Abstract
Aims
Co‐morbid atrial fibrillation (AF) increases both mortality and N‐terminal pro brain natriuretic peptide (NT‐proBNP) concentrations in patients with chronic heart failure (CHF). It is unclear whether AF worsens prognosis independently from NT‐proBNP concentrations. If AF was an independent risk factor, NT‐proBNP levels for outcome prediction would need to be adjusted in patients with AF. We aimed to analyse the influence of AF on the prognostic value of NT‐proBNP in patients with CHF.
Methods and results
A total of 2541 consecutive CHF patients with sinus rhythm (SR) or AF were identified in the outpatients' CHF registry of the University of Heidelberg, Germany. Of these, 250 patients with SR were individually matched to 250 patients with AF with respect to NT‐proBNP, New York Heart Association functional class, sex, age, and aetiology of CHF. In the general sample, both AF and NT‐proBNP were associated with all‐cause mortality [hazard ratio (HR) = 1.96, 95% confidence interval (CI) 1.61–2.39, P < 0.001; and HR = 1.03 per 1000 ng/L increase, 95% CI 1.02 to 1.04, P < 0.001, respectively]. After matching, NT‐proBNP retained its prognostic power (HR = 1.13 per 1000 ng/L increase, 95% CI 1.10 to 1.16, P < 0.001), but AF did not (HR = 0.91, 95% CI 0.66 to 1.25, P = 0.56). Despite similar prognosis, matched patients with SR were in more advanced CHF than were AF patients as indicated by a lower left ventricular ejection fraction (30 ± 13% vs. 34 ± 14%, P < 0.001).
Conclusions
The prognostic value of NT‐proBNP in CHF is not influenced by concomitant AF. AF, in return, might be a surrogate of a worse cardiac condition rather than an independent risk factor.
Keywords: Chronic heart failure, Atrial fibrillation, NT‐proBNP, Prognosis
Introduction
Chronic heart failure (CHF) and atrial fibrillation (AF) are both highly prevalent in industrial countries.1, 2, 3, 4 Co‐morbidity of CHF and AF is common and conveys a poor prognosis.5, 6, 7, 8, 9 Whether AF is an independent risk factor in CHF or a surrogate of advanced CHF disease, however, is still under debate.10
In patients with CHF, AF is associated with higher levels of N‐terminal pro brain natriuretic peptide (NT‐proBNP).11, 12, 13 Elevated NT‐proBNP concentrations, in turn, are related to increased mortality.14, 15, 16 On the other hand, the prognostic power of NT‐proBNP may differ by patient characteristics and hence compromise its use as a prognostic marker for CHF in daily clinical practice.17 The effect of AF on the prognostic value of NT‐proBNP regarding all‐cause mortality in patients with CHF remains unclear.18, 19, 20, 21, 22
If AF was an independent risk factor—leading by itself to increased levels of NT‐proBNP—patients with similar NT‐proBNP but different in rhythm would differ in prognosis. Therefore, cut‐off levels for outcome prediction would need to be adjusted in patients with AF.21 For example, if patients with AF had better prognosis than had sinus rhythm (SR) patients with similar NT‐proBNP concentrations, NT‐proBNP cut‐off levels for risk prediction would need to be increased. If AF was rather reflecting more advanced disease—potentially indicated by elevated NT‐proBNP levels—the prognosis of these patients would be similar. We addressed the influence of AF on the prognostic value of NT‐proBNP in a sample of 2541 ambulatory patients with stable CHF.
Methods
Patient selection
Patients' data were extracted from the Heidelberg Heart Failure Registry as described subsequently. All patients with stable CHF who attended the specialized heart failure outpatient clinic of the University Hospital Heidelberg, Germany, for evaluation of heart failure were offered inclusion into the local heart failure registry. Less than 1% refused to participate. Inclusion into the heart failure registry is continuous and ongoing. To be eligible for this study, patients were selected from the registry according to the following inclusion criteria:
diagnosis of CHF,
presentation at the specialized heart failure outpatient clinic of the University Hospital Heidelberg, Germany, for evaluation of heart failure between 1995 and 2014,
signature of written informed consent for inclusion into the Heidelberg Heart Failure Registry,
diagnosis of either SR or AF, and
measurement of NT‐proBNP at the time of presentation.
