- Email: [email protected]
Value of Routine Admission Laboratory Tests to Predict Thirty-Day Mortality in Patients With Acute Myocardial Infarction
Value of Routine Admission Laboratory Tests to Predict Thirty-Day Mortality in Patients With Acute Myocardial Infarction Krischan Daniël Sjauw, MDa, Iwan Cornelis Clemens van der Horst, MDa,*, Maarten Willem Nicolaas Nijsten, MDb, Wybe Nieuwland, MDa, and Felix Zijlstra, MDa Most risk-stratification instruments that have been developed to predict outcome after myocardial infarction do not make use of laboratory parameters, although several laboratory parameters have been shown to be predictors of adverse outcome. To assess the prognostic value of routine admission laboratory tests, we studied a sample of 264 of 3,746 patients with myocardial infarction from a coronary care unit database of 12,043 patients for differences between survivors and nonsurvivors at 30 days. In multivariate analyses, higher white blood cell count, higher levels of serum creatinine, glucose, and lactate dehydrogenase, and lower platelet count were identified as independent risk factors for 30-day mortality. The model that incorporated these risk factors (added laboratory parameters model) had a 17% higher predictive power than did the model that contained only conventional risk factors (conventional risk factor model). The added laboratory parameters model showed better discriminative ability than the conventional risk factor model according to the area under the curve (0.87 vs 0.80). In conclusion, routine admission laboratory tests hold significant prognostic information, with value in addition to conventional risk factors. Incorporating these tests in risk-stratification instruments will further improve risk assessment of patients with myocardial infarction. © 2006 Elsevier Inc. All rights reserved. (Am J Cardiol 2006;97:1435–1440) Several risk stratification instruments have been developed to predict outcome after myocardial infarction, such as the multivariable risk assessment model derived from the First Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries (GUSTO-I) trial,1 the Thrombolysis In Myocardial Infarction risk score,2 the Gruppo Italiano per lo Studio della Sopravvivenza nell’Infarto Miocardico (GISSI) risk index,3 and the Global Registry of Acute Coronary Events (GRACE)4,5 risk models. Derived from findings of large clinical trials and registries, these instruments are based on demographic and clinical risk factors, electrocardiographic changes, and measurements of heart failure or left ventricular dysfunction. Only the GISSI risk index and the GRACE risk models make some use of laboratory values. From routine laboratory tests obtained at admission, white blood cell count,6,7 glucose,8,9 hemoglobin,10 serum creatinine,11,12 serum urea nitrogen,13 and sodium14 have been shown to be independent predictors of adverse outcome. In most studies, these were evaluated as isolated predictors of mortality, and as such, were incorporated in some of the risk-stratification instruments. In this study we investigated the incremental prognostic value of combined routine hematology and bio-
The Departments of aCardiology and bIntensive Care Medicine, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands. Manuscript received August 18, 2005; revised manuscript received and accepted December 1, 2005. * Corresponding author: Tel: 31-50-361-2355; fax: 31-50-361-1347. E-mail address: [email protected] (I.C.C. van der Horst). 0002-9149/06/$ – see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.amjcard.2005.12.034
chemistry laboratory values in addition to conventional risk factors. •••
Data were drawn from a database of all patients who had been consecutively admitted from 1993 to 2003 to a single university center coronary care unit (University Medical Center Groningen, Groningen, The Netherlands). The database contained the laboratory results of all admissions. Laboratory results were obtained from the central laboratory database and checked for inconsistency. Cases with acute myocardial infarction were selected, with acute myocardial infarction defined as a creatine kinase level ⬎250 U/L (5 times the upper limit of normal used by our clinical laboratory) within the first 4 days of admission. No selection was made based on the existence of ST-segment elevation or type of treatment, i.e., thrombolysis or primary angioplasty. From the selected cases with myocardial infarction, we took a stratified randomized sample. A primary division was made in 30-day mortality and a secondary division was made in year of admission; i.e., 12 patients were randomly selected per year from among survivors and nonsurvivors at 30 days, resulting in a total sample of 264 patients, with 132 patients in each group (12 ⫻ 11 ⫻ 2 ⫽ 264). Baseline characteristics, including demographic, clinical presentation and electrocardiographic data, hospital procedures, and admission medications, were obtained by review of electronic medical records and hospital case records. Characteristics that were not described were considered absent. Except for family history of coronary artery disease www.AJConline.org
1436
The American Journal of Cardiology (www.AJConline.org)
Table 1 Baseline characteristics of patients according to 30-day mortality Variables Age (yrs) Men Hypertension Diabetes mellitus Current smoker Family history of coronary artery disease Previous myocardial infarction Previous percutaneous coronary intervention Previous coronary artery bypass grafting Stroke/transient ischemic attack Peripheral arterial disease Dyslipidemia Clinical presentation and ECG data Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg) Heart rate (beats/min) Sinus rhythm Atrial fibrillation Bundle branch block Anterior myocardial infarction Initial laboratory results Sodium (mmol/L) Potassium (mmol/L) Alanine aminotransferase (U/L) Aspartate aminotransferase (U/L) Lactate dehydrogenase (U/L) Serum creatinine (mol/L) Serum urea nitrogen (mmol/L) Total creatinine kinase (U/L) Creatinine kinase-MB (U/L) Hemoglobin (mmol/L) White blood cell count (10 E9/L) Platelet count (10 E9/L) Glucose (mmol/L) Hospital procedures Reanimation or defibrillation Revascularization Thrombolysis Percutaneous coronary intervention Coronary artery bypass grafting
Survivors (n ⫽ 132)
Nonsurvivors (n ⫽ 132)
p Value
60.9 ⫾ 12.8 64.4% 30.3% 9.1% 42.4% 30.3%
72.6 ⫾ 10.5 60.6% 34.1% 19.7% 22.0% 12.9%
⬍0.0001 0.53 0.51 0.01 ⬍0.0001 0.001
20.5%
35.6%
0.006
6.8%
8.3%
0.64
9.1%
9.1%
1.000
5.3%
21.2%
⬍0.0001
8.3% 30.3%
14.4% 17.4%
0.12 0.01
130.9 ⫾ 25.3
122.9 ⫾ 26.5
0.01
77.0 ⫾ 16.5
72.2 ⫾ 15.6
0.02
79.1 ⫾ 22.1 91.9% 5.3% 11.4% 29.5%
86.3 ⫾ 23.2 85.1% 9.8% 16.7% 31.1%
0.004 0.10 0.16 0.22 0.79
138.6 ⫾ 3.2 4.2 ⫾ 0.41 50.4 ⫾ 35.4
137.2 ⫾ 3.9 4.6 ⫾ 0.85 106.1 ⫾ 230
0.003 0.001 0.002
161.3 ⫾ 209.6
261.5 ⫾ 342.1
0.001
521.7 ⫾ 401.4
812.0 ⫾ 802.9 ⬍0.0001
94.6 ⫾ 30.3 6.9 ⫾ 2.5
145.4 ⫾ 96.0 11.3 ⫾ 7.0
725.8 ⫾ 1167.3 717.3 ⫾ 921.9
⬍0.0001 ⬍0.0001 0.37
36.7 ⫾ 38.8
47.3 ⫾ 52.1
0.19
8.0 ⫾ 1.3 11.4 ⫾ 3.6
7.5 ⫾ 1.5 14.9 ⫾ 9.4
0.007 ⬍0.0001
238.6 ⫾ 87.1 7.9 ⫾ 2.6
211.9 ⫾ 67.7 9.7 ⫾ 3.6
0.03 ⬍0.0001
3.0%
18.2%
⬍0.0001
51.5% 15.9% 38.6%
30.3% 12.9% 21.2%
⬍0.0001 0.48 0.002
14.4%
14.4%
1.000
Values are means ⫾ SDs or percentages. ECG ⫽ electrocardiographic.
