Survival from death of all causes.
Omland T, Samuelsson A, Hartford M, Herlitz J, Karlsson T, Christensen B, Caidahl K. Serum Homocysteine Concentration as an Indicator of Survival in Patients With Acute Coronary Syndromes. Arch Intern Med. 2000;160(12):1834–1840. doi:10.1001/archinte.160.12.1834
Circulating homocysteine levels are predictive of survival in patients with stable coronary artery disease. The prognostic value of serum homocysteine levels, obtained in the acute phase in patients with myocardial infarction or unstable angina, is unknown.
To test the hypothesis that circulating homocysteine levels, obtained during the first 24 hours following hospital admission in patients with acute coronary syndromes, are predictive of long-term mortality.
To test this hypothesis we performed a prospective inception cohort study at a teaching hospital in Gothenburg, Sweden. A total of 579 patients (179 women and 400 men; median age, 67 years) were included (Q-wave myocardial infarction in 163 patients, non– Q-wave myocardial infarction in 210 patients, unstable angina pectoris in 206 patients).
Main Outcome Measure
During a median follow-up of 628 days, 65 patients died. The serum homocysteine level (mean [SD]) was significantly lower in long-term survivors (n=514) than in nonsurvivors (n=65) (12.3 [7.0] vs 14.3 [5.9] µmol/L; P=.003). The relative risk (all-cause mortality) for patients with homocysteine levels in the upper quartile was 2.4 (95% confidence interval, 1.5-4.0) compared with that of patients in the 3 lower quartiles. After adjustment for relevant confounders, the relative risk estimate remained significant (relative risk=1.69; 95% confidence interval, 1.02-2.80). In a stepwise model the homocysteine level provided prognostic information additional to that of patient age, diabetes mellitus, and diuretic usage prior to hospital admission (P=.03).
The serum homocysteine level on hospital admission is an independent predictor of long-term survival in patients with acute coronary syndromes.
UNSTABLE ANGINA pectoris and acute myocardial infarction are closely related pathogenetically and clinically, and the term "acute coronary syndrome" is commonly used to signify both disease entities.1,2 Pathophysiologically, the acute coronary syndrome is characterized by rupture of an atherosclerotic plaque, alterations in coronary vasomotor tone secondary to endothelial dysfunction, platelet aggregation, and an active thrombotic process.3,4 Although many patients are initially successfully medically stabilized, most have extensive atherosclerotic disease and a high risk of subsequent development or recurrence of myocardial infarction, the leading cause of early and late mortality in this patient group.1
Mildly to moderately elevated circulating levels of homocysteine are associated with an increased incidence of coronary heart disease,5- 8 peripheral artery disease,7- 9 stroke,8,10,11 and venous thrombosis.12 Recently, a relation between the plasma homocysteine level and long-term mortality was demonstrated in patients with angiographically documented, predominantly stable, coronary artery disease.13 Prognostic data from patients with stable coronary artery disease cannot be extrapolated to patients with acute coronary syndromes, partly because the pathophysiological determinants of survival may markedly differ and partly because small-scale studies measuring the levels of homocysteine serially after myocardial infarction have indicated that homocysteine levels increase significantly following the acute phase.14,15
Experimental evidence suggests a potential role for homocysteine in the pathogenesis of recurrent events in patients with acute coronary syndromes. In addition to having procoagulant and platelet aggregating effects,16- 22 homocysteine may be toxic to the endothelium,16 reduce the bioavailability of nitric oxide resulting in endothelial dysfunction,23- 26 and stimulate the production of proinflammatory cytokines27,28 and smooth muscle cells,29 leading to progression of atherosclerosis and plaque destabilization.
The demonstration of a relation between the level of homocysteine and survival in patients with acute coronary syndromes, in addition to providing clues to pathophysiological mechanisms, may be of considerable clinical interest, considering that the time of hospitalization may be the time when the level of homocysteine is first measured in many patients, and the decision to initiate homocysteine-lowering therapy may be based on the result. Accordingly, the primary objective of this prospective, inception cohort study was to test the hypothesis that circulating homocysteine levels, obtained during the first 24 hours following hospital admission in patients with acute coronary syndromes, are predictive of long-term mortality. A secondary objective was to investigate whether the predictive value of the homocysteine level differed in the early (day 30 or earlier) and late (after day 30) convalescent phase following the index event.
