Author + information
- Received October 11, 2006
- Revision received July 12, 2007
- Accepted August 14, 2007
- Published online November 13, 2007.
- Stephan von Haehling, MD⁎,†,5,⁎ (, )
- Ewa A. Jankowska, MD‡,
- Nils G. Morgenthaler, MD§,1,
- Corrado Vassanelli, MD∥,
- Luisa Zanolla, MD∥,
- Piotr Rozentryt, MD¶,
- Gerasimos S. Filippatos, MD#,
- Wolfram Doehner, MD, PhD⁎,
- Friedrich Koehler, MD⁎⁎,
- Jana Papassotiriou, PhD§,
- Dimitrios T. Kremastinos, MD, PhD#,
- Waldemar Banasiak, MD‡,
- Joachim Struck, PhD§,1,2,
- Piotr Ponikowski, MD, PhD‡,
- Andreas Bergmann, PhD§,1,2,3 and
- Stefan D. Anker, MD, PhD⁎,†,4
- ↵⁎Reprint requests and correspondence:
Dr. Stephan von Haehling, Applied Cachexia Research, Department of Cardiology, Charité Medical School, Campus Virchow-Klinikum, Augustenburger Platz 1, 13353 Berlin, Germany.
Objectives Our aim was assess the prognostic value of midregional pro-atrial natriuretic peptide (MR-proANP) using a new immunoassay in patients with chronic heart failure (HF).
Background Assessment of natriuretic peptides represents a useful addition in establishing the diagnosis of chronic HF. Their plasma values are powerful predictors of survival in chronic HF.
Methods We assessed MR-proANP in 525 chronic HF patients (derivation study: age 61 ± 12 years, New York Heart Association (NYHA) functional class I/II/III/IV 6%/44%/41%/9%, N-terminal pro-B-type natriuretic peptide (NT-proBNP) 3,637 ± 6,362 pg/ml) and validated our findings in 249 additional chronic HF patients (age 63 ± 9 years, NYHA functional class I/II/III/IV 14%/50%/33%/3%, NT-proBNP 1,116 ± 1,991 pg/ml).
Results The MR-proANP levels (mean 339 ± 306 pmol/l, range 24.5 to 2,280 pmol/l) increased with NYHA funcitonal class (p < 0.0001). During follow-up (>6 months in survivors), 171 patients (33%) died. Increasing MR-proANP was a predictor of poor survival (risk ratio 1.35 per increase in standard deviation, 95% confidence interval 1.17 to 1.57; p = 0.0061), adjusted for NT-proBNP, age, left ventricular ejection fraction, NYHA functional class, creatinine, and body mass index (BMI). In receiver operating characteristic curve analysis of 12-month survival, the area under the curve for MR-proANP was 0.74 and that of NT-proBNP was 0.75 (p = 0.7). In a validation study, MR-proANP levels above the optimal prognostic cutoff value from the validation cohort remained a significant independent predictor of death. In chronic HF patients in NYHA functional class II to III and all subgroups of BMI and kidney function, MR-proANP added prognostic value to NT-proBNP. In patients with BMI ≥30 kg/m2, MR-proANP had higher prognostic power than NT-proBNP.
Conclusions Midregional proANP is an independent predictor of mortality in patients with chronic HF. Midregional proANP adds prognostic information to NT-proBNP.
Chronic heart failure (HF) is increasing in prevalence and carries a devastating prognosis. Early diagnosis of chronic HF can improve patient outcome through timely preventive and therapeutic measures (1). A wealth of data suggests that the assessment of natriuretic peptides represents a useful addition to the chest X-ray, electrocardiogram, and Doppler echocardiography in verifying the clinical diagnosis of suspected heart failure (2). The most important members of the family of natriuretic peptides are atrial natriuretic peptide (ANP) and B-type natriuretic peptide (BNP). Both are markers of cardiac function and predict survival in patients with chronic HF. In addition to their use in helping to establish the diagnosis of chronic HF (3,4), they are used in the risk stratification of patients with stable or recently decompensated chronic HF (5–7) and are considered for guidance in chronic HF therapy (8).
Both BNP and its precursor N-terminal pro-BNP (NT-proBNP) are currently used in the assessment of patients with chronic HF. Currently, both are considered superior to ANP for diagnostic and prognostic assessment of patients with heart failure, mostly because the assessment of ANP is considered less reproducible. However, the concentration of ANP in the circulation is approximately 10 times as high as that of BNP (9,10). Atrial natriuretic peptide is derived from the cleavage of its precursor pro-atrial natriuretic peptide (proANP), which is significantly more stable in the circulation than the mature peptide. Therefore, proANP has been suggested to be a more reliable analyte (11). The N- and C-terminal regions of propeptides still can undergo enzymatic degradation (12). Therefore, we developed a new sandwich immunoassay that recognizes a midregional sequence of proANP (MR-proANP) (13). We determined blood levels of MR-proANP in unselected stable patients with chronic HF from 4 European centers to find its clinical correlates and prognostic usefulness. We compared all results to those obtained with NT-proBNP and validated our findings in a similar cohort of patients from 1 additional European center.
We analyzed MR-proANP in a cohort of 525 patients with chronic HF enrolled at 4 European centers: Athens (Greece; n = 59), London (United Kingdom; n = 90), Wroclaw (Poland; n = 247), and Zabrze (Poland; n = 129). A total of 249 patients with chronic HF from Verona (Italy) served as the validation sample. All patients participated in projects designed to investigate novel prognostic biochemical markers and provided written informed consent. The respective projects were approved by the local ethics committee at each center. The diagnosis of chronic HF was based on symptoms and clinical signs according to guidelines issued by the European Society of Cardiology (14) and the American College of Cardiology (15). All patients had a history of chronic HF of at least 3 months’ duration and documented left ventricular impairment with a left ventricular ejection fraction (LVEF) ≤45%. Patients were stable on medication for at least 4 weeks before the study. The patients’ clinical characteristics and medication are presented in Table 1,and biochemical markers are shown in Table 2.
The MR-proANP was analyzed from plasma (78%) or serum samples immediately frozen at −80°C until analysis. Detection of MR-proANP was performed using a sandwich immunoassay (MR-proANP LIA, B.R.A.H.M.S, Hennigsdorf/Berlin, Germany) as described in detail elsewhere (13). The functional assay sensitivity (interassay coefficient of variance <20%) is 20 pmol/l. The stability of MR-proANP at room temperature is >24 h. After 7 days at room temperature, the degradation of MR-proANP does not exceed 20%. This assay allows measurement of MR-proANP in serum and plasma (with EDTA, heparin, or citrate) (11). Median MR-proANP in 325 healthy individuals in previous investigations was 45 pmol/l (95% confidence interval [CI] 43 to 49 pmol/l) (13). The NT-proBNP levels were determined by a electrochemiluminescence immunoassay (ELICIA, Roche Diagnostics, Basel, Switzerland).
The participating centers followed study subjects for a mean of 28 months or until death. The mean follow-up in the validation study was 60 months. All survivors were followed for at least 8 months. The primary end point of the study was cardiovascular mortality.
Data are expressed as mean ± SD. The MR-proANP and NT-proBNP data were non-normally distributed and, therefore, log-transformed before analysis. Student unpaired ttest and analysis of variance with Fisher post hoc test were used to compare differences between groups. Univariate and multivariate regression analyses were applied to assess factors that independently predicted MR-proANP levels. A value of p < 0.05 was considered to be significant. To compare different predictive values, areas under the receiver-operating characteristic curve (AUCs) for sensitivity and specificity were constructed. The best prognostic cutoff for survival status was defined as the highest product of sensitivity and specificity. To contrast prognostic accuracy, statistical comparison of receiver-operating characteristic (ROC) curves was performed using the method for paired ROC curves described by Hanley and McNeil (16). The relationship of baseline variables with survival was assessed by Cox proportional hazards analysis (single predictor and multivariable analysis). Risk ratio (RR) and 95% CI for risk factors and significance level for chi-square (likelihood ratio test) are given. To estimate the influence of risk factors on 24-month survival, Kaplan-Meier cumulative survival curves were constructed for illustrative purposes and compared by the Mantel-Haenszel log rank test. All statistics, except analyses of ROC curves, were performed using StatView 5.0 software for Windows (Abacus Concepts, Berkeley, California). The ROC curves were constructed using MedCalc for Windows version 126.96.36.199 (Broekstraat, Mariakerke, Belgium).
A total of 525 patients with chronic HF (30 women) were analyzed. The MR-proANP in this study population ranged from 24.5 to 2,280 pmol/l, with a median of 241 pmol/l (25th percentile 141.75 pmol/l, 75th percentile 429.75 pmol/l) and a mean of 339 ± 306 pmol/l. Median MR-proANP levels increased with increasing New York Heart Association (NYHA) functional class (Fig. 1).Using simple regression, we found that MR-proANP correlated with NT-proBNP (r = 0.80), serum creatinine (r = 0.40), LVEF (r = −0.27; all p < 0.0001), and age (r = 0.15; p = 0.0006), but not with chronic HF etiology (r = 0.001; p = 0.98). The chronic HF patients’ medication (Table 1) with angiotensin-converting enzyme inhibitors/angiotensin receptor blockers or beta-blockers did not affect MR-proANP levels (all p > 0.4). However, patients on loop diuretics or spironolactone presented with higher MR-proANP levels than those not on the drug (loop diuretics: 378 ± 335 pmol/l vs. 257 ± 212 pmol/l; spironolactone: 404 ± 355 pmol/l vs. 298 ± 263 pmol/l; both p < 0.0001). In multivariate regression, MR-proANP correlated with NT-proBNP (standardized coefficient [SC] = 0.68), NYHA functional class (SC = 0.13), and creatinine (SC = 0.14; all p < 0.0001), weakly with age (SC = 0.07; p = 0.02), but not with gender, etiology (all p > 0.2), or LVEF (SC = −0.05; p = 0.08).
A total of 171 patients (32.6%) died during follow-up. Kaplan-Meier analyses for mortality revealed a 24-month mortality rate of 29.1% (95% CI 24.8% to 33.4%). Survivors had a mean MR-proANP of 271 ± 219 pmol/l, and nonsurvivors 481 ± 398 pmol/l (p < 0.0001). The 24-month mortality rates according to ascending quartiles of MR-proANP were 5.6% (95% CI 1.5% to 9.7%), 20.7% (95% CI 13.1% to 28.3%), 32.8% (95% CI 23.6% to 42.0%), and 54.2% (95% CI 43.9% to 64.5%), respectively (Fig. 2).Several baseline variables were investigated to determine their impact on the end point in single-predictor Cox regression analysis. Clinically relevant classes were defined for continuous variables to further characterize their impact on outcome. Besides MR-proANP (continuous), the following were found to have statistically significant effects on mortality (Table 3): NT-proBNP (continuous), age (continuous), LVEF (continuous and categorized), NYHA functional class (categorized), and serum creatinine (continuous and categorized). Gender (categorized) and chronic HF etiology (categorized) had no such effect (Table 3). Adjustment for center had no effect (p > 0.4).
To define optimal prognostic accuracy of MR-proANP values regarding 24-month survival in patients with chronic HF, we performed ROC curve analyses (Fig. 3).For comparison, the same ROC analyses were performed with NT-proBNP. At 24 months, the AUC for MR-proANP was 0.79 and that for NT-proBNP was 0.76 (p = 0.12) (Table 4).The optimal prognostic accuracy for MR-proANP was 295 pmol/l. At this cutoff, the sensitivity was 71% and the specificity 75% (Table 4). In comparison, the optimal prognostic cutoff-value for NT-proBNP was 1,770 pg/ml at which the sensitivity reached 76% and the specificity 64%. We calculated ROC curves for different time points during follow-up, and it emerged that the AUC values for MR-proANP and NT-proBNP were similar during the first 12 months of follow-up but had began to split at 24 months (Table 4). The difference became significant at 48 months, with an AUC of 0.79 for MR-proANP and 0.75 for NT-proBNP (p = 0.03).
These 249 patients with chronic HF presented with somewhat less severe disease status than those in the derivation study (Table 1). The MR-proANP in the validation sample ranged from 27.9 to 1,310 pmol/l, with a median of 148 pmol/l (25th percentile 98.80 pmol/l, 75th percentile 224.75 pmol/l) and a mean of 190 ± 146 pmol/l. Levels of MR-proANP correlated with NT-proBNP (r = 0.85), serum creatinine (r = 0.45), LVEF (r = 0.36), age (r = 0.41), and chronic HF etiology (r = 0.28; all p < 0.0001). During follow-up, a total of 68 deaths (27.3%) were observed (cardiovascular mortality rate at 24-months: 14.5%, 95% CI 10.2% to 18.8%). Survivors had a mean MR-proANP of 161 ± 108 pmol/l, and nonsurvivors 270 ± 199 pmol/l (p < 0.0001). The 24-month mortality rates according to ascending quartiles of MR-proANP were 7.9% (95% CI 1.3% to 14.5%), 6.5% (95% CI 0.5% to 12.5%), 9.8% (95% CI 2.4% to 17.2%), and 33.8% (95% CI 11.7% to 45.5%), respectively.
We validated our findings from the derivation cohort using the optimal cutoff values for both MR-proANP and NT-proBNP derived at 12 months’ follow-up. An MR-proANP above the optimal cutoff value of 296 pmol/l remained a significant independent predictor of death in single-predictor analysis (RR 2.837, 95% CI 1.702 to 4.728; chi-square = 16.0; p < 0.0001) and after adjusting for NT-proBNP, LVEF, NYHA functional class, and creatinine (RR 2.10, 95% CI 1.01 to 4.34, chi-square = 4.0; p = 0.046). Gender, age, etiology, and body mass index (BMI) were left out of this model, because they did not predict survival in single-predictor analysis in the validation cohort (all p > 0.1). The NT-proBNP showed borderline significance in single-predictor analysis (RR 2.430, 95% CI 0.976 to 6.054; p = 0.057) but failed to do so after adjusting for the aforementioned covariables (RR 0.99, 95% CI 0.32 to 3.07; chi-square <0.1; p = 0.98). Other independent prognosticators in this model were LVEF (p < 0.0001) and NYHA functional class (p = 0.004).
Subgroup analysis and added prognostic value
After combining the derivation and the validation sample (774 patients), we analyzed the prognostic value of MR-proANP in several important subgroups of patients with chronic HF. We repeated this analysis for NT-proBNP. We focused on patients with moderate disease (because their prognosis is more difficult to assess than that of patients with very mild or very severe disease status), on patients in different BMI subgroups, and those with different stages of kidney function (Table 5).The MR-proANP predicted survival in all subgroups that we analyzed, having higher prognostic power than NT-proBNP in patients with obesity. The combination of the 2 natriuretic peptides showed that the assessment of MR-proANP on top of NT-proBNP adds prognostic value in all subgroups of patients beyond assessing the latter peptide only (Table 5).
This is the first study to evaluate a newly developed immunoassay for MR-proANP in the follow-up of a cohort of patients with chronic HF. In chronic HF, elevated levels of MR-proANP were associated with an increased risk of death, and this effect was independent of other established prognosticators, including NT-proBNP, LVEF, NYHA functional class, creatinine, age, and gender. The AUCs for MR-proANP and NT-proBNP are overlapping to a great extent (Table 4), although longer follow-up (4 years) favors MR-proANP. Subgroup analyses revealed that MR-proANP predicts mortality in several subgroups of patients, including several important so-called “gray zones” such as those with moderate disease, the obese, and those with mildly impaired kidney function (Table 5). Overall, MR-proANP adds prognostic information in all subgroups that we analyzed if assessed in addition to NT-proBNP.
It is increasingly important to establish the added prognostic value of a new parameter over or on top of the established parameter. We documented the added prognostic value of MR-proANP versus NT-proBNP in 2 different ways. First, we documented the prognostic power of both parameters in relevant subgroups (Table 5). Second, we calculated the percentage change in overall prognostic power (i.e., the percentage change in chi-squared value) for Cox proportional hazard analysis results for NT-proBNP alone versus NT-proBNP plus MR-proANP. All subgroups that we analyzed in this way showed added prognostic power derived from the additional assessment of MR-proANP. However, we were able to document a >50% added prognostic value for MR-proANP on top of NT-proBNP in chronic HF patients with a normal kidney function and, importantly, in the obese (Table 5).
Both tests used to assess natriuretic peptide levels in this study concern the assessment of propeptides. The N-terminal regions of both proANP and pro-B-type natriuretic peptide are subject to structural changes that are dependent on a number of external influences. All sandwich immunoassays developed for the detection of proANP use an antibody against the N-terminal region of the peptide. This is combined with a secondary antibody against either the midregion or the C-terminal region (13). However, proteolytic degradation of both proANP and pro-B-type natriuretic peptide seems to be mainly directed to the N- and the C-terminal parts of the respective precursors. The midregion is significantly more stable (17). In the present study we therefore used a new sandwich immunoassay, which uses antibodies designed to detect the midregion of proANP (Fig. 4).A recent study compared the MR-proANP assay used in the present study with NT-proBNP and BNP in the diagnosis of acutely decompensated HF patients presenting to an emergency department (18). That study demonstrated that the diagnostic information obtained by MR-proANP measurements was similar to that obtained with either BNP or NT-proBNP measurements (18). This finding indicates that some of the previously reported discrepancies on the diagnostic value between ANP and BNP or their precursors could be attributed to differences in assay design and may not be present if stable and sensitive assay formats are used. Another important factor is sample stability, which is >24 h for MR-proANP at room temperature (13).
The release of ANP is mostly triggered by changes in atrial wall transmural pressure and yields increases in sodium and water excretion, suppression of renin and aldosterone release, and dilatation of the venous and arterial systems (19). An increase in hemodynamic load yields increased secretion of both ANP and BNP (“stretch-secretion coupling” ). However, the amount of natriuretic peptides secreted from the atria under those conditions remains higher than that secreted from the ventricular myocardium (21). Moreover, only secretion of BNP, but not ANP secretion, is significantly increased by tumor necrosis factor-alpha (22), a proinflammatory cytokine frequently up-regulated in patients with chronic HF (23). Tumor necrosis factor-α may therefore contribute to increases in BNP plasma values.
When comparing the performance of 2 well-performing tests to assess prognosis, it is important to construct ROC curves, compare sensitivity and specificity, use a derivation and validation study approach, and adjust results for relevant covariates. Mostly, the prognostic value cannot be compared by analysis of nominal risk ratios alone. Risk ratios refer to the change in risk relative to a given change in the parameter of interest. The concentration of natriuretic peptides in the blood of chronic HF patients can vary over a wide range of levels, and therefore results are typically expressed per unit change in peptide. This reference unit change can be freely chosen depending on the overall range of plasma levels and personal preference, and therefore RRs can be very different without any change in the true prognostic power of the given variable. We therefore used the standard deviation to compare the 2 markers analyzed in the present study. Another important issue is that when strong prognosticators are compared, the p value is of little help in comparing these prognosticators unless it is calculated to far more than 4 decimals. The comparative overall prognostic power can therefore best be seen in the chi-squared value as derived from Cox proportional hazard analysis.
First, the present study populations consisted mainly of male patients, and it is unknown whether the findings can be extrapolated to female patients. Moreover, the number of patients in NYHA functional class I was small (9% of all patients). This has to be kept in mind when the data are being interpreted.
We have documented the independent prognostic power of MR-proANP as assessed with a newly developed immunoassay in two cohorts of patients with chronic HF. The MR-proANP levels strongly correlate with disease severity. The power of MR-proANP to predict prognosis is at least as large as that of NT-proBNP, and MR-proANP adds prognostic information to NT-proBNP across all disease stages and subgroups of chronic HF analyzed in this study.
↵1 Drs. Morgenthaler, Struck, and Bergmann are employed by B.R.A.H.M.S, a biotech company that developed the midregional pro-atrial natriuretic peptide (MR-proANP) assay.
↵2 Drs. Struck and Bergmann hold patent applications on the use of MR-proANP for diagnostics.
↵3 Dr. Bergmann is a member of the board of directors and shareholder of B.R.A.H.M.S.
↵4 Dr. Anker has received consultant honoraria from B.R.A.H.M.S.
↵5 Drs. von Haehling and Anker have received honoraria for presentations on MR-proANP.
The first 3 authors contributed equally to this manuscript.
- Abbreviations and Acronyms
- area under the receiver-operating characteristic curve
- heart failure
- left ventricular ejection fraction
- midregional pro-atrial natriuretic peptide
- N-terminal pro-B-type natriuretic peptide
- New York Heart Association
- receiver-operating characteristic
- Received October 11, 2006.
- Revision received July 12, 2007.
- Accepted August 14, 2007.
- American College of Cardiology Foundation
- Jortani S.A.,
- Prabhu S.D.,
- Valdes R. Jr..
- Cowie M.R.,
- Jourdain P.,
- Maisel A.,
- et al.
- Omland T.,
- Aakvaag A.,
- Bonarjee V.V.,
- et al.
- Anand I.S.,
- Fisher L.D.,
- Chiang Y.T.,
- et al.
- Cheng V.,
- Kazanagra R.,
- Garcia A.,
- et al.
- Yoshibayashi M.,
- Saito Y.,
- Nakao K.
- Ala-Kopsala M.,
- Magga J.,
- Peuhkurinen K.,
- et al.
- Morgenthaler N.G.,
- Struck J.,
- Thomas B.,
- Bergmann A.
- Swedberg K.,
- Cleland J.,
- Dargie H.,
- et al.,
- Task Force for the Diagnosis and Treatment of Chronic Heart Failure of the European Society of Cardiology
- Hunt S.A.,
- Abraham W.T.,
- Chin M.H.,
- et al.
- Ala-Kopsala M.,
- Magga J.,
- Peuhkurinen K.,
- Leipala J.,
- Ruskoaho H.,
- Leppaluoto J.,
- Vuolteenaho O.
- Gegenhuber A.,
- Struck J.,
- Poelz W.,
- et al.
- McMurray J.,
- Pfeffer M.A.
- de Bold A.J.,
- Bruneau B.G.,
- Kuroski de Bold M.L.
- Ma K.K.,
- Ogawa T.,
- de Bold A.J.