(direct blood relatives with history of coronary artery disease at ⬍60 years of age) and dyslipidemia (diagnosed/ treated dyslipidemia or 2 separate measurements of a total cholesterol level ⬎6.5 mmol/L, which is the upper limit of normal by our clinical laboratory), definitions of the data elements were adapted from the American College of Cardiology clinical data standards for patients with acute coronary syndromes. The laboratory results obtained at admission were used for analyses. The primary end point was all-cause mortality within 30 days after admission. Deaths were classified as death due to sudden cardiac death, recurrent myocardial infarction, stroke, progressive heart failure or cardiogenic shock, cardiac arrhythmia, other cardiovascular causes, and noncardiovascular causes. Data are presented as means ⫾ SDs for continuous variables and as frequencies for categorical variables. Normal distribution of variables was assessed visually and by the Shapiro-Wilk test. Differences between survivors and nonsurvivors in continuous variables were tested with Student’s t test or the Mann-Whitney U statistic test, as appropriate. Differences in categorical variables were tested by chi-square test. All tests were 2-tailed, and a p value ⬍0.05 was considered statistically significant. Multivariate stepwise logistic regression analyses were performed to identify independent predictors of 30-day mortality. Candidate variables were those that proved to be univariately associated with outcome. Initially, only conventional risk factors were used in the analyses; subsequently, laboratory parameters were added. Significance at an ␣ value ⱕ0.1 was used as the criterion for entry or removal from the models. Interaction and multicollinearity were investigated. The discriminative ability of the derived models was compared by constructing the receiver-operating characteristic curves and calculating the area under them. Calibration, i.e., the extent to which the predicted probabilities are similar to what is observed, of the models was evaluated by the Hosmer-Lemeshow statistic. Odds ratios are reported with logistic regression models that adjust for factors that are independently associated with outcome. SPSS 11.0.1 (SPSS, Inc., Chicago, Illinois) was used for statistical analysis. The total database consisted of 12,043 patients, 3,746 of whom had a creatinine kinase level ⬎250 U/L. In these patients with acute myocardial infarction, mortality was 9.2%. The study sample (n ⫽ 264) consisted of 62.5% men who were 19.8 to 94.4 years of age. Baseline characteristics for survivors and nonsurvivors of myocardial infarction at 30 days are presented in Table 1. Nonsurvivors were more likely to be older. They were also more likely to have a history of diabetes, previous myocardial infarction, and stroke/transient ischemic attack. On presentation, systolic and diastolic blood pressures were lower in nonsurvivors; heart rate was higher. Nonsurvivors underwent reanimation or defibrillation more frequently and revascularization procedures less frequently. Death due to progressive heart failure or cardiogenic shock was the most frequent cause of
Coronary Artery Disease/Routine Laboratory Tests in Myocardial Infarction Table 4 Derived multivariate logistic models
Table 2 Causes of death in study population Variable Sudden cardiac death Recurrent myocardial infarction Stroke Progressive heart failure/cardiogenic shock Cardiac arythmia Other cardiovascular causes Noncardiovascular causes
12.7% 7.9% 5.6% 39.7% 11.9% 11.1% 11.1%
Percentages based on 126 patients because of 6 cases with missing causes of death. Table 3 Candidate predictor variables in univariate analysis OR Conventional risk factors Age (yrs) History of stroke/transient ischemic attack Current smoking Family history of coronary artery disease Previous myocardial infarction Heart rate (beats/min) History of dyslipidemia Systolic blood pressure (mm Hg) History of diabetes mellitus Diastolic blood pressure (mm Hg) Laboratory parameters Serum creatinine (mol/L) Blood urea nitrogen (mmol/L) White blood cell count (10E9/L) Glucose (mmol/L) Potassium (mmol/L) Lactate dehydrogenase (U/L) Alanine aminotransferase (U/L) Sodium (mmol/L) Aspartate aminotransferase (U/L) Platelet count (10E9/L) Hemoglobin (mmol/L)
1437
95% CI
p Value
Wald’s Statistic
1.089 1.062–1.116 ⬍0.0001 4.808 2.018–11.454 ⬍0.0001
43.696 12.568
0.382 0.223–0.654 0.340 0.181–0.639
⬍0.0001 0.001
12.307 11.257
2.150 1.237–3.738
0.007
7.364
1.014 1.003–1.026 2.060 1.150–3.692 0.988 0.979–0.998
0.01 0.02 0.02
6.212 5.904 5.864
2.453 1.179–5.101
0.02
5.768
0.981 0.966–0.997
0.02
5.647
1.022 1.014–1.031
⬍0.0001
28.061
1.271 1.162–1.390
⬍0.0001
27.647
1.142 1.070–1.218
⬍0.0001
16.157
1.205 1.091–1.318 2.392 1.516–3.775 1.001 1.000–1.002
⬍0.0001 ⬍0.0001 ⬍0.0001
14.255 14.051 12.159
1.010 1.004–1.017
0.001
10.262
0.893 0.828–0.964 1.002 1.000–1.003
0.004 0.007
8.361 7.164
0.995 0.992–0.999 0.793 0.659–0.954
0.01 0.01
6.318 6.033
For continuous variables ORs are noted per unit of increment. CI ⫽ confidence interval; OR ⫽ odds ratio.