The diagnosis of CHF was established according to the European Society of Cardiology guideline criteria5 on the basis of typical symptoms in the presence of objective abnormalities of cardiac function on echocardiography or left heart catheterization. Inclusion into the present study was independent from underlying left ventricular ejection fraction (LVEF). Thus, the study cohort comprised patients with reduced, mid‐range, or preserved ejection fraction. The diagnosis of SR or AF was based on a 12‐lead electrocardiogram (ECG) recorded at the same day as the blood sample was taken. Interpretation of the ECG was performed with respect to guideline recommendations.6 Every patient with AF on 12‐lead ECG at baseline was classified as AF. Blood samples for NT‐proBNP analyses were taken from a peripheral venous catheter using EDTA vacutainers. NT‐proBNP was measured according to the standard protocol of the fully automated Elecsys® Roche Diagnostics 2010 analyser. Assay precision, analytical sensitivity, interferences, and stability for this method have been described previously.23
Baseline characteristics included medical history, physical examination, 12‐lead ECG, LVEF, blood count and standard chemistry, and medication. Medication was at the discretion of the referring physician.
If multiple visits were available, only the first visit was chosen. The study conformed to the principles of the Declaration of Helsinki and was approved by the local ethics committee.
Follow‐up and endpoint
Surviving patients were followed up for a minimum of 365 days. Information on survival status was collected by scheduled visits to the outpatient clinic, telephone calls either to the patients' homes or to their physicians, or hospital electronic records.
Primary endpoint of this study was all‐cause mortality. Patients who received cardiac transplantation were followed up until transplantation and censored thereafter.
Matching
In order to analyse the impact of AF on the prognostic value of NT‐proBNP, CHF patients with AF were individually matched with patients with SR with respect to NT‐proBNP levels. For NT‐proBNP concentrations, a deviation of ±15% was permitted. This range was chosen as it represents half of the intra‐individual biological variation of NT‐proBNP described in prior studies.14, 24 To account for possible confounders, additional matching criteria were: sex, age, New York Heart Association (NYHA) functional class, and aetiology of CHF. We chose these variables because they have an influence on both mortality5, 25, 26 and NT‐proBNP concentrations14 in patients with CHF. With regard to age, a variation of ±5 years was tolerated.
Automated matching was performed using a ‘visual basic for applications’ code, as follows: first, pseudonymous data of CHF patients with AF and of those with SR were listed separately into two Excel® files. Both lists were put in random order, using Excel® random number generator. Then, beginning with the first patient with AF, the first patient on the SR list who fulfilled every single matching criterion was chosen. The pair was then removed from both lists, and the matching routine continued with the next patient on the AF list. In case no matching patient with SR could be identified, the AF patient was removed from the AF sample, and the matching cycle continued with the next AF patient. In addition, we performed a reverse matching starting with the first patient with SR. As results of the reverse matching were similar to those of the original matching, only results from the original matching are presented.
Subgroups
All analyses were repeated in subgroups with respect to age (≤65 vs. >65), sex (male vs. female), aetiology (ischaemic, cardiomyopathy, and others), NYHA (I/II vs. III/IV), LVEF (≤30% vs. >30%), estimated glomerular filtration rate (eGFR) (≤60 vs. >60 mL/min/1.73 m2) calculated using the Modification of Diet in Renal Disease equation,27 chronic obstructive pulmonary disease (COPD) (yes vs. no), diabetes (yes vs. no), heart rate (≤75 vs. >75 b.p.m.), and systolic blood pressure (SBP) (≤120 vs. >120 mmHg). Cut‐offs for age, LVEF, eGFR, heart rate, and SBP represent the median of the respective variables in the matched sample.