This investigation is part of an ongoing prospective risk stratification study in patients with acute coronary syndromes, admitted to the coronary care unit of Sahlgrenska University Hospital, Gothenburg, Sweden.
Blood collection for homocysteine determination commenced November 15, 1995. Patients who consented to participate in the risk stratification program and who also consented to blood sampling were included consecutively during the recruitment period that ended January 1, 1998. During this period a total of 1013 admissions for acute myocardial infarction, and 548 admissions for unstable angina pectoris were registered. The main reasons for exclusion were older than 80 years, residence outside the city of Gothenburg, and prior hospital admission resulting in inclusion in the study. The primary outcome measure was all-cause mortality from the time of inclusion in the study until November 10, 1998. The minimum duration of follow-up was 302 days. Survival status and date of death were obtained from the Death Registry of Western Sweden. The cause of death was classified by one of us (M.H.), who was blinded with regard to homocysteine levels, based on hospital records, death certificates, and autopsy findings, when available.
Patients living in Gothenburg, not previously included in the study, admitted to the hospital with an acute coronary syndrome, defined as a diagnosis of unstable angina pectoris or acute myocardial infarction, were eligible for participation in this study. The diagnosis of unstable angina was based on 1 or more of the following 3 historical features: (1) crescendo angina (more severe, prolonged, or frequent) superimposed on a preexisting pattern of relatively stable, exertion-related angina pectoris, (2) angina pectoris at rest or on minimal exertion, or (3) angina pectoris of new onset brought on by minimal exertion. The diagnosis of acute myocardial infarction was based on the combination of typical chest pain of at least 15 minutes duration, typical electrocardiographic (ECG) changes, and elevation of creatine kinase MB and/or troponin T levels.
Ineligible for participation were patients younger than 18 years or older than 80 years, patients with noncoronary artery disease associated with a life expectancy of less than 1 year, and patients unable or unwilling to provide informed consent. The patients were prospectively classified as current smokers vs nonsmokers. Electrocardiographic findings were classified according to the (1) presence or absence of new pathological Q-waves in the electrocardiogram, (2) presence or absence of ST-segment elevation, and (3) as normal vs abnormal. Based on hospital medical records and personal interview the patients were classified as having or not having a medical history of myocardial infarction, angina pectoris, congestive heart failure, diabetes mellitus, and arterial hypertension. The study protocol was approved by the regional ethics committee prior to the initiation of the study. Informed consent was obtained from all participating patients.
A total number of 579 patients, 179 women and 400 men, were included in this study. The median age of the patients was 67 years (age range, 32-79 years). Q-wave myocardial infarction was diagnosed in 163 patients (28%), non–Q-wave myocardial infarction in 210 patients (36%), whereas unstable angina pectoris was diagnosed in the remaining 206 patients (36%). ST-segment elevation was diagnosed in 172 patients (30%), ST-segment depression in 142 patients (25%), whereas a normal electrocardiogram on hospital admission was seen in 129 patients (22%). Based on interviews and hospital medical records, 141 patients (24%) had a history of myocardial infarction; 293 patients (51%), a history of angina pectoris; 54 patients (9%), a history of congestive heart failure; 94 patients (16%), a history of diabetes mellitus; and 242 patients (42%), a history of arterial hypertension. One hundred seventy-three patients (30%) were current smokers. Preadmission therapy included aspirin in 201 patients (35%), oral nitrates in 215 patients (37%), β-adrenoceptor blockers in 211 patients (36%), calcium channel blockers in 85 patients (15%), and diuretics in 80 patients (14%). In-hospital therapy included thrombolytics in 109 patients (19%), β-adrenoceptor blockers in 539 patients (93%), diuretics in 206 patients (36%), and angiotensin-converting enzyme inhibitors in 182 patients (31%). Technically satisfactory echocardiograms, permitting estimation of left ventricular ejection fraction, were obtained in a subset of 488 patients (84%). The median left ventricular ejection fraction was 54% (range, 10%-88%). Revascularization procedures were performed in 176 patients (30%) during the primary hospitalization, of whom 21% underwent percutaneous transluminal coronary angioplasty and 9% coronary artery bypass grafting.
