Author + information
- Received February 26, 2002
- Revision received April 8, 2002
- Accepted April 17, 2002
- Published online August 7, 2002.
- Tomas Jernberg, MD, PhD*,* (, )
- Mats Stridsberg, MD, PhD†,
- Per Venge, MD, PhD† and
- Bertil Lindahl, MD, PhD*
- ↵*Reprint requests and correspondence:
Dr. Tomas Jernberg, Department of Cardiology, Cardiothoracic Center, University Hospital, 751 85 Uppsala, Sweden.
Objectives The study evaluated the prognostic value of single measurement of N-terminal pro brain natriuretic peptide (NT-proBNP) obtained on admission in patients with symptoms suggestive of an acute coronary syndrome and no ST-segment elevation.
Background Patients with symptoms suggestive of an acute coronary syndrome and no ST-segment elevation constitute a large and heterogeneous population. Early risk stratification has been based on clinical background factors, electrocardiography (ECG) and biochemical markers of myocardial damage. The neurohormonal activation has, so far, received less attention.
Methods The NT-proBNP was analyzed on admission in 755 patients admitted because of chest pain and no ST-segment elevation. Patients were followed concerning death for 40 months (median).
Results The median NT-proBNP level was 400 (111 to 1646) ng/l. Compared to the lowest quartile, patients in the second, third and fourth quartiles had a relative risk of subsequent death of 4.2 (1.6 to 11.1), 10.7 (4.2 to 26.8) and 26.6 (10.8 to 65.5), respectively. When NT-proBNP was added to a Cox regression model including clinical background factors, ECG and troponin T, the NT-proBNP levels were independently associated with prognosis.
Conclusions A single measurement of NT-proBNP on admission will substantially improve the early risk stratification of patients with symptoms suggestive of an acute coronary syndrome and no ST-segment elevation. A combination of clinical background factors, ECG, troponin T and NT-proBNP obtained on admission will provide a highly discerning tool for risk stratification and further clinical decisions.
Patients presenting with chest pain or other symptoms suggestive of an acute coronary syndrome amount today to about 20% of all visits to the medical emergency department (1). Of the two-thirds who will be admitted, about 90% will have an electrocardiogram (ECG) nondiagnostic of acute myocardial infarction (AMI) and, thereby, constitute a heterogeneous group concerning both the underlying pathophysiology and future risk of cardiac events (2). An early risk stratification of these patients is important for several reasons. Those identified as being high-risk patients might need a more intense pharmacologic treatment and be considered for intervention early (3). Patients with a low risk, in contrast, may benefit more from conservative management with a low risk of side effects. Moreover, considerable economic gains may be achieved by early identification of patients who are at sufficiently low risk for early transfer to a lower level of care and early discharge.
Brain natriuretic peptide (BNP) is a circulating cardiac hormone that is released mainly from the ventricles in response to increased stretch or wall tension (4). It is broadly involved in the regulation of blood pressure, blood volume and sodium balance; it has in several studies been shown useful in detecting left ventricular (LV) dysfunction after AMI and to be related to poor outcome (5–8). Recently, it was shown that BNP, obtained in the first few days after the onset of symptoms, also provided important prognostic information in patients with non–ST-segment elevation AMI or unstable angina pectoris (9). The BNP is produced as a prohormone, proBNP, which upon secretion is split into BNP and N-terminal pro brain natriuretic peptide (NT-proBNP). A close correlation exists between these peptides. However, in patients with LV dysfunction, the proportional and absolute increase of NT-proBNP exceeds that of BNP, suggesting that NT-proBNP may be a more discerning marker of LV dysfunction (10).
The aim of the present study was therefore to evaluate the prognostic value of NT-proBNP analyzed on admission in patients admitted to the coronary care unit because of chest pain and an ECG nondiagnostic of AMI.
Patients admitted to the coronary care unit at the Uppsala University Hospital were eligible for participation between March 1997 and February 1998. The inclusion criteria were a history of chest pain or other symptoms suggestive of an acute coronary syndrome. Exclusion criteria were prehospital thrombolysis, presence of ST-segment elevation on admission ECG and previous enrollment in the study. After a run-in period of two months, consecutive patients were included.
