d

CLINICAL RESEARCH STUDY

B-type Natriuretic Peptide in the Early Diagnosis andRisk Stratification of Acute Chest PainPhilip Haaf, MD,a* Tobias Reichlin, MD,a* Nils Corson, MD,a Raphael Twerenbold, MD,a Miriam Reiter, MD,a

Stephan Steuer, MD,b Stefano Bassetti, MD,c Katrin Winkler, MD,d,e Claudia Stelzig, MSc,a Corinna Heinisch, MD,a

Beatrice Drexler, MD,a Heike Freidank, MD,f Christian Mueller, MD, FESCa

aDepartment of Internal Medicine, University Hospital, Basel, Switzerland; bDepartment of Emergency Medicine, Spital Limmattal,Schlieren, Switzerland; cDepartment of Internal Medicine, Kantonsspital Olten, Olten, Switzerland; dServicio de Pneumologia, Hospitalel Mar - IMIM, UPF, CIBERES, ISC III and eServicio de Urgencias, Hospital del Mar - IMIM, Barcelona, Spain; fDepartment of

Laboratory Medicine, University Hospital, Basel, Switzerland.

of Internal Medicin

0002-9343/$ -see fdoi:10.1016/j.amjm

ABSTRACT

BACKGROUND: Myocardial ischemia is a strong trigger of B-type natriuretic peptide (BNP) release. Asischemia precedes necrosis in acute myocardial infarction, we hypothesized that BNP might be useful inthe early diagnosis and risk stratification of patients with acute chest pain.METHODS: In a prospective, international multicenter study, BNP was measured in 1075 unselectedpatients with acute chest pain. The final diagnosis was adjudicated by 2 independent cardiologists. Patientswere followed long term regarding mortality.RESULTS: Acute myocardial infarction was the adjudicated final diagnosis in 168 patients (16%). BNPlevels at presentation were significantly higher in acute myocardial infarction as compared with patientswith other diagnoses (median 224 pg/mL vs. 56 pg/mL, P �.001). The diagnostic accuracy of BNP for thediagnosis of acute myocardial infarction as quantified by the area under the receiver operating character-istic curve (AUC) (0.74; 95% confidence interval [CI], 0.70-0.78) was lower compared with cardiactroponin T at presentation (AUC 0.88; 95% CI, 0.84-0.92; P �.001). Cumulative 24-month mortality rateswere 0.5% in the first, 2.1% in the second, 7.0% in the third, and 22.9% in the fourth quartile of BNP(P �.001). BNP predicted all-cause mortality independently of and more accurately than cardiac troponinT: AUC 0.81 (95% CI, 0.76-0.86) versus AUC 0.70 (95% CI, 0.62-0.77; P �.001). Net reclassificationimprovement for BNP was 0.10 (P � .04), and integrated discrimination improvement 0.068 (P � .01).CONCLUSIONS: BNP accurately predicts mortality in unselected patients with acute chest pain indepen-dently of and more accurately than cardiac troponin T, but does not seem to help in the early diagnosis ofacute myocardial infarction.© 2011 Elsevier Inc. All rights reserved. • The American Journal of Medicine (2011) 124, 444-452

KEYWORDS: Acute Coronary Syndrome; Brain Natriuretic Peptide; Chest Pain; Diagnosis; Mortality

raria from Brahms and Roche. All other authors declare that they have noconflict of interest.

Authorship: The authors designed the study, gathered and analyzedthe data, vouch for the data and analysis, wrote the paper, and decided topublish. The sponsors had no role in designing or conducting the study andno role in gathering or analyzing the data or writing the manuscript. Allauthors had access to the data and a role in writing the manuscript.

*P. Haaf and T. Reichlin contributed equally to this work.Requests for reprints should be addressed to Christian Müller, MD,

Department of Internal Medicine, University Hospital Basel, Petersgraben4, Basel CH-4031, Switzerland.

Funding: The study was supported by research grants from the SwissNational Science Foundation (PP00B-102853), the Swiss Heart Founda-tion, Abbott, Roche, Siemens, and the Department of Internal Medicine,University Hospital Basel.

Conflict of Interest: Dr. Mueller has received research support fromthe Swiss National Science Foundation (PP00B-102853), the Swiss HeartFoundation, the Novartis Foundation, the Krokus Foundation, Abbott,Astra Zeneca, Biosite, Brahms, Roche, Siemens, and the Department ofInternal Medicine, University Hospital Basel, as well as speaker honorariafrom Abbott, Biosite, Brahms, Roche, and Siemens. Dr. Reichlin hasreceived research grants from the University of Basel and the Department

e, University Hospital Basel as well as speaker hono- E-mail address: chmueller@uhbs.ch

ront matter © 2011 Elsevier Inc. All rights reserved.ed.2010.11.012

pcB

popep

aascsiddoawcc

onin

445Haaf et al BNP for Diagnosis and Risk Stratification of AMI

B-type natriuretic peptide (BNP) has emerged as a quanti-tative marker of ventricular wall stress1-3 and importantclinical help in the early diagnosis of heart failure.4-6 Ex-erimental and clinical studies have demonstrated that myo-ardial ischemia and acute hypoxia are strong triggers ofNP release.7,8 These observa-

tions were confirmed by clinicalstudies demonstrating an immedi-ate increase in natriuretic peptidelevels after transient myocardialischemia induced by cardiac exer-cise testing.9-11

The early diagnosis of acutemyocardial infarction is currentlylimited due to a delayed increaseof circulating cardiac troponinlevels within the first 4 to 6 hoursafter the onset of symptoms.12

Therefore, it has been speculatedthat the additional use of natri-uretic peptides as a marker ofmyocardial ischemia might im-prove the diagnostic accuracy atpresentation and the managementof patients.13-15 Several studiesshowed increased levels of natriuretic peptides in patientswith acute myocardial infarction but failed to convincinglydemonstrate a clinically meaningful benefit in the diagnosisof acute myocardial infarction through the additional use ofnatriuretic peptides.16-20 Because ischemia precedes necro-sis in acute myocardial infarction, we hypothesized thatBNP might be useful in the early diagnosis of patients withacute chest pain.