Peripheral blood samples for serum homocysteine determination were obtained within 24 hours of hospital admission by direct venipuncture of an antecubital vein after the patient had been in the supine position for longer than 30 minutes. After coagulation at room temperature for 1 hour, samples were centrifuged at 3000 rpm, and serum was aspirated. Serum samples were stored at −70°C pending analysis.
Total homocysteine levels in serum were determined by high-pressure liquid chromatography, using the method developed by Ubbink and coworkers.30 Briefly, prior to reverse-phase high-pressure liquid chromatographic analysis, serum thiols were derivatized with 7-fluoro-benzo-2-oxa-1,3-diazole-γ-sulfonate (SBD-F), a thiol-specific fluorogenic reagent. Retention of the SBD-F-thiol adducts is sensitive to pH. A mobile phase pH of 2.1 enabled baseline separation of homocysteine in serum within 4 minutes 18 seconds. The system was calibrated with aliquots of the same thiol standard solution prior to each run, and serum aliquots from one control batch were used as controls throughout the study. The intra-assay coefficient of variation was 6.3%. The analyses were performed by one of us (B.C.) blinded with regard to the endpoint status of the patients. Serum concentrations of creatinine, creatine kinase–MB fraction, troponin T, low-density lipoprotein– and high-density lipoprotein–cholesterol were determined by standard laboratory methods.
Echocardiographic investigation was performed by an experienced operator (K.C. and colleagues) within 5 days following hospital admission, using a 2- or 3.5-MHz transducer and an ultrasound scanner (128/XP10; Acuson, Mountain View, Calif). Patients were placed in the left lateral supine position on a bed with an apical cutout. Biplane left ventricular ejection fraction was calculated online by the disk sum method, and tracings checked in motion mode for accuracy.
Continuous data are presented as mean±SD or as median and interquartile range, as indicated. The Mann-Whitney and the Fisher exact tests were used to contrast ordered/continuous and categorical variables, as appropriate. Spearman rank correlations were used to assess correlations between variables. Multivariate linear regression analysis, with the appropriate logarithmic transformation of continuous variables with a skewed distribution, was used to identify predictors of serum homocysteine levels. Kaplan-Meier plots were generated and the log rank test was used for comparison of the resulting survival curves. The Cox proportional hazards regression model was used in 2 different multivariate analyses. First, we simultaneously included all variables that separately altered the relative risk (RR) ratio for homocysteine levels regarding all-cause mortality by at least 10%. Second, we performed a forward stepwise selection procedure of all variables with nonmissing data and univariately related to mortality with P<.05 and adding homocysteine, while, thereafter, forcing homocysteine levels into the final model.
The median value of serum homocysteine concentration was 11.4 µmol/L (interquartile range, 8.9-14.1 µmol/L). Serum homocysteine concentrations correlated significantly with the age of the patient (r=0.18, P<.001) and was significantly higher in men than in women (12.8 [7.5] vs 11.9 [5.2] µmol/L; P=0.03). As detailed in Table 1, the serum homocysteine concentration was significantly higher in patients with a history of myocardial infarction than in those without such a history. Likewise, serum homocysteine levels were higher in patients with a history of angina pectoris or congestive heart failure. No significant difference in serum homocysteine levels was observed between patients with or without a history of arterial hypertension or diabetes mellitus. Current smokers had significantly higher homocysteine concentrations than nonsmokers. No association was observed between serum homocysteine and serum low-density lipoprotein– or high-density lipoprotein–cholesterol levels (r=−0.03; P=.53 and r=−0.08, P=.10, respectively), or between serum homocysteine levels and peak levels of creatine kinase–MB (r=−0.08; P=.07) or troponin T (n=419; r=−0.02; P=.73). However, a significant association between serum homocysteine and serum creatinine concentrations was demonstrable (r=0.27; P<.001). No significant association was evident between the level of serum homocysteine and left ventricular systolic function, expressed as ejection fraction (n=484; r=0.06; P=.16). As detailed in Table 1, the level of serum homocysteine was significantly higher in users than in nonusers of diuretics, aspirin, nitrates and β-blockers, whereas no difference was observed between patients using or not using calcium channel blockers. In a multivariate linear regression model, patient age (P=.01), serum creatinine level (P<.001), and a history of myocardial infarction (P=.007) were identified as independent predictors of serum homocysteine concentration.
