Stress-Delta B-Type Natriuretic Peptide Does Not Exclude ACS in the ED

Abstract

Background:

There are many detectable changes in circulating biomarkers in the setting of myocardial ischemia. We hypothesize that there are associated changes in circulating B-type natriuretic peptide (BNP) level after stress-induced myocardial ischemia, which can be used for Emergency Department (ED) acute coronary syndrome (ACS) risk stratification.

Methods:

In a prospective study, we enrolled 340 patients over the age of 30 receiving an exercise echocardiography stress test in an ED observational unit for suspected ACS. We collected blood samples at baseline and at2 and 4 hours post–stress test, measuring the relative and absolute changes (stress-delta) in plasma BNP concentrations. In addition, patients were contacted at 90 days and at 1 year post-test for a follow-up. We calculated the diagnostic test characteristics of stress-delta BNP for a composite outcome of ischemic imaging on stress echocardiogram, non-elective PCI, CABG surgery, subsequent AMI, or cardiac death at 1 year via a logistic regression. We analyzed the 2-hour BNP concentrations using an ANOVA model to adjust for the baseline BNP level.

Results:

Baseline and 2-hour post-stress BNP were both higher in the positive outcome group, but the stress-delta BNP was not. Stress-delta BNP had a sensitivity and specificity respectively of 53% and 76% at 2 hours and 67% and 68% at 4 hours. It was noted that patients with the composite outcome had a higher baseline BNP level.

Conclusions:

BNP stress-deltas are poor diagnostic means for ACS risk stratification, but resting BNP remains a promising prognostic tool for ED patients with suspected ACS.

INTRODUCTION

According to the American Heart Association(1), an estimated 18.2 million Americans over the age of 20 have coronary heart disease, with an overall prevalence of myocardial infarction (MI) at 8.4 million. Because of the high morbidity and mortality associated with acute coronary syndrome (ACS), it is imperative for it to be diagnosed promptly during the medical evaluation. Due to the difficulty in determining long-term risk in suspected cases, patients with suspected ACS are often admitted for cardiac stress testing, wherein the patient undergoes either an exercise or chemically induced increase in oxygen demand, and the presence of resulting cardiac ischemia is then identified via some diagnostic imaging technique. This unfortunately requires expensive equipment with highly trained personnel to operate and interpret and is not available at all hours of the day, which decreases its availability.

Because of these limitations, some have proposed the use of a biomarker-based stress test in place of the current diagnostic method. In this form of test, serial blood samples would be drawn before and after the stress test to determine whether changes between the two timepoints (stress-delta) could identify inducible myocardial ischemia. Troponin T(2) showed no significant change post–stress test using a high-sensitivity assay, whereas pilot work demonstrated that some amino acids and acyl carnitines(3) may change based on stress-induced ischemia. We examined N-terminal B-type natriuretic peptide (NT-proBNP)(4) and found significantly different stress-delta values at 2 and 4 hours post–stress test between patients with ischemia on imaging and those without. Previously, we explored this stress testing model in a meta-analysis(5), failing to establish a relationship between either stress delta B-type natriuretic peptide (BNP) or NT-proBNP and stress-induced ischemia due to the heterogeneous and small sample sizes in the studies analyzed. Thus, it remains to be seen whether BNP has predictive value for MI in emergency department (ED) patients. It is further unclear whether a stress-delta biomarker can risk stratify patients with suspected ACS.

In this study, we sought to determine the sensitivity and specificity of 2-hour stress-delta BNP for the prediction of a composite outcome of abnormal imaging on index visit stress test, non-elective percutaneous coronary intervention (PCI), coronary artery bypass graft (CABG) surgery, subsequent acute myocardial infarction (AMI), or cardiac-related death within 1 year. The composite outcome includes a positive stress echo because in an ED at the point of decision, an abnormal echo indicates that the patient is not safe to discharge without a cardiology consult, regardless of whether CAD is later confirmed. Additionally, we sought to conduct a subgroup analysis on patients with a high-sensitivity troponin I (hsTnI) level consistent with a low–moderate risk for ACS. We hypothesized that stress-delta BNP can predict the occurrence of our composite outcome, and secondarily, stress-delta BNP patients can safely rule out composite outcomes in patients with low but detectable troponin levels.

