Objectives Due to different release mechanisms, mid-regional pro-atrial natriuretic peptide (MR proANP) may be superior to N-terminal pro-B-type natriuretic peptide (NT proBNP) in the diagnosis of acute heart failure (AHF) in patients with atrial fibrillation (AF). We compared MR proANP and NT proBNP for their diagnostic value in patients with AF and sinus rhythm (SR).
Design Prospective cohort study.
Setting University hospital, emergency department.
Patients 632 consecutive patients presenting with acute dyspnoea.
Main outcome measures MR proANP and NT proBNP plasma levels were determined. The diagnosis of AHF was adjudicated by two independent cardiologists using all available data. Patients received long-term follow-up.
Results AF was present in 151 patients (24%). MR proANP and NT proBNP levels were significantly higher in the AF group compared with the SR group (385 (258–598) versus 201 (89–375) pmol/l for MR proANP, p<0.001 and 4916 (2169–10285) versus 1177 (258–5166) pg/ml, p<0.001 for NT proBNP). Diagnostic accuracy in AF patients was similar for MR proANP (0.90, 95% CI 0.84 to 0.95) and NT proBNP (0.89, 95% CI 0.81 to 0.96). Optimal cut-off levels in AF patients were significantly higher compared with the optimal cut-off levels for patients in SR (MR proANP 240 vs 200 pmol/l; NT proBNP 2670 vs 1500 pg/ml respectively). After adjustment in multivariable Cox proportional hazard analysis, MR proANP strongly predicted one-year all-cause mortality (HR=1.13 (1.09–1.17), per 100 pmol/l increase, p<0.001).
Conclusion In AF patients, NT proBNP and MR proANP have similar diagnostic value for the diagnosis of AHF. The rhythm at presentation has to be taken into account because plasma levels of both peptides are significantly higher in patients with AF compared with SR.
Statistics from Altmetric.com
- Atrial fibrillation
- MR proANP
- NT proBNP
- heart failure
- sensitivity and specificity
- endocardial map
- implantable cardioverter defibrillator
- coronary artery disease
- chest pain clinic
- mortality statistics
- congestive heart failure
- valvular disease
- mitral regurgitation
- myocardial ischaemia and infarction
- interventional cardiology
- acute coronary syndrome
- intensive care
- natriuretic peptides
- acute myocardial infarction
Acute heart failure (AHF) and atrial fibrillation (AF) are both common diagnoses in patients presenting to the emergency department (ED). Accurate and rapid diagnosis of AHF is crucial for effective treatment and outcome. Used in conjunction with patient history, clinical examination and chest x-ray, natriuretic peptides improve the diagnosis of AHF in patients presenting with acute dyspnoea.1–3 In patients with AF, concentrations of B-type natriuretic peptide (BNP) seem to be slightly less accurate for the diagnosis of AHF as compared with patients with sinus rhythm (SR).4–7
Plasma concentration of atrial natriuretic peptide (ANP) is approximately 50- to 100-fold higher than that of BNP.8 Mid-regional proANP (MR proANP) is a stable precursor of the mature peptide and a more reliable substrate for analysis.9 Two recent studies have consistently shown that MR proANP has similar accuracy in the diagnosis of AHF as N-terminal proBNP (NT proBNP).1 ,10 In patients without heart failure, MR proANP is produced in atrial and ventricular tissue. In contrast with NT proBNP which is produced predominately in ventricular cardiomyocytes, and to a smaller amount in the atria, MR proANP is produced to a larger amount in atrial tissue.8 ,11–13
Plasma levels of both peptides are known to increase significantly with heart failure due to increased production in the ventricular tissue and, to a lesser extent because of the smaller muscle mass in the atrial tissue.13
The primary aim of this study was to compare the diagnostic accuracy of MR proANP versus NT proBNP in the diagnosis of AHF in patients with AF and dyspnoea.