No patient was lost to follow-up. After a median follow-up of 628 days, 65 patients (11%) had died. Nineteen deaths (3%) occurred during the first 30 days following hospital admission. Forty-eight deaths were classified as cardiac, 5 as vascular, and 12 as noncardiovascular. Among the latter, 9 were related to cancer and 3 to miscellaneous causes. Four patients suffered a fatal stroke—2 ischemic and 2 hemorrhagic—1 of each occurring with a myocardial infarction. One patient died of rupture of an aortic abdominal aneurysm. In 23 cases an acute myocardial infarction was the cause of death. A further 6 patients died within minutes or a few hours after development of chest pain or dyspnea. Eight patients suffered an instantaneous death, in most cases owing to verified ventricular arrhythmia. In 6 patients death occurred in association with coronary bypass surgery, 4 during the index hospitalization. Progressive congestive heart failure was the cause of death in 5 cases. The serum homocysteine level was significantly lower in long-term survivors than in nonsurvivors (12.3 [7.0] vs 14.3 [5.9] µmol/L; P=.003). Using the lower quartile of homocysteine as reference, no significant increase in mortality was observed for the second and third quartiles (relative risk [RR], 1.2 [95% confidence interval (CI), 0.5-2.5] and 0.8 [95% CI, 0.3-1.8], respectively). However, the mortality rate in the upper quartile was significantly increased (RR, 2.4 [95% CI, 1.2-4.8]). After pooling the 3 lower quartiles and using them as reference, the RR, estimate for the fourth quartile was 2.4 (95% CI, 1.5-4.0). Owing to the apparent nonlinear relation between the level of homocysteine and survival, in subsequent analyses we used homocysteine as a dichotomous variable with the 75 percentile as the cutoff. The Kaplan-Meier survival plot for mortality, subdivided according to the 75th percentile of homocysteine in serum, is shown (Figure 1).
To identify confounders of the association between the level of homocysteine and survival, we first conducted a series of 2-factor Cox proportional hazards regression analyses. The results of these 2-factor analyses are summarized in Table 2. A series of 3-factor analyses with the level of homocysteine and the potential confounder as independent variables was subsequently performed. Variables without missing data that altered the RR estimate for homocysteine concentration by more than 10%, ie, patient age, a history of congestive heart failure, serum creatinine level, and preadmission therapy with diuretics, were included in a multivariate model. Although the association between the level of homocysteine and all-cause mortality was weakened by the adjustment for these confounders, the RR estimate remained significant (1.69 [95% CI, 1.02-2.80]).
Owing to co-linearity between patient age and serum creatinine level (r=0.19) and between a history of congestive heart failure and preadmission diuretic use (φ coefficient=0.51), we also performed a stepwise selection procedure in which we included all variables with nonmissing data significantly related to survival in univariate analyses and forced homocysteine into the model. The final model, which included the homocysteine level, patient age, and diuretic usage prior to hospital admission is given in Table 3.
To assess whether the prognostic value of serum homocysteine concentration differed for early (day 30 or earlier) vs late deaths (after day 30), the RR estimates for the upper quartile of homocysteine in these 2 periods were compared. No substantial difference was observed (early vs late deaths: 2.3 [95% CI, 0.9-5.7] vs 2.5 [95% CI, 1.4-4.5]).
The relation between the level of homocysteine and mortality was not markedly altered when the end point analyzed was cardiovascular rather than total mortality.
The new and salient finding of our prospective inception cohort study is that the serum homocysteine concentration is a significant predictor of mortality in patients admitted to the hospital with acute coronary syndromes. The relation was consistent for early and late deaths and seemed to be nonlinear, with a threshold corresponding to the 75th percentile. After adjustment for significant confounders the relation was weakened, but remained significant. Our findings provide important complementary information to previous prospective studies of the prognostic value of homocysteine concentration in subjects presumed to be free of coronary heart disease5,6,31- 35 and in patients with predominantly stable coronary artery disease.13 Moreover, it provides novel information regarding confounders of the relation between the level of homocysteine and survival. Finally, our findings may have implications for the design and conduct of future clinical trials aimed at reducing circulating homocysteine levels.
