trials with homocysteine-lowering therapy. 


This study looked at all-cause mortality in a group of patients with angiographically-documented coronary artery disease. Patients with higher plasma homocysteine levels at baseline had a higher mortality. Homocysteine did not correlate with the degree of coronary disease, however. Furthermore, serum lipids, which did correlate with the extent of coronary disease, did not correlate well with mortality. The authors conclude that homocysteine has a role in the development of thrombotic events, whereas the role of lipids is more in the development of atherosclerosis, per se.

This study is thus interesting for two reasons: it provides another piece of evidence that moderately elevated homocysteine levels contribute to vascular morbidity and mortality, and it also hypothesizes a mechanism for the respective roles of homocysteine and lipids in the pathogenesis of these conditions.

Evidence that homocysteine is involved in vascular pathology has been accumulating for many years; interest in its role has increased greatly over the past few years. There have been numerous studies published, most of which confirm an association. Many of these studies were case-control. The study presented here is a cohort follow-up study, which is somewhat less prone to selection bias.

Obviously, an association neither guarantees causality nor ensures that lowering elevated homocysteine levels will decrease morbidity, but it is an important first step. Controlled intervention trials are urgently needed and will hopefully be undertaken soon.

The hypothesis that homocysteine contributes to the thrombotic side of the story, whereas lipids contribute to the atherosclerotic side is fascinating. Not all previous studies confirm this but, as the authors point out, the pathogenesis of atherosclerosis is a complex process. Thrombotic events can lead to a progression of atherosclerosis and might explain some studies that have found a correlation between the extent of vascular disease and homocysteine levels. Certainly, if the mechanism hypethesized here turns out to be correct, one could envision a dual-pronged approach to the problem of vascular disease: combatting the atherosclerotic process with lipid-lowering therapy while decreasing the incidence of acute events by lowering homocysteine levels. The statin-manufacturers are probably breathing a little sigh of relief, here.

I suspect that we will be seeing and hearing much more about this topic in the near future -- the Sunday New York Times Magazine had an article on it a few weeks ago. 

September 2, 1997


References related to this article from the NLM's PubMed database. 

Reader Comments

Date: Wed, 03 Sep 1997
From: Colin Rose <>

Actually, the randomized trials have already been done but it wasn't realized at the time that the intervention was in part hypohomocysteinemic. I refer of course to all those diet trials showing reduction of coronary events and regression of atherosclerosis (Ornish, Renaud, STARS, etc.) in which meat was reduced and fruits and vegetables increased. This is exactly the type of diet which would reduce blood homocysteine.

Homocysteine is a metabolite of methionine, an essential amino acid which can only enter to body via food. The food which contains the most methionine is animal protein. Folate, which is required for the metabolism of homocysteine, comes from vegetables. Ergo, less meat and more vegetables is the answer. If you also make the diet low fat you also reduce cholesterol. On this type of diet almost nobody should get CAD and those who have it should almost eliminate events and regress their disease.

Dr. Colin Rose
Montreal, QC

    This is an excellent point. Of course, the only way to tell how much event reduction is due to homocysteine reduction and how much to cholesterol reduction would be to measure both and do regression analysis or adjustment. I wonder if any of these trials have samples left over in which homocysteine could be measured... -- mj 

Date: Tue, 11 Nov 1997
From: "Aloyzio Achutti" <>

Another confounding factor to be discussed on the meat & homocysteine x cholesterol issue, is the role of iron for people with enhanced intestinal absorption of this element. Heterozygotic hemochromatosis is said to be one of the most frequent chromosomal anomalies in the general population. Meat (particularly red meat) is also one of the sources of iron. Exaggerated iron deposits are implicated in the free radicals mechanism and in atherosclerosis.

Aloyzio Achutti
E-Mail: ''
Home page: 

January 5, 1998

Letters to the editor about this article, from the November 27, 1997 New England Journal of Medicine. 

Submit a comment

Journal Club home page

Site Meter