Study population and design
We prospectively enrolled 684 consecutive patients presenting to the ED of the University Hospital, Basel, with dyspnoea as the most prominent symptom, from April 2006 to March 2008. Patients under 18 years of age, patients on haemodialysis, and trauma patients were excluded. The presence of AF (including atrial flutter) or SR was prospectively defined by ECG at presentation by a cardiologist blinded to all other patient data. Patients without an ECG at presentation, or rhythms other than AF or SR on the initial ECG were excluded from this analysis. Six hundred and thirty-two of the 684 patients had an ECG available at presentation and were considered as the study population. The study was carried out according to the principles of the Declaration of Helsinki, and approved by the local ethics committee. Written informed consent was obtained from all participating patients.
Clinical evaluation of patients
Upon presentation to the ED, all patients underwent a standard clinical assessment (clinical history, physical examination, ECG, pulse oximetry, blood tests including BNP and chest x-ray). Echocardiography and pulmonary function tests were strongly recommended, and additional testing was performed if necessary.
Reference (‘Gold’) standard definition of heart failure
Two independent cardiologists reviewed all medical records pertaining to the patient, and classified the diagnosis as dyspnoea due to AHF, or dyspnoea due to causes other than AHF. Both cardiologists had access to all available medical records pertaining to the patient from the time of ED presentation to the results of the 90-day follow-up. This information included BNP levels, chest x-ray, medical history not available at the time of admission, echocardiography, left ventriculography (performed at the time of cardiac catheterisation), pulmonary function test, CT scan, right heart catheterisation, hospital course, response to therapy, autopsy data for deceased patients, and information about clinical events or readmissions during the 90-day follow-up period. The BNP level was considered as a quantitative marker of AHF and, therefore, interpreted as a continuous variable to make best use of the information provided by this test. BNP cut-off levels for the diagnosis of AHF were adjusted for the presence of both kidney disease and obesity in order to maximise diagnostic accuracy.14–16 NT proBNP and MR proANP levels were not available to the cardiologist at the time of classification.
End points and follow-up
The primary objective was to compare the accuracy of MR proANP with that of NT proBNP to diagnose AHF in patients with AF compared with SR. As BNP levels were available for the reference (‘Gold’) standard diagnosis of AHF, the comparison of the MR proANP with BNP was considered biased in favour of BNP and, therefore, comparisons were made only between MR proANP and NT proBNP. As a secondary end point, the prognostic value for the prediction of death during long-term follow-up was assessed. All patients were contacted by telephone interview performed by trained researchers blinded to the results of laboratory testing. Follow-up was scheduled after 90, 180 and 360 days. In case of uncertainties regarding vital status, referring physicians and administrative database of the respective hometown were contacted.
Measurement of natriuretic peptides
At presentation to the ED, blood samples for determination of MR proANP and NT proBNP levels were collected into tubes containing potassium EDTA. After centrifugation, each plasma sample was aliquoted and immediately frozen at −80°C until later analysis. Detection of MR proANP was performed using an automated sandwich chemiluminescence immunoassay on the KRYPTOR System (BRAHMS AG, Hennigsdorf/Berlin, Germany) as described earlier in detail,9 and used earlier in other studies.10 The assay contains an antibody directed against epitopes in the mid-region of proANP, and so, quantifies the level of the pro-hormone. 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).17 Median MR proANP in 325 healthy individuals in previous investigations was 45 pmol/l (95% CI 43 to 49 pmol/l).9 NT proBNP levels were determined by a quantitative electrochemiluminescence immunoassay (Elecsys proBNP, Roche Diagnostics AG, Zug, Switzerland). All blood samples were processed by personnel blinded for any patient data.
The data is presented as proportions, means (±SD) and in case of non-normal distribution as median with IQR. Comparisons were made using the t test for normally distributed continuous variables, Mann–Whitney U test for non-normally distributed continuous variables, Fisher's exact test for categorical variables with any field including <6 patients, and χ2 test for the other categorical variables. Kolmogorov–Smirnov test was used to test for normality. Receiver operating characteristic (ROC) curves were constructed to assess the diagnostic accuracy for the diagnosis of AHF. The comparison of areas under the ROC curves (AUC) was performed as recommended by DeLong.18 Optimal cut-off values were derived from ROC curves as described by Youden,19 and sensitivity, specificity, negative predictive value and positive predictive value were calculated. Kaplan–Meier analysis was performed for survival, and log-rank values to assess statistical significance. We used Cox proportional hazard models adjusted for age, sex, New York Heart Association (NYHA) class, systolic blood pressure, diastolic blood pressure, body mass index, estimated glomerular filtration rate (eGFR), AF, MR proANP, NT proBNP, all on admission, and final diagnosis of AHF to compute HRs and 95% CIs of predictors of 1-year all-cause mortality. All hypothesis testing was two-tailed. The eGFR was calculated using the Modification of Diet in Renal Disease formula.20 The statistical analyses were performed using the SPSS/PC (V. 19.0, SPSS Inc.) software package. A statistical significance level of 0.05 was used.