The single previous prospective study of the association between the level of homocysteine and survival in patients with established coronary artery disease by Nygård et al,13 demonstrated a strong relation between homocysteine levels and mortality. That study, however, included predominantly patients with stable coronary artery disease, who pathophysiologically and prognostically differ substantially from patients with acute coronary syndromes. Although the main results of the 2 studies concur, notable differences are also evident. In contrast with our study, where a threshold corresponding to the 75th percentile was evident, no threshold effect was apparent in the study by Nygård et al. However, the previous study used arbitrary cutoff points corresponding to the 22nd, 86th, and 96th percentiles, resulting in patient groups of varying size. Although a significant linear trend was observed, in the multivariate analysis a significant increase in risk was observed only for the group with homocysteine values greater than the 96th percentile. If patients in our study had been grouped according to the cutoff values used by Nygård et al rather than in quartiles, it is possible that a linear relation would have been observed.
Whether a potential prognostic factor should be regarded as being independent depends critically on the potential confounding factors adjusted for in the multivariate model. In our study, the RR ratio point estimate for the upper quartile of homocysteine levels was reduced from 2.4 to 1.7 after adjustment for the confounders identified. In accord with our findings, the relation between homocysteine levels and mortality in the study by Nygård et al,13 encompassing patients with predominantly stable coronary artery disease, was somewhat weakened after adjustment for serum creatinine concentration. In the current analysis we also identified preadmission diuretic treatment as an important confounder of the relation between homocysteine level and mortality, a variable not adjusted for in the previous study.13 The reason why diuretic use and a history of congestive heart failure were associated with elevated homocysteine levels is not entirely clear. The observation that the level of homocysteine was increased in patients with a history of myocardial infarction and the lack of correlation between homocysteine concentration and left ventricular ejection fraction suggest that the association is based on the extent of atherosclerosis rather than on ventricular dysfunction per se. Alternatively, diuretic use may potentially affect homocysteine production or metabolism. However, diuretic use and a history of previous congestive heart failure showed some degree of co-linearity and, in a stepwise analysis, heart failure seemed to be the stronger determinant of homocysteine levels. To determine whether diuretic use affects homocysteine levels and to determine whether hyperhomocysteinemia is related to adverse prognosis in patients with congestive heart failure, should be objectives of future investigations.
Based on the results of epidemiologic studies showing an association between homocysteine level and the incidence of coronary heart disease,5,6,35 combined with the well-documented homocysteine-lowering effects of folic acid and vitamins B6 and B12,36 a number of clinical trials with the primary objective of examining the effect of vitamin therapy on the incidence of clinical end points have recently been initiated or planned. Despite the lack of previous data showing a relation between homocysteine levels and clinical end points in patients with acute coronary syndromes, at least one of these studies will selectively include patients with acute myocardial infarction. Because patients with acute coronary syndromes have a higher probability of adverse outcome, including the development and recurrence of myocardial infarction and sudden cardiac death, than patients with stable angina pectoris, the a priori probability of demonstrating a beneficial effect of medical interventions is likely to be greater in the former group. However, the pathophysiology of the acute coronary syndrome is complex with the activation of platelets, the coagulation system, inflammatory processes, and the neuroendocrine axes.1- 3 Accordingly, the contribution of a single factor, like homocysteine, may be obscured by other competing pathogenetic mechanisms. Despite the potential dilution of the effect of homocysteine by other pathophysiological processes, our observations suggest that it is possible to identify patients at high risk of premature death by measurement of the homocysteine concentration in the acute phase. This view is substantiated by unpublished data from our group (B.C., oral communication, June 12, 1999). In 123 patients with an acute myocardial infarction, plasma homocysteine levels on the 16th day of hospital admission (12.46 [4.29] mmol/L) were unchanged compared with those observed 3 months later (12.12 [4.52] mmol/L) (B.C., oral communication, June 12, 1999). Our findings provide further support for the hypotheses that early initiation of homocysteine-lowering therapy may improve the prognosis of patients with acute coronary syndromes, and that intervention with homocysteine-lowering agents in patients with an acute coronary syndrome may be more likely to yield a positive result if patients with elevated homocysteine levels, rather than unselected patients, are targeted.