Of the 1,194 eligible patients, 131 were excluded due to prehospital thrombolysis or ST-segment elevation on admission ECG and 170 because of previous enrollment in the study. Frozen plasma samples from 118 randomly selected patients were used for other purposes. Thus, 775 patients had an admission-sample available for analysis. Another sample at 6 h from admission was available in 650 patients.
For the diagnosis of AMI at the index event, one of the following should be fulfilled: 1) pathologic Q-waves developing in at least two leads; 2) symptoms suggestive of AMI and typical elevated levels of biochemical markers with creatine kinase-MB fraction (mass) ≥10 μg/l; and 3) signs of AMI at autopsy. All clinical data were prospectively collected and entered into the local database of the Swedish Register of Cardiac Intensive Care (11). The treatment of individual patients was left to the discretion of the individual physician. The study was approved by the local ethics committee.
A 12-lead ECG was obtained on admission. Patients were divided into four mutually exclusive groups: 1) those with normal ECG (absence of pathological Q-waves, bundle branch block, signs of left ventricular hypertrophy, ST-segment depression, T-wave inversion, nonsinus rhythm and tachycardia [≥100 beats/min]); 2) pathologic changes on ECG other than ST-segment depression or left bundle branch block (LBBB); 3) ST-segment depression ≥0.05 mV; and 4) LBBB.
Blood samples were collected in ethylenediamine-tetraacetic acid-containing tubes. The samples were then centrifuged, and plasma was stored frozen in aliquots at −70°C within 30 min. Plasma cardiac troponin T (cTnT) was later determined by the third-generation cTnT assay on an Elecsys 2010 (Roche Diagnostics, Mannheim, Germany). The analytical range extends from 0.01 to 25 μg/l, and the upper reference level, defined as the 99th percentile of healthy controls, is below the lower limit of the analytical range. At the Department of Clinical Chemistry at the Uppsala Hospital, the intra-assay CV was 7.9% (n = 19) and the interassay was 11.2% (n = 8), at the low end of the analytical range (cTnT = 0.037 μg/l). Plasma NT-proBNP was determined using Elecsys proBNP sandwich immunoassay on an Elecsys 2010. The analytical range extends from 20 to 35,000 ng/l. At our laboratory the total CV was 3.3% (n = 21) at a level of 209 ng/l and 3.0% (n = 21) at a level of 7,431 ng/l. The P-creatinine was assayed at the Department of Clinical Chemistry on a Hitachi 911 machine (Roche Diagnostics).
The end point was death. All in-hospital events were registered in the local database of the Swedish Register of Cardiac Intensive Care (11). Out of hospital, information about death was obtained from the National Registry on Mortality. The median follow-up time was 40 months (3.3 years), ranging from 35 to 47 months.
All data analysis was performed using the Statistical Package for Social Sciences (SPSS 10.1) software (SPSS Inc., Chicago, Illinois). Differences in proportions were judged by chi-square analysis. If not stated otherwise, continuous data are given as median-value (25th to 75th percentile). For continuous data in two independent groups, the Mann-Whitney U test was used. To compare paired continuous data, the Wilcoxon matched-pairs signed rank-sum test was used. The Kruskal-Wallis one-way analysis of variance was used to test the equality of distributions in the four NT-proBNP groups. A significant difference was considered to exist with p < 0.05. To evaluate the correlation between the level of NT-proBNP and cTnT, the Spearman rank-correlation coefficient was calculated. The Kaplan-Meier method was employed to analyze the timing of events during follow-up. Statistical assessment was performed using the log-rank test with p < 0.05 considered as significant.
To compare the predictive value of NT-proBNP measured on admission and that of NT-proBNP measured at 6 h, receiver operating characteristic (ROC) curves were generated, and the area under the curves calculated. This approach was also used to investigate the prognostic value of the relative change of NT-proBNP, defined as (NT-proBNP at 6 h − NT-proBNP on admission)/NT-proBNP on admission. For the ROC analysis, death at 35 months was used as the end point. To identify predictors of death, the univariate Cox regression analyses were used. All variables with a p < 0.10 were then tested in a multivariate Cox regression analysis using forward stepwise selection. Variables were entered if p < 0.05 and removed if p > 0.10.