Several studies have established the prognostic value ofBNP in patients with acute myocardial infarction and un-stable angina.21-23 The value of BNP in the risk stratificationof unselected patients with acute chest pain, however, hasbeen analyzed in only few studies for subgroups withoutST-elevation myocardial infarction.24 The second aim of theresent study, therefore, was to assess the prognostic valuef BNP in these patients and compare it with establishedrognostic markers such as cardiovascular risk factors, withstablished risk-stratification models25 and with cardiac tro-onin T (cTnT).

METHODS

Study Design and PopulationAdvantageous Predictors of Acute Coronary SyndromeEvaluation (APACE) is an ongoing prospective interna-tional multicenter study designed and coordinated by theUniversity Hospital, Basel, Switzerland.26,27 From April2006 to June 2009, a total of 1247 consecutive patientspresenting to the Emergency Department with symptomssuggestive of acute myocardial infarction such as acutechest pain and angina pectoris were recruited. Of these,

CLINICAL SIGNIF

● B-type natriuretof patients withfarction are highpain patients.

● The diagnostic ais moderate.

● BNP is a powerfuin chest pain pa

● Prediction of mpatients with BNwith cardiac trop

BNP values at presentation were available in 1075. Patients o

with terminal kidney failure requiring dialysis were ex-cluded. The study was carried out according to the princi-ples of the Declaration of Helsinki and approved by thelocal ethics committees. Written informed consent was ob-tained from all patients.

Routine ClinicalAssessmentAll patients underwent an initialclinical assessment including clin-ical history, physical examina-tion, 12-lead electrocardiogram(ECG), pulse oximetry, standardblood tests, and chest radiogra-phy. cTnT, the MB fraction ofcreatine kinase (CK-MB), andmyoglobin were measured atpresentation, and thereafter aslong as clinically indicated. Tim-ing and treatment of patients wereleft to the discretion of the attend-ing physicians.

Adjudicated FinalDiagnosis

To determine the causal diagnosis for each patient, 2independent cardiologists reviewed all available medicalrecords (including patient history, physical examination,results of laboratory and radiologic testing, ECG, echo-cardiography, cardiac exercise test, coronary angiogra-phy) pertaining to the patient from the time of EmergencyDepartment presentation to 60-day follow-up. Cases werereviewed with a third cardiologist in situations of diag-nostic disagreement.

Acute myocardial infarction was defined as recom-mended in current guidelines.28 In brief, acute myocardialinfarction was diagnosed when there was evidence of myo-cardial necrosis in a clinical setting consistent with myocar-dial ischemia. Necrosis was diagnosed by a rising or falling(or both) pattern of cTnT with at least one value above the99th percentile with an imprecision of �10%.26,27 Unstablengina was diagnosed in patients with normal cTn levelsnd typical angina at rest, a deterioration of a previouslytable angina, in cases of positive cardiac exercise testing orardiac catheterization with coronary arteries found to havetenosis �70%, and in ambiguous cases in which follow-upnformation revealed acute myocardial infarction or a sud-en unexpected cardiac death within 60 days. Further, pre-efined diagnostic categories included cardiac but not cor-nary symptoms (eg, perimyocarditis, tachyarrhythmias)nd noncardiac symptoms. If acute myocardial infarctionas excluded in the Emergency Department, but no suffi-

ient further diagnostic procedures were performed for con-lusive diagnosis, symptoms were classified as of unknown

CE

tide (BNP) levelste myocardial in-an in other chest

cy of BNP for AMI

ictor of mortality.

ty of chest painore accurate thanT.

ICAN

ic pepacu

er th

ccura

l predtients

ortaliP is m

rigin.

�

mitattm

cdwptsowypahipa(fS

wp

q

446 The American Journal of Medicine, Vol 124, No 5, May 2011

Biochemical AnalysisBlood samples for determination of cTnT and BNP werecollected at presentation. Additional samples were collectedat 1, 2, 3, and 6 hours. Serial sampling was discontinuedwhen the diagnosis of acute myocardial infarction was cer-tain and treatment required transferring the patient to thecatheter laboratory or coronary care unit. After centrifuga-tion, samples were frozen at �80°C until assayed in ablinded fashion in a dedicated core laboratory. BNP wasmeasured in ethylenediaminetetraacetic acid-plasma usingthe AxSYM BNP assay (Abbott Laboratories, Abbott Park,Ill). The analytical range as reported by the manufacturerextends from 15 to 20,000 pg/mL for the AxSYM assay.CTnT (4th generation) was measured on Elecsys 2010(Roche Diagnostics, Mannheim, Germany) with a limit ofdetection of 0.01 �g/L, a 99th-percentile cutoff point of

0.01 �g/L, and a coefficient of variation of �10% at0.035 �g/L.