The predictive value of homocysteine was similar for early and late mortality, consistent with the theory that the harmful effect of homocysteine is not mediated exclusively via procoagulant or exclusively via atherogenic mechanisms. This observation is supported by experimental data suggesting that the homocysteine concentration not only may induce activation of platelets and of the coagulation system,16- 22 but also cause reduced nitric oxide bioavailability with secondary endothelial dysfunction,23- 26 as well as activation of endothelial cells and monocytes by stimulation of proinflammatory cytokines,27,28 resulting in the progression of atherosclerosis and plaque destabilization.3,4
Despite a significant association between homocysteine levels and prognosis, it should be underscored that even a prospective study like this one does not provide final evidence for a causal relation between the homocysteine level and survival. For instance, it is conceivable that atherosclerosis results in a secondary increase in homocysteine levels. Although subjects in prospective cohort studies may be free of clinical vascular disease at the start of the study, the extent of subclinical disease may explain increased levels of homocysteine in incident cases. Likewise, in cohort studies of patients with established coronary artery disease, the extent of atherosclerosis may differ considerably between subjects, thereby explaining the association between the homocysteine level and risk of subsequent events. Accordingly, the demonstration of a causal relation between the homocysteine level and cardiovascular disease must await the results of ongoing clinical trials.37
In our study the concentration of homocysteine was measured in serum rather than in plasma samples. Owing to a continuous production and release of homocysteine from blood cells, even at room temperature, homocysteine levels measured in serum samples tend to be somewhat higher than in corresponding plasma samples.38,39 Although the use of serum samples theoretically may have affected our results, because of the standardized handling of the samples, the magnitude of the increase is likely to be the same in all subjects. Moreover, if a bias has been introduced, it is likely to have resulted in an underestimation rather than an overestimation of the prognostic value of homocysteine concentration.
In this prospective study of an inception cohort of patients, admitted to a single coronary care unit with acute coronary syndromes, serum homocysteine levels were significantly related to long-term survival. Despite previous reports of an increase in circulating homocysteine levels in the convalescent phase after acute myocardial infarction, homocysteine measured in the acute phase provided independent prognostic information with regard to the end point all-cause mortality. Accordingly, early measurement of homocysteine levels in patients with acute coronary syndromes may prove useful for risk stratification. Our findings suggest that homocysteine levels measured on hospital admission in patients with acute coronary syndromes, represent a valid criterion for stratifying patients into groups that are more or less likely to benefit from homocysteine-lowering therapy and may have important implications for the planning and design of future clinical trials evaluating the effect of therapy aimed at reducing circulating homocysteine levels.
Accepted for publication December 8, 1999.
The study was supported by grants from the Swedish Heart and Lung Foundation (Drs Hartford and Caidahl), the Vårdal Foundation (Dr Hartford), grant K99-04RM-13192-01 from the Medical Research Council Stockholm; and the Sahlgrenska Foundations (Dr Caidahl), Gothenburg Medical Society (Dr Caidahl), and the Local Government of Gothenburg (Drs Hartford, Herlitz, and Caidahl), Gothenburg, Sweden.
We are grateful to the staff at the Departments of Cardiology (Olof Wiklund, MD, PhD) and Clinical Physiology (Marie Beckman Suurküla, MD, PhD) for making this study possible. In particular we acknowledge the contributions of Irene Ditmansen Schanche, Margareta Sjölin, Elisabeth Perers, MD, Ann-Sofi Petersson, MSc, Natalia Kharitonova, and Mona From Attebring, as well as the participation in echocardiographic investigations by Odd Bech-Hanssen, MD, PhD, Annika Berggren, MD, and Vuk Kujacic, MD, PhD. The Heart and Lung Institute, Gothenburg University, is also acknowledged (Peter Friberg, MD, PhD, and Åke Hjalmarson, MD, PhD).
Corresponding author: Kenneth Caidahl, MD, PhD, Department of Clinical Physiology, Sahlgrenska University Hospital, Gothenburg, Sweden (e-mail: email@example.com).