The baseline characteristics, findings on admission ECG and the diagnoses within 24 h of admission are listed in Table 1. Of 407 patients with AMI or unstable angina, 68 (17%) and 95 (23%) patients were revascularized in hospital and 30 days after admission, respectively. During follow-up, 172 (22%) deaths occurred.
ECG and troponin T
On admission, 242 (31%) patients had a normal ECG, 252 (33%) ST-segment depression ≥0.05 mV, 62 (8%) LBBB, and 219 (28%) other pathologic ECG changes. Compared to those with normal ECG, patients with other pathologic ECG changes had a relative risk (RR) of subsequent death of 4.24 (95% confidence interval [CI]: 2.23 to 8.06), whereas those with ST-segment depression ≥0.05 mV and LBBB had an RR of 8.17 (95% CI: 4.46 to 15.0) and 13.4 (95% CI: 6.83 to 26.1), respectively. There were 247 (32%) and 271 (42%) patients with at least one sample of cTnT ≥0.01 μg/l on admission and after 6 h, respectively. The median cTnT level on admission was significantly higher in patients who died during follow-up than in those who survived (0.02 vs. <0.01 μg/l, p < 0.001). Patients with a cTnT ≥0.01 μg/l on admission had a 3.2-fold increased risk of death compared to those without.
The median value of NT-proBNP on admission was 400 (111 to 1,646) ng/l. A total of 650 patients had NT-proBNP analyzed also at 6 h, with a median value of 525 (122 to 2,228) ng/l. The relationship between NT-proBNP level on admission and clinical background factors is shown in Table 1. Patients with a diagnosis of AMI or other cardiac causes had the highest levels of NT-proBNP, whereas those with noncardiac or unknown causes had the lowest levels (Table 2). At 6 h, there was a significant increase of NT-proBNP in all groups, with the largest increase in the group with AMI. The group with a cTnT <0.01 μg/l on admission had a median level of NT-proBNP of 204 (72 to 738) ng/l. In patients with a cTnT ≥0.01 μg/l, the median level was 1,639 (491 to 5,382). In the latter group, a moderate correlation existed between the level of cTnT and that of NT-proBNP (rho 0.49, p = 0.01).
Predictive value of NT-proBNP
The risk of death during follow-up increased with increasing levels of NT-proBNP on admission (Fig. 1). Compared to the lowest quartile, patients in the second, third, and fourth quartiles had an RR (95% CI) of subsequent death of 4.2 (1.6 to 11.1), 10.7 (4.2 to 26.8) and 26.6 (10.8 to 65.5), respectively. Patients with elevated NT-proBNP had an increased risk irrespective of diagnosis (Fig. 2). When the predictive value of NT-proBNP measured on admission and the value of NT-proBNP measured at 6 h were compared in an ROC analysis, the areas under the curves were almost identical (Fig. 3). As compared to patients who survived, patients who died had a larger increase of NT-proBNP between admission and at 6 h (median: 334 vs. 26 ng/l, p < 0.001). The relative change did not differ significantly between the groups (19% vs. 16%, p = 0.07) and carried no prognostic information according to the ROC analysis (Fig. 3). The RR (95% CI) of subsequent death was 1.00 (0.56 to 1.8) and 1.28 (0.75 to 2.18) in patients with unchanged (±10%) and with increasing NT-proBNP levels between admission and 6 h, respectively, compared to patients with decreasing levels.
In a Cox regression model, increasing age, diabetes, ECG changes, elevated cTnT and P-creatinine proved to be independent predictors of mortality (Table 3, model 1). When proBNP measured on admission was added, NT-proBNP levels, but not ECG changes, were independently associated with prognosis (Table 3, model 2). When time of delay from start of symptoms to admission and diagnosis was forced into the final model, the association between NT-proBNP levels and prognosis remained unchanged. The predictive value of NT-proBNP was apparent both in patients with and without an elevated cTnT and in patients with different risks according to the ECG on admission (Figs. 4 and 5). ⇓⇓
The aim was to study a nonselected population, similar to the one dealt with in clinical practice. Accordingly, consecutive patients admitted to the coronary care unit were included with no requirements of ECG changes on admission or raised biochemical markers of myocardial damage, excluding only patients with definite diagnosis based on ST-segment elevation. This was reflected by a fairly low incidence of cTnT ≥0.01 μg/l (32%) and by the fact that patients were older, with a higher mortality compared to studies that included a more selected population (9,12). Patients were enrolled from March 1997 to February 1998, during which a conservative approach, one with few early revascularizations, was still in use in our institution.