Methods

We conducted a prospective observational study of a diagnostic test. We have previously reported(2) on the methods for the collection and analysis of these blood samples, but briefly overview them here for clarity. From November 2012 to September 2014, 413 patients were recruited in the ED observation unit of an academic hospital with 80,000 visits per year. We included ED patients older than 30 years undergoing a stress test as part of their standard of care who could provide informed consent. The study protocol was approved by the Institutional Review Board of Duke University Medical Center. Abbott Laboratories provided funding for the collection and analysis of these samples as well as support for salary of research staff. The authors retained control of the decision of whether and how to report the results. Patients were excluded from the observation unit, and therefore our study, if they had any ventricular arrhythmias, any narrow spectrum complex with a rapid ventricular rate, unstable vital signs—including persistent (defined as >2 readings over 2 hours) hypotension (systolic BP <80 mmHg) or tachycardia (pulse >110 bpm)—aortic aneurysm or dissection, severe aortic stenosis, active myocarditis, or pericarditis, acute or decompensated heart failure or pulmonary edema, or pulmonary embolism. Patients were also excluded from the study if they were scheduled for stress test modality other than exercise echocardiogram.

Patients received a standard of care exercise stress test echocardiogram if the patient’s serial electrocardiogram and cardiac troponin testing—tested via the Roche Elecsys 4^th gen cardiac troponin T test—was below the institutional cutoff (<100 ng/L, 10% CV concentration = <30 ng/L, which is the manufacturer’s recommended cutoff for clinical care purposes). Patients underwent symptom-limited exercise testing, and all beta-blocking medication was discontinued at least 24 hours before the test. Per the Bruce Protocol, the test was stopped for angina, >/= 3 mm ST-segment depression, >/= 20 mmHg drop in systolic BP, ventricular arrhythmias, any narrow spectrum complex arrhythmia with a rapid ventricular rate, fatigue, or severe dyspnea. Before the test, we collected additional patient information concerning demographics, including variables in the thrombosis in myocardial infarction (TIMI) risk score, electrocardiogram characteristics, height and weight, medical comorbidities, smoking history, illicit drug use history, medications, serum creatinine, and results of any prior cardiac stress testing or angiography performed.

For research purposes we drew a peripheral blood sample <1 hour before the stress test (baseline) and 2 hours afterward (2-h post-stress value). When possible, we also collected a 4-hour post-stress blood sample for exploratory analysis (4-h post-stress value). When possible, samples were drawn via an existing intravenous catheter. Blood was then centrifuged within 1 hour for collection of plasma and aliquoted into 0.5 mL aliquots. Samples were banked and frozen at −80°C and tested on the Abbott Architect I2000SR platform (Abbott Laboratories, Abbott Park, IL) in two batches at Baylor University and University College Dublin Clinical Research Centre Scientific Services & Laboratories for both BNP and hsTnI. The limit of detection of the BNP assay was 10 ng/L and was 3.4 ng/L for hsTnI. For assay values below this level, we used 9.9 ng/L for stress-delta calculation purposes per laboratory recommendations. At a concentration of 87.8 ng/L, the Coefficient of Variation was 4.7% for BNP and the coefficient of variation was 10% at a concentration of 5.2 for hsTnI(6,7). We excluded those patients who had samples that were missing, hemolyzed, or had an incalculable level of BNP or hsTnI due to assaying technical difficulties. In both laboratories, the researchers were blinded to all clinical data, including the stress test outcomes.

Inducible ischemia was defined as the occurrence of new segmental wall motion abnormalities during stress. Stress echocardiograms were initially interpreted by board-certified cardiologists who were blinded to the results of the BNP blood tests. Two of the investigators also reviewed >75% of the stress test reports to confirm abnormal test results and differentiate indeterminant tests into either “normal” or “ischemic” categories. Any disagreements were settled by a third reviewer, a board-certified cardiologist using the stress test report and clinical records. All echocardiogram report reviewers were blinded to the results of the BNP blood samples at the time of adjudication.

We contacted all patients at 90 days and 1 year to follow up, as well as checking their medical records, to discover any adverse events, testing, or hospitalizations. Other events reported include any subsequent MI, abnormal stress testing, significant coronary disease by angiography (lesions >50% in a major epicardial coronary artery), PCI, CABG, or death (cardiovascular or otherwise). This data was used to determine the composite outcome analysis.