ECG AnalysisAll 12-lead ECGs recorded at presentation were assessed asrecommended in current guidelines28 in a core laboratory byinternal medicine specialists blinded to the clinical andbiochemical patient details.

Follow-up and Clinical EndpointsAfter hospital discharge, patients were followed after 3, 12,and 24 months by telephone or in written form. Any clinical(cardiovascular) events since presentation to the EmergencyDepartment were collected by establishing contact with thepatient and his family physician. Information about deathalso was obtained from the national registry on mortality.The primary endpoint was death from all causes.29 Thesecondary endpoint was a composite of all-cause death orfirst acute myocardial infarction in chest pain patients with-out acute myocardial infarction as the index event.

Statistical AnalysisComparisons between groups were made using chi-squaredmethod for categorical and Mann-Whitney U test for con-tinuous variables. Receiver operating characteristic (ROC)curves (and their 95% confidence intervals [CI]) were con-structed to assess the sensitivity and specificity throughoutthe concentrations of cTnT and BNP to compare their in-dividual or combined ability to diagnose acute myocardialinfarction and to predict the probability of all-cause deathand acute myocardial infarction during follow-up. Compar-ison of areas under the ROC curves was performed asrecommended by DeLong et al.30 The Kaplan-Meier

ethod was employed to analyze the timing of events dur-ng follow-up. Statistical assessment was performed usinghe log-rank test. We used univariate Cox proportional haz-rd analysis to compute hazard ratios and 95% CI of poten-ial predictors of death. All significant variables were thenested in a multivariable model using the forward stepwise

ethod. v

Values were adjusted for age and established cardiovas-ular risk factors (hypertension, dyslipidemia, smoking, andiabetes mellitus). Family history of cardiovascular diseaseas not considered because it was not collected in allatients. Net reclassification improvement (NRI) compareshe proportions moving up or down in clinical categories inurvivors and nonsurvivors. We chose risk categories tobtain clinically meaningful results. For further verificatione used integrated discrimination improvement (IDI) anal-sis, which is not dependent on certain risk groups becauserobability differences are used instead of categories.31 Tossess the incremental value of BNP over and above theistory, physical examination, and ECG, patients were strat-fied into low, intermediate, or high risk based on history,hysical examination, initial 12-lead ECG, and troponin Ts recommended in the American College of CardiologyACC)/American Heart Association (AHA) 2007 guidelinesor the management of patients with unstable angina/nonT-elevation myocardial infarction.25

All hypothesis testing was 2-tailed and a P-value of �.05was considered statistically significant. All statistical anal-yses were performed using SPSS for Windows 15.0 (SPSSInc, Chicago, Ill) and MedCalc 9.6.4.0 (MedCalc Software,Mariakerke, Belgium).

RESULTS

Characteristics of PatientsBaseline characteristics of the 1075 patients enrolled areshown in Table 1. The adjudicated final diagnosis was acutemyocardial infarction in 168 (16%) patients (25% ST-ele-vation myocardial infarction [STEMI], 75% non-ST-eleva-tion myocardial infarction [NSTEMI]), unstable angina in15%, cardiac symptoms of origin other than coronary arterydisease in 13%, noncardiac symptoms in 47%, and symp-toms of unknown origin in 9%.

Levels of BNP in Different Groups of ChestPainLevels of BNP are displayed in Figure 1. Patients with acutemyocardial infarction had higher presentation BNP levels(median, 224 pg/mL, interquartile range [IQR] 73-625 pg/mL) as compared with patients with unstable angina (me-dian, 104 pg/mL, IQR 53-209 pg/mL) and patients withoutacute coronary syndrome (median, 56 pg/mL, IQR 19-145pg/mL, not shown) (P �.001). In patients with cardiac butnot coronary disease, median BNP values were 121 pg/mL(IQR 44-494 pg/mL), which was inferior compared withpatients with acute myocardial infarction (P � .02) and sim-ilar compared with unstable angina patients (P � .12). There

ere no significant differences regarding BNP values inatients with STEMI and NSTEMI (P � .78).

The distribution of several baseline characteristics acrossuartiles of BNP is shown in Table 2. Higher baseline

alues of BNP were associated with most cardiovascular

H

V

E

re of t

447Haaf et al BNP for Diagnosis and Risk Stratification of AMI

risk factors, previous acute myocardial infarction, andrevascularization.

BNP for the Diagnosis of Acute MyocardialInfarctionAs displayed in Figure 2A, the diagnostic accuracy of BNPat presentation for the diagnosis of acute myocardial infarc-tion as quantified by the area under the ROC curve (AUC)was 0.74 (95% CI, 0.70-0.78). Evaluating the value of BNPas a diagnostic test for acute myocardial infarction, ROCcurves yielded an optimal cutoff value of 132 pg/mL. Usingthis cutoff value yielded sensitivities, specificities, negativepredictive values, and positive predictive values of 63%,73%, 91%, and 30%, respectively. Serial measurements ofBNP, including changes of BNP during the first hour, didnot significantly alter its diagnostic accuracy (data notshown). The AUC of cTnT at presentation (available in1026 of 1075 [95%] patients) was significantly higher ascompared with BNP (0.88; 95% CI, 0.84-0.92; P �.001).The additional use of BNP in combination with cTnT didnot significantly increase the diagnostic accuracy providedby cTnT alone (AUC of combination with BNP 0.89; 95%

Table 1 Baseline Characteristics of the Patients

CharacteristicsAll Pat(n � 1

Age – years 64 (Male sex – no. (%) 717 (Risk factors – no. (%)