death (39.7%), followed by death due to sudden cardiac death and cardiac arrhythmia (12.7% and 11.9%; Table 2). Baseline characteristics were tested as univariate predictors in logistic regression models (Table 3). First, only the conventional risk factors, which were significant in univariate analysis, were entered into multivariate models. Sys-
CRF model Age (yrs) Systolic blood pressure (mm Hg) History of stroke/transient ischemic attack Heart rate (beats/min) ALP model Age (yrs) White blood cell count (109/L) History of stroke/transient ischemic attack Serum creatinine (mol/L) Glucose (mmol/L) Lactate dehydrogenase (U/L) Platelet count (109/L)
OR
95% CI
p Value
Wald’s Statistic
1.086 0.984
1.058–1.115 0.973–0.995
⬍0.0001 0.006
38.720 7.520
3.119
1.214–8.011
0.02
5.586
1.016
1.002–1.029
0.02
5.387
1.091 1.152
1.058–1.125 1.050–1.265
⬍0.0001 0.003
30.625 8.898
4.490
1.506–13.385
0.007
7.263
1.011
1.003–1.020
0.008
7.069
1.147 1.001
1.018–1.292 1.000–1.001
0.025 0.068
5.045 3.336
0.996
0.991–1.001
0.097
2.760
For continuous variables, the ORs are noted per unit of increment. The Nagelkerke R2 Values, a measurement of predictive capability, for the CRF and ALP models are ⫺0.34 and 0.51, respectively. 1 ⫺ R2 is the proportion of predictive capability that is attributable to causal factors not contained in the model, a different model form, and/or random effects. Abbreviations as in Table 3.
tolic and diastolic blood pressures seemed to be strongly correlated (r ⫽ 0.741 p ⬍0.0001) by attenuating each other’s effect. Based on Wald’s statistic and review of the literature,1–5 we excluded diastolic blood pressure from the model. The final conventional risk factor (CRF) model consisted of age, history of stroke/transient ischemic attack, systolic blood pressure, and heart rate (Table 4). From these variables, age was the strongest independent predictor of 30-day mortality. In a second step, laboratory parameters were added to the CRF model. A strong correlation existed between serum creatinine and serum urea nitrogen (r ⫽ 0.738, p ⬍0.0001) and across lactate dehydrogenase, aspartate aminotransferase, and alanine aminotransferase (lactate dehydrogenase/ aspartate aminotransferase r ⫽ 0.901, p ⬍0.0001; lactate dehydrogenase/alanine aminotransferase r ⫽ 0.794, p ⬍0.0001; aspartate aminotransferase/alanine aminotransferase r ⫽ 0.789, p ⬍0.0001). We excluded serum urea nitrogen and aspartate and alanine aminotransferases from the model, rather than creating combined variables, to simplify the final model. The final added laboratory parameters (ALP) model consisted of age, history of stroke/transient ischemic attack, white blood cell count, serum creatinine, glucose, lactate dehydrogenase, and platelet count (Table 4). Comparing the 2 models showed that the ALP model had a 17% higher predictive capability than did the CRF model (Nagelkerke R2 0.51 vs 0.34). Receiver-operating
1438
The American Journal of Cardiology (www.AJConline.org)
1
Sensitivity
0.8
0.6
0.4
0.2
0 0
0.2
0.4
0.6
0.8
1
1 - Specificity Figure 1. Receiver-operating characteristic curves for the derived multivariate models. The ALP model (black line) showed better discriminative ability for 30-day mortality than the CRF model (gray line). Areas under receiver-operating characteristic curves were 0.87 and 0.80, respectively.
characteristic curves (Figure 1) for the 2 models reflected this increment, with the ALP model having better discriminative ability for 30-day mortality (area under the receiveroperating characteristic curve 0.87 vs 0.80). According to their Hosmer-Lemeshow statistic, the 2 models were well calibrated. •••
This study reflects an overall representative population of patients with myocardial infarction who were admitted to a coronary care unit over a long range of years (1993 to 2003). It extends to previous studies, which mostly were done in clinical trial populations, about the prognostic value on short-term mortality of several routine laboratory parameters. As expected, many laboratory parameters and other baseline characteristics were significant predictors of 30day mortality in univariate logistic regression analyses. However, when fitted together in multivariate models, many parameters and baseline characteristics lost their significance or attenuated each other’s effect. Analyses showed that many variables hold prognostic information, but that their importance varies. Investigators of the GUSTO-I trial noted this and concluded that the relations of clinical determinants of mortality are complex and multifactorial.1 In the array of baseline characteristics of the GUSTO-I trial, age, lower systolic blood pressure, higher Killip’s class, increased heart rate, and anterior infarction contained 90% of the prognostic information. However, they did not examine the effects of laboratory parameters on other variables. The GISSI risk index and the GRACE risk models did examine laboratory parameters.3–5 They incorporated white
blood cell count, fibrinogen and high-density lipoprotein concentration, and serum creatinine, respectively, but these laboratory parameters were not adjusted for other laboratory parameters. In this study, a new observation was the amount of prognostic information that was held by white blood cell count, serum creatinine, glucose, lactate dehydrogenase, and platelet count together. In the multivariate (ALP) model of conventional risk factors and laboratory parameters, aside from age and history of stroke/transient ischemic attack, they were found to independent predictors of 30-day mortality. In addition, although heart rate and systolic blood pressure were removed from the ALP model, it contained more prognostic information than the CRF model. This finding is interesting, because, until presently, the risk scores based on findings of large clinical trials and registries were primarily based on conventional risk factors. In the ALP model, white blood cell count, serum creatinine, and glucose appeared as the strongest independent laboratory parameters, aside from platelet count and lactate dehydrogenase. Conclusions about the pathophysiologic and etiologic relations between the found independent laboratory parameters and mortality are beyond the scope of this study, but the following is noteworthy to address. Evidence is increasing that white blood cell count is not merely a marker of a general state of inflammation but plays a role in the pathogenesis of microvascular injury. Further, white blood cell count on admission appears to be associated with outcome in myocardial infarction, which may be explained by the association with decreased epicardial and myocardial blood flow, thromboresistance, and a higher incidence of new-onset chronic heart failure.6,15 Interestingly, death due to progressive heart failure/cardiogenic shock was the most frequent cause of death in our study. The relation between serum creatinine, reflecting renal function, and adverse outcome is well established.11,12 Mild and chronic renal dysfunction is strongly associated with adverse outcomes after myocardial infarction.16 Moreover, the relation between hyperglycemia and adverse outcome in patients with myocardial infarction is well established in diabetic and nondiabetic patients.8,9 Hyperglycemia was recently found to be a strong predictor of impaired epicardial flow, possibly due to prothrombotic properties and increased inflammatory response associated with hyperglycemia.17 Libby and Simon18 addressed the role of platelets in the complex relation between inflammation and thrombosis in vascular pathology. Upon activation in the process of plaque rupture and thrombus formation, platelets release several inflammatory modulators. However, they also respond to chemoattractant cytokines. Leukocytes bind to platelets at sites of intravascular platelet coverage and circulating leukocyte-platelet aggregates can be found in coronary artery disease. The extent of usage of platelets may be an explanation for the found association between lower platelet count and adverse outcome. Although not direct