Statistics
Calculations were performed using Excel® 2013 (Microsoft, Redmond, USA), IBM SPSS® Statistics version 21 (IBM, Ehningen, Germany), and MedCalc® (MedCalc® Software bvba, Ostend, Belgium). Figures were created with GraphPad® PRISM 6 (GraphPad® Software Inc., La Jolla, USA). Results are presented as numbers (%), mean ± standard deviation, or median [inter‐quartile range], as appropriate. To test for significant differences between groups, Student's t‐test, χ 2 test, odds ratio (OR), or Mann–Whitney U‐test were used, where appropriate. To estimate cut‐off levels of NT‐proBNP as a predictor of 1 year mortality, receiver operating characteristic (ROC) curves were used. The Youden index of ROC was used to identify the best cut‐off levels of NT‐proBNP as a predictor of 1 year mortality.28 Survival was analysed in the general as well as matched sample using Cox proportional hazard models and displayed by Kaplan–Meier product limit method. As matched patients with SR and AF differed in a number of baseline variables, we performed a sensitivity analysis using stepwise multivariable Cox regression analysis in the matched sample. Covariates included all variables that significantly differed between matched patients with AF and SR. In addition, we performed additional matching procedures with respect to NT‐proBNP (±15%), eGFR (±10%), and LVEF (±5%), and NT‐proBNP (±15%), age, sex, and eGFR (±10%). Univariable and multivariable survival analyses were repeated in the additional matched samples.
All tests are two tailed. A P‐value of <5% was regarded as statistically significant.
Results
Baseline characteristics and follow‐up
A total of 2541 patients fulfilled the inclusion criteria outlined previously. Of these, 2110 patients were with SR, while 431 patients had AF. After matching with respect to NT‐proBNP, age, sex, NYHA functional class, and aetiology of CHF, 250 pairs of CHF patients with AF or SR were identified. In 181 AF patients, no matching SR patient could be identified. These patients were therefore excluded from the matched analyses.
Baseline characteristics of the general and matched samples are shown in Table 1. In the general sample, patients with SR and AF differed in most characteristics, whereas patients in the matched sample were similar with regard to the matching criteria age, sex, NYHA functional class, NT‐proBNP, and aetiology of CHF. However, matched patients with AF were more likely to have diabetes (P = 0.025) or hypertension (P = 0.025) than were patients with SR. In addition, body mass index, diastolic blood pressure, and resting heart rate were higher in CHF patients with AF than in those with SR (P < 0.001, P = 0.007, and P < 0.001, respectively). On the other hand, patients with SR had a lower LVEF (P < 0.001) and a lower eGFR (P = 0.012). A total of 1506 (59%) and 362 (72%) patients in the general and matched sample had an LVEF ≤ 40%.
Table 1.
Baseline characteristics of the general chronic heart failure patient sample and of chronic heart failure patients with atrial fibrillation matched to chronic heart failure patients with sinus rhythm with respect to N‐terminal pro brain natriuretic peptide, age, sex, New York Heart Association functional class, and aetiology
| Characteristics | General sample | Matched sample | ||||
|---|---|---|---|---|---|---|
| AF (n = 431) | SR (n = 2110) | P‐value | AF (n = 250) | SR (n = 250) | P‐value | |
| NT‐proBNP (ng/L) | 2051 [985–4246] | 466 [142–1508] | <0.001 | 1621 [858–3017] | 1555 [846–3014] | 0.774 |
