Distribution of N-terminal pro–B-natriuretic peptide (NT-proBNP) in quartiles and proportion of patients (N = 1051) without secondary fatal and nonfatal cardiovascular disease (CVD) events during follow-up.
Rothenbacher D, Koenig W, Brenner H. Comparison of N-Terminal Pro–B-Natriuretic Peptide, C-Reactive Protein, and Creatinine Clearance for Prognosis in Patients With Known Coronary Heart Disease. Arch Intern Med. 2006;166(22):2455-2460. doi:10.1001/archinte.166.22.2455
The purpose of this study was to investigate the prognostic role of N-terminal pro–B-natriuretic peptide (NT-proBNP) serum level compared with C-reactive protein (CRP) level and creatinine clearance (CrCl) for the subsequent risk of cardiovascular events in a large cohort of patients with stable coronary heart disease (CHD).
Serum concentrations of NT-proBNP and CRP and CrCl were measured at baseline in a cohort of 1051 patients aged 30 to 70 years with CHD. The Cox proportional hazards model was used to determine the prognostic value of NT-proBNP, CRP, and CrCl on a combined cardiovascular disease (CVD) end point (fatal and nonfatal myocardial infarction and stroke).
During follow-up (mean of 48.7 months), 95 patients (9.0%) experienced a secondary CVD event. Patients in the top quartile of the NT-proBNP distribution at baseline had a hazard ratio (HR) of 3.34 (95% confidence interval [CI], 1.74-6.45) for subsequent secondary CVD events compared with those in the bottom quartile (P for trend <.001) after controlling for age, sex, smoking status, history of diabetes mellitus, initial management of CHD, rehabilitation clinic, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, and treatment with lipid-lowering drugs. For CRP, the corresponding HR was 1.76 (95% CI, 0.96-3.24) (P value for trend, .06). Patients with CrCl levels lower than 60 mL/min had an HR of 2.39 (95% CI, 1.06-5.40) compared with patients with a CrCl of 90 mL/min or higher (P for trend, .002). If all 3 markers were included simultaneously in 1 model, NT-proBNP still showed predictive ability for recurrent CVD events.
N-terminal proBNP may be a clinically useful marker weeks after an acute coronary event and may provide complementary prognostic information to established risk determinants.
Prevalence of coronary heart disease (CHD) will increase within the next years, and CHD will be the number 1 cause of disability and death by the year 2020 worldwide.1 Therefore, beside new avenues and concepts of primary prevention, improved secondary preventive strategies of CHD are necessary to reduce the burden of disease and recurrent complications for the individual as well as for society.
The use of biological markers for diagnostic purposes and risk stratification represents one promising area in cardiology. Meanwhile convincing evidence suggests that CHD is an inflammatory process, and a variety of inflammatory markers such as C-reactive protein (CRP) may play a role in disease risk assessment.2 In addition, renal impairment is also an independent risk factor for CHD.3
B-type natriuretic peptide (BNP) and the N-amino terminal (NT)-proBNP (the latter with a much longer half-life than BNP4) have been initially introduced as markers to aid in the diagnosis of congestive heart failure (CHF), to estimate the severity of cardiac dysfunction, and to guide the management and adjustment of pharmacotherapy.4- 7 In addition, evidence is accumulating that raised BNP and NT-proBNP plasma/serum levels are of prognostic value because they are independently associated with adverse prognosis in several cardiac disorders, including CHF and acute CHD,5,8,9 and they may allow identifying subjects at risk for adverse cardiovascular disease (CVD) events to tailor therapy.10 Recent studies also demonstrated their prognostic value in predicting all-cause mortality in patients with prevalent CHD.11- 13 Whether elevated NT-proBNP serum levels in patients with CHD reflect a specific pathomechanism of cardiac disease, which is independent from other basic concepts of atherogenesis such as vascular inflammation and renal disease, is under debate.
In the present study, we investigated the prognostic role of NT-proBNP serum levels for the subsequent risk of cardiovascular events in a large cohort of patients with stable CHD weeks after the first clinical manifestation or diagnosis, after controlling for potential confounders, and compared it with the prognostic value of a serum marker of systemic inflammation (CRP) and renal function (creatinine clearance [CrCl]).