Hypertension 687 (Hypercholesterolemia 475 (Diabetes 203 (Current smoking 265 (History of smoking 378 (

istory – no. (%)Coronary artery disease 405 (Previous myocardial infarction 274 (Previous revascularization 301 (Peripheral artery disease 80 (7Previous stroke 65 (6

ital statusHeart rate – beats per minute 76 (6Systolic blood pressure – mm Hg 143 (1Diastolic blood pressure – mm Hg 84 (7Body mass index* 26.4 (2

lectrocardiographic findings – no. (%)Left bundle branch block 43 (4ST-segment elevation 56 (5ST-segment depression 107 (1T-wave inversion 79 (7No significant electrocardiographic abnormalities 790 (7Glomerular filtration rate – mL/min/1.73 m2 89 (

*The body mass index is the weight in kilograms divided by the squa

CI, 0.87-0.91; P � .33).

BNP for the Diagnosis of Acute MyocardialInfarction in Early PresentersIn patients presenting within 3 hours of chest pain onset, theAUC for BNP was 0.66 (95% CI, 0.57-0.75) and 0.74 (95%CI, 0.65-0.83; P � .11) for cTnT. Again, the combination ofBNP and cTnT did not improve the diagnostic accuracy ascompared with cTnT alone (AUC 0.70; 95% CI, 0.60-0.81;P � .24).

Prognostic Value of BNP for the Prediction ofDeathDuring a mean follow-up time of 18 � 10 months, therewere 73 deaths in the whole cohort with a median time todeath of 121 days. Median BNP values in deceased patients(400 pg/mL, IQR 125-976) were significantly higher ascompared with those in survivors (61 pg/mL, IQR 21-168,P �.001).

Kaplan-Meier analyses based on BNP at presentationshowed a much higher 24-month mortality rate for patientsin the fourth quartile (22.9%) than in the other quartiles(7.0% in quartile 3, 2.1% in quartile 2, 0.5% in quartile 1,P �.001, Figure 3A). The same predictive value of BNPquartiles for all-cause mortality in Kaplan-Meier analyses

Acute Myocardial Infarction

P ValueYes (n � 168) No (n � 907)

73 (59-81) 63 (50-75) �.001120 (71) 597 (66) .16

127 (76) 560 (62) .00182 (49) 393 (43) .1939 (23) 164 (18) .1150 (30) 215 (24) .1055 (33) 323 (36) .46

76 (45) 329 (36) .0355 (33) 219 (24) .0250 (30) 251 (28) .5821 (13) 59 (7) .00721 (13) 44 (5) �.001

81 (69-94) 75 (65-89) .0030) 145 (125-163) 143 (128-159) .95

85 (5-96) 84 (74-93) .71.6) 26.2 (23.9-29.0) 26.4 (24.0-29.7) .70

17 (10) 26 (3) �.00138 (23) 18 (2) �.00141 (24) 66 (7) �.00116 (10) 63 (7) �.00156 (33) 734 (81) �.001

) 79 (60-101) 91 (73-108) �.001

he height in meters.

ients075)

51-76)67)

64)44)19)25)35)

38)25)28)))

6-90)28-164-93)4.0-29

))0))4)

71-107

was observed in the important subgroups of patients with

dhw[

HHCHN

G

448 The American Journal of Medicine, Vol 124, No 5, May 2011

acute myocardial infarction (P �.001) as well as in patientswith noncardiac chest pain (P � .004).

Assessing the prognostic value for the prediction ofeath by ROC-curve analysis showed a significantlyigher AUC for presentation values of BNP comparedith troponin T (AUC 0.81 [95% CI, 0.76-0.86]) vs. 0.70

95% CI, 0.62-0.77]; P �.001 for comparison, Figure2B). The prognostic value of the 6-hour value of cTnT(AUC 0.67 [95% CI, 0.55-0.80]) did not differ signifi-cantly from that of the cTnT at presentation and wasinferior compared with that of BNP as well. A BNP value

Figure 1 Levels of B-type natriuretic peptide (BNP) at pre-sentation to the Emergency Department in all patients accord-ing to the adjudicated final diagnosis. Boxes represent inter-quartile ranges (IQRs), while whiskers display ranges (withoutoutliers further than 1.5 IQRs from the end of the box).CAD � coronary artery disease.

Table 2 Baseline Characteristics of the Patients in Relation to

CharacteristicsBNP Quartile 1�23.1 pg/mL

Age (years) 50 (41-59)Male sex 207 (77)

istory of hypertension 118 (44)istory of diabetes 32 (12)urrent smoking 102 (38)istory of smoking 80 (30)ever smokers 86 (32)

History of coronary artery disease 39 (14)Previous myocardial infarction 23 (9)Previous revascularization 29 (11)Body mass index (kg/m2) 26.6 (24.2-29.9)lomerular filtration rate (mL/min/1.73 m2) 101 (88-114)

Acute myocardial infarction 14 (5)Unstable angina 16 (6)No acute coronary syndrome 240 (89)

of 207 pg/mL emerged as the best prognostic cutoff valueof BNP to predict death with a sensitivity of 69.4% anda specificity of 79.1%.

Multivariable Cox proportional hazard analysis dis-played that BNP and cTnT at presentation independentlypredicted mortality (Table 3). Reclassification tables forpredicting death with presentation values of cTnT alone andthe additional use of BNP yielded an NRI of 0.10 (P � .04,Table 4). IDI was 0.068 (P � .011), supporting the findingsof NRI.