| Age (years) | 65.9 ± 11.6 | 55.9 ± 14.8 | <0.001 | 64.3 ± 10.7 | 64.2 ± 10.6 | 0.860 |
| Men, n (%) | 341 (79.1) | 1518 (71.8) | 0.002 | 210 (84.0) | 210 (84.0) | 1.000 |
| NYHA | <0.001 | 1.000 | ||||
| I, n (%) | 95 (22.0) | 861 (40.8) | 56 (22.4) | 56 (22.4) | ||
| II, n (%) | 140 (32.4) | 687 (32.5) | 92 (36.8) | 92 (36.8) | ||
| III, n (%) | 189 (43.8) | 545 (25.8) | 101 (40.4) | 101 (40.4) | ||
| IV, n (%) | 7 (1.6) | 17 (0.8) | 1 (0.4) | 1 (0.4) | ||
| Aetiology | 0.001 | 1.000 | ||||
| Ischaemic, n (%) | 207 (47.9) | 1041 (49.3) | 156 (62.4) | 156 (62.4) | ||
| Cardiomyopathy, n (%) | 133 (30.8) | 771 (36.5) | 71 (28.4) | 71 (28.4) | ||
| Other, n (%) | 92 (21.3) | 301 (14.3) | 23 (9.2) | 23 (9.2) | ||
| BMI (kg/m2) | 27.8 ± 4.9 | 27.3 ± 5.1 | 0.107 | 28.5 ± 5.0 | 27.0 ± 4.4 | <0.001 |
| HR (1 b.p.m.) | 80 ± 19 | 71 ± 14 | <0.001 | 81 ± 18 | 71 ± 13 | <0.001 |
| Hypertension, n (%) | 366 (84.7) | 1708 (80.8) | 0.066 | 219 (87.6) | 195 (78.0) | 0.025 |
| SBP (mmHg) | 120 ± 19 | 120 ± 19 | 0.779 | 120 ± 19 | 118 ± 20 | 0.265 |
| DBP (mmHg) | 74 ± 12 | 75 ± 12 | 0.178 | 75 ± 11 | 72 ± 11 | 0.007 |
| LVEF (%) | 35 ± 15 | 37 ± 15 | 0.030 | 34 ± 14 | 30 ± 13 | <0.001 |
| 6MWT (m) | 420 ± 127 | 478 ± 117 | <0.001 | 430 ± 120 | 438 ± 120 | 0.532 |
| Diabetes, n (%) | 146 (33.8) | 438 (20.7) | <0.001 | 96 (38.4) | 68 (27.2) | 0.025 |
| COPD, n (%) | 32 (7.4) | 106 (5.0) | 0.045 | 19 (7.6) | 18 (7.2) | 0.864 |
| eGFR (mL/min/1.73 m2) | 73 ± 27 | 83 ± 29 | <0.001 | 77 ± 26 | 71 ± 23 | 0.012 |
| Haemoglobin (g/dL) | 13.9 ± 0.8 | 13.8 ± 1.9 | 0.83 | 14.1 ± 1.7 | 13.6 ± 1.9 | 0.008 |
| Serum cholesterol (mg/dL) | 171 ± 41 | 186 ± 46 | <0.001 | 175 ± 43 | 183 ± 48 | 0.140 |
| ACEI, n (%) | 310 (71.9) | 1462 (69.2) | 0.278 | 182 (72.8) | 177 (70.8) | 0.619 |
| ARB, n (%) | 128 (29.7) | 535 (25.3) | 0.061 | 80 (32.0) | 68(27.2) | 0.240 |
| Beta‐blocker, n (%) | 382 (88.6) | 1760 (83.3) | 0.006 | 222 (88.8) | 216 (86.4) | 0.416 |
| MRA, n (%) | 211 (49.0) | 795 (37.6) | <0.001 | 131 (52.4) | 129 (51.6) | 0.858 |
| Digitalis, n (%) | 237 (55.0) | 384 (18.2) | <0.001 | 140 (56.0) | 55 (22.0) | <0.001 |
| Statin, n (%) | 246 (57.1) | 1187 (56.2) | 0.754 | 150 (60.0) | 165 (66.0) | 0.165 |
| Oral antidiabetics, n (%) | 77 (17.9) | 250 (11.8) | 0.001 | 52 (20.8) | 29 (11.6) | 0.005 |
| Anticoagulation, n (%) | 392 (91.0) | 803 (38.0) | <0.001 | 230 (92.0) | 126 (50.4) | <0.001 |
P‐values < 0.05 are printed in bold letters.
6MWT, 6 min walk test; ACEI, angiotensin‐converting enzyme inhibitor; AF, atrial fibrillation; ARB, angiotensin receptor blocker; BMI, body mass index; COPD, chronic obstructive pulmonary disease; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate calculated using the Modification of Diet in Renal Disease equation; HR, heart rate; LVEF, left ventricular ejection fraction; MRA, mineralocorticoid receptor antagonist; NT‐proBNP, N‐terminal pro brain natriuretic peptide; NYHA, New York Heart Association functional class; SBP, systolic blood pressure; SR, sinus rhythm.
In the general sample, the diagnosis of AF was associated with higher levels of NT‐proBNP as compared with SR (2051 ng/L [985–4246 ng/L] vs. 466 ng/L [142–1508 ng/L], P < 0.001). Median NT‐proBNP concentrations with respect to LVEF and heart rhythm are shown in Table S1. By definition, NT‐proBNP levels did not differ between patients with AF and SR in the matched sample (1621 ng/L [858–3017 ng/L] vs. 1555 ng/L [846–3014 ng/L], P = 0.774).