Coronary Artery Disease/Routine Laboratory Tests in Myocardial Infarction
evidence for this association, Frossard et al19 found that platelet function was independently associated with severity of myocardial infarction. Assays of lactate dehydrogenase are widely performed in the early phase of suspected ischemic myocardial injury. Elsman et al20 found that infarct size as determined by cumulative lactate dehydrogenase at 36 hours is a good method for early risk stratification for cardiac death in patients who are treated with primary angioplasty for acute ST-elevation myocardial infarction. This finding is in concordance with the association of early lactate dehydrogenase measurements with 30-day mortality in this study. In short, the found independent laboratory parameters are biomarkers of the inflammatory, metabolic, neuroendocrine, and tissue destructive processes in acute coronary syndrome. The parameters constitute a diverse set of markers that represent different pathophysiologic processes and may explain their additive predictive value for 30-day mortality. There are several important limitations to the study. First, it concerns a retrospective analysis. However, selection bias is unlikely to play an important role because the studied population is an unbiased representative population of patients with myocardial infarction. Second, we did not have data on the entire array of variables used in previous risk-stratification instruments; namely, we did not have information on Killip’s class, which in the model based on the GUSTO-I trial and the Thrombolysis In Myocardial Infarction risk score, which was a strong predictor of adverse outcome. Further, in this study, Killip’s class could have offered important prognostic information because progressive heart failure/cardiogenic shock was the primary cause of death. However, the prognostic information of Killip’s class is partly accounted for in blood pressure and heart rate. Third, we did not have data on some of the newer biomarkers, such as C-reactive protein, brain natriuretic peptide, and Nterminal pro–B-type natriuretic peptide. These have been shown to hold prognostic information on adverse outcome in patients with myocardial infarction. Whether they hold prognostic information in addition to other risk factors remains unclear and should be tested by adjusting not only for conventional risk factors but also for several predictive laboratory parameters, as shown in this study. Brain natriuretic peptide and N-terminal pro–B-type natriuretic peptide, which reflect myocardial ischemia, could especially provide additive prognostic information because the information they contain differs from other biomarkers. Fourth, the ALP model offers excellent discriminative ability in the study sample, but the model should be tested in a validation cohort. A planned analysis in a prospective gathered database is currently under investigation.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14. 1. Lee KL, Woodlief LH, Topol EJ, Weaver WD, Betriu A, Col J, Simoons M, Aylward P, van de Werf F, Califf RM. Predictors of 30-day mortality in the era of reperfusion for acute myocardial infarc-
1439
tion. Results from an international trial of 41,021 patients. GUSTO-I Investigators. Circulation 1995;91:1659 –1668. Morrow DA, Antman EM, Charlesworth A, Cairns R, Murphy SA, de Lemos JA, Giugliano RP, McCabe CH, Braunwald E. TIMI risk score for ST-elevation myocardial infarction: a convenient, bedside, clinical score for risk assessment at presentation: an intravenous nPA for treatment of infarcting myocardium early II trial substudy. Circulation 2000;102:2031–2037. Marchioli R, Avanzini F, Barzi F, Chieffo C, Di Castelnuovo A, Franzosi MG, Geraci E, Maggioni AP, Marfisi RM, Mininni N, et al. Assessment of absolute risk of death after myocardial infarction by use of multiple-risk-factor assessment equations: GISSI-Prevenzione mortality risk chart. Eur Heart J 2001;22:2085–2103. Granger CB, Goldberg RJ, Dabbous O, Pieper KS, Eagle KA, Cannon CP, van de Werf F, Avezum A, Goodman SG, Flather MD, Fox KA. Predictors of hospital mortality in the global registry of acute coronary events. Arch Intern Med 2003;163:2345–2353. Eagle KA, Lim MJ, Dabbous OH, Pieper KS, Goldberg RJ, van de Werf F, Goodman SG, Granger CB, Steg PG, Gore JM, et al. A validated prediction model for all forms of acute coronary syndrome: estimating the risk of 6-month postdischarge death in an international registry. JAMA 2004;291:2727–2733. Barron HV, Cannon CP, Murphy SA, Braunwald E, Gibson CM. Association between white blood cell count, epicardial blood flow, myocardial perfusion, and clinical outcomes in the setting of acute myocardial infarction: a Thrombolysis In Myocardial Infarction 10 substudy. Circulation 2000;102:2329 –2334. Grzybowski M, Welch RD, Parsons L, Ndumele CE, Chen E, Zalenski R, Barron HV. The association between white blood cell count and acute myocardial infarction in-hospital mortality: findings from the National Registry of Myocardial Infarction. Acad Emerg Med 2004; 11:1049 –1060. Foo K, Cooper J, Deaner A, Knight C, Suliman A, Ranjadayalan K, Timmis AD. A single serum glucose measurement predicts adverse outcomes across the whole range of acute coronary syndromes. Heart 2003;89:512–516. Pinto DS, Skolnick AH, Kirtane AJ, Murphy SA, Barron HV, Giugliano RP, Cannon CP, Braunwald E, Gibson CM. U-shaped relationship of blood glucose with adverse outcomes among patients with STsegment elevation myocardial infarction. J Am Coll Cardiol 2005;46: 178 –180. Sabatine MS, Morrow DA, Giugliano RP, Burton PB, Murphy SA, McCabe CH, Gibson CM, Braunwald E. Association of hemoglobin levels with clinical outcomes in acute coronary syndromes. Circulation 2005;111:2042–2049. Gibson CM, Pinto DS, Murphy SA, Morrow DA, Hobbach HP, Wiviott SD, Giugliano RP, Cannon CP, Antman EM, Braunwald E. Association of creatinine and creatinine clearance on presentation in acute myocardial infarction with subsequent mortality. J Am Coll Cardiol 2003;42:1535–1543. Yamaguchi J, Kasanuki H, Ishii Y, Yagi M, Ogawa H, Fujii SY, Koganei H, Okada H, Kimura H, Horie T, et al. Prognostic significance of serum creatinine concentration for in-hospital mortality in patients with acute myocardial infarction who underwent successful primary percutaneous coronary intervention (from the Heart Institute of Japan Acute Myocardial Infarction [HIJAMI] Registry). Am J Cardiol 2004;93:1526 –1528. Kirtane AJ, Leder DM, Waikar SS, Chertow GM, Ray KK, Pinto DS, Karmpaliotis D, Burger AJ, Murphy SA, Cannon CP, et al. Serum blood urea nitrogen as an independent marker of subsequent mortality among patients with acute coronary syndromes and normal to mildly reduced glomerular filtration rates. J Am Coll Cardiol 2005;45:1781–1786. Goldberg A, Hammerman H, Petcherski S, Zdorovyak A, Yalonetsky S, Kapeliovich M, Agmon Y, Markiewicz W, Aronson D. Prognostic importance of hyponatremia in acute ST-elevation myocardial infarction. Am J Med 2004;117:242–248.