Of the 632 patients included, 151 patients had documented AF at presentation to the ED, and 481 were in SR. Diagnosis of AHF was significantly more frequent in the group of patients with AF (92 vs 51% in SR, p<0.001). Patients in the AF group were older (80 (75–85) vs 75 (63–82), p<0.001) more often suffered from chronic kidney disease (38% vs 26%, p=0.002) and had a previous stroke (17% vs 10%, p=0.041). Patients with AF were more often treated with β blockers, diuretics, digoxin, antiarrhythmic drugs and coumarin, and presented with more clinical signs of AHF and a higher heart rate (104±31/min vs 90±21/min, p<0.001, online table 1). Patients in the SR group more often suffered from chronic obstructive pulmonary disease (COPD, 23% vs 37%, p=0.002). The left ventricular ejection fraction was significantly lower in the AF group (47 (35–60) vs 60 (35–60) %, in SR, p=0.013).
MR proANP levels were significantly increased at presentation in the AF group compared with the SR group (385 (258–598) vs 201 (89–375) pmol/l, p<0.001). A comparable difference between patients in AF and patients in SR could be found for NT proBNP (4916 (2169–10285) vs 1177 (258–5166) pg/ml, p<0.001). This significant difference was mainly attributable to higher MR proANP and NT proBNP levels in AF patients without AHF compared with SR patients without AHF. No significant difference in natriuretic peptide plasma levels were found between the AF and the SR group in AHF patients (online table 1, figure 1A,B).
An additional analysis was performed to test for the impact of multiple diagnoses on the natriuretic peptide plasma levels in patients with the diagnosis of AHF. Therefore, the group of patients with AHF (n=362) was subdivided in patients with AHF only (n=240), and patients with one or more additional diagnoses (n=122). NT proBNP and MR proANP levels were higher in the group with AHF only, but no additional diagnostic information derived from this calculation (online table 2).
Diagnostic performance of MR proANP and NT proBNP
The diagnostic performance of different cut-off values for MR proANP and NT proBNP in the diagnosis of AHF subdivided according to AF or SR are presented in online table 3. ROC analyses to compare the diagnostic performance for the diagnosis of AHF of both peptides in patients with AF and SR were performed (figure 2). The areas under the curve for NT proBNP and MR proANP (all patients) were 0.92 (CI 0.91 to 0.94) and 0.92 (CI 0.90 to 0.94) respectively (p=0.50). Divided for the subgroups of patients with AF and SR they were 0.89 (CI 0.81 to 0.96) and 0.91 (CI 0.89 to 0.94) for NT proBNP and 0.90 (CI 0.84 to 0.95) and 0.92 (CI 0.89 to 0.94) for MR proANP, respectively.
In patients with AF, the optimal cut-off point derived from the ROC curve, for MR proANP was 240 pmol/l, which was significantly higher as compared with the optimal cut-off point of 200 pmol/l in patients with SR. For NT proBNP, the optimal cut-off point for patients in AF in our study was 2670 pg/ml, which was significantly higher as compared with the optimal cut-off point of 1500 pg/ml in patients with SR.
The diagnostic accuracy was similar for SR and AF for MR proANP, but different for NT proBNP, with a lower accuracy in AF (online table 3).
Multivariate regression analysis was performed to detect the determinants of elevated natriuretic peptide levels (MR proANP >120 pmol/l or NT proBNP >900 pg/ml). Age, eGFR, coronary artery disease and AF were independent determinants of both, higher MR proANP and NT proBNP levels. Systolic and diastolic blood pressure and low body mass index were only determinants for higher NT proBNP levels.