Total follow‐up in the general sample was 9717 patient‐years. During that time, 507 (20.0%) patients died, 132 (30.6%) with AF and 375 (17.8%) with SR. One‐year mortality was 9.3% and 4.0% [OR = 2.04, 95% confidence interval (CI) 1.62 to 2.58, P < 0.001] for patients with AF and SR, respectively.
Total follow‐up in matched CHF patients was 1847 patient‐years. During that time, 157 (31.4%) patients died, 72 (28.8%) with AF and 85 (34.0%) with SR. One‐year mortality was 7.6% and 9.2% (OR = 0.81, 95% CI 0.43 to 1.53, P = 0.52) for patients with AF and SR, respectively.
Prognostic significance of atrial fibrillation and N‐terminal pro brain natriuretic peptide
General sample
In the general sample, the diagnosis of AF was associated with a higher all‐cause mortality than the diagnosis of SR (HR = 1.96, 95% CI 1.61 to 2.39, P < 0.001, for AF vs. SR). The Kaplan–Meier survival curves of patients with AF and SR in the general sample are depicted in Figure 1.
Figure 1.
Kaplan–Meier survival curves for 5 year survival for CHF outpatients with respect to heart rhythm (general sample).
Increasing levels of NT‐proBNP were associated with increasing mortality (HR = 1.03 per 1000 ng/L increase, 95% CI 1.02 to 1.04, P < 0.001). This was true for patients with AF as well as for those with SR (HR = 1.02 per 1000 ng/L increase, 95% CI 1.01 to 1.03, P < 0.001; and HR = 1.03 per 1000 ng/L increase, 95% CI 1.02 to 1.04, P < 0.001, for patients with AF and SR in the general sample, respectively).
The prognostic information of rhythm and NT‐proBNP was independent one from the other in the common Cox proportional hazard model (HR = 1.82, 95% CI 1.49 to 2.22, P < 0.001, for AF vs. SR; and HR = 1.03, 95% CI 1.02 to 1.03, P < 0.001, per 1000 ng/L increase of NT‐proBNP).
Matched sample
After matching for age, sex, NYHA functional class, NT‐proBNP, and aetiology of CHF, no difference in mortality was noted between CHF patients with AF and those with SR (HR = 0.91, 95% CI 0.66 to 1.25, P = 0.56, for AF vs. SR). The Kaplan–Meier survival curves of patients with AF and SR in the matched sample are shown in Figure 2.
Figure 2.
Kaplan–Meier survival curves for 5 year survival for CHF outpatients with respect to heart rhythm (matched sample).
In contrast, NT‐proBNP retained its prognostic power in the matched sample (HR = 1.13 per 1000 ng/L increase, 95% CI 1.10 to 1.16, P < 0.001). Again, this was true for patients with AF as well as for those with SR (HR = 1.12 per 1000 ng/L increase, 95% CI 1.08 to 1.16, P < 0.001; and HR = 1.14 per 1000 ng/L increase, 95% CI 1.10 to 1.18, P < 0.001, for patients with AF and SR in the matched sample, respectively).
The prognostic information of NT‐proBNP remained independent from rhythm in the common Cox proportional hazard model (HR = 0.89, 95% CI 0.65 to 1.23, P = 0.48 for AF vs. SR; and HR = 1.13, 95% CI 1.10 to 1.16, P < 0.001, per 1000 ng/L increase of NT‐proBNP).
Sensitivity analyses
Results were confirmed after adjusting for covariates in multivariable Cox regression analyses of the matched sample. Again, there was no significant difference in survival between patients with AF and SR (P = 0.66), while increasing NT‐proBNP levels were associated with increasing mortality (P < 0.001). In addition, survival analyses were repeated in another sample of patients who were matched with respect to NT‐proBNP, eGFR, and LVEF. Baseline characteristics of the second matched sample are presented in Table S2. Similar to the results from our main analyses, no difference in mortality was noted between CHF patients with AF and those with SR (HR = 1.04, 95% CI 0.68 to 1.59, P = 0.86, for AF vs. SR), whereas NT‐proBNP retained its prognostic power (HR = 1.13 per 1000 ng/L increase, 95% CI 1.07 to 1.19, P < 0.001). Again, results were confirmed after adjusting for covariates in a multivariable Cox regression analyses: there was no significant difference in survival between patients with AF and SR (P = 0.89), while increasing NT‐proBNP levels were associated with increasing mortality (P = 0.03). Finally, survival analyses were repeated in a third sample of patients who were matched with respect to NT‐proBNP, age, sex, and eGFR. Baseline characteristics of the third matched sample are presented in Table S3. Again, no difference in mortality was noted between CHF patients with AF and those with SR (HR = 0.735, 95% CI 0.51 to 1.05, P = 0.09, for AF vs. SR), whereas NT‐proBNP retained its prognostic power (HR = 1.19 per 1000 ng/L increase, 95% CI 1.13 to 1.24, P < 0.001).