BNP for the Prediction of Death or AcuteMyocardial Infarction in Chest Pain Patientswith Diagnoses Other Than Acute MyocardialInfarctionIn patients without acute myocardial infarction (n � 907,84%) there were 38 (4.2%) deaths and 40 (4.4%) first acutemyocardial infarctions during follow-up, resulting in a com-posite endpoint of death or acute myocardial infarction in 71(7.8%) patients. Median BNP values of patients reachingthe composite endpoint were significantly higher comparedwith patients that did not (median, 123 pg/mL [IQR 69-318pg/mL] vs 52 pg/mL [IQR 18-132 pg/mL]; P �.001).

Whereas only 2.5% of chest pain patients without acutemyocardial infarction in the first quartile reached this com-bined endpoint of death or acute myocardial infarction, itwas 5.3% in the second, 13.3% in the third, and 18.8% inthe fourth BNP quartile (P �.001, Figure 3B).

Incremental Value of BNP Over and Above theHistory, Physical Examination, and ECGBased on the ACC/AHA 2007 guidelines for the manage-ment of patients with UA/NSTEMI,25 patients were strati-fied into low, intermediate, and high risk based on history,

e Natriuretic Peptide (BNP) Quartiles

Quartile 2 BNP Quartile 3 BNP Quartile 4P Value-68.3 pg/mL 68.4-202.4 pg/mL �202.4 pg/mL

(49-70) 70 (59-77) 77 (68-82) �.001(63) 181 (67) 161 (60) �.001(53) 204 (76) 223 (84) �.001(16) 61 (23) 66 (25) �.001(28) 51 (19) 38 (14) �.001(31) 106 (39) 110 (41) .004(40) 110 (41) 110 (41) .054(29) 126 (47) 163 (61) �.001(19) 80 (30) 120 (45) �.001(22) 99 (37) 113 (42) �.001(23.8-29.4) 27.1 (24.2-30.0) 26.1 (23.9-29.0) .162(80-110) 84 (67-103) 71 (52-90) �.001(8) 47 (17) 85 (32) �.001(12) 64 (24) 44 (16) �.001(80) 158 (59) 140 (52) �.001

B-typ

BNP23.1

59168142447482

107775160

26.3982233

214

stt3[ta

(spc(ia[c

d BNP

449Haaf et al BNP for Diagnosis and Risk Stratification of AMI

physical examination, initial 12-lead ECG, and troponin T.Patients in the high-risk group (n � 246) who reached theecondary endpoint of death or nonfatal myocardial infarc-ion (n � 25) had significantly higher BNP values than pa-ients who did not reach the secondary endpoint (median18 pg/mL [IQR 140-945 pg/mL] vs median 154 pg/mLIQR 77-382 pg/mL]; P � .02 for comparison). For pa-ients in the intermediate-risk group (n � 332), there was

trend towards higher BNP values in the event group

Figure 2 Diagnostic and prognostic performanin receiver operating characteristic (ROC) curvperformance of troponin T and BNP at presentat(B) the prognostic performance of troponin T an

Figure 3 Kaplan-Meier curves for (A) survivalmyocardial infarction in nonacute myocardial infcumulative survival during follow-up in all chnatriuretic peptide (BNP) quartiles. (B) Kaplaninfarction at presentation (n � 907) displaying th

dial infarction during follow-up according to BNP qua

n � 12) compared with patients who did not reach theecondary endpoint (median 108 pg/mL [IQR 67-143g/mL] vs median 67 pg/mL [24-145 pg/mL]; P � .08 foromparison). Six patients in the low-risk categoryn � 329) reached the secondary endpoint, and these sim-larly had higher BNP levels compared with patients with

better prognosis (median BNP value 46 pg/mL, IQR26-62 pg/mL] vs 22 pg/mL, IQR [15-51]; P � .10 foromparison) (Figure 4).

roponin T and B-type natriuretic peptide (BNP)sis. ROC curves describing (A) the diagnostic

the diagnosis of acute myocardial infarction andin determining all-cause mortality.

chest pain patients and (B) survival free of acutepatients. (A) Kaplan-Meier analysis displaying

in patients (n � 1075) subdivided into B-typeanalysis of patients without acute myocardialined endpoint of survival free of acute myocar-

ce of te analyion in

in allarctionest pa-Meiere comb

rtiles.

oiTa

cspdi

cuudpfpp

pdrca

nhcscsldpwnshnrpd

P

P

450 The American Journal of Medicine, Vol 124, No 5, May 2011

DISCUSSIONIn this prospective, international multicenter study of 1075consecutive patients presenting with acute chest pain to theEmergency Department, we examined the diagnostic andprognostic value of BNP. We report 4 major findings:

First, BNP levels were significantly higher in patientswith acute myocardial infarction than in patients with otherdiagnoses. We hereby confirm earlier pilot studies.18-20 Sec-nd, the diagnostic accuracy of BNP for acute myocardialnfarction was moderate and inferior compared with cTnT.here was no additive diagnostic value by combining BNPnd cTnT, even in early presenters. Therefore, our data

Table 3 Cox Proportional Hazard Analysis for Prediction of All

Variable

Univariate A

Hazard Ratio

BNP (per increase of 100 pg/mL) 1.115 (1.092Cardiac troponin T (per increase of 0.1 �g/L) 1.064 (1.037Age (per year) 1.100 (1.076Female sex 0.841 (0.510Body mass index (per increase of 1 unit) 0.948 (0.898Diabetes mellitus 1.964 (1.191History of coronary artery disease 4.628 (2.767Active smoking 0.709 (0.396Systolic blood pressure (per mm Hg) 0.987 (0.977Estimated glomerular filtration rate 0.970 (0.962

BNP � B-type natriuretic peptide; CI � confidence interval.