1440
The American Journal of Cardiology (www.AJConline.org)
15. Furman MI, Gore JM, Anderson FA, Budaj A, Goodman SG, Avezum A, Lopez-Sendon J, Klein W, Mukherjee D, Eagle KA, et al. Elevated leukocyte count and adverse hospital events in patients with acute coronary syndromes: findings from the Global Registry of Acute Coronary Events (GRACE). Am Heart J 2004;147:42– 48. 16. Anavekar NS, McMurray JJ, Velazquez EJ, Solomon SD, Kober L, Rouleau JL, White HD, Nordlander R, Maggioni A, Dickstein K, et al. Relation between renal dysfunction and cardiovascular outcomes after myocardial infarction. N Engl J Med 2004;351:1285– 1295. 17. Timmer JR, Ottervanger JP, de Boer MJ, Dambrink JH, Hoorntje JC, Gosselink AT, Suryapranata H, Zijlstra F, van’t Hof AW. Hyperglycemia is an important predictor of impaired coronary flow before
reperfusion therapy in ST-segment elevation myocardial infarction. J Am Coll Cardiol 2005;45:999 –1002. 18. Libby P, Simon DI. Inflammation and thrombosis: the clot thickens. Circulation 2001;103:1718 –1720. 19. Frossard M, Fuchs I, Leitner JM, Hsieh K, Vlcek M, Losert H, Domanovits H, Schreiber W, Laggner AN, Jilma B. Platelet function predicts myocardial damage in patients with acute myocardial infarction. Circulation 2004;110:1392–1397. 20. Elsman P, Zijlstra F, Miedema K, Hoorntje JC, Dikkeschei LD, Slingerland RJ, Reiffers S, de Boer MJ, Suryapranata H. The predictive value of cumulative lactate dehydrogenase release within the first 72 h of acute myocardial infarction in patients treated with primary angioplasty. Ann Clin Biochem 2004;41:142–148.
The Departments of aCardiology and bIntensive Care Medicine, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands. Manuscript received August 18, 2005; revised manuscript received and accepted December 1, 2005. * Corresponding author: Tel: 31-50-361-2355; fax: 31-50-361-1347. E-mail address: [email protected] (I.C.C. van der Horst). 0002-9149/06/$ – see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.amjcard.2005.12.034
chemistry laboratory values in addition to conventional risk factors. •••
Data were drawn from a database of all patients who had been consecutively admitted from 1993 to 2003 to a single university center coronary care unit (University Medical Center Groningen, Groningen, The Netherlands). The database contained the laboratory results of all admissions. Laboratory results were obtained from the central laboratory database and checked for inconsistency. Cases with acute myocardial infarction were selected, with acute myocardial infarction defined as a creatine kinase level ⬎250 U/L (5 times the upper limit of normal used by our clinical laboratory) within the first 4 days of admission. No selection was made based on the existence of ST-segment elevation or type of treatment, i.e., thrombolysis or primary angioplasty. From the selected cases with myocardial infarction, we took a stratified randomized sample. A primary division was made in 30-day mortality and a secondary division was made in year of admission; i.e., 12 patients were randomly selected per year from among survivors and nonsurvivors at 30 days, resulting in a total sample of 264 patients, with 132 patients in each group (12 ⫻ 11 ⫻ 2 ⫽ 264). Baseline characteristics, including demographic, clinical presentation and electrocardiographic data, hospital procedures, and admission medications, were obtained by review of electronic medical records and hospital case records. Characteristics that were not described were considered absent. Except for family history of coronary artery disease www.AJConline.org
1436
The American Journal of Cardiology (www.AJConline.org)
Table 1 Baseline characteristics of patients according to 30-day mortality Variables Age (yrs) Men Hypertension Diabetes mellitus Current smoker Family history of coronary artery disease Previous myocardial infarction Previous percutaneous coronary intervention Previous coronary artery bypass grafting Stroke/transient ischemic attack Peripheral arterial disease Dyslipidemia Clinical presentation and ECG data Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg) Heart rate (beats/min) Sinus rhythm Atrial fibrillation Bundle branch block Anterior myocardial infarction Initial laboratory results Sodium (mmol/L) Potassium (mmol/L) Alanine aminotransferase (U/L) Aspartate aminotransferase (U/L) Lactate dehydrogenase (U/L) Serum creatinine (mol/L) Serum urea nitrogen (mmol/L) Total creatinine kinase (U/L) Creatinine kinase-MB (U/L) Hemoglobin (mmol/L) White blood cell count (10 E9/L) Platelet count (10 E9/L) Glucose (mmol/L) Hospital procedures Reanimation or defibrillation Revascularization Thrombolysis Percutaneous coronary intervention Coronary artery bypass grafting
Survivors (n ⫽ 132)
Nonsurvivors (n ⫽ 132)
p Value
60.9 ⫾ 12.8 64.4% 30.3% 9.1% 42.4% 30.3%
72.6 ⫾ 10.5 60.6% 34.1% 19.7% 22.0% 12.9%
⬍0.0001 0.53 0.51 0.01 ⬍0.0001 0.001
20.5%
35.6%
0.006
6.8%
8.3%
0.64
9.1%
9.1%
1.000
5.3%
21.2%
⬍0.0001
8.3% 30.3%
14.4% 17.4%
0.12 0.01
130.9 ⫾ 25.3
122.9 ⫾ 26.5
0.01
77.0 ⫾ 16.5
72.2 ⫾ 15.6
0.02
79.1 ⫾ 22.1 91.9% 5.3% 11.4% 29.5%
86.3 ⫾ 23.2 85.1% 9.8% 16.7% 31.1%
0.004 0.10 0.16 0.22 0.79
138.6 ⫾ 3.2 4.2 ⫾ 0.41 50.4 ⫾ 35.4
137.2 ⫾ 3.9 4.6 ⫾ 0.85 106.1 ⫾ 230
0.003 0.001 0.002
161.3 ⫾ 209.6
261.5 ⫾ 342.1
0.001
521.7 ⫾ 401.4
812.0 ⫾ 802.9 ⬍0.0001
94.6 ⫾ 30.3 6.9 ⫾ 2.5
145.4 ⫾ 96.0 11.3 ⫾ 7.0
725.8 ⫾ 1167.3 717.3 ⫾ 921.9
⬍0.0001 ⬍0.0001 0.37
36.7 ⫾ 38.8
47.3 ⫾ 52.1
0.19
8.0 ⫾ 1.3 11.4 ⫾ 3.6
7.5 ⫾ 1.5 14.9 ⫾ 9.4
0.007 ⬍0.0001
238.6 ⫾ 87.1 7.9 ⫾ 2.6
211.9 ⫾ 67.7 9.7 ⫾ 3.6
0.03 ⬍0.0001
3.0%
18.2%
⬍0.0001
51.5% 15.9% 38.6%
30.3% 12.9% 21.2%
⬍0.0001 0.48 0.002
14.4%
14.4%
1.000
Values are means ⫾ SDs or percentages. ECG ⫽ electrocardiographic.