Area under the curve and cut‐off
In the matched sample, the area under the ROC curve for NT‐proBNP as a predictor of 1 year mortality was 0.79 in AF patients and 0.80 in SR patients (P = 0.88). Both for AF and SR, cut‐off values for risk prediction were comparable (Table 2). The ROC curves for patients with AF and SR are depicted in Figure 3.
Table 2.
Best cut‐off values of N‐terminal pro brain natriuretic peptide for the prediction of 1 year mortality for matched chronic heart failure patients with atrial fibrillation and sinus rhythm, respectively
| AF | SR | ||||||
|---|---|---|---|---|---|---|---|
| NT‐proBNP (ng/L) | Sensitivity | Specificity | YI | NT‐proBNP (ng/L) | Sensitivity | Specificity | YI |
| 1751 | 0.95 | 0.57 | 0.51 | 2684 | 0.74 | 0.77 | 0.51 |
| 1730 | 0.95 | 0.56 | 0.51 | 1912 | 0.87 | 0.63 | 0.50 |
| 1703 | 0.95 | 0.56 | 0.51 | 2666 | 0.74 | 0.76 | 0.50 |
| 1685 | 0.95 | 0.55 | 0.50 | 1904 | 0.87 | 0.63 | 0.50 |
AF, atrial fibrillation; NT‐proBNP, N‐terminal pro brain natriuretic peptide; SR, sinus rhythm; YI, Youden index.
Figure 3.
ROC curves for patients with AF and SR, respectively (matched sample).
Subgroups
Subgroup analyses with respect to age, sex, aetiology, NYHA functional class, LVEF, eGFR, COPD, diabetes, heart rate, and SBP did not detect any statistically significant difference in survival between matched patients with AF and SR in any of the predefined subgroups (Table 3).
Table 3.
Cox regression analyses for all‐cause mortality in matched chronic heart failure patients with respect to heart rhythm (atrial fibrillation vs. sinus rhythm) in subgroups
| Subgroup | HR | 95% CI | P‐value | |||
|---|---|---|---|---|---|---|
| Age (years) | >65 | 0.89 | 0.57 | — | 1.40 | 0.62 |
| ≤65 | 0.90 | 0.57 | — | 1.41 | 0.64 | |
| Sex | Male | 0.92 | 0.66 | — | 1.29 | 0.62 |
| Female | 0.82 | 0.32 | — | 2.09 | 0.67 | |
| Aetiology | Ischaemic | 0.97 | 0.66 | — | 1.44 | 0.89 |
| Cardiomyopathy | 0.86 | 0.46 | — | 1.64 | 0.65 | |
| Other | 1.30 | 0.42 | — | 3.99 | 0.65 | |
| NYHA | III/IV | 1.05 | 0.66 | — | 1.66 | 0.83 |
| I/II | 0.80 | 0.51 | — | 1.25 | 0.32 | |
| LVEF (%) | >30 | 0.81 | 0.47 | — | 1.42 | 0.47 |
| ≤30 | 1.11 | 0.75 | — | 1.65 | 0.61 | |
| eGFR (mL/min/1.73 m2) | >60 | 0.95 | 0.60 | — | 1.49 | 0.82 |
| ≤60 | 0.94 | 0.55 | — | 1.62 | 0.82 | |
| COPD | Yes | 1.33 | 0.37 | — | 4.73 | 0.66 |
| No | 0.90 | 0.65 | — | 1.26 | 0.55 | |
| Diabetes | Yes | 0.72 | 0.43 | — | 1.22 | 0.22 |
| No | 0.99 | 0.65 | — | 1.51 | 0.96 | |
| Heart rate (b.p.m.) | >75 | 0.71 | 0.45 | — | 1.12 | 0.14 |
| ≤75 | 0.96 | 0.60 | — | 1.54 | 0.87 | |
| SBP (mmHg) | >120 | 0.98 | 0.58 | — | 1.66 | 0.95 |
| ≤120 | 0.86 | 0.57 | — | 1.29 | 0.46 | |
Cut‐offs for age, LVEF, heart rate, and SBP represent the median of the respective variables in the matched sample.
