Author + information
- Received June 22, 2005
- Revision received September 13, 2005
- Accepted September 19, 2005
- Published online February 7, 2006.
- Renate Schnabel, MD⁎,⁎ (, )
- Edith Lubos, MD⁎,
- Hans J. Rupprecht, MD⁎,
- Christine Espinola-Klein, MD⁎,
- Christoph Bickel, MD§,
- Karl J. Lackner, MD†,
- François Cambien, MD‡,
- Laurence Tiret, PhD‡,
- Thomas Münzel, MD, FAHA⁎ and
- Stefan Blankenberg, MD⁎
- ↵⁎Reprint requests and correspondence:
Dr. Renate Schnabel, Johannes Gutenberg-University, Cardiology, Langenbeckstrasse 1, Mainz, Rheinland-Pfalz 55131, Germany.
Objectives The aim of this study was to assess the predictive value of the cardiac hormone B-type natriuretic peptide (BNP) for long-term outcome in a large cohort of stable angina patients.
Background Recent data suggest a role of BNP in stable ischemic heart disease beyond its known value in heart failure and acute coronary syndromes.
Methods In 1,085 patients with coronary artery disease (CAD) baseline levels of BNP were prospectively associated with cardiovascular (CV) events during a mean follow-up of 2.5 years.
Results BNP concentrations were significantly elevated in patients with future CV events (median [25th/75th interquartile range] 119.2 [43.6/300.4] pg/ml vs. 36.2 [11.3/94.6] pg/ml; p < 0.001). Kaplan-Meier survival analysis showed a stepwise decrease in event-free survival across quartiles of BNP baseline concentration (plog rank< 0.001). Patients in the highest quartile revealed a 6.1-fold increased risk (p = 0.001) compared to patients in the lowest quartile after adjustment for potential confounders. For a cut-off value of 100 pg/ml, an independently increased risk of adverse outcome (hazard ratio [HR] 4.4; p < 0.001) could be demonstrated. One standard deviation (SD) decrease in ejection fraction implied the most prominent increase in risk of future CV events (HR 1.69; p < 0.001) followed by one SD increase in BNP (HR 1.53; p < 0.001). The highest prognostic accuracy could be demonstrated for BNP (area under the curve 0.671).
Conclusions The data of this large group of CAD patients provide independent evidence that BNP is a strong predictor of cardiovascular risk in patients with stable angina independent of left ventricular systolic performance and known risk factors.
The degree of cardiac neurohormonal activation as indicated by B-type natriuretic peptide (BNP) has been extensively investigated in cardiac disease, primarily in mechanistic disorders of cardiac function. As a hormone indicator of myocardial stretch (1) BNP is an excellent marker of heart failure and has already entered international guidelines. Its reliable characteristics as a biomarker have spurred further investigations in other cardiovascular disease entities. Indeed, there is an increasing body of evidence for the concept that BNP might be an indicator of hypoxia and ischemia itself which may result in myocyte stress under ischemic conditions despite constancy in measurable hemodynamic parameters (2–4). For patients with acute coronary syndromes, impressive data have been generated for BNP in the prediction of outcome. Under these conditions BNP provides information on survival and incident heart failure incremental to that of anthropometric data and clinical variables (5,6). For long-term mortality assessment BNP has proved to be superior to necrosis markers (7). Elevated BNP obviously does not depend on acute necrosis but may also indicate stable states of coronary artery disease (CAD) characterized by repetitive microischemia under stress without significant rise of creatine kinase or troponins (8).
There is also evidence that N-terminal B-type natriuretic peptide (Nt-proBNP), the inactive fragment of the pro-hormone, might also be a robust indicator of cardiovascular risk in patients with stable coronary disease (9–13).
The aim of the current study was to examine the prognostic impact of BNP in a large group of consecutively enrolled stable angina patients on short-term as well as long-term cardiovascular (CV) outcome to evaluate the potential clinical applicability of BNP measurements in CAD.
During the enrollment phase 1,085 consecutive patients presenting with stable angina and at least one stenosis >30% in the larger coronary arteries entered the observational AtheroGeneregistry at the two recruitment centers: Department of Medicine II of the Johannes Gutenberg University, Mainz, and the Bundeswehrzentralkrankenhaus, Koblenz, Germany.
Further details on the concept of the AtheroGenestudy have been previously described (14). In the present substudy, exclusion criteria were clinical signs of instability (unstable angina Braunwald classification class B or C, acute ST-segment elevation, and non–ST-segment elevation myocardial infarction). Further reasons for exclusion were clinical or echocardiographic signs of severe heart failure and additional known hemodynamically relevant cardiac abnormalities which might generate myocyte stress in addition to overt CAD, including severe valvular heart disease, surgery or trauma within the previous month, known cardiomyopathy, manifest carcinoma, chronic inflammatory disease states and febrile conditions, and use of oral anticoagulant therapy within the previous four weeks. These exclusion criteria were met by up to 30% of the eligible patients.
The history of classic risk factors was assessed as follows. Patients who had received antihypertensive treatment or who had received a diagnosis of hypertension (blood pressure above 160/90 mm Hg) were considered to have hypertension. Patients were classified as currently smoking, as having smoked in the past (if they had stopped more than 4 weeks and less than 40 years earlier), or as never having smoked (if they had never smoked or had stopped 40 or more years earlier). Patients who were receiving dietary treatment or medication for diabetes or who presented with fasting blood glucose levels above 125 mg/dl were defined as diabetic.
In 769 patients, left ventricular ejection fraction (EF) was determined by angiography and off-line analysis according to the area-length method (15).
Among the 1,085 patients, survival status remained unknown in 11 subjects (1%) who were lost during follow-up, and the data of 2 patients for whom EF was not available were excluded from analysis owing to a BNP concentration over 1,500 pg/ml, which might indicate severe heart failure. The final study population consisted of 1,072 individuals who were followed over a median of 2.5 ± 1.2 years (skewness <1). During this time, 35 cardiovascular deaths, 15 deaths from other causes, and 17 nonfatal myocardial infarctions were registered. The primary end point was nonfatal myocardial infarction and cardiovascular death. Follow-up information was obtained from patient charts and death certificates. All data were evaluated by an independent adjudication committee consisting of experienced physicians who were blinded to BNP concentrations.
The study was approved by the local ethics committee of the University of Mainz. All patients were Caucasian. Participation was voluntary, and patients were enrolled after written informed consent was obtained.
Blood samples were drawn under standardized conditions before coronary angiography was performed when the patients entered the catheterization lab after a minimum 12-h fast. Serum lipids and creatinine were measured immediately by routine methods; low-density lipoprotein was calculated by the Friedewald formula. For all other biomarkers measured in the study population, plasma and serum were stored at −80°C immediately after centrifugation.
Plasma B-type natriuretic peptide was determined using a fluorescence immunoassay (Biosite, San Diego, California). The detection limit reported is <5 pg/ml, the upper limit 5,000 pg/ml. The assay has an interassay coefficient of variation of near 10%, and a recovery of 100% of added peptide was found. Cross-reactivity with other natriuretic peptides is negligible (16). C-reactive protein (CRP) was determined by a highly sensitive, latex particle-enhanced immunoassay (detection range 0 to 20 mg/l; Roche Diagnostics, Mannheim, Germany).
The mean values and proportions of baseline cardiovascular risk factors, clinical variables, and biomarkers were calculated for study participants according to occurrence of future cardiovascular events. The statistical significance of differences between the means for the two subgroups was assessed with Student ttest, and the significance of differences in proportions was tested with the chi-squared statistic. Variables with a skewed distribution were presented as medians and Wilcoxon rank sum test was applied. Patients were divided into four subgroups on the basis of their BNP level at the time of enrollment. The cumulative event plots according to quartiles of BNP concentration were estimated by the Kaplan-Meier method and compared with use of the log rank test. All survival analyses were conducted for the primary end point of nonfatal myocardial infarction and cardiovascular death. Data from patients who died from causes not related to cardiovascular disease were censored at the time of death. Hazard ratios for future CV events were estimated by Cox regression models adjusted for potential confounders. All models included the adjustment for age and sex. Further adjustment was performed for classic risk factors (body mass index, high-density lipoprotein cholesterol levels, a history of hypertension, diabetes, and smoking). The second model further comprised clinical variables such as presence of multivessel disease and cardiac medication with angiotensin-converting enzyme inhibitors and statins and the inflammatory marker CRP. The final analyses included EF into the regression models. To compare the predictive power of BNP to known risk measures in stable angina patients, the hazard ratios (HR) found for one standard deviation increase of BNP, CRP, fibrinogen, and EF were presented for two Cox regression models. A backward stepwise Cox regression approach was taken for the multivariate analyses, with p = 0.10 as the critical value for entering and excluding variables in the model.
The HR and 95% confidence interval (CI) are reported. The p values are two-sided; a p value of less than 0.05 was considered to indicate statistical significance. All computations were carried out with SPSS software 11.5 (SPSS Inc., Chicago, Illinois).
The mean age of the study population was 61.6 ± 9.4 years, and 79.7% were male. During follow-up, 52 CV events were documented (17 nonfatal myocardial infarctions and 35 CV deaths). BNP levels exhibited a skewed distribution with a median (25th/75th interquartile range) of 38.3 (11.8/100.35) pg/ml. BNP concentration was significantly higher in patients who suffered from future CV events than in those who remained free of adverse events (119.2 [43.6/300.4] vs. 36.2 [11.3/94.6] pg/ml; p < 0.001). Baseline levels of BNP in patients who died from non-CV causes (n = 15) were not significantly different from those in event-free individuals (45.5 [5.0/171.4] vs. 36.1 [11.5/93.7] pg/ml; p = 0.83). For overall mortality (n = 50) it could be demonstrated that BNP concentrations were significantly higher in the CV event group (119.2 [20.3/315.3] vs. 36.31 [11.6/97.8] pg/ml in event-free subjects; p < 0.001). In Table 1the baseline characteristics of the study population according to the occurrence of future CV events are outlined. In the event group, the proportion of females was higher and diabetes was significantly more often seen. Left ventricular ejection fraction was lower in subjects with adverse outcome, although the mean value was within the normal range in both groups. Both groups did not relevantly differ in general hemodynamic parameters measured invasively at study entry. Of the inflammatory biomarkers determined, CRP and fibrinogen were higher in the event group, as expected (p < 0.001 for both). Age at baseline was not significantly different in both groups.
Figure 1shows the Kaplan-Meier survival curves according to quartiles of baseline BNP concentration. The unadjusted event-free survival decreased in a stepwise fashion across increasing quartiles (plog rank< 0.001). A threshold effect appeared for patients in the upper quartile, where a rapid decline in survival rate can be observed. To assess the independent strength of BNP for cardiovascular risk prediction in comparison with known cardiovascular risk factors, three Cox regression models were developed for log-transformed BNP concentration. The age- and gender-adjusted HR showed a 3.36-fold (95% CI 2.01 to 5.61; p < 0.001) increased risk. After adjustment for risk factors (HR 3.16, 95% CI 1.82 to 5.48; p < 0.001) and EF (HR 2.87, 95% CI 1.47 to 5.58; p = 0.002), which was available in 769 subjects, this association remained independently significant. These models were similarly applied to the quartile analysis of BNP (Table 2).For patients in the upper quartile of BNP baseline concentration, an 8.1-fold (p < 0.001) increased risk for future CV events could be observed in comparison with subjects in the first quartile. This predictive power was only slightly diminished after the adjustment for potential confounders such as classic risk factors, clinical variables, CRP, and even EF.
Owing to the observation of the fourth quartile border of 100.35 pg/ml, a cut-off value of 100 pg/ml was chosen and several Cox regression models were examined for patients who presented with BNP concentrations above this value at enrollment (Fig. 2).In comparison with patients whose baseline levels did not exceed this concentration, these subjects revealed a 4.4-fold (95% CI 2.4 to 7.8; p < 0.001) increased risk of adverse outcome. This association remained stable even after full adjustment for classic risk factors, clinical variables, CRP, and EF.
The 13 clinical and laboratory variables that predicted (p ≤ 0.1) univariate risk were entered into multivariate Cox regression analyses (Table 3).One SD increase of BNP (HR 1.4, 95% CI 1.1 to 1.7; p = 0.004) was selected as an independent predictor of cardiovascular events, together with one SD decrease in EF (HR 1.4, 95% CI 1.1 to 1.9; p = 0.012). The final model further included female gender (HR 0.5, 95% CI 0.2 to 1.0; p = 0.04) and statin therapy (HR 0.6, 95% CI 0.3 to 1.1; p = 0.08). These results did not differ significantly when CRP was also introduced into the initial model.
To further pose BNP into relation to known risk markers in patients with manifest CAD, the predictive power of one SD increase in BNP levels, CRP, and fibrinogen and one SD decrease in EF are illustrated in Figure 3.In this group of patients without overt systolic heart failure, the decrease in left ventricular function implied the most prominent increase in risk of future CV events (HR 1.69; p < 0.001), followed by BNP (HR 1.53; p < 0.001), fibrinogen (HR 1.49; p = 0.001), and CRP (HR 1.22; p = 0.002). Similar relations were found after adjustment for classic risk factors and clinical variables. The highest prognostic accuracy could be demonstrated for BNP (maximum area under the receiver operating characteristics curve of 0.671) and the lowest value for EF (0.623). When analyses were performed in the subgroup of patients (n = 769) for whom EF data were available, the results were similar.
This large-scale prospective study demonstrates the high prognostic value of BNP for long-term outcome in stable angina patients in a Caucasian cohort of consecutive CAD subjects independent of left ventricular systolic function and inflammatory biomarkers.
Evidence is accumulating that BNP may serve as an indicator of ischemic disease. Apart from its strong diagnostic and prognostic characteristics in patients with left ventricular dysfunction (11,17), ischemia seems to be a trigger of BNP expression and release (2,4). In acute ischemia the magnitude of BNP increase mirrors the extent of jeopardized myocardium and the tendency toward adverse remodeling more closely than any other current marker. It has been shown that BNP concentration provides information independent of necrosis markers and more adequately reflects myocardial territory at risk under acute ischemia. Similar conditions seem to apply for reversible ischemia under stress testing, because the proportion of ischemic territory without overt necrosis is consistently reflected in BNP levels (18).
However, generalization has to be considered with caution, because BNP is not specific enough as a measure of ischemia. In community-based cohorts, a prospective association between BNP levels and a range of CV outcomes and overall mortality has been reported in asymptomatic individuals with BNP thresholds well below those used to diagnose heart failure (19). In the primary prevention screening it will be difficult to provide therapeutically clear directives facing the multitude of causes leading to a moderate increase in BNP. Under the precondition of diagnosed CAD the situation might be different.
First evidence thus suggests an advantageous role of BNP in secondary risk stratification. BNP seems to reliably diagnose those patients with adverse outcome with a four-fold increased risk for individuals with BNP levels elevated above a potential cut-off value of 100 pg/ml. Although median BNP concentrations in the general CAD population were found to be far below those observed in acute heart failure (20), the concentration to rule out heart failure might get a new dimension, because stable angina patients who present with BNP above this level were at high risk to experience an adverse cardiovascular outcome during follow-up and should attract special attention.
Further investigations have to be undertaken to standardize Nt-proBNP and BNP data, determine cut-off values, and derive therapeutic implications.
The value of additional BNP determination becomes obvious in comparison with anthropometric and metabolic variables. B-type natriuretic peptide retained its strong and independent predictive power after adjustment for age, gender, and renal function, which are associated with BNP concentrations and are potent risk factors themselves (21–23). The data from our cohort of consecutive unselected stable angina pectoris patients show that routine applicability seems to be practicable without loss of validity.
Importantly, the risk information provided by BNP proved to be independent of angiographically defined severity of CAD and cardiac functional parameters, which further underlines the value of additional BNP determination. Backward regression analysis revealed BNP and EF as the strongest predictors of future CV events, which demonstrates that risk information gained by BNP determination is complementary and obviously incremental to systolic left ventricular performance (6).
As a limitation it must be mentioned that for this cohort only single measurements at enrollment can be investigated. Serial measurements of BNP concentrations would be of great interest for elucidating effectiveness of therapeutic strategies and evaluating whether initially determined cardiac risk can be modified by therapeutic interventions.
Furthermore, as expected, only a small number of events were registered during follow-up in this intermediate-risk population, which reduces statistical power. Despite the relatively low event rate, BNP provided strong results after adjustment for known risk predictors, underlining the strength of this novel biomarker.
The present data support the view that BNP might develop into a biomarker for “many cardiac seasons” (24). Responsible for the fine tuning of cardiac homeostasis, it is involved in a variety of cardiac abnormalities. Until its pathophysiologic background is entirely understood, an increasing number of novel clinical indications and benefits can be expected. The results of the present study, taken together with recent data on Nt-proBNP, provide a promising basis for further investigations into risk stratification in stable CAD where BNP has been shown to improve prognosis assessment in this intermediate-risk group independent of and incremental to classical risk factors.
The test assay for BNP was kindly performed by Biosite, San Diego, California. We are indebted to Margot Neuser for her graphical work.
- Abbreviations and Acronyms
- B-type natriuretic peptide
- coronary artery disease
- C-reactive protein
- left ventricular ejection fraction
- hazard ratio
- Received June 22, 2005.
- Revision received September 13, 2005.
- Accepted September 19, 2005.
- American College of Cardiology Foundation
- Toth M.,
- Vuorinen K.H.,
- Vuolteenaho O.,
- et al.
- Goetze J.P.,
- Christoffersen C.,
- Perko M.,
- et al.
- Jernberg T.,
- Stridsberg M.,
- Venge P.,
- Lindahl B.
- James S.K.,
- Lindahl B.,
- Siegbahn A.,
- et al.
- Schnabel R.,
- Rupprecht H.J.,
- Lackner K.J.,
- et al.
- Richards A.M.,
- Doughty R.,
- Nicholls M.G.,
- et al.,
- Australia-New Zealand Heart Failure Group
- Richards A.M.,
- Doughty R.,
- Nicholls M.G.,
- et al.,
- Australia-New Zealand Heart Failure Group
- Talwar S.,
- Squire I.B.,
- Davies J.E.,
- Barnett D.B.,
- Ng L.L.
- Yerkey M.W.,
- Kernis S.J.,
- Franklin B.A.,
- Sandberg K.R.,
- McCullough P.A.