(direct blood relatives with history of coronary artery disease at ⬍60 years of age) and dyslipidemia (diagnosed/ treated dyslipidemia or 2 separate measurements of a total cholesterol level ⬎6.5 mmol/L, which is the upper limit of normal by our clinical laboratory), definitions of the data elements were adapted from the American College of Cardiology clinical data standards for patients with acute coronary syndromes. The laboratory results obtained at admission were used for analyses. The primary end point was all-cause mortality within 30 days after admission. Deaths were classified as death due to sudden cardiac death, recurrent myocardial infarction, stroke, progressive heart failure or cardiogenic shock, cardiac arrhythmia, other cardiovascular causes, and noncardiovascular causes. Data are presented as means ⫾ SDs for continuous variables and as frequencies for categorical variables. Normal distribution of variables was assessed visually and by the Shapiro-Wilk test. Differences between survivors and nonsurvivors in continuous variables were tested with Student’s t test or the Mann-Whitney U statistic test, as appropriate. Differences in categorical variables were tested by chi-square test. All tests were 2-tailed, and a p value ⬍0.05 was considered statistically significant. Multivariate stepwise logistic regression analyses were performed to identify independent predictors of 30-day mortality. Candidate variables were those that proved to be univariately associated with outcome. Initially, only conventional risk factors were used in the analyses; subsequently, laboratory parameters were added. Significance at an ␣ value ⱕ0.1 was used as the criterion for entry or removal from the models. Interaction and multicollinearity were investigated. The discriminative ability of the derived models was compared by constructing the receiver-operating characteristic curves and calculating the area under them. Calibration, i.e., the extent to which the predicted probabilities are similar to what is observed, of the models was evaluated by the Hosmer-Lemeshow statistic. Odds ratios are reported with logistic regression models that adjust for factors that are independently associated with outcome. SPSS 11.0.1 (SPSS, Inc., Chicago, Illinois) was used for statistical analysis. The total database consisted of 12,043 patients, 3,746 of whom had a creatinine kinase level ⬎250 U/L. In these patients with acute myocardial infarction, mortality was 9.2%. The study sample (n ⫽ 264) consisted of 62.5% men who were 19.8 to 94.4 years of age. Baseline characteristics for survivors and nonsurvivors of myocardial infarction at 30 days are presented in Table 1. Nonsurvivors were more likely to be older. They were also more likely to have a history of diabetes, previous myocardial infarction, and stroke/transient ischemic attack. On presentation, systolic and diastolic blood pressures were lower in nonsurvivors; heart rate was higher. Nonsurvivors underwent reanimation or defibrillation more frequently and revascularization procedures less frequently. Death due to progressive heart failure or cardiogenic shock was the most frequent cause of
Coronary Artery Disease/Routine Laboratory Tests in Myocardial Infarction Table 4 Derived multivariate logistic models
Table 2 Causes of death in study population Variable Sudden cardiac death Recurrent myocardial infarction Stroke Progressive heart failure/cardiogenic shock Cardiac arythmia Other cardiovascular causes Noncardiovascular causes
12.7% 7.9% 5.6% 39.7% 11.9% 11.1% 11.1%
Percentages based on 126 patients because of 6 cases with missing causes of death. Table 3 Candidate predictor variables in univariate analysis OR Conventional risk factors Age (yrs) History of stroke/transient ischemic attack Current smoking Family history of coronary artery disease Previous myocardial infarction Heart rate (beats/min) History of dyslipidemia Systolic blood pressure (mm Hg) History of diabetes mellitus Diastolic blood pressure (mm Hg) Laboratory parameters Serum creatinine (mol/L) Blood urea nitrogen (mmol/L) White blood cell count (10E9/L) Glucose (mmol/L) Potassium (mmol/L) Lactate dehydrogenase (U/L) Alanine aminotransferase (U/L) Sodium (mmol/L) Aspartate aminotransferase (U/L) Platelet count (10E9/L) Hemoglobin (mmol/L)
1437
95% CI
p Value
Wald’s Statistic
1.089 1.062–1.116 ⬍0.0001 4.808 2.018–11.454 ⬍0.0001
43.696 12.568
0.382 0.223–0.654 0.340 0.181–0.639
⬍0.0001 0.001
12.307 11.257
2.150 1.237–3.738
0.007
7.364
1.014 1.003–1.026 2.060 1.150–3.692 0.988 0.979–0.998
0.01 0.02 0.02
6.212 5.904 5.864
2.453 1.179–5.101
0.02
5.768
0.981 0.966–0.997
0.02
5.647
1.022 1.014–1.031
⬍0.0001
28.061
1.271 1.162–1.390
⬍0.0001
27.647
1.142 1.070–1.218
⬍0.0001
16.157
1.205 1.091–1.318 2.392 1.516–3.775 1.001 1.000–1.002
⬍0.0001 ⬍0.0001 ⬍0.0001
14.255 14.051 12.159
1.010 1.004–1.017
0.001
10.262
0.893 0.828–0.964 1.002 1.000–1.003
0.004 0.007
8.361 7.164
0.995 0.992–0.999 0.793 0.659–0.954
0.01 0.01
6.318 6.033
For continuous variables ORs are noted per unit of increment. CI ⫽ confidence interval; OR ⫽ odds ratio.
death (39.7%), followed by death due to sudden cardiac death and cardiac arrhythmia (12.7% and 11.9%; Table 2). Baseline characteristics were tested as univariate predictors in logistic regression models (Table 3). First, only the conventional risk factors, which were significant in univariate analysis, were entered into multivariate models. Sys-
CRF model Age (yrs) Systolic blood pressure (mm Hg) History of stroke/transient ischemic attack Heart rate (beats/min) ALP model Age (yrs) White blood cell count (109/L) History of stroke/transient ischemic attack Serum creatinine (mol/L) Glucose (mmol/L) Lactate dehydrogenase (U/L) Platelet count (109/L)
OR
95% CI
p Value
Wald’s Statistic
1.086 0.984
1.058–1.115 0.973–0.995
⬍0.0001 0.006
38.720 7.520
3.119
1.214–8.011
0.02
5.586
1.016
1.002–1.029
0.02
5.387
1.091 1.152
1.058–1.125 1.050–1.265
⬍0.0001 0.003
30.625 8.898
4.490
1.506–13.385
0.007
7.263
1.011
1.003–1.020
0.008
7.069
1.147 1.001
1.018–1.292 1.000–1.001
0.025 0.068
5.045 3.336
0.996
0.991–1.001
0.097
2.760
For continuous variables, the ORs are noted per unit of increment. The Nagelkerke R2 Values, a measurement of predictive capability, for the CRF and ALP models are ⫺0.34 and 0.51, respectively. 1 ⫺ R2 is the proportion of predictive capability that is attributable to causal factors not contained in the model, a different model form, and/or random effects. Abbreviations as in Table 3.