Predictors of outcome
In accordance with previous studies (12–14), several clinical background factors were associated with poor outcome. In a multivariate analysis, increasing age, diabetes, previous AMI, elevated P-creatinine, ECG changes and elevated cTnT were independent predictors of mortality. When adjusted for baseline characteristics and the level of cTnT, only a small difference was seen between the predictive value of ST-segment depression and that of other pathologic signs. This may indicate that, in patients without ST-segment elevation or LBBB, the absence of a normal ECG, rather than the presence of ST-segment depression, is the main determinant of outcome. However, both ECG and cTnT were probably underestimated as predictors in the present study. Continuous monitoring and repeated measurements have been shown to improve the prognostic performance of ECG and cTnT, respectively (15,16).
The present study is the first one to demonstrate that a single measurement of NT-proBNP, obtained on admission, provides important prognostic information in the large population of patients with symptoms suggestive of an acute coronary syndrome but with an ECG without ST-segment elevation. When patients were divided according to diagnosis into those with AMI, unstable angina, other cardiac causes and other noncardiac or unknown causes, the prognostic value of NT-proBNP was evident in all groups. Moreover, when the level of NT-proBNP was tested in a multivariate analysis including well-known predictors of outcome, such as age, diabetes, history of previous myocardial infarction, congestive heart failure, ECG changes and cTnT, the NT-proBNP was still an important independent predictor of outcome.
Natriuretic peptides have been shown to be useful for detection of LV dysfunction and to be related to poor outcome in patients with AMI (5–8,17–19). However, there are still few studies that have examined this area, including patients with unstable angina or non-ST-segment elevation AMI. In a recent study by de Lemos et al. (9), BNP was measured in a cohort including 1,698 patients with unstable angina or non-ST-segment elevation AMI. In that study, the 10-month mortality increased with increasing levels of BNP, with a mortality of approximately 1% in the lowest quartile and 7% to 15% in the highest quartile. In another study (20), our group measured N-terminal pro atrial natriuretic peptide (NT-proANP) in 104 patients with unstable angina or non-ST-segment elevation AMI. In that study, patients with elevated NT-proANP had a fivefold increased risk of death during a median follow-up time of 48 months. Further support has been provided by Omland et al. (21) reporting higher levels of both NT-proANP and NT-proBNP in patients who died after an episode of acute coronary syndrome compared to a group of matched controls. In all three studies, an elevated level of the natriuretic peptide was an independent predictor of mortality when adjusted for clinical background factors, ECG changes and levels of troponin. However, in contrast to the present study, the two former studies measured natriuretic peptide approximately two days after the index event.
Furthermore, NT-proBNP and BNP have been suggested to be better indicators of LV dysfunction and poor outcome than NT-proANP and ANP (7,8,19). Although NT-proBNP and BNP are highly correlated, the proportional and absolute increase above normal levels of NT-proBNP exceeds that for BNP in patients with LV dysfunction (10). This may indicate that NT-proBNP is a more discerning marker of cardiac dysfunction and poor outcome.
One measurement of NT-proBNP on admission identified a low-risk group with cumulative probability of death of 3%, and a high-risk group with cumulative probability of death of 55%. In a majority of patients, the level of NT-proBNP increased at 6 h from admission, with the highest increase in patients with AMI. However, when the predictive values of NT-proBNP measured on admission and that of NT-proBNP measured at 6 h were compared in an ROC analysis, the areas under the curves were almost identical. Furthermore, the relative change of NT-proBNP did not carry any prognostic information. This may indicate that the incremental value of serial measurements is limited.
At least two possible mechanisms can explain the relation between NT-proBNP level and prognosis in the present population. First, an elevated NT-proBNP may reflect a permanent LV dysfunction established prior to or during the current episode of instability, which is an important predictor of outcome in patients with acute coronary syndromes (22). Second, it can also reflect a temporary LV dysfunction secondary to transient ischemic episodes jeopardizing a large part of the myocardium. Previous studies have demonstrated elevation of BNP shortly after percutaneous coronary intervention (23). Thus, an elevated NT-proBNP does not necessarily reflect a neurohormonal activation or cell leakage in response to myocardial necrosis. The fact that the event curves continued to separate up to four years supports the theory that raised levels of NT-proBNP reflect LV dysfunction or a large area of myocardium at jeopardy.