Prediction of death by MR proANP and NT proBNP
One-year follow-up was complete in 631 patients (99.9%). Of these, 172 patients (27%) died within 1 year. Non-survivors had higher MR proANP levels than survivors (422 (205–709) vs 215 (96–363) pmol/l, p<0.001). Similar differences were seen for NT proBNP (5879 (1676–17318) vs 1322 (286–4167) pg/ml, p<0.001). The pattern of higher MR proANP and NT proBNP concentrations in non-survivors versus survivors remained when analysis were repeated in patients with SR or AF (both p<0.001).
Kaplan–Meier curve analysis based on MR proANP at presentation showed a much higher 1-year mortality rate in the fourth quartile (Q4) (50%) than in the first three quartiles (Q1 (13%), Q2 (20%) and Q3 (22%), p<0.001 by log-rank test). For NT proBNP, the mortality was 51% in Q4 and 11% in Q1, 18% in Q2 and 27% in Q3 (p<0.001 by log-rank test). The AUC to predict 360-day mortality was 0.71 (95% CI 0.66 to 0.76) for MR proANP and 0.73 (CI 0.68 to 0.78) for NT proBNP. Univariate Cox proportional hazard analysis showed that concentrations of MR proANP, NT proBNP, diagnosis of AHF, NYHA class, age and eGFR, body mass index, systolic blood pressure, NT proBNP and MR proANP all assessed on presentation to the ED, were predictors of 1-year all-cause mortality. Importantly, AF was not an independent predictor for mortality (online table 4). Other independent predictors were age, NYHA class, body mass index and systolic blood pressure. In multivariate analysis, MR proANP and NT proBNP independently predicted 1-year all-cause mortality in separate models. The inclusion of both MR proANP and NT proBNP in one multivariate model is not recommended because of the high correlation with possible colinearity issues in the multivariate analysis.
Natriuretic peptides have become an important aid in the diagnosis of AHF.6 ,7 ,21 ,22 Breidthardt et al demonstrated that despite increased baseline levels, the use of BNP as a diagnostic marker improves the initial management of patients presenting with acute dyspnoea and AF to the ED.7 In this study, we focused on the use of MR proANP, an atria-derived peptide possibly reflecting atrial stress more adequately, and directly compared the accuracy of MR proANP versus NT proBNP in the diagnosis of AHF in patients with AF.
We report four major findings: First, MR proANP and NT proBNP levels of unselected patients presenting to the ED with dyspnoea are significantly higher in patients with AF compared with patients with SR. Second, in patients with AF, the diagnostic accuracy quantified by the area under the ROC curve for AHF is similar for MR proANP and NT proBNP. Third and most important, the optimal cut-off levels for the diagnosis of AHF in patients with AF are significantly higher as compared with the optimal cut-off levels for patients in SR. Fourth: After adjustment in multivariable Cox proportional hazard analysis, MR proANP and NT proBNP strongly predicted one-year all-cause mortality.
These findings extend and corroborate previous studies of MR proANP, and help to appropriately use NT proBNP in clinical practice.1 ,10 While the BACH study defined the optimal diagnostic cut-off level for MR proANP in unselected patients with dyspnoea (120 pmol/l), this analysis showed that a higher cut-off level should be used in patients with AF. Remarkably, the cut-off level calculated for all patients of this cohort was higher (206 pmol/l) than in the BACH cohort. We would state that the significant difference in age between the BACH Cohort (64 year) and this cohort (72 year) can be seen as the main reason for these differing cut-off values.
Rienstra and coworkers studied NT-ANP levels in 354 patients (76 in AF) with advanced chronic heart failure and evaluated the prognostic value of NT-ANP and NT proBNP. They concluded that NT-ANP levels are affected by AF, while NT proBNP levels are not.21 Interestingly, MR proANP and NT proBNP levels were both significantly affected by the presence of AF in our study. This can be explained by the different study population, consisting of chronic heart failure patients in the study of Rienstra and an unselected population presenting with dyspnoea to the ED in our work. In our study the increased levels of MR proANP and NT proBNP in the AF group could further be related to the compromised renal function in our patients with AF, which was not present in the study of Rienstra. In a multivariate regression analysis, we could identify renal function as an independent risk factor for increased MR proANP and NT proBNP levels.