CI, confidence interval; COPD, chronic obstructive pulmonary disease; eGFR, estimated glomerular filtration rate calculated using the Modification of Diet in Renal Disease equation; HR, hazard ratio; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association functional class; SBP, systolic blood pressure.
Discussion
The objective of the present study was to clarify the impact of heart rhythm on the prognostic value of NT‐proBNP in patients with CHF. We therefore analysed survival of 2541 CHF patients with respect to underlying heart rhythm. We repeated our analyses in a sample of 500 matched CHF patients.
Our main findings are as follows:
High levels of NT‐proBNP were associated with higher mortality independent of underlying heart rhythm.
Survival, however, was independent of underlying rhythm after matching for NT‐proBNP, sex, age, NYHA functional class, and aetiology of CHF.
Results were confirmed in multivariable analyses as well as in predefined subgroups with respect to age, sex, aetiology, NYHA functional class, LVEF, eGFR, COPD, diabetes, heart rate, and SBP.
While patients different in rhythm but similar (matched) in NT‐proBNP, sex, age, NYHA functional class, and aetiology of CHF had a similar prognosis, they were still different in severity of heart failure as indicated by a lower LVEF in patients with SR.
We confirmed the known independent prognostic value of NT‐proBNP in patients with CHF.14, 15, 16 Individual patient characteristics such as AF, however, may influence NT‐proBNP levels and consequently affect the prognostic value of NT‐proBNP.14, 17 Although some studies have reported a positive association between AF and increased NT‐proBNP levels,11, 12, 13 the effect of AF on the prognostic value of NT‐proBNP is still under debate.13, 19, 20, 21, 22, 29
This is because AF has been described as an independent determinant of increased NT‐proBNP levels in patients with CHF.13, 20, 21, 29 Although no specific NT‐proBNP cut‐off levels have been established for risk prediction in CHF, some authors conclude that the threshold level for outcome prediction of NT‐proBNP is higher in patients with AF than in patients with SR.21 Our data do not support this notion.
The present study significantly extends the data from Rienstra et al.19 in terms of cohort size, extent of disease covered, follow‐up duration, and scrutiny of methodology. We identified NT‐proBNP as an independent determinant of prognosis in advanced CHF irrespective of the underlying rhythm. Then again, Linssen et al. observed that AF was an independent determinant of elevated NT‐proBNP levels in hospitalized heart failure patients with preserved ejection fraction but not in patients with heart failure with reduced ejection fraction (HFrEF).22 Accordingly, the risk of hospitalization for heart failure or death was independent of underlying rhythm in the HFrEF group.22 Interestingly, NT‐proBNP concentrations were similar between HFrEF patients with AF and those with SR in their study. This would rather support our observation that NT‐proBNP retains its prognostic power in CHF patients irrespective of underlying rhythm.
In the present study, the presence of AF was associated with higher mortality in the general sample. However, after matching for NT‐proBNP, age, sex, NYHA functional class, and aetiology, survival was similar in patients with AF and SR, while NT‐proBNP retained its prognostic power. Results were confirmed after adjusting for additional covariates including LVEF as well as in subgroup analyses.
Interestingly, matched patients with AF had less severe heart failure than had those with SR as indicated by a lower LVEF in patients with SR. At the same time, NT‐proBNP retained its prognostic power independent of the underlying heart rhythm. This finding would support the interpretation that both the presence of AF and a low LVEF represent certain complementary aspects of ‘advanced state’ in CHF. It is this advanced state that is reflected by an increase of NT‐proBNP levels. NT‐proBNP release has been associated with cardiac stress such as wall stress, neuroendocrine activation, vascular stiffness, high blood pressure, wall motion abnormalities, and AF. It may therefore integrate the ‘risk sum’ of progress/advance in the course of CHF—with LVEF or AF as (to a certain extent) mutually complementary indicators of its state. Our study therefore also argues against the adjustment of NT‐proBNP cut‐off levels in patients with AF.