Data Analysis

We defined a composite outcome of an abnormal imaging on index visit stress test or unexpected (not ordered during the index visit workup) non-elective PCI, CABG surgery, subsequent AMI, or cardiac-related death within 1 year. Abnormal imaging on a stress test may represent a surrogate outcome; however, since a stress-delta BNP test could be used as a screening test for conventional stress testing with imaging, we included it in the composite outcome.. We calculated our sample size so that, at a 95% CI of roughly +/− 15%, the sensitivity of the BNP stress-delta could be determined, given an 85% true-sensitivity for a myocardial infarction or death. We found that if at least 25 outcomes of interest were within our sample size, we would be able to detect a significant difference at alpha <0.05 at 85% power.

In our first aim, we determined the sensitivity and specificity of stress-delta BNP for the prediction of our composite outcome. For each patient, we determined the absolute and relative changes in BNP; we then compared the distributions of each in patients with or without composite outcome via the Wilcoxon Rank-Sum Test. A logistic regression model using these patients was used to assess the association between change in 2-hour post-stress BNP from baseline and the odds of the composite outcome, with and without adjusting for baseline measurements. We then composed an AUROC curve to assess the predictive performance of the test, as well as the sensitivity, specificity, positive predictive value, and negative predictive value. We considered a 2-sided P-value of <0.05 as statistically significant. We repeated this analysis with the 4-hour data.

In our second aim, we repeated the 2-hour and 4-hour analysis with the subset of patients with an hsTnI score at the “low-moderate” risk range per the package insert(8), which is <10 ng/L for women and <12 ng/L for men.

The data capture tool used for data collection and management was REDCap (Research Electronic Data Capture), hosted at Duke University, Durham, NC. For analysis, we used SAS statistical software, version 9.4, produced by the SAS Institute (Cary, NC), and R version 3.6.3 (R Core Team, 2020).

Results

We enrolled 508 patients. Of these patients, 42 did not undergo a stress test and 61 were missing baseline BNP values, resulting in 414 total patients who underwent a stress test and had baseline BNP values. From this group, 145 were missing 2-hour BNP values and 319 were missing 4-hour BNP values, resulting in 345 patients with at least one of 2- or 4-hour BNP values. An additional 5 in this group had indeterminate stress test results, which excluded them from the study, resulting in 340 patients.

The demographics of our study are shown below on Table 1. Because the patients in the study were receiving stress tests to risk-stratify for discharge, the prevalence of coronary artery disease (CAD) and CAD risk factors was relatively low. From the 340 patients who underwent stress testing for suspected ACS and who had both baseline BNP and either 2-hour or 4-hour BNP values, 31 had a composite outcome. From this group, only 3 (9.7%) showed no evidence of ischemia on the stress-test imaging interpretation. In addition, no patients in the study experienced either an AMI or cardiac-related death within 1 year of the initial ED visit. Patients in the composite outcomes group were slightly older and more likely to have a history of hypertension, hyperlipidemia, MI, Congestive Heart Failure (CHF), or PCI.

Table 1:

Demographics and Clinical Characteristics

Characteristics	No Composite Outcome at 1 Year (N=309)	Any Composite Outcome at 1 Year (N=31)	Total (N=340)
Age
N	308	31	339
Mean (SD)	52.9 (10.9)	57.1 (9.9)	53.3 (10.9)
Median	52	55	52
Q1, Q3	45.0, 60.0	49.0, 65.0	46.0, 61.0
Range	(32.0-89.0)	(42.0-79.0)	(32.0-89.0)
Gender
Missing	1 (0.3%)	0 (0.0%)	1 (0.3%)
Male	140 (45.3%)	15 (48.4%)	155 (45.6%)
Female	168 (54.4%)	16 (51.6%)	184 (54.1%)
Race
Missing	1 (0.3%)	0 (0.0%)	1 (0.3%)
Asian	3 (1.0%)	0 (0.0%)	3 (0.9%)
Black or African American	106 (34.3%)	8 (25.8%)	114 (33.5%)
White / Caucasian	192 (62.1%)	23 (74.2%)	215 (63.2%)
Other	6 (1.9%)	0 (0.0%)	6 (1.8%)
Unknown/Not Reported/Declined	1 (0.3%)	0 (0.0%)	1 (0.3%)
Ethnicity
Missing	3 (1.0%)	0 (0.0%)	3 (0.9%)
Hispanic or Latino	12 (3.9%)	1 (3.2%)	13 (3.8%)
Not Hispanic or Latino	294 (95.1%)	30 (96.8%)	324 (95.3%)
Hypertension
Missing	1 (0.3%)	0 (0.0%)	1 (0.3%)
Yes	136 (44.0%)	21 (67.7%)	157 (46.2%)
No	171 (55.3%)	10 (32.3%)	181 (53.2%)
Unknown	1 (0.3%)	0 (0.0%)	1 (0.3%)
Diabetes
Missing	2 (0.6%)	0 (0.0%)	2 (0.6%)
Yes	52 (16.8%)	8 (25.8%)	60 (17.6%)
No	253 (81.9%)	23 (74.2%)	276 (81.2%)
Unknown	2 (0.6%)	0 (0.0%)	2 (0.6%)
Smoking
Missing	8 (2.6%)	1 (3.2%)	9 (2.6%)
No	177 (57.3%)	16 (51.6%)	193 (56.8%)
Yes	124 (40.1%)	14 (45.2%)	138 (40.6%)
Hyperlipidemia
Missing	1 (0.3%)	0 (0.0%)	1 (0.3%)
Yes	101 (32.7%)	21 (67.7%)	122 (35.9%)
No	205 (66.3%)	10 (32.3%)	215 (63.2%)
Unknown	2 (0.6%)	0 (0.0%)	2 (0.6%)
Cocaine
Missing	6 (1.9%)	1 (3.2%)	7 (2.1%)
Current user	2 (0.6%)	0 (0.0%)	2 (0.6%)
Past (none for past month)	20 (6.5%)	2 (6.5%)	22 (6.5%)
Never	281 (90.9%)	28 (90.3%)	309 (90.9%)
Renal disease/insufficiency
Missing	2 (0.6%)	0 (0.0%)	2 (0.6%)
Yes	4 (1.3%)	2 (6.5%)	6 (1.8%)
No	302 (97.7%)	29 (93.5%)	331 (97.4%)
Unknown	1 (0.3%)	0 (0.0%)	1 (0.3%)
Myocardial infarction
Missing	1 (0.3%)	0 (0.0%)	1 (0.3%)
Yes	6 (1.9%)	6 (19.4%)	12 (3.5%)
No	301 (97.4%)	25 (80.6%)	326 (95.9%)
Unknown	1 (0.3%)	0 (0.0%)	1 (0.3%)
Coronary artery disease
Missing	1 (0.3%)	0 (0.0%)	1 (0.3%)
Yes	14 (4.5%)	11 (35.5%)	25 (7.4%)
No	290 (93.9%)	20 (64.5%)	310 (91.2%)
Unknown	4 (1.3%)	0 (0.0%)	4 (1.2%)
Coronary intervention
Missing	3 (1.0%)	0 (0.0%)	3 (0.9%)
Yes	14 (4.5%)	6 (19.4%)	20 (5.9%)
No	291 (94.2%)	25 (80.6%)	316 (92.9%)
Unknown	1 (0.3%)	0 (0.0%)	1 (0.3%)
Congestive heart failure
Missing	1 (0.3%)	0 (0.0%)	1 (0.3%)
Yes	6 (1.9%)	1 (3.2%)	7 (2.1%)
No	300 (97.1%)	30 (96.8%)	330 (97.1%)
Unknown	2 (0.6%)	0 (0.0%)	2 (0.6%)
Chest pain
Missing	2 (0.6%)	1 (3.2%)	3 (0.9%)
Yes	285 (92.2%)	28 (90.3%)	313 (92.1%)
No	21 (6.8%)	2 (6.5%)	23 (6.8%)
Unknown	1 (0.3%)	0 (0.0%)	1 (0.3%)
Peak pain score
N	274	28	302
Mean (SD)	6.6 (2.8)	7.0 (1.8)	6.6 (2.7)
Median	7	7	7
Q1, Q3	5.0, 9.0	6.0, 8.0	5.0, 9.0
Range	(0.0-10.0)	(2.0-10.0)	(0.0-10.0)
Worse chest pain when walking, climbing stairs?
Missing	24 (7.8%)	3 (9.7%)	27 (7.9%)
Yes	68 (22.0%)	4 (12.9%)	72 (21.2%)
No	132 (42.7%)	16 (51.6%)	148 (43.5%)
Unknown	85 (27.5%)	8 (25.8%)	93 (27.4%)
Worse chest pain when taking a deep breath or coughing?
Missing	24 (7.8%)	3 (9.7%)	27 (7.9%)
Yes	84 (27.2%)	8 (25.8%)	92 (27.1%)
No	156 (50.5%)	13 (41.9%)	169 (49.7%)
Unknown	45 (14.6%)	7 (22.6%)	52 (15.3%)
Worse chest pain when pressed on chest wall?
Missing	24 (7.8%)	3 (9.7%)	27 (7.9%)
Yes	52 (16.8%)	5 (16.1%)	57 (16.8%)
No	172 (55.7%)	16 (51.6%)	188 (55.3%)
Unknown	61 (19.7%)	7 (22.6%)	68 (20.0%)
Nausea or vomiting
Missing	24 (7.8%)	3 (9.7%)	27 (7.9%)
Yes	104 (33.7%)	9 (29.0%)	113 (33.2%)
No	180 (58.3%)	19 (61.3%)	199 (58.5%)
Unknown	1 (0.3%)	0 (0.0%)	1 (0.3%)
Dyspnea
Missing	24 (7.8%)	3 (9.7%)	27 (7.9%)
Yes	142 (46.0%)	15 (48.4%)	157 (46.2%)
No	142 (46.0%)	13 (41.9%)	155 (45.6%)
Unknown	1 (0.3%)	0 (0.0%)	1 (0.3%)
Diaphoresis
Missing	25 (8.1%)	3 (9.7%)	28 (8.2%)
Yes	83 (26.9%)	13 (41.9%)	96 (28.2%)
No	200 (64.7%)	15 (48.4%)	215 (63.2%)
Unknown	1 (0.3%)	0 (0.0%)	1 (0.3%)
Syncope
Missing	24 (7.8%)	3 (9.7%)	27 (7.9%)
Yes	83 (26.9%)	5 (16.1%)	88 (25.9%)
No	200 (64.7%)	23 (74.2%)	223 (65.6%)
Unknown	2 (0.6%)	0 (0.0%)	2 (0.6%)
Non-elective PCI within 1 year
No	309 (100.0%)	21 (67.7%)	330 (97.1%)
Yes	0 (0.0%)	10 (32.3%)	10 (2.9%)
CABG surgery within 1 year
No	309 (100.0%)	29 (93.5%)	338 (99.4%)
Yes	0 (0.0%)	2 (6.5%)	2 (0.6%)
AMI within 1 year
No	309 (100.0%)	31 (100.0%)	340 (100.0%)
Cardiac-related death within 1 year
No	309 (100.0%)	31 (100.0%)	340 (100.0%)
Imaging interpretation
Stress-induced ischemia	0 (0.0%)	28 (90.3%)	28 (8.2%)
No evidence of ischemia	309 (100.0%)	3 (9.7%)	312 (91.8%)