We hypothesise that cardiac rhythm has an impact on the diagnostic accuracy of MR proANP and NT proBNP. Natriuretic peptide plasma levels are influenced by the rhythm present at the time point of blood sampling, and less by the history of past arrhythmias. As a consequence thereof, MR proANP and NT proBNP plasma levels of patients with paroxysmal AF in SR at the time point of venipuncture should be comparable with SR patients.
This hypothesis is supported by work of Mookherjee who demonstrated a significant intra-individual reduction of ANP plasma levels after cardioversion of atrial flutter and AF,23 but it is challenged by a study of Knudsen who did not find significant differences in BNP levels between patients with paroxysmal AF in SR and patients with AF.6
Deftereos et al reported a significant increase of NT proBNP with a maximum level within the first 24–36 h after the onset of AF.24 They further could demonstrate that patients with NT proBNP levels above predefined age-adapted cut-off values had significantly less left atrial thrombi than patients below the cut-off levels, which supported their findings of increased NT proBNP levels in recent onset AF.25
Yoshihara et al analysed plasma ANP levels in 62 patients with AF who underwent a modified maze procedure.26 They correlated the collagen volume fraction as a marker of atrial fibrosis with plasma ANP levels, and described a negative correlation. In this very specific patient population, this was interpreted in the way that atrial fibrosis reduces ANP production. We think that in our patient population, significantly less atrial fibrosis should be expected than in the average patient with AF undergoing cardiac surgery, which would explain the different findings.
The difference in left ventricular ejection fraction between the AF and the SR group of our study could result from tachymyopathy putatively present in some patients with AF at presentation to the ED. This hypothesis is supported by the tachycardia and the more pronounced clinical symptoms of AHF present in the AF group. In line with these findings, diagnosis of AHF was much more frequent in this group compared with the SR group.
Our results demonstrate that NT proBNP and MR proANP can be used reliably as diagnostic markers for the diagnosis of AHF in patients with dyspnoea and AF on admission to the ED. It is important to realise though, that the rhythm at the time point of presentation to the ED has to be taken into account.
This study has limitations. First, the group of patients with AF may have been too small to detect subtle differences in the diagnostic performance of MR proANP and NT proBNP. Second, we cannot comment on the utility of MR proANP in patients with terminal renal failure as these patients were excluded from this study. Third, we did not discriminate between patients without AF and patients with paroxysmal AF in SR at time of presentation, a difference that should not significantly influence the performance of the biomarkers tested.
In AF patients, NT proBNP and MR proANP have similar diagnostic values for the diagnosis of AHF. The rhythm at presentation has to be taken into account because plasma levels of both peptides are significantly higher in patients with AF compared with SR, which requires the use of higher diagnostic cut-off levels.
We thank the patients who participated in the study, the staff of the emergency department, the laboratory technicians, and particularly Claudia Stelzig. Michael Freese, Irina Klimmeck, Melanie Wieland, Selina Christen, Janine Vögele, Beate Hartmann, and Fausta Chiaverio for their most valuable efforts, and we thank Drs C Schindler and K Denhaerynck for expert statistical advice.
JE and MP equal contribution to this work.
Funding Christian Mueller, as principal investigator, was supported by grants from the Swiss National Science Foundation (PP00B-102853), the Swiss Heart Foundation, the University of Basel, Abbott, ALERE, BRAHMS, and Pronota. In addition, Christian Mueller received lecture fees from Abbott, ALERE, BRAHMS, Novartis and Roche. Jens Eckstein was supported by research grants from the Swiss National Science Foundation, the Marie Curie Program and the University of Basel, and travel grants from Medtronic, St Jude Medical and Boehringer Ingelheim. Mihael Potocki has received speaker honoraria from Abbott, BRAHMS and Roche. For the other authors no conflict of interest exists.
Competing interests None.
Ethics approval Ethics approval provided by Local Ethics Committee University Basel.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement At present, there is no additional unpublished data of this particular study available to any other scientists/journals.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.