There is an ongoing discussion about NT‐proBNP‐guided management of patients with stable CHF.30, 31, 32 One major requirement for a practicable clinical use of NT‐proBNP as a leading instrument in the therapy of CHF would be a universal interpretability. As our work shows, the prognostic information of NT‐proBNP is independent of the underlying heart rhythm. The present study therefore adds important information to this discussion.
Limitations
Owing to the observational nature of our study, the reported relationships between heart rhythm, NT‐proBNP concentrations, and mortality are associative and not causal. As with any non‐randomized design, we cannot rule out hidden bias due to unmeasured confounders. However, our study includes a well‐characterized patient sample with continuous inclusion and close surveillance. As we analysed data from a university hospital setting, patient selection may be biased, and thus, included patients may differ from the general population. For example, the matched sample comprised only 80 (16%) women. Therefore, results in this subgroup may only be interpreted with caution. Moreover, our data do not allow differentiation between paroxysmal, permanent, and persistent AF in patients classified as having AF. We therefore cannot comment on differences between these types of AF. Then again, it was reported that the different types of AF do not significantly affect NT‐proBNP concentrations.19, 33 Then again, we found that 22 (1.0%) patients with SR in the general sample and 6 (2.4%) patients with SR in the matched sample had a history of paroxysmal AF. As these patients were with SR at the index visit, they were classified in the SR group. As the proportion of patients with paroxysmal AF in the SR groups was thus very low, this may not have a major impact on our results.
In the matching process, 181 AF patients without a matching SR partner were removed from the matched sample. This may have caused a skewed selection, because AF patients with very high NT‐proBNP levels may have been excluded from further analyses. As patients with NT‐proBNP levels >10 000 ng/L are underrepresented in the matched sample (3.2% of the AF patients in the matched sample vs. 9.5% of the AF patients in the general sample), our findings have to be interpreted carefully in this subgroup. All‐cause mortality was the primary endpoint of the present analysis. As we do not have any information about implantation of extracorporeal membrane oxygenation or left ventricular assist devices during follow‐up, we cannot comment on results in this subgroup.
Conclusions
In patients with CHF, NT‐proBNP predicts all‐cause mortality independent from underlying heart rhythm. As the prognostic value of NT‐proBNP is not influenced by concomitant AF, there is no need to adjust NT‐proBNP‐levels in these patients. Presence of AF, on the other hand, balances the negative prognostic effect of advanced CHF as indicated by a lower LVEF. Our findings strengthen the importance of NT‐proBNP as a prognostic marker in patients with CHF, while they would favour the interpretation that AF is a surrogate of a worse cardiac condition rather than an independent cardiac risk factor.
Conflict of interest
None declared.
Funding
We acknowledge financial support by Deutsche Forschungsgemeinschaft within the funding programme Open Access Publishing, by the Baden‐Württemberg Ministry of Science, Research and the Arts and by Ruprecht‐Karls‐Universität Heidelberg
Supporting information
Table S1. Median NT‐proBNP levels (ng/l) with respect to LVEF and heart rhythm in the matched and general sample, respectively.
Table S2. Baseline characteristics of the general patient sample and of CHF patients matched with respect to NT‐proBNP, eGFR and LVEF (stratified by rhythm).
Table S3. Baseline characteristics of the general CHF patient sample and of CHF patients matched with respect to NT‐proBNP, age, sex and eGFR (stratified by rhythm).
Acknowledgements
We thank Ms. Maike Hornig and Ms. Karin Hornig for their great commitment and expertise in organizing and managing the Heidelberg Heart Failure Registry.
Schnorbach J., Fröhlich H., Täger T., Corletto A., Katus H. A., and Frankenstein L. (2019) N‐terminal pro brain natriuretic peptide eliminates the prognostic effect of atrial fibrillation in patients with chronic heart failure, ESC Heart Failure, 6, 640–648. 10.1002/ehf2.12464.
References
- 1. Ho KK, Pinsky JL, Kannel WB, Levy D, Pitt B. The epidemiology of heart failure: the Framingham Study. J Am Coll Cardiol 1993; 22: A6–A13. [DOI] [PubMed] [Google Scholar]
- 2. Kannel WB, Wolf PA, Benjamin EJ, Levy D. Prