Most patients presenting to the emergency department with possible cardiac symptoms have low cardiac troponin (cTn) concentrations. A combination of biomarkers that improves risk stratification in patients at very low risk for major adverse cardiovascular events (MACEs) would be beneficial. In this multicenter prospective cohort study, specimens from 598 subjects presenting to 5 emergency departments with suspected acute coronary syndromes were collected on arrival and serially for traditional and novel biomarkers. Subjects were evaluated for MACEs, defined as death, myocardial infarction, or revascularization at 30 and 365 days. Classification and regression tree analysis assessed biomarker and clinical factors associated with MACEs. The 1-year rate of MACE was 10.5% (47 of 449). Rates of death, myocardial infarction, and revascularization were 4.2%, 1.6%, and 4.7%, respectively. The combination of B-type natriuretic peptide (BNP), placental growth factor (PlGF), and estimated glomerular filtration rate (eGFR) was the most accurate predictor of MACEs compared to any other biomarker or clinical factors including cTnI. If BNP was ≤65 ng/L and PlGF was ≤19.5 ng/L, the negative predictive value for 1-year MACEs was 99.1%. Conversely, BNP >150 ng/L and eGFR ≤68 ml/min/1.73 m 2 predicted a very high (36.5%) MACE rate. Prognostic values of BNP and PlGF were incremental (none increased, 2 of 212, 0.9%; only PlGF increased, 30 of 170, 17.6%; only BNP increased, 33 of 153, 21.6%; BNP and PlGF increased, 18 of 86, 20.9%). Considering only initial emergency department samples, 97% and 96% of patients with normal PlGF, BNP, and cTnI levels were event-free at 30 and 365 days, respectively. In conclusion, the combination of BNP, PlGF, and eGFR is the most accurate in risk-stratifying patients with suspected acute coronary syndrome.
There is extensive historical research on novel biomarkers of myocardial ischemia, early myocardial necrosis, and risk stratification in the absence of necrosis. Although many have seemed promising, all have ultimately failed, largely because of limitations in clinical specificity. In addition, most of these biomarkers are limited by only proof-of-principle studies with less clinically relevant patient populations. Thus, a large, prospective multicenter investigation simultaneously evaluating a multimarker strategy predicated on the absence of myonecrosis is needed. The aim of this study was to determine if a variety of novel candidate biomarkers improve emergency department risk stratification.
Methods
The Acute Risk and Related Outcomes Assessed With Cardiac Biomarkers (ARROW) study was a prospective blinded multicenter cohort investigation of the diagnostic and prognostic clinical utility of multiple biomarkers. Eligible patients with ischemic symptoms suggestive of a new ST-segment elevation acute coronary syndrome were recruited at emergency department presentation from May 2006 to August 2007 at 5 centers in the United States. Institutional review boards at each site approved the study protocol and written informed consent was obtained from all subjects. Blood (plasma and serum) samples were obtained at enrollment, 4 to 8 hours later, and, if still hospitalized, 12 to 16 hours later. Plasma was stored on ice for no more than 2 hours before processing. Serum was stored for no more than 1 hour at room temperature before centrifugation and clot removal and then refrigerated for no more than 1 hour. Samples were places in aliquots and frozen at ≤−60°C until analyzed and did not undergo more than 1 freeze–thaw cycle. All analyses were performed at centralized testing sites.
Data regarding clinical characteristics, cardiac procedures, and cardiac events during hospitalization were collected. Demographic factors, medical history, and medication use were recorded. Telephone follow-up occurred 2 times to assess for primary end points at 30 days and 1 year after emergency department presentation. The major adverse cardiovascular event (MACE) primary end point was a combination of death, nonfatal myocardial infarction, and revascularization within 1 year of presentation. A local blinded committee using the Joint European Society of Cardiology/American College of Cardiology Foundation/American Heart Association/World Heart Federation Task Force definition of myocardial infarction adjudicated in-hospital myocardial infarction.
Patients who were ≥18 years of age and presented to the emergency department with ≥20 minutes of chest pain or ischemic equivalent symptoms within the preceding 12 hours were eligible for inclusion. Those subjects presenting with persistent ST-segment elevation myocardial infarction and those unable to provide informed consent were excluded.
The following biomarkers were tested: cardiac troponin I (cTnI), B-type natriuretic peptide (BNP), homocysteine, C-reactive protein, placental growth factor (PlGF), myeloperoxidase, choline, soluble CD40 ligand, ischemia-modified albumin, and lipoprotein-associated phospholipase A2. Commercially available cTnI, BNP, homocysteine, and C-reactive protein concentrations were determined using the Architect instrument as described by the manufacturer (Abbott Laboratories, Abbott Park, Illinois). Analytic characteristics for each of the 4 commercially available assays are available in the manufacturer’s assay package insert. The published 99th percentile in the manufacturer’s assay package insert of the cTnI assay is 0.028 μg/L.
Prototype assays were used to determine concentrations of PlGF and myeloperoxidase using the Architect instrument and had total imprecisions of 1.8% to 6.7% at 12 and 990 ng/L and 3.3% to 9.8% at 177 and 7,730 pmol/L, respectively. Choline was determined using liquid chromatography/tandem mass spectrometry with an imprecision of 2.6% to 5.4% at 6.1 and 22.5 μmol/L. Soluble CD40 ligand was measured by enzyme-linked immunosorbent assay (R&D Systems Research, Minneapolis, Minnesota) and had an imprecision of 6.0% to 6.2% at 437 and 2,612 ng/L. Ischemia-modified albumin was measured with an albumin cobalt binding method (Inverness Medical Innovations, Waltham, Massachusetts) on a Hitachi analyzer (Roche Diagnostics, Indianapolis, Indiana) and had an imprecision of 4.3% to 7.1% at 70.6 and 125.9 U/ml. Lipoprotein-associated phospholipase A2 was measured by enzyme-linked immunosorbent assay using the PLAC test (DiaDexus, Inc., South San Francisco, California) and had an imprecision of 6.0% to 12.8% at 143 and 830 μg/L. Creatinine was measured locally using the standard clinical laboratory method and estimated glomerular filtration rate (eGFR) was calculated by the Modification of Diet in Renal Disease equation.
Cardiac TnI, C-reactive protein, homocysteine, and lipoprotein-associated phospholipase A2 were analyzed in lithium heparin plasma, and BNP, PlGF, and myeloperoxidase were analyzed in ethylenediaminetetra-acetic acid plasma. Ischemia-modified albumin and soluble CD40 ligand were analyzed in serum. Choline was analyzed in whole blood and lithium heparin plasma.
Descriptive statistical analyses were used to define demographic and clinical characteristics for the entire the study population, comparing patients with and without the primary outcome by 1 year after presentation to the emergency department. Categorical characteristics were compared using exact binomial tests, chi-square test, or Fisher’s exact test and continuous variables were analyzed using Student’s t test.
To determine the prognostic ability for the primary end point at 1 year, classification and regression tree analysis was used to identify the biomarkers and clinical characteristics that were most strongly associated with the primary end point. As potential cutoffs for numerical variables, rounded centiles were used. Maximum biomarker concentration values from serial draws were used for analysis. Kaplan–Meier analyses were performed to determine the association of event-free survival and cTnI, BNP, and PlGF between groups and were compared using log-rank test. All tests of significance were 2-sided, with a p value <0.05 indicating statistical significance. Statistical analyses were conducted using SAS 9.1.3 (SAS Institute, Cary, North Carolina) and JMP 8.0.1 (SAS Institute).
Results
Subjects without follow-up (n = 55) or incomplete biomarker data (n = 54), arrival revascularization or death (n = 10), or with prevalent myocardial infarction (n = 30), defined as myocardial infarction within the first 8 hours of presentation, were excluded from analysis for 1-year events. Of the total study population of 449 patients, 85% presented within 3 hours of onset of symptoms of ischemia. In 30 of 31 patients with prevalent myocardial infarction, myocardial infarction was identified by an initial increased cTnI concentration (>99th percentile of 0.028 μg/L) at emergency department arrival. In the remaining single patient, myocardial infarction was diagnosed within 8 hours. Baseline demographic and clinical characteristics for the study population, stratified by 1-year MACEs, are listed in Table 1 . Expected differences between patients with and without events were seen, including a higher prevalence of coronary artery disease risk factors in the population with the primary event.
| Characteristic | Total (n = 449) | Patients Without MACEs (n = 402) | Patients With MACEs (n = 47) | p Value |
|---|---|---|---|---|
| Age (years), mean ± SD | 54 ± 13.8 | 53 ± 13.4 | 65 ± 12.6 | <0.0001 |
| Age ≥65 years | 107 (23.8%) | 81 (20.1%) | 26 (55.3%) | 0.2066 |
| Women | 209 (46.5%) | 194 (48.3%) | 15 (31.9%) | 0.0335 |
| Diabetes mellitus | 103 (22.9%) | 81 (20.1%) | 22 (46.8%) | <0.0001 |
| Hypertension | 270 (60.5%) | 235 (58.9%) | 35 (74.5%) | 0.0389 |
| Current tobacco user | 146 (32.9%) | 132 (33.2%) | 14 (29.8%) | 0.6328 |
| Lung disease | 33 (7.4%) | 24 (6.0%) | 9 (19.1%) | 0.0011 |
| History of hyperlipidemia | 172 (39.2%) | 146 (37.2%) | 26 (56.5%) | 0.0109 |
| Body mass index (kg/m 2 ), mean ± SD ⁎ | 31.8 ± 15.44 | 32.3 ± 16.30 | 28.9 ± 6.78 | 0.0252 |
| Estimated glomerular filtration rate, mean ± SD | 87.4 ± 32.05 | 89.5 ± 31.75 | 69.7 ± 29.33 | <0.0001 |
| Previous myocardial infarction | 80 (17.8%) | 64 (15.9%) | 16 (34.0%) | 0.0021 |
| Previous congestive heart failure | 63 (14.2%) | 47 (11.8%) | 16 (34.0%) | <0.0001 |
| Thrombolysis In Myocardial Infarction risk score ⁎ | <0.0001 | |||
| 0–2 | 281 (70.4%) | 264 (73.9%) | 17 (40.5%) | |
| 3–4 | 111 (27.8%) | 88 (24.6%) | 23 (54.8%) | |
| 5–7 | 7 (1.8%) | 5 (1.4%) | 2 (4.8%) | |
| Medication use at emergency department presentation | ||||
| Aspirin | 179 (40.0%) | 154 (38.5%) | 25 (53.2%) | 0.0518 |
| Statin | 142 (31.8%) | 116 (29.0%) | 26 (55.3%) | 0.0002 |
| Angiotensin-converting enzyme inhibitor | 125 (28.0%) | 103 (25.8%) | 22 (46.8%) | 0.0024 |
Thirty-day MACE rate was 4.5% (20 of 449). Death, myocardial infarction, and revascularization occurred in 0.2%, 1.1%, and 4.0%, respectively. In addition to cTnI, maximal concentrations of BNP and PlGF were associated with 30-day MACEs ( Table 2 ). Patients with increased PlGF were significantly more likely to have a 30-day adverse event, as did patients with increased BNP (>65 ng/L). In patients with normal concentrations of PlGF, BNP, and cTnI, 99.5% were event-free at 30 days. This is in contrast to 30-day event-free survival in 96% of patients when using cTnI alone based on the 99th percentile cut-off value.
| Variable | MACEs | |
|---|---|---|
| Odds Ratio (95% CI) | p Value | |
| Initial biomarker values | ||
| Troponin I >0.028 μg/L | 3.43 (1.18–9.95) | 0.0234 |
| B-type natriuretic peptide >65 ng/L | 3.18 (1.29–7.88) | 0.0122 |
| Placental growth factor >19.5 ng/L | 1.93 (0.77–4.85) | 0.1604 |
| Maximum biomarker values | ||
| Troponin I >0.028 μg/L | 3.07 (1.06–8.87) | 0.0382 |
| B-type natriuretic peptide >65 ng/L | 3.06 (1.22–7.67) | 0.0167 |
| Placental growth factor >19.5 ng/L | 5.30 (1.89–14.87) | 0.0015 |
At 1 year, after exclusion of prevalent myocardial infarctions, overall MACE rate was 10.5% (47 of 449). Rates of death, myocardial infarction, and revascularization were 4.2%, 1.6%, and 4.7%, respectively, for this same time frame. Several biomarkers had significant predictive values for the primary end point at 1 year ( Table 3 ). The combination of BNP, PlGF, and eGFR was a more accurate 1-year MACE predictor than any other biomarker or clinical factor based on classification and regression tree analysis of maximal biomarker values ( Figure 1 ). If BNP was ≤65 ng/L and PlGF ≤19.5 ng/L, then the negative predictive value of a primary event within 1 year was 99.1%. Conversely, in those patients with BNP values >150 ng/L, eGFR was an accurate predictor of a poor outcome. An eGFR value ≤68 ml/min/1.73 m 2 predicted a 36.5% likelihood of the primary event. Furthermore, if BNP was ≤150 ng/L and PlGF was increased >19.5 ng/L or if BNP was ≤150 ng/L and eGFR was >68 ml/min/1.73 m 2 , there was a 14.5% and a 14.3% likelihood of the primary event, respectively.
| Item | Cut-Off Value | Incidence of Event (number) If Higher Than Cutoff | Incidence of Event (number) If Lower Than or Equal to Cutoff | Relative Risk (95% CI) | p Value |
|---|---|---|---|---|---|
| B-type natriuretic peptide | 150 ng/L | 27.6% (87) | 6.4% (362) | 4.3 (2.6–7.3) | <0.00001 |
| Age | 65 years | 26.0% (96) | 6.2% (353) | 4.2 (2.5–7.1) | <0.00001 |
| Troponin I | 0.015 μg/L | 26.4% (87) | 6.6% (362) | 4.0 (2.4–6.7) | <0.00001 |
| Homocysteine | 12.5 μmol/L | 22.8% (114) | 6.3% (335) | 3.6 (2.1–6.2) | <0.00001 |
| Estimated glomerular filtration rate ⁎ | 68 ml/min/1.73 m 2 | 6.3% (319) | 20.8% (130) | 3.3 (1.9–5.7) | <0.0001 |
| Placental growth factor | 20 ng/L | 18.4% (158) | 6.2% (291) | 3.0 (1.7–5.2) | <0.001 |
| Choline | 15.3 μmol/L | 20.5% (112) | 7.1% (337) | 2.9 (1.7–4.9) | <0.001 |
| C-reactive protein | 13 mg/L | 20.0% (90) | 8.1% (359) | 2.5 (1.4–4.3) | <0.01 |
| Gender | male | 13.3% (240) | 7.2% (209) | 1.9 (1.0–3.3) | <0.01 |
| Ischemia-modified albumin | 144 U/ml | 27.3% (22) | 9.6% (427) | 2.8 (1.4–6.0) | <0.05 |
| Lipoprotein-associated phospholipase A2 | 240 μg/L | 13.9% (230) | 6.8% (219) | 2.0 (1.1–3.6) | <0.05 |
| Myeloperoxidase | 200 pmol/L | 12.0% (291) | 7.6% (158) | 1.6 (0.8–3.0) | 0.15 |
| Soluble CD40 ligand | 760 ng/L | 7.0% (114) | 11.6% (335) | 0.6 (0.3–1.3) | 0.21 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