Table 4 Reclassification of Patients Who Survived and WhoDied during the First Year

Prognostic Modelwith cTnT atPresentation

Prognostic Model cTnT and B-typeNatriuretic Peptide (BNP) atPresentation

�5.0%Risk

5%-10%Risk

�10%Risk

TotalNo.

atients who survived�5.0% risk 665 5 3 6735%-10% risk 58 80 13 151�10% risk 0 42 103 145Total no. 723 127 119 969

atients who died�5.0% risk 9 1 0 105%-10% risk 2 6 3 11�10% risk 0 1 35 36Total no. 11 8 38 57

In survivors, 100 patients (10%) correctly moved downward and 21(2%) incorrectly moved upward in the classification. In nonsurvivors, 4patients (7%) correctly moved upward and 3 (5%) incorrectly moveddownward in the classification. The net reclassification improvement wascalculated at 0.10 (P � .038).

Both models are adjusted for age and cardiovascular risk factors(cvRF). cvRF used in the prognostic models are: hypertension, dyslipi-

cdemia, smoking and diabetes mellitus.

learly reject the hypothesis generated by the initial pilottudies.18-20 Third, BNP at presentation was a powerfulredictor of mortality, irrespective of the adjudicated finaliagnosis. Fourth, the prediction of mortality with BNP wasndependent of and more accurate than with cTnT.

These results are of major clinical importance: BNPonstitutes a strong and valuable prognostic indicator innselected patients with acute chest pain regardless of thenderlying disease. It identifies patients at a higher risk ofeath who might benefit from additional cardiac diagnosticrocedures and more intensive medical treatment. All ef-orts should be undertaken to detect the underlying patho-hysiology causing the elevation of BNP in order to treatatients accordingly.

Our results support measuring both cTn and BNP inatients presenting with acute chest pain, the former for theiagnosis of acute myocardial infarction and the latter forisk stratification. Obviously, BNPs can only be used inonjunction with other clinical information to guide man-gement in patients with acute chest pain.

The pathophysiology of BNP provides a valuable expla-ation for our findings. BNP is a quantitative marker ofemodynamic cardiac stress that is primarily produced byardiomyocytes in both ventricles. Myocardial ischemia is atrong trigger of BNP release, as are many other cardiaconditions that lead to increased cardiac volumes or pres-ures.9-11 The BNP value summarizes and integrates botheft ventricular systolic and diastolic dysfunction, valvularysfunction, and right ventricular dysfunction.32 This ex-lains the occurrence of high BNP values both in patientsith acute myocardial infarction and cardiac, but not coro-ary disease (in particular, heart failure). Therefore, BNPecretion is not specific for myocardial ischemia and of littleelp in the diagnosis of acute myocardial infarction. Thisonspecificity of BNP, however, makes it well suited forisk stratification in the heterogeneous group of chest painatients, as increased hemodynamic cardiac stress seems toetermine disease severity and therefore mortality in most

Death during Follow-up

Multivariate Analysis

CI) P-Value Hazard Ratio (95% CI) P-Value

) �.001 1.073 (1.038-1.108) �.001) �.001 1.041 (1.012-1.072) .006) �.001 1.084 (1.056-1.113) �.001) .498) .055) .008 1.546 (0.912-2.621) .106) �.001 2.393 (1.399-4.094) .001) .247) .008 0.991 (0.981-1.000) .062) �.001 0.996 (0.986-1.006) .445

-cause

nalysis

(95%

-1.140-1.092-1.125-1.387-1.001-3.283-7.742-1.270-0.997-0.978

ardiac conditions.32,33

dptttdppctacmi(t

cptwlmwo

iiousufwc

451Haaf et al BNP for Diagnosis and Risk Stratification of AMI

In terms of prognosis, our study confirms and corrobo-rates the prognostic role of BNP in patients with acutemyocardial infarction and unstable angina.22,23 Our studyiffers from these previous studies by assessing a real-worldopulation of consecutive chest pain patients presenting tohe Emergency Department instead of a subgroup of pa-ients from a randomized acute coronary syndrome trial. Wehereby for the first time extend these previous findings byemonstrating the prognostic value of BNP not only inatients with an acute coronary syndrome, but in all chestain patients regardless of the underlying disease and ECGhanges. A BNP level of 202 pg/mL (4th quartile) formed ahreshold with marked increase in mortality for valuesbove this level with a 2-year mortality rate of 22.9%,ompared with 0.5% in the first quartile. In addition toultivariate Cox regression analysis showing BNP as an

ndependent predictor in chest pain patients, both NRI0.10, P � .04) and IDI (0.068, P � .011) analysis indicated

Figure 4 B-type natriuretic peptide (BNP) levels in patientswithout myocardial infarction at presentation according to dif-ferent risk categories. BNP levels in patients without myocar-dial infarction at presentation according to events (death andnonfatal myocardial infarction) during follow-up and accordingto different risk categories. Risk categories are based on his-tory, physical examination, initial 12-lead electrocardiogram,and troponin T as proposed in the American College of Car-diology/American Heart Association (ACC/AHA) 2007 guide-lines for the management of patients with unstable angina/non-ST-elevation myocardial infarction.25 Event rates in thehigh- (n � 246), intermediate- (n � 332), and low-risk catego-ries (n � 329) were 10%, 4%, and 2%, respectively.

hat adding BNP to cTnT significantly adds prognostic ac-

uracy. Furthermore, the incremental value of BNP—inarticular for high-risk patients—was illustrated after pa-ients without acute myocardial infarction at presentationere risk-stratified according to ACC/AHA 2007 guide-

ines. Of important note, the prognostic role of BNP re-ained significant in the important subgroup of patientsith noncardiac chest pain, which constitutes roughly 50%f all patients in daily practice.34 Furthermore, our results

are supported by data from Jernberg and co-workers, whofound a prognostic value of N-terminal proBNP in 755 chestpain patients regardless of the underlying disease.17

BNP has been used for diagnosing heart failure,4-6 guid-ng drug treatment for cardiac failure, and as a prognosticndex in acute coronary syndrome.21-23 Based on the resultsf this analysis derived from an unselected chest pain pop-lation, we propose to extend the use of BNP for risktratification to all chest pain patients irrespective of thenderlying disease. These data need to be confirmed inurther studies. Furthermore, BNP needs to be comparedith other novel markers proposed for risk stratification in

hest pain patients, such as ischemia-modified albumin,35

myeloperoxidase,36 copeptin,27,37 or growth differentiationfactor 15.38

Potential limitations of the current study merit consider-ation. First, as a prospective observational study, we cannotquantify exactly the clinical benefit associated with im-proved risk stratification. Second, given that escalating BNPlevels were linked to factors also linked to increased mor-bidity and mortality, residual confounding may possibly bepresent in the reported data. Third, we cannot comment onthe diagnostic and prognostic accuracy among patients withterminal kidney failure requiring dialysis, because such pa-tients were excluded from our study.

CONCLUSIONBNP accurately predicts mortality in unselected patientswith acute chest pain independently of and more accuratelythan cTnT regardless of the adjudicated final diagnosis.However, BNP does not seem to help in the early diagnosisof acute myocardial infarction.

ACKNOWLEDGMENTWe are indebted to the patients who participated in the studyand to the Emergency Department staff as well as thelaboratory technicians for their most valuable efforts, andwe thank Drs. C. Schindler and K. Denhaerynck for expertstatistical advice.

References1. Yoshimura M, Yasue H, Okumura K, et al. Different secretion patterns

of atrial natriuretic peptide and brain natriuretic peptide in patientswith congestive heart failure. Circulation. 1993;87:464-469.

2. Yasue H, Yoshimura M, Sumida H, et al. Localization and mechanismof secretion of B-type natriuretic peptide in comparison with those ofA-type natriuretic peptide in normal subjects and patients with heart

failure. Circulation. 1994;90:195-203.

452 The American Journal of Medicine, Vol 124, No 5, May 2011

3. Maeda K, Tsutamoto T, Wada A, Hisanaga T, Kinoshita M. Plasmabrain natriuretic peptide as a biochemical marker of high left ventric-ular end-diastolic pressure in patients with symptomatic left ventric-ular dysfunction. Am Heart J. 1998;135(5 Pt 1):825-832.

4. Maisel AS, Krishnaswamy P, Nowak RM, et al. Rapid measurement ofB-type natriuretic peptide in the emergency diagnosis of heart failure.N Engl J Med. 2002;347:161-167.

5. McCullough PA, Nowak RM, McCord J, et al. B-type natriureticpeptide and clinical judgment in emergency diagnosis of heart failure:analysis from Breathing Not Properly (BNP) Multinational Study.Circulation. 2002;106:416-422.

6. Mueller C, Scholer A, Laule-Kilian K, et al. Use of B-type natriureticpeptide in the evaluation and management of acute dyspnea. N EnglJ Med. 2004;350:647-654.

7. Goetze JP, Gore A, Moller CH, Steinbruchel DA, Rehfeld JF, NielsenLB. Acute myocardial hypoxia increases BNP gene expression.FASEB J. 2004;18:1928-1930.

8. Hama N, Itoh H, Shirakami G, et al. Rapid ventricular induction ofbrain natriuretic peptide gene expression in experimental acute myo-cardial infarction. Circulation. 1995;92:1558-1564.

9. Sabatine MS, Morrow DA, de Lemos JA, et al. Acute changes incirculating natriuretic peptide levels in relation to myocardial isch-emia. J Am Coll Cardiol. 2004;44:1988-1995.

10. Weber M, Dill T, Arnold R, et al. N-terminal B-type natriuretic peptidepredicts extent of coronary artery disease and ischemia in patients withstable angina pectoris. Am Heart J. 2004;148:612-620.

11. Staub D, Nusbaumer C, Zellweger MJ, et al. Use of B-type natriureticpeptide in the detection of myocardial ischemia. Am Heart J. 2006;151:1223-1230.

12. Melanson SE, Morrow DA, Jarolim P. Earlier detection of myocardialinjury in a preliminary evaluation using a new troponin I assay withimproved sensitivity. Am J Clin Pathol. 2007;128:282-286.

13. Mueller C. The use of B-type natriuretic peptides in coronary arterydisease: utile or futile? J Am Coll Cardiol. 2007;50:215-216.

14. Stewart RA. Broader indications for B-type natriuretic peptide testingin coronary artery disease. Eur Heart J. 2005;26:207-209.

15. Body R, Roberts C. Best evidence topic report. Brain natriureticpeptide as a potential marker of acute coronary syndromes. EmergMed J. 2006;23:403-407.

