Skip to main content
. 2006 May;23(5):403–407. doi: 10.1136/emj.2006.036608

Table 3.

Author, date and country Patient group Study type (level of evidence) Outcomes Key results Study weaknesses
Mukoyama et al, 1991, Japan 13 consecutive patients with AMI <12 hours from symptom onset. Prospective diagnostic cohort BNP levels Rose after AMI (within “hours” of onset of AMI). Small numbers
Blood samples at enrollment and every 4–24 hours over 4 days. Correlation of BNP with PCWP and CIn No correlation with PCWP. Highly correlated inversely with CIn (r = −0.81, P<0.01). Subopimal reporting of results. Little statistical analysis. Meaningful conclusions cannot be drawn about the value of BNP as a diagnostic test.
Horio et al, 1993, Japan 16 patients admitted to CCU with AMI within 9 hours of symptom onset. All patients had right‐sided cardiac catheterisation, coronary angiography and primary angioplasty. 16 normal subjects. Prospective observational cohort BNP levels on day 1 Higher in AMI patients than controls (4.5 fold on day 1). No P value. Small numbers
BNP levels at follow up Significantly elevated in AMI patients at 14 days; still “abnormally elevated” at 4 weeks (no P values). Method of identification of normal subjects not described.
Blood taken on admission and serially on days 3, 7, 14 and 28 Correlation between BNP and haemodynamic variables No correlation with PCWP, right atrial pressure or CIn. Significant inverse correlation with LVEF (r = −0.67, P<0.01). Not all pertinent P values given.
This study suggests raised BNP in AMI but, because of the design, data cannot be used to evaluate BNP as a diagnostic test for AMI.
Morita et al, 1993, Japan 50 consecutive patients with AMI within 8 hours of symptom onset. 30 age‐ and sex‐matched controls. Prospective diagnostic cohort BNP level on admission Significantly increased with AMI v. controls (92+/−28 v. 5.2+/−0.5 pg/ml; P<0.01) Diagnosis of MI had already been ruled in at enrollment. Investigating application in an undifferentiated group would be more clinically relevant.
Peak BNP level in MI patients Peak level at 16.4+/−0.7 hours after admission
Blood taken on admission. All patients had coronary angiography on admission. 10 had intracoronary thrombolysis, 11 had IV thrombolysis; 26 had PTCA. BNP levels at four weeks Still significantly higher in MI group than controls (149+/−47 v. 5.2+/−0.5 pg/ml; P<0.001).
Correlation between BNP and haemodynamic parameters No correlation with PCWP; No correlation with CIn in 1st 2 days; Significant inverse correlation with CIn at time of peak BNP level (r = −0.476, P<0.01).
Kikuta et al, 1995, Japan 73 patients (already diagnosed) with either UA (n = 33), SA (n = 20) or atypical chest pain with normal coronary angiogram, stress test and hyperventilation test (n = 20). Prospective diagnostic cohort BNP levels Significantly higher in UA group compared with SA and controls (P<0.01 for each). No significant difference between SA and controls. 16 of UA group had ST elevation on ECG. CK‐MB < twice normal but no troponin testing. These patients may actually have had MI.
BNP levels according to ST elevation or depression No significant difference
Blood taken within 24 hours of last attack in UA group. BNP levels following treatment BNP decreased significantly (P<0.01) in UA group but not SA group. Results not useful for clinical evaluation of BNP as a diagnostic test (sensitivities, specificities, etc, cannot be calculated; patients had already been diagnosed at enrollment).
Regional wall motion abnormalities on echocardiography Significantly higher BNP if +ve for this outcome (P<0.01).
de Lemos et al, 2001, United States 2525 patients with ACS (825 STEMI, 565 NSTEMI, 1133 UA) enrolled into another study investigating GPIIb/IIIa inhibitors. Prospective observational cohort Baseline characteristics Patients with higher BNP more likely to be older (P<0.001), male (P<0.001), hypertensive (P<0.003), have CHF (P<0.001, hypercholesterolaemia (P<0.001), smokers (P<0.001), to have reduced creatinine clearance (P<0.001), CK‐MB > upper limit of normal (P<0.001) or ST segment depression (P<0.001). Very late sampling time.
Correlation of BNP with angiography/stress test Patients with higher BNP more likely to have >50% stenosis (P<0.001) or +ve stress test (P<0.01) than patients with low levels ACS had already been ruled in for this patient group at time of inclusion (by ECG or cardiac marker testing).
Blood sent at a mean of 40+/−20 hours from symptom onset. BNP as an independent predictor of death (after adjustment using logistic regression) Adjusted OR's for death at 10 months in 2nd, 3rd & 4th quartiles of BNP were 3.8 (95% CI 1.1–13.3), 4.0 (1.2–13.7) and 5.8 (1.7–19.7), respectively. Inadequate reporting of data to allow calculation of sensitivities, specificities, PPVs, NPVs or likelihood ratios.
Value of a BNP cutoff of 80 pg/ml as an independent predictor of mortality (adjusted) BNP remain significantly associated with increased 10‐month mortality (P = 0.04).