All patients with CHD (International Classification of Diseases, Ninth Revision [ICD-9] codes 410-414) aged 30 to 70 years and participating in an in-hospital rehabilitation program between January 1999 and May 2000 in 2 cooperating clinics (Schwabenland-Klinik, Isny, and Klinik am Südpark, Bad Nauheim, Germany) were enrolled in the study, and clinically relevant CHD was defined by a greater than 50% diameter stenosis of 1 or more major epicardial coronary arteries. In Germany, every post–acute coronary syndrome patient by law has the right to receive institutionalized cardiac rehabilitation in a specialized center, which is used by most patients. This in-hospital rehabilitation program after acute coronary syndrome usually begins within the first weeks after discharge from the acute care hospital. Although we recruited patients only in 2 such in-patient rehabilitation centers (one in the south of Germany and the other in the middle of Germany), these specialized centers serve a large geographic area with a radius of up to 200 km, and therefore, patients were referred from a large number of different acute care hospitals. In the present study, only patients who were admitted within 3 months after the acute event or coronary artery bypass grafting (CABG) were included. Of all eligible patients, 58% agreed to participate. The mean interval from the acute event to recruitment was 43 days (interquartile range, 36-51 days).
All subjects gave written informed consent. The study was approved by the ethics boards of the Universities of Ulm and Heidelberg and of the physicians' chamber of the States of Baden-Württemberg and Hessen (Germany).
At the beginning of the in-hospital rehabilitation program, all participants filled out a standardized questionnaire containing sociodemographic information and medical history. In addition, information was taken from the patients' hospital medical charts, which also included information from the acute care hospital. From all patients, a 12-lead electrocardiogram (ECG) was recorded at the beginning of the rehabilitation and evaluated according to a standardized protocol by a trained investigator. Intrarater (interrater) reliability was determined in 100 randomly chosen ECGs by a blinded second rating from the same investigator 4 weeks later and from an independent second investigator. The κ coefficients were 1 (0.88) for sinus tachycardia (>100/min), 0.85 (1.0) for atrial flutter or fibrillation, and 0.96 (0.73) for infarction location.
Information on left ventricular function (LVF) was derived from (1) the most recent left ventricular cineangiography (available in 680 patients [67%]) or (2) from echocardiographic study when available (available in 821 patients [78%]). This way, LVF could be assessed in 964 patients (91.7%). Data on LVF were collected from patients' medical charts and were documented on a 4-point semiquantitative scale as normal (ejection fraction [EF], >65%), as mild depression (EF, 50%-65%), moderate depression (EF, 35%-50%), or severe depression (EF, <35%).
In all patients, active follow-up was conducted 1, 3, and 4.5 years after discharge from the rehabilitation center. Information regarding secondary cardiovascular events and treatment since discharge was obtained from the primary care physician by means of a standardized questionnaire. If a patient had died during follow-up, the death certificate was obtained from the local public health department, and the main cause of death was coded according to the ICD-9. Secondary cardiovascular events were defined either as CVD as the main cause of death (as stated in the death certificate), nonfatal myocardial infarction (MI), or ischemic stroke. All nonfatal secondary events were reported by the primary care physicians.
Blood was drawn at baseline at the end of the rehabilitation phase in a fasting state under standardized conditions and stored at −80°C until analysis. N-terminal proBNP was measured from plasma by means of a 1-step enzyme immunoassay based on electrochemiluminescence (Elecsys; Roche Diagnostics, Mannheim, Germany). The interassay coefficient was between 3.1% and 7.2%. C-reactive protein was determined by a high-sensitivity assay (N Latex CRP mono; Dade Behring, Marburg). For measurement of creatinine, in one hospital the conventional kinetic Jaffe method (interassay coefficient of variation, 2.4%-5.7%) was used, whereas in the other hospital measurement was done by an enzymatic creatinine p-aminophenazone method (interassay coefficient of variation, 1.2%-2.2%). Creatinine clearance was calculated according the Cockcroft-Gault formula.14 Serum lipid measurements were performed by routine methods in both participating clinics. All markers were measured in a blinded fashion.
Median levels of NT-proBNP were calculated according to levels of various sociodemographic characteristics, various cardiovascular risk factors, and levels of ECG findings. Associations of these factors with the NT-proBNP distribution were tested for statistical significance by the nonparametric Kruskal-Wallis test. A (partial) Spearman rank correlation coefficient between NT-proBNP, CRP, and CrCl was calculated after adjustment for age and sex.
The relation of NT-proBNP, CRP, and CrCl levels with CVD events during follow-up was assessed by the Kaplan-Meier method. The Cox proportional hazards model was used to assess the independent association of NT-proBNP distribution with the risk of secondary CVD events. Besides a model adjusting for age and sex, the following potential confounders were considered in multivariable analyses: age, sex, body mass index (BMI [calculated as weight in kilograms divided by height in meters squared]), smoking status, duration of school education, family status, history of MI, history of hypertension, history of diabetes mellitus, severity of CHD, intake of β-blockers, intake of angiotensin-converting enzyme inhibitors, intake of diuretics, intake of lipid-lowering drugs, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, atrial flutter or fibrillation, left ventricular hypertrophy, anterior infarction location, posterior infarction location, and hospital site. To avoid overadjustment, the latter covariates were added only if they were significant predictors of a secondary event at an α level of .10 or if their inclusion changed the parameter estimates for the main variables (NT-proBNP) by more than 10%.
To compare their prognostic value with that of NT-proBNP, CRP and CrCl levels were included instead of NT-proBNP in additional analyses in the adjusted model (model 2), and, finally, NT-proBNP, CRP, and CrCl categories were included simultaneously (model 3). Finally, hazard ratios (HRs) were additionally adjusted for LVF (no or only little impairment, modest or severe impairment) (model 4).
A receiver operating curve was constructed after adjustment for covariates, and the area under the curve with its 95% confidence interval (CI) was calculated. In addition, Somers D, a measure of association that provides a rank correlation between predicted and observed probabilities, was calculated for the various models. Somers D ranges between −1 and +1; 0 reflects no association at all. All statistical procedures were carried out with the SAS statistical software package (release 8.2 ; SAS Institute Inc, Cary, NC).
Overall, 1206 patients with a diagnosis of CHD within the past 3 months were included at baseline during the in-hospital rehabilitation program. Four-year follow-up information was complete for 1051 patients (87.2%).
Table 1 gives the distribution (median) of NT-proBNP values according to various sociodemographic factors, cardiovascular risk factors, and ECG findings. Women and patients in the older age groups had statistically significantly higher NT-proBNP values, as did those with a BMI less than 25, patients with a history of diabetes, MI, and hypertension, those with a high angiography score, those who underwent CABG, and those with impaired LVF.
As determined from 12-lead routine ECG, subjects with atrial flutter or fibrillation, signs of left ventricular hypertrophy, ECG signs of anterior infarction location and, to a lesser degree, ECG signs of posterior infarction location also had higher NT-proBNP serum concentrations compared with patients without these findings.
Spearman rank correlation coefficient (r) was r = 0.22 (P<.001) between NT-proBNP and CRP and r = −0.24 (P<.001) between NT-proBNP and CrCl. C-reactive protein and CrCl were not correlated with each other in a statistically significant way.
During follow-up (mean [SD], 48.7 [15.9] months)95 patients (9.0%) experienced a secondary CVD event. Thirty patients (2.9%) died from CVD, 35 patients (3.3%) had a nonfatal MI, and in 30 patients (2.9%), a stroke was diagnosed.
Of patients in the top quartile of the NT-proBNP distribution, 16.5% experienced an event compared with 4.5%, 4.7%, and 9.1% in the first, second, and third quartile, respectively (P<.001) (Figure). Of patients in the top quartile of the CRP distribution,12.6% experienced an event compared with 7.6%, 7.2%, and 8.7% in the first, second, and third quartile, respectively (P = .09). Of patients with a CrCl lower than 60 mL/min,13.6% experienced an event compared with 6.7% and 11.7% in the categories of 90 mL/min or higher and 60 to lower than 90 mL/min, respectively (P = .002).
Table 2 gives the results of multivariable analysis. Compared with patients in the bottom quartile of the NT-proBNP distribution at baseline, patients in the top quartile had an HR of 3.34 (95% CI, 1.74-6.45) for a CVD event during follow-up after adjustment for multiple covariates (P for trend, <.001). The respective HR for CRP was 1.76 (95% CI, 0.96-3.24) (P for trend, .06). Compared with patients with a CrCl of 90 mL/min or higher at baseline, patients with a CrCl lower than 60 mL/min had an HR of 2.39 (95% CI, 1.06-5.40) (P for trend, .002) after adjustment for multiple covariates.
If all 3 markers were considered simultaneously and adjustment was done for potential confounders, NT-proBNP level showed by far the strongest association with secondary CVD events. Results were similar if all 3 markers were included as continuous variables. If the final model was controlled for LVF, the HR decreased; however, NT-proBNP values were still strong predictors of CVD events.
Finally, in Table 3 we quantified the incremental contribution of all 3 markers to risk prediction in the presence of classic risk factors. According to receiver operating curve analyses, the addition of NT-proBNP to the basic model improved the predictive accuracy of the model (area under the curve from 0.69 to 0.71) more than inclusion of CRP and CrCl; it did not improve further if all 3 variables were included simultaneously.
This prospective cohort study including 1051 patients aged 30 to 70 years with stable CHD at baseline demonstrates that increased NT-proBNP levels are strongly and independently associated with secondary CVD events and show a superior predictive utility when directly compared with markers of renal function and inflammation. Notably, the relationship persisted after adjustment for LVF.
B-type natriuretic peptide is a hormone that is produced and secreted mainly from ventricular cardiomyocytes as a response to an increased wall stress and has biological effects that counterbalance the effects of the renin-angiotensin and the sympathetic nervous system.7 Compared with BNP, NT-proBNP has a much longer half-life. It has been reported that in patients with acute MI, plasma levels of NT-proBNP could be divided in 2 phases: the first peak appearing days after the acute event as an acute-phase response to the injured tissue5,15 and the second peak appearing weeks thereafter as a result of impaired LVF.5
All subjects in the present study were included weeks after the acute manifestation of CHD or CABG procedure and were in a clinically stable condition. Higher values were seen in women and in older subjects as described by others.16 We found indicators of the severity of CHD (number of affected epicardial vessels) positively associated with NT-proBNP levels. Furthermore, patients with a history of MI or diabetes, various ECG findings such as atrial flutter or fibrillation, and signs of anterior or posterior infarction had higher NT-proBNP levels. The latter might be related to previous myocardial damage in response to ischemia, which most likely leads to elevated left ventricular diastolic filling pressure and in consequence, to the secretion of BNP.
Despite adjusting for several potential confounding factors, the prognostic value of NT-proBNP concentrations was still statistically significant. It remained so even after adjustment for CRP and CrCl levels, both relevant markers associated with secondary cardiovascular events in patients with CHD.2,3 A recent study in patients with acute coronary syndromes showed that NT-proBNP added prognostic information to measures of inflammation and renal impairment for subsequent mortality.17 The fact that the prognostic information of NT-proBNP level was still evident if controlled for LVF at baseline suggests that additional prognostic information (besides the ones related to CHF or left ventricular hypertrophy) might play a role. This is supported by a recent population-based study including 3346 middle-aged subjects without CHF, in whom plasma natriuretic peptide levels were of prognostic relevance for subsequent atrial fibrillation, stroke, and all-cause mortality.18 Notably, in the latter study the prognostic value was already evident at relatively low levels, which are usually not related to CHF.
Our findings extend those of a recent report in which NT-proBNP in 1034 patients with CHD provided prognostic information on all-cause mortality independent of LVF.11 Our data show that especially the risk for secondary CVD events is increased in patients with CHD and high NT-proBNP levels, and therefore NT-proBNP measurement may be useful for better risk stratification in patients with manifest CHD, suggesting a possible shift in emphasis from CRP level to natriuretic peptide levels in the future. The independence of NT-proBNP level from CRP level in relation to future cardiovascular events has also been suggested in a recent analysis including 570 patients with stable CHD.12 In a substudy from the Heart Outcomes Prevention Evaluation (HOPE) trial, only the inclusion of NT-proBNP improved the predictive ability by increasing the area under the curve, whereas various inflammatory markers did not, although they all showed a statistically significant association with a combined CVD end point.19 Although CRP levels were associated tentatively with CVD events in our study, they did not significantly predict secondary CVD events. This finding is in accordance with recent observations by Danesh et al20 and Campbell et al.21 C-reactive protein level may be a better predictor in primary prevention and younger subjects.
Whereas other studies reporting on the association between inflammatory markers and study end points represent highly selected populations from randomized clinical trials,22,23 our study consisted of “everyday” patients from 1 large geographic area; for example, the Aggrastat-to-Zocor (A-to-Z) Trial was conducted internationally, and subjects originated from over 40 different countries and had a special risk profile at baseline (mean CRP level, 20 mg/L; current smokers, approximately 40%).23
B-type natriuretic peptide synthesis occurs in the necrotic as well as in the healthy myocardium.24 Furthermore, BNP levels were shown to increase after exercise testing in patients with stable CHD, and levels reflected the size of the ischemic territory.25 In 112 patients undergoing exercise testing, it was demonstrated that transient ischemia may result in an immediate rise in BNP level and, to a lesser degree, in NT-proBNP levels.26 The elevation of NT-proBNP level in patients with CHD may be a result of a recurrent MI even if necrosis has not occurred10,27; however, in the context of our study, it is difficult to separate ischemia from impaired LVF.
The following limitations of our study should be considered. Although we had a large sample of patients with CHD, fatal CVD events were rare in this population. This is explained by the fact that case fatality is highest before hospitalization and during the early in-hospital phase. Furthermore, we could successfully follow up 87.2% of the patients; patients lost to follow-up were in general younger and had lower BMI, NT-proBNP, and CRP values.
Not all patients were willing or able to participate in an in-hospital rehabilitation program. This may be a further reason for the slight underrepresentation of more severely ill patients in our sample, but on the other side, almost half of the patients had extensive (3 vessel) disease. Underrepresentation of severely ill patients does not adequately explain the positive findings between NT-proBNP serum concentration and CVD events but suggests that the true prognostic value of this marker may even be stronger than shown in our study. Information of global LVF was available for most patients; however, when this semiquantitative assessment was compared with quantitative ejection fraction (available for 471 patients) derived from left ventricular cineangiograms, excellent agreement was found (mean ejection fraction in semiquantitative global LVF categories: normal, 69.6%; mild depression, 54.9%; moderate depression, 44.1%; and severe depression, 30.0%).
In conclusion, despite these limitations, our data are in support of an important prognostic value of NT-proBNP among patients with known CHD and suggest that NT-proBNP level is a useful prognostic marker that provides complementary information to markers of inflammation and renal function and to other established risk determinants.
Correspondence: Wolfgang Koenig, MD, Department of Internal Medicine II–Cardiology, University of Ulm Medical Center, Robert-Koch Str 8, D-89081 Ulm, Germany (firstname.lastname@example.org).
Accepted for Publication: August 30, 2006.
Author Contributions: Dr Rothenbacher had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Rothenbacher and Brenner. Acquisition of data: Rothenbacher, Koenig, and Brenner. Analysis and interpretation of data: Rothenbacher and Koenig. Drafting of the manuscript: Rothenbacher. Critical revision of the manuscript for important intellectual content: Koenig and Brenner. Statistical analysis: Rothenbacher and Brenner. Obtained funding: Rothenbacher, Koenig, and Brenner. Administrative, technical, and material support: Rothenbacher and Brenner. Study supervision: Rothenbacher and Brenner.
Financial Disclosure: None reported.
Acknowledgment: We highly appreciate the excellent technical assistance of Gerlinde Trischler. We thank Roche, Mannheim (Germany), for providing the NT-proBNP reagents.