16. Morita E, Yasue H, Yoshimura M, et al. Increased plasma levels ofbrain natriuretic peptide in patients with acute myocardial infarction.Circulation. 1993;88:82-91.

17. Jernberg T, Stridsberg M, Venge P, Lindahl B. N-terminal pro brainnatriuretic peptide on admission for early risk stratification of patientswith chest pain and no ST-segment elevation. J Am Coll Cardiol.2002;40:437-445.

18. Bassan R, Potsch A, Maisel A, et al. B-type natriuretic peptide: a novelearly blood marker of acute myocardial infarction in patients withchest pain and no ST-segment elevation. Eur Heart J. 2005;26:234-240.

19. Connolly SB, Collier T, Khugputh R, et al. Brain natriuretic peptidetesting for angina in a rapid-access chest pain clinic. QJM. 2007;100:779-783.

20. Brown A, George J, Murphy MJ, Struthers A. Could BNP screening ofacute chest pain cases lead to safe earlier discharge of patients withnon-cardiac causes? A pilot study. QJM. 2007;100:755-761.

21. Arakawa N, Nakamura M, Aoki H, Hiramori K. Plasma brain natri-uretic peptide concentrations predict survival after acute myocardialinfarction. J Am Coll Cardiol. 1996;27:1656-1661.

22. de Lemos JA, Morrow DA, Bentley JH, et al. The prognostic value ofB-type natriuretic peptide in patients with acute coronary syndromes.

N Engl J Med. 2001;345:1014-1021.

23. Morrow DA, de Lemos JA, Sabatine MS, et al. Evaluation of B-typenatriuretic peptide for risk assessment in unstable angina/non-ST-elevation myocardial infarction: B-type natriuretic peptide and prog-nosis in TACTICS-TIMI 18. J Am Coll Cardiol. 2003;41:1264-1272.

24. Bassan R, Tura BR, Maisel AS. B-type natriuretic peptide: a strongpredictor of early and late mortality in patients with acute chest painwithout ST-segment elevation in the emergency department. CoronArtery Dis. 2009;20:143-149.

25. Anderson JL, Adams CD, Antman EM, et al. ACC/AHA 2007 guide-lines for the management of patients with unstable angina/non ST-elevation myocardial infarction: a report of the American College ofCardiology/American Heart Association Task Force on PracticeGuidelines (Writing Committee to Revise the 2002 Guidelines for theManagement of Patients With Unstable Angina/Non ST-ElevationMyocardial Infarction): developed in collaboration with the AmericanCollege of Emergency Physicians, the Society for CardiovascularAngiography and Interventions, and the Society of Thoracic Surgeons:endorsed by the American Association of Cardiovascular and Pulmo-nary Rehabilitation and the Society for Academic Emergency Medi-cine. Circulation. 2007;116:e148-e304.

26. Reichlin T, Hochholzer W, Bassetti S, et al. Early diagnosis of myo-cardial infarction with sensitive cardiac troponin assays. N Engl J Med.2009;361:858-867.

27. Reichlin T, Hochholzer W, Stelzig C, et al. Incremental value ofcopeptin for rapid rule out of acute myocardial infarction. J Am CollCardiol. 2009;54:60-68.

28. Thygesen K, Alpert JS, White HD. Universal definition of myocardialinfarction. J Am Coll Cardiol. 2007;50:2173-2195.

29. Pratt CM, Greenway PS, Schoenfeld MH, Hibben ML, Reiffel JA.Exploration of the precision of classifying sudden cardiac death. Im-plications for the interpretation of clinical trials. Circulation. 1996;93:519-524.

30. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areasunder two or more correlated receiver operating characteristic curves:a nonparametric approach. Biometrics. 1988;44:837-845.

31. Romanens M, Ackermann F, Spence JD, et al. Improvement of car-diovascular risk prediction: time to review current knowledge, debates,and fundamentals on how to assess test characteristics. Eur J Cardio-vasc Prev Rehabil. 2010;17:18-23.

32. Maisel A, Mueller C, Adams K Jr, et al. State of the art: usingnatriuretic peptide levels in clinical practice. Eur J Heart Fail. 2008;10:824-839.

33. Mueller C. Acute respiratory failure: back to the roots! Intensive CareMed. 2008;34:787-789.

34. Hamm CW, Goldmann BU, Heeschen C, Kreymann G, Berger J,Meinertz T. Emergency room triage of patients with acute chest painby means of rapid testing for cardiac troponin T or troponin I. N EnglJ Med. 1997;337:1648-1653.

35. Anwaruddin S, Januzzi JL Jr, Baggish AL, Lewandrowski EL, Le-wandrowski KB. Ischemia-modified albumin improves the usefulnessof standard cardiac biomarkers for the diagnosis of myocardial isch-emia in the emergency department setting. Am J Clin Pathol. 2005;123:140-145.

36. Brennan ML, Penn MS, Van Lente F, et al. Prognostic value ofmyeloperoxidase in patients with chest pain. N Engl J Med. 2003;349:1595-1604.

37. Khan SQ, Dhillon OS, O’Brien RJ, et al. C-terminal provasopressin(copeptin) as a novel and prognostic marker in acute myocardialinfarction: Leicester Acute Myocardial Infarction Peptide (LAMP)study. Circulation. 2007;115:2103-2110.

38. Eggers KM, Kempf T, Allhoff T, Lindahl B, Wallentin L, Wollert KC.Growth-differentiation factor-15 for early risk stratification in patients

with acute chest pain. Eur Heart J. 2008;29:2327-2335.