Morrow et al, 2003, USA 1676 patients (of 2220), enrolled in TACTICS‐TIMI 18 (early invasive v. conservative strategy), with non‐ST elevation acute coronary syndromes (NSTACS) and symptoms in the last 24 hours. Nested prospective observational cohort Elevated plasma BNP in UA (>80 pg/ml) Raised in 15.6% (n = 167) [i.e. sensitivity 15.6%], and 10.1% (n = 67) of those with ‐ve baseline TnI. Elevated BNP in 13.6% (n = 135) of patients with UA and no history or current evidence of HF. Patients enrolled having already been diagnosed with ACS (an undifferentiated group of patients would have enabled more clinically relevant conclusions to be drawn).
Blood taken “at enrolment”. Follow‐up at 6 months. Elevated plasma BNP in NSTEMI (>80 pg/ml) Raised in 25.2% [i.e. sensitivity 25.2%] (n = 153) Insufficient data reported to allow calculation of specificities, PPV's, NPV's or LR's.
Death at 30 days (BNP cut‐off 80 pg/ml) 5.0% raised v. 1.2% not raised (P<0.0001). Calculated sensitivity 80.6%.
Death at 6 months (BNP cut‐off 80 pg/ml) 8.4% raised v. 1.8% not raised (P<0.0001). Calculated sensitivity 82%.
Death at 7 days (BNP cut‐off 80 pg/ml) 2.5% raised v. 0.74% not raised. Calculated sensitivity 77%.
Association between BNP and 6/12 mortality after adjustment for important clinical predictors available at presentation Remained an independent predictor of mortality (OR 3.3, 95% CI 1.7–6.3).
BNP for prediction of recurrent MI Not predictive (5.3% v. 5.2%, p = 1.0)
BNP for prediction of hospitilisation with recurrent ACS Not predictive (13.4% v. 12.2%, p = 0.6).
BNP for prediction of new or worsening CHF 30 days: Significantly increased risk with raised BNP (5.9% v. 1.0%, p<0.0001). 6 months: Risk persisted (9.1% v. 1.8%, P<0.0001).
BNP + TnI for prediction of 6/12 mortality Both ‐ve: 0.7% risk of death.
Death with conservative/early invasive treatment according to BNP results No appreciable difference (p = 0.6 for 6/12 mortality).
Mega et al, 2004, USA 438 eligible (of 483) patients enrolled in the ENTIRE‐TIMI 23 trial (full v. half dose thrombolysis + abciximab and LMWH or heparin in STEMI <6 hours). Blood sampling on admission, before thrombolysis. Nested prospective observational cohort 30‐day mortality (reported results) Significantly higher BNP among those who died (P<0.0001). BNP>80 pg/ml associated with significantly higher risk of death (P<0.0001). All patients had STEMI already ruled in by ECG ‐ the study is not helpful to evaluate BNP as a diagnostic test for MI.
30‐day mortality (values calculated using reported data) BNP>80 pg/ml: Sensitivity 53%, specificity 91%, PPV 17%, NPV 98.2% (i.e. 1.8% chance of death despite ‐ve BNP); LR+ 5.9; LR‐ 0.5
New or worsening CHF at 30 days More frequent if BNP>80 pg/ml (8.7% v. 3.3%, p = 0.09).
New or worsening CHF at 30 days (calculated) BNP>80 pg/ml: Sensitivity 23.5%; Specificity 90%; PPV 8.7%; NPV 96.7% (i.e. probability of CHF despite ‐ve BNP 3.3%); LR+ 2.35; LR‐ 0.7.
Independent predictors of mortality following logistic regression BNP remained independently associated with mortality (OR 7.2, 95% CI 2.1–24.5, P = 0.001).
BNP as a predictor of successful myocardial reperfusion Elevated BNP associated with incomplete reperfusion (impaired flow, P = 0.04; poor myocardial perfusion, P = 0.06; Failed ST‐segment resolution, P = 0.005).
Bassan et al, 2005, Brazil 631 consecutive patients presenting to the Emergency Department with suspected cardiac chest pain <12 hours and no ST elevation on ECG. Blood taken on admission. Prospective observational cohort Association between BNP and diagnosis Significantly higher BNP in MI patients (P<0.0001). Increasing risk of MI with increasing BNP (divided into quartiles; P<0.0001). No follow‐up following hospital discharge. 30‐day and 6‐month follow‐up would be desirable.
Diagnostic accuracy of BNP for MI (optimal 100 pg/ml cut‐off) Area under ROC curve 0.710. Sensitivity 70.8%; Specificity 68.9%; PPV 22.7%; NPV 94.8%; LR+ 2.28; LR‐ 0.42. RR 4.38
Raised BNP, CK‐MB or TnI (on admission) for diagnosis of MI Sensitivity 87.3%; Specificity 65.7%; PPV 27.0%; NPV 97.3%; LR+ 2.55; LR‐ 0.19. RR 9.91
Follow‐up through in‐patient stay. Logistic regression model for association with MI BNP an independent predictor for diagnosis of MI (P = 0.0026). No firm diagnosis in 202 (32%) patients (MI excluded but not UA).
Clinical utility of BNP to rule in MI Authors' note: 72 patients had MI, 37 of whom had raised CK‐MB or TnI on admission. Raised BNP would have allowed detection of 22 more MI's [albeit at a cost of 163 false +ve diagnoses!]
Value of BNP in ruling out MI (calculated) For every 100 patients treated according to BNP levels on admission (cut‐off 100 pg/ml), 3 MI's would be missed.

AMI, acute myocardial infarction; CCU, coronary care unit; PTCA, per cutaneous coronary angioplasty; IV, intravenous; UA, unstable angina; BNP, brain natriuretic peptide; PCWP, pulmonary capillary wedge pressure; MI, myocardial infarction; SA, stable angina