tolic and diastolic blood pressures seemed to be strongly correlated (r ⫽ 0.741 p ⬍0.0001) by attenuating each other’s effect. Based on Wald’s statistic and review of the literature,1–5 we excluded diastolic blood pressure from the model. The final conventional risk factor (CRF) model consisted of age, history of stroke/transient ischemic attack, systolic blood pressure, and heart rate (Table 4). From these variables, age was the strongest independent predictor of 30-day mortality. In a second step, laboratory parameters were added to the CRF model. A strong correlation existed between serum creatinine and serum urea nitrogen (r ⫽ 0.738, p ⬍0.0001) and across lactate dehydrogenase, aspartate aminotransferase, and alanine aminotransferase (lactate dehydrogenase/ aspartate aminotransferase r ⫽ 0.901, p ⬍0.0001; lactate dehydrogenase/alanine aminotransferase r ⫽ 0.794, p ⬍0.0001; aspartate aminotransferase/alanine aminotransferase r ⫽ 0.789, p ⬍0.0001). We excluded serum urea nitrogen and aspartate and alanine aminotransferases from the model, rather than creating combined variables, to simplify the final model. The final added laboratory parameters (ALP) model consisted of age, history of stroke/transient ischemic attack, white blood cell count, serum creatinine, glucose, lactate dehydrogenase, and platelet count (Table 4). Comparing the 2 models showed that the ALP model had a 17% higher predictive capability than did the CRF model (Nagelkerke R2 0.51 vs 0.34). Receiver-operating
1438
The American Journal of Cardiology (www.AJConline.org)
1
Sensitivity
0.8
0.6
0.4
0.2
0 0
0.2
0.4
0.6
0.8
1
1 - Specificity Figure 1. Receiver-operating characteristic curves for the derived multivariate models. The ALP model (black line) showed better discriminative ability for 30-day mortality than the CRF model (gray line). Areas under receiver-operating characteristic curves were 0.87 and 0.80, respectively.
characteristic curves (Figure 1) for the 2 models reflected this increment, with the ALP model having better discriminative ability for 30-day mortality (area under the receiveroperating characteristic curve 0.87 vs 0.80). According to their Hosmer-Lemeshow statistic, the 2 models were well calibrated. •••
This study reflects an overall representative population of patients with myocardial infarction who were admitted to a coronary care unit over a long range of years (1993 to 2003). It extends to previous studies, which mostly were done in clinical trial populations, about the prognostic value on short-term mortality of several routine laboratory parameters. As expected, many laboratory parameters and other baseline characteristics were significant predictors of 30day mortality in univariate logistic regression analyses. However, when fitted together in multivariate models, many parameters and baseline characteristics lost their significance or attenuated each other’s effect. Analyses showed that many variables hold prognostic information, but that their importance varies. Investigators of the GUSTO-I trial noted this and concluded that the relations of clinical determinants of mortality are complex and multifactorial.1 In the array of baseline characteristics of the GUSTO-I trial, age, lower systolic blood pressure, higher Killip’s class, increased heart rate, and anterior infarction contained 90% of the prognostic information. However, they did not examine the effects of laboratory parameters on other variables. The GISSI risk index and the GRACE risk models did examine laboratory parameters.3–5 They incorporated white
blood cell count, fibrinogen and high-density lipoprotein concentration, and serum creatinine, respectively, but these laboratory parameters were not adjusted for other laboratory parameters. In this study, a new observation was the amount of prognostic information that was held by white blood cell count, serum creatinine, glucose, lactate dehydrogenase, and platelet count together. In the multivariate (ALP) model of conventional risk factors and laboratory parameters, aside from age and history of stroke/transient ischemic attack, they were found to independent predictors of 30-day mortality. In addition, although heart rate and systolic blood pressure were removed from the ALP model, it contained more prognostic information than the CRF model. This finding is interesting, because, until presently, the risk scores based on findings of large clinical trials and registries were primarily based on conventional risk factors. In the ALP model, white blood cell count, serum creatinine, and glucose appeared as the strongest independent laboratory parameters, aside from platelet count and lactate dehydrogenase. Conclusions about the pathophysiologic and etiologic relations between the found independent laboratory parameters and mortality are beyond the scope of this study, but the following is noteworthy to address. Evidence is increasing that white blood cell count is not merely a marker of a general state of inflammation but plays a role in the pathogenesis of microvascular injury. Further, white blood cell count on admission appears to be associated with outcome in myocardial infarction, which may be explained by the association with decreased epicardial and myocardial blood flow, thromboresistance, and a higher incidence of new-onset chronic heart failure.6,15 Interestingly, death due to progressive heart failure/cardiogenic shock was the most frequent cause of death in our study. The relation between serum creatinine, reflecting renal function, and adverse outcome is well established.11,12 Mild and chronic renal dysfunction is strongly associated with adverse outcomes after myocardial infarction.16 Moreover, the relation between hyperglycemia and adverse outcome in patients with myocardial infarction is well established in diabetic and nondiabetic patients.8,9 Hyperglycemia was recently found to be a strong predictor of impaired epicardial flow, possibly due to prothrombotic properties and increased inflammatory response associated with hyperglycemia.17 Libby and Simon18 addressed the role of platelets in the complex relation between inflammation and thrombosis in vascular pathology. Upon activation in the process of plaque rupture and thrombus formation, platelets release several inflammatory modulators. However, they also respond to chemoattractant cytokines. Leukocytes bind to platelets at sites of intravascular platelet coverage and circulating leukocyte-platelet aggregates can be found in coronary artery disease. The extent of usage of platelets may be an explanation for the found association between lower platelet count and adverse outcome. Although not direct
Coronary Artery Disease/Routine Laboratory Tests in Myocardial Infarction