In our first aim, we analyzed the efficacy of stress-delta BNP for predicting composite outcomes in patients. Table 2 shows the statistics for hsTnI and BNP measurements separately for patients with and without the occurrence of the composite outcomes. We calculated p-values for differences in the distribution of the two groups using the Wilcoxon rank-sum test for the continuous measures and a chi-square test for the categorical measures.

Table 2:

BNP and Troponin Measurements

	No Composite Outcomes at 1 Year (N=309)	Any Composite Outcomes at 1 Year [AG1] (N=31)	Total (N=340)	p-value
Baseline high-sensitivity troponin (ng/L)				0.0381
N	293	29	322
Mean (SD)	89.0 (196.7)	161.7 (326.4)	95.5 (212.0)
Median	5.2	14.8	5.6
Q1, Q3	2.4, 81.8	4.4, 117.9	2.5, 90.8
Range	(0.1-1403.1)	(0.7-1378.0)	(0.1-1403.1)
Baseline BNP (ng/L)				0.0068
N	309	31	340
Mean (SD)	26.9 (49.7)	41.3 (36.9)	28.2 (48.8)
Median	12.2	26.6	13.2
Q1, Q3	9.9, 26.6	9.9, 58.5	9.9, 30.0
Range	(9.9-682.5)	(9.9-134.3)	(9.9-682.5)
Baseline BNP <10 ng/L[AG2] [MKP3]				0.1079
No	173 (56.0%)	22 (71.0%)	195 (57.4%)
Yes	136 (44.0%)	9 (29.0%)	145 (42.6%)
2-hour post-stress BNP (ng/L)				0.0016
N	308	30	338
Mean (SD)	26.5 (44.7)	41.8 (40.3)	27.8 (44.5)
Median	12.6	28.6	13.2
Q1, Q3	9.9, 29.3	12.1, 48.8	9.9, 31.1
Range	(9.9-661.7)	(9.9-151.1)	(9.9-661.7)
4-hour post-stress BNP (ng/L)				0.0146
N	156	18	174
Mean (SD)	28.0 (58.3)	36.8 (31.7)	28.9 (56.1)
Median	11.9	26.6	12.9
Q1, Q3	9.9, 26.9	12.1, 47.3	9.9, 29.4
Range	(9.9-686.5)	(9.9-126.0)	(9.9-686.5)
2-hour delta BNP				0.7501
N	308	30	338
Mean (SD)	−0.5 (26.5)	0.7 (18.9)	−0.4 (25.9)
Median	0	0	0
Q1, Q3	−1.0, 1.6	−2.6, 5.5	−1.0, 1.6
Range	(−404.3-114.3)	(−50.6-64.6)	(−404.3-114.3)
2-hour percentage delta BNP				0.5105
N	308	30	338
Mean (SD)	7.0 (48.1)	8.8 (34.3)	7.1 (47.0)
Median	0	0	0
Q1, Q3	−3.7, 5.8	−4.1, 20.2	−3.9, 8.2
Range	(−95.6-571.5)	(−77.5-91.3)	(−95.6-571.5)
4-hour delta BNP				0.0915
N	156	18	174
Mean (SD)	0.8 (13.1)	3.8 (14.1)	1.1 (13.2)
Median	0	1.6	0
Q1, Q3	−0.7, 2.1	0.0, 6.9	−0.6, 2.6
Range	(−80.2-103.5)	(−18.0-39.5)	(−80.2-103.5)
4-hour percentage delta BNP				0.0728
N	156	18	174
Mean (SD)	14.9 (90.4)	17.6 (32.1)	15.1 (86.2)
Median	0	11.7	0
Q1, Q3	−3.8, 9.4	0.0, 45.7	−2.8, 14.1
Range	(−58.0-1045.5)	(−29.2-90.6)	(−58.0-1045.5)

Patients with no composite outcomes had significantly lower BNP levels at baseline and both 2 hours and 4 hours post-stress. These patients also had lower baseline levels of hsTnI. There was no significant difference in BNP stress-deltas between the patients with composite outcomes and the patients without composite outcomes. It should be noted that a large number of patients had blood values below the threshold of detection in the instruments, with 2 ng/L for hsTnI and 10 ng/L for BNP.

Of the 340 patients included in this study, 338 had 2-hour post-stress BNP measurements, 30 of whom experienced a composite outcome. The odds ratios with 95% confidence intervals and p-values, as well as the area under the receiver-operating characteristic curve for each model is shown in Table 3. The odds ratios are for the 1-unit increase in baseline and 2-hour delta BNP. For a 10-unit increase in 2-hour delta BNP, the unadjusted odds ratio is 1.02 (95% CI: 0.85, 1.23) and the adjusted odds ratio is 1.06 (95% CI: 0.90, 1.25). The AUC for the unadjusted model is 0.52 while the adjusted model has an AUC of 0.66, indicating poor to moderate predictive performance(9). The optimal cut-point for the adjusted model is shown in Table 4. This was found using Youden’s J-statistic, which aims to optimize the sensitivity and specificity jointly. This method leads to a sensitivity of 0.53 and a specificity of 0.76. Together, this leads to a positive likelihood ratio of 2.19 and a negative likelihood ratio of 0.62. The results are shown with the 4-hour test characteristics in Table 4.

Table 3:

Estimated Odds Ratio for 2-Hour Delta BNP, with and without Adjusting for Baseline BNP

Variable	Estimated OR (95% Confidence Interval)	P-Value	Estimated OR (95% Confidence Interval)	P-Value
Baseline BNP	-	-	1.00 (1.00, 1.01)	0.137
2-hour delta BNP	1.00 (0.98, 1.02)	0.809	1.01 (0.99, 1.02)	0.524
	AUC = 0.52		AUC = 0.66

Table 4:

Diagnostic Test Characteristics of 2- and 4-Hour Stress-Delta BNP for Composite Cardiac Outcomes

Variable	Sensitivity (%)	Specificity (%)	LR+	LR−
2-hour stress-delta BNP (cut-off p = 0.088)	53	76	2.19	0.62
4-hour stress-delta BNP (cut-off p = 0.099)	67	68	2.08	0.49