To be useful, a clinical risk indicator should not only identify patients with increased risk, but also patients who benefit from a certain treatment, and thereby be able to guide treatment. Troponin and ECG changes have previously been shown to identify patients with a high risk of subsequent cardiac events and who benefit most from intensive antithrombotic treatment and early revascularization (24–27). A previous study (28) showed that treatment of heart failure guided by NT-proBNP concentrations resulted in a better outcome than treatment guided by clinical assessment. It is therefore tempting to suggest that patients with elevated NT-proBNP will benefit from an early aggressive treatment with neurohormonal antagonism (such as angiotensin-converting enzyme inhibitors and beta-blockade) and early revascularization in addition to an intense antithrombotic treatment. However, this needs to be evaluated in future studies. Conversely, patients with a low level of NT-proBNP on admission and a low risk of subsequent events will probably benefit most from a conservative management with a low risk of side effects.
There are some limitations with the present study. There was no systematic evaluation of LV function. It is generally acknowledged that LV function is one of the most important predictors of outcome in patients with acute coronary syndrome (22). Previous studies have suggested BNP and NT-proBNP to be better predictors of outcome (7,19). Therefore, it would have been interesting to compare the LV ejection fraction determined by echocardiography with measurement of NT-proBNP in the present population. Another limitation is that NT-proBNP was only measured during the first 6 h from admission. In patients with AMI, NT-proBNP continues to increase during the first 24 h (29,30). Thus, although the level of NT-proBNP at 6 h from admission did not add any prognostic information, measurement later in the course of instability might still be of additive value.
The early risk stratification of patients with symptoms suggestive of an acute coronary syndrome and an ECG without ST-segment elevation has been based on clinical background factors, the acute clinical presentation, ECG measurements and biochemical markers of myocardial damage. The neurohormonal activation has, so far, received less attention. This study shows that a single measurement of NT-proBNP on admission will substantially improve this process, and that a combination of clinical background factors, ECG, troponin T and NT-proBNP obtained on admission will provide a highly discerning tool for risk stratification and further clinical decisions.
☆ This study was supported by grants from the Swedish Heart and Lung Foundation, the Uppsala County Association Against Heart and Lung Diseases and Roche Diagnostics. Gottlieb C. Friesinger, II, MD, acted as the Guest Editor for this paper.
- acute myocardial infarction
- brain natriuretic peptide
- confidence interval
- cardiac troponin T
- left bundle branch block
- left ventricular
- N-terminal pro atrial natriuretic peptide
- N-terminal pro brain natriuretic peptide
- receiver operating characteristic
- relative risk
- Received February 26, 2002.
- Revision received April 8, 2002.
- Accepted April 17, 2002.
- American College of Cardiology Foundation
- ↵The Swedish Register of Cardiac Intensive Care, Annual Report 2000
- Choy A.M.,
- Darbar D.,
- Lang C.C.,
- et al.
- Omland T.,
- Aakvaag A.,
- Bonarjee V.V.,
- et al.
- PURSUIT investigators,
- Boersma E.,
- Pieper K.S.,
- Steyerberg E.W.,
- et al.
- Jernberg T.,
- Lindahl B.,
- Wallentin L.
- GUSTO-IIa investigators,
- Newby L.K.,
- Christenson R.H.,
- Ohman E.M.,
- et al.
- TIMI II investigators,
- Hall C.,
- Cannon C.P.,
- Forman S.,
- Braunwald E.
- Hall C.,
- Rouleau J.L.,
- Moyé L.,
- et al.
- Richards A.M.,
- Nicholls M.G.,
- Yandle T.G.,
- et al.
- St John Sutton M.,
- Pfeffer M.A.,
- Plappert T.,
- et al.
- Lindahl B.,
- Venge P.,
- Wallentin L.
- TACTICS-TIMI 18 investigators,
- Cannon C.P.,
- Weintraub W.S.,
- Demopoulos L.A.,
- et al.
- Morita E.,
- Yasue H.,
- Yoshimura M.,
- et al.