evidence for this association, Frossard et al19 found that platelet function was independently associated with severity of myocardial infarction. Assays of lactate dehydrogenase are widely performed in the early phase of suspected ischemic myocardial injury. Elsman et al20 found that infarct size as determined by cumulative lactate dehydrogenase at 36 hours is a good method for early risk stratification for cardiac death in patients who are treated with primary angioplasty for acute ST-elevation myocardial infarction. This finding is in concordance with the association of early lactate dehydrogenase measurements with 30-day mortality in this study. In short, the found independent laboratory parameters are biomarkers of the inflammatory, metabolic, neuroendocrine, and tissue destructive processes in acute coronary syndrome. The parameters constitute a diverse set of markers that represent different pathophysiologic processes and may explain their additive predictive value for 30-day mortality. There are several important limitations to the study. First, it concerns a retrospective analysis. However, selection bias is unlikely to play an important role because the studied population is an unbiased representative population of patients with myocardial infarction. Second, we did not have data on the entire array of variables used in previous risk-stratification instruments; namely, we did not have information on Killip’s class, which in the model based on the GUSTO-I trial and the Thrombolysis In Myocardial Infarction risk score, which was a strong predictor of adverse outcome. Further, in this study, Killip’s class could have offered important prognostic information because progressive heart failure/cardiogenic shock was the primary cause of death. However, the prognostic information of Killip’s class is partly accounted for in blood pressure and heart rate. Third, we did not have data on some of the newer biomarkers, such as C-reactive protein, brain natriuretic peptide, and Nterminal pro–B-type natriuretic peptide. These have been shown to hold prognostic information on adverse outcome in patients with myocardial infarction. Whether they hold prognostic information in addition to other risk factors remains unclear and should be tested by adjusting not only for conventional risk factors but also for several predictive laboratory parameters, as shown in this study. Brain natriuretic peptide and N-terminal pro–B-type natriuretic peptide, which reflect myocardial ischemia, could especially provide additive prognostic information because the information they contain differs from other biomarkers. Fourth, the ALP model offers excellent discriminative ability in the study sample, but the model should be tested in a validation cohort. A planned analysis in a prospective gathered database is currently under investigation.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14. 1. Lee KL, Woodlief LH, Topol EJ, Weaver WD, Betriu A, Col J, Simoons M, Aylward P, van de Werf F, Califf RM. Predictors of 30-day mortality in the era of reperfusion for acute myocardial infarc-
1439
tion. Results from an international trial of 41,021 patients. GUSTO-I Investigators. Circulation 1995;91:1659 –1668. Morrow DA, Antman EM, Charlesworth A, Cairns R, Murphy SA, de Lemos JA, Giugliano RP, McCabe CH, Braunwald E. TIMI risk score for ST-elevation myocardial infarction: a convenient, bedside, clinical score for risk assessment at presentation: an intravenous nPA for treatment of infarcting myocardium early II trial substudy. Circulation 2000;102:2031–2037. Marchioli R, Avanzini F, Barzi F, Chieffo C, Di Castelnuovo A, Franzosi MG, Geraci E, Maggioni AP, Marfisi RM, Mininni N, et al. Assessment of absolute risk of death after myocardial infarction by use of multiple-risk-factor assessment equations: GISSI-Prevenzione mortality risk chart. Eur Heart J 2001;22:2085–2103. Granger CB, Goldberg RJ, Dabbous O, Pieper KS, Eagle KA, Cannon CP, van de Werf F, Avezum A, Goodman SG, Flather MD, Fox KA. Predictors of hospital mortality in the global registry of acute coronary events. Arch Intern Med 2003;163:2345–2353. Eagle KA, Lim MJ, Dabbous OH, Pieper KS, Goldberg RJ, van de Werf F, Goodman SG, Granger CB, Steg PG, Gore JM, et al. A validated prediction model for all forms of acute coronary syndrome: estimating the risk of 6-month postdischarge death in an international registry. JAMA 2004;291:2727–2733. Barron HV, Cannon CP, Murphy SA, Braunwald E, Gibson CM. Association between white blood cell count, epicardial blood flow, myocardial perfusion, and clinical outcomes in the setting of acute myocardial infarction: a Thrombolysis In Myocardial Infarction 10 substudy. Circulation 2000;102:2329 –2334. Grzybowski M, Welch RD, Parsons L, Ndumele CE, Chen E, Zalenski R, Barron HV. The association between white blood cell count and acute myocardial infarction in-hospital mortality: findings from the National Registry of Myocardial Infarction. Acad Emerg Med 2004; 11:1049 –1060. Foo K, Cooper J, Deaner A, Knight C, Suliman A, Ranjadayalan K, Timmis AD. A single serum glucose measurement predicts adverse outcomes across the whole range of acute coronary syndromes. Heart 2003;89:512–516. Pinto DS, Skolnick AH, Kirtane AJ, Murphy SA, Barron HV, Giugliano RP, Cannon CP, Braunwald E, Gibson CM. U-shaped relationship of blood glucose with adverse outcomes among patients with STsegment elevation myocardial infarction. J Am Coll Cardiol 2005;46: 178 –180. Sabatine MS, Morrow DA, Giugliano RP, Burton PB, Murphy SA, McCabe CH, Gibson CM, Braunwald E. Association of hemoglobin levels with clinical outcomes in acute coronary syndromes. Circulation 2005;111:2042–2049. Gibson CM, Pinto DS, Murphy SA, Morrow DA, Hobbach HP, Wiviott SD, Giugliano RP, Cannon CP, Antman EM, Braunwald E. Association of creatinine and creatinine clearance on presentation in acute myocardial infarction with subsequent mortality. J Am Coll Cardiol 2003;42:1535–1543. Yamaguchi J, Kasanuki H, Ishii Y, Yagi M, Ogawa H, Fujii SY, Koganei H, Okada H, Kimura H, Horie T, et al. Prognostic significance of serum creatinine concentration for in-hospital mortality in patients with acute myocardial infarction who underwent successful primary percutaneous coronary intervention (from the Heart Institute of Japan Acute Myocardial Infarction [HIJAMI] Registry). Am J Cardiol 2004;93:1526 –1528. Kirtane AJ, Leder DM, Waikar SS, Chertow GM, Ray KK, Pinto DS, Karmpaliotis D, Burger AJ, Murphy SA, Cannon CP, et al. Serum blood urea nitrogen as an independent marker of subsequent mortality among patients with acute coronary syndromes and normal to mildly reduced glomerular filtration rates. J Am Coll Cardiol 2005;45:1781–1786. Goldberg A, Hammerman H, Petcherski S, Zdorovyak A, Yalonetsky S, Kapeliovich M, Agmon Y, Markiewicz W, Aronson D. Prognostic importance of hyponatremia in acute ST-elevation myocardial infarction. Am J Med 2004;117:242–248.
1440
The American Journal of Cardiology (www.AJConline.org)
15. Furman MI, Gore JM, Anderson FA, Budaj A, Goodman SG, Avezum A, Lopez-Sendon J, Klein W, Mukherjee D, Eagle KA, et al. Elevated leukocyte count and adverse hospital events in patients with acute coronary syndromes: findings from the Global Registry of Acute Coronary Events (GRACE). Am Heart J 2004;147:42– 48. 16. Anavekar NS, McMurray JJ, Velazquez EJ, Solomon SD, Kober L, Rouleau JL, White HD, Nordlander R, Maggioni A, Dickstein K, et al. Relation between renal dysfunction and cardiovascular outcomes after myocardial infarction. N Engl J Med 2004;351:1285– 1295. 17. Timmer JR, Ottervanger JP, de Boer MJ, Dambrink JH, Hoorntje JC, Gosselink AT, Suryapranata H, Zijlstra F, van’t Hof AW. Hyperglycemia is an important predictor of impaired coronary flow before
reperfusion therapy in ST-segment elevation myocardial infarction. J Am Coll Cardiol 2005;45:999 –1002. 18. Libby P, Simon DI. Inflammation and thrombosis: the clot thickens. Circulation 2001;103:1718 –1720. 19. Frossard M, Fuchs I, Leitner JM, Hsieh K, Vlcek M, Losert H, Domanovits H, Schreiber W, Laggner AN, Jilma B. Platelet function predicts myocardial damage in patients with acute myocardial infarction. Circulation 2004;110:1392–1397. 20. Elsman P, Zijlstra F, Miedema K, Hoorntje JC, Dikkeschei LD, Slingerland RJ, Reiffers S, de Boer MJ, Suryapranata H. The predictive value of cumulative lactate dehydrogenase release within the first 72 h of acute myocardial infarction in patients treated with primary angioplasty. Ann Clin Biochem 2004;41:142–148.