We analyzed the 4-hour post-stress change from baseline in BNP in the same way. Of the 340 patients in this study, 174 had 4-hour delta-BNP measurements, of whom 18 experienced a composite outcome. The results from the adjusted and unadjusted models are shown below in Table 5. Once again, the odds ratios reported are for a 1-unit increase in 4-hour delta BNP. For a 10-unit increase in 4-hour delta BNP, the unadjusted odds ratio is 1.16 (95% CI: 0.85, 1.58) with the adjusted odds ratio being the same at 1.16 (95% CI: 0.85, 1.58). We found similar results between the adjusted and unadjusted models for the 4-hour delta BNP, with the AUC being nearly equal at 0.62 and 0.63 for the unadjusted and adjusted models, respectively. Using Youden’s J-statistic model, we again found the optimal cut-point to estimate the sensitivity and specificity, which came to 0.67 and 0.68, respectively. This results in a positive likelihood ratio of 2.08 and a negative likelihood ratio of 0.49. Table 4 shows these results with the test characteristics of the 2-hour stress-delta data. Shown in Table 4 is the predictive performance of the 2-hour and 4-hour adjusted stress-delta models.

Table 5:

Estimated Odds Ratio for 4-Hour Delta BNP, with and without Adjusting for Baseline BNP

Variable	Estimated OR (95% Confidence Interval)	P-Value	Estimated OR (95% Confidence Interval)	P-Value
Baseline BNP	-	-	1.00 (1.00, 1.01)	0.635
4-hour delta BNP	1.02 (0.98, 1.05)	0.363	1.02 (0.98, 1.05)	0.35
	AUC = 0.62		AUC = 0.63

For our second aim, we replicated the analyses of Aim 1 in the subset of patients with low baseline troponin levels, defined as <10 ng/L for women and <12 ng/L for men. Of the 340 patients in the analysis, 194 (57%) had low troponin levels, with 13 experiencing a composite outcome. In this group of 194 with low troponin, 193 had 2-hour post-stress BNP measurements, with 13 experiencing a composite outcome. Due to the low count of events in this subset, the models did not adjust for the baseline BNP. As a result, the models shown in Table 6 cannot be directly compared to the analysis for Aim 1.

Table 6:

Odds Ratios and Diagnostic Characteristics for Low Troponin Patients, without Adjusting for Baseline BNP

Variable	Estimated OR (95% Confidence Interval)	P-Value	AUC	Sensitivity (%)	Specificity (%)	LR+	LR−
2-Hour Delta BNP	1.00 (0.95, 1.05)	0.978	0.56	46	71	1.6	0.75
4-Hour Delta BNP	0.99 (0.94, 1.03)	0.545	0.56	38	83	2.2	0.75

Again, the odds ratios are for a 1-unit increase in the 2-hour and 4-hour stress-delta BNP. For a 10-unit increase, the odds ratios are 0.99 (95% CI: 0.61, 1.63) and 0.86 (95% CI: 0.53, 1.40) for the 2-hour and 4-hour stress-delta BNP tests, respectively. It can be seen that these models have a very poor predictive performance

We performed a sensitivity analysis using only patients with confirmed CAD in the outcome, leaving 14 who experienced any composite outcome within 1 year. In this subgroup analysis, a 10-unit increase in 2-hour delta BNP was associated with an unadjusted odds-ratio of 1.02 (95% CI: 0.78, 1.33) and the adjusted odds-ratio of 1.03 (95% CI: 0.8, 1.34). Using Youden’s J-statistic to choose the cut-point, as seen on Supplementary Table 1, we obtained a sensitivity for the 2-hour stress-delta test of 0.77 and a specificity of 0.50. This resulted in a positive likelihood ratio of 1.52 and a negative likelihood ratio of 0.47. For the 4-hour stress delta BNP test, the sensitivity and specificity was .64 and .68, with a LR+ and LR− of 1.99 and 0.53 respectively.

Discussion

Due to the high morbidity and mortality of ACS, it is very important to detect suspected cases as early as possible, but the current stress test used for patient risk stratification in ED patients is expensive and not available at all times during the day. Because of these limitations, there is a need for a reliable biomarker-based stress test which could be used instead. In previous papers, our group has examined various stress test biomarkers including high-sensitivity Troponin T, amino acids and acyl carnitines, and NT-proBNP, some of which showed promising results. In this study, we investigated the possible use of BNP as a possible stress test biomarker. We found that stress-delta BNP has low sensitivity for a composite cardiovascular outcome.

This was the largest prospective study of the utility of stress delta BNP for risk-stratification for cardiac events of acutely symptomatic patients in the ED with suspected ACS. Other studies have examined stress-delta BNP in outpatients. In addition, this was the only paper that did a sub study examining stress delta BNP in patients with low troponin levels. The result of our analysis is that BNP stress-deltas hold low predictive value for adverse events in ED patients with suspected ACS, even in populations with low baseline troponin I. Our analysis did confirm, however, that patients without our composite outcome had significantly lower BNP levels at baseline as well as significantly lower baseline troponin levels.

The results of our first Aim corresponds well with the findings of a recent systematic review(5), which examined a total of 15 heterogeneous studies, six of which investigated stress-delta BNP for prediction of inducible ischemia by stress testing. While some of these studies reported higher stress-delta BNP measurements in the ischemic group, these findings were inconsistent. Thus, the systematic review concluded that BNP and NT-proBNP stress-deltas were insufficient for ACS risk stratification, noting that patients without inducible ischemia appeared to have a lower baseline BNP and NT-proBNP compared with patients with inducible ischemia by stress testing. This contrasts with several previous studies, such as Foote et al(10) which found a four-fold increase in BNP stress-deltas in patients with diagnosed CAD. This was measured by an exercise stress test and SPECT myocardial imaging, which has a higher sensitivity for detecting ischemia than stress echocardiography, the imaging modality in our study.

This utility of resting baseline BNP levels for risk stratification for patients with suspected ACS is similar to results in several previous papers which found baseline BNP and NT-proBNP values to be a significant predictor for major adverse cardiovascular events in patients with confirmed stable angina(11) as well as after Non-ST elevation MI ACS(12). BNP was also found to predict mortality in patients with acute chest pain, even more accurately than cardiac troponin T(13), though notably without helping in early diagnosis of acute MI. Additionally, a previous meta-analysis(14) found a relatively high diagnostic potential for BNP and NT-proBNP resting baseline levels for predicting inducible myocardial ischemia on reference stress testing. Despite evidence for baseline BNP and NT-proBNP having predictive value for ACS outcomes, and even being recommended in the AHA/ACC Non-ST elevation MI a Class IIb biomarker, very few EDs routinely use resting BNP measurements to risk stratify patients.

A few important limitations in our research must be addressed. To start, the study was only performed in a single hospital system with patients selected for relatively low risk for ACS. As an observational study, there is risk for ascertainment bias, which we sought to avoid by using a reference standard as an inclusion criterion for admittance.

We conducted a sensitivity analysis using only patients with confirmed CAD but this was underpowered due to small numbers of such patients in our cohort. Although future studies could use additional diagnostic imaging techniques such as a cardiac MRI or angiography to confirm CAD, we chose to use the stress echocardiogram since this was the most commonly ordered stress test in our clinical practice at the time. While this results in a significant risk of false positives, a positive stress echo in the ED would represent patients that we would not feel safe discharging without a cardiology consult, which is the same function as our biomarker-based stress test. This results in us not being able to comment on adverse outcomes due to the limited number of confirmed CAD patients in our study. Our paper does, however, add to a growing body of evidence showing that a stress-BNP is not sufficient to exclude ACS in the ED.

It should also be noted that the patients in our study were physically healthy enough to undergo an exercise stress test, which may not be representative of the majority of patients presenting with acute chest pain. Due to a small sample size with a relatively low event rate, we could not control for other risk factors besides BNP to avoid overfitting our data.

In conclusion, our data shows little use of BNP stress-deltas for risk stratification for patients with suspected ACS. The pattern of lower resting BNP in patients without our composite outcomes does show promise for further use in the ED to rule out ACS in patients with acute chest pain.

IMPACT STATEMENT.

This study was made in response to a need for a biomarker-based stress test that could be used for ACS risk stratification of patients in the emergency department, which could be cheaper and available at all hours. Our paper is unique in its investigation of the use of changes in BNP levels due to a cardiac stress test for practical ACS risk assessment—specifically including positive post-stress echocardiograms in our composite outcome group, which reflects patients an ED would not feel comfortable discharging. Our negative findings are important to disseminate to guide future investigations into alternative biomarkers for this application.

Abbreviations:

ACS

acute coronary syndrome

stress-delta

change in concentration due to stress

BNP

B type Natriuretic Peptide

ED

emergency department

PCI

percutaneous coronary intervention

CABG

coronary artery bypass graft

NT-proBNP

N-terminal pro-BNP

TIMI

Thrombosis in Myocardial Infarction

AMI

acute myocardial infarction

CV

coefficient of variation

ng

nanogram

L

liter

mL

milliliter

dL

deciliter

BP

blood pressure

bpm

beats per minute

CI

confidence interval

ANOVA

analysis of variance

ROC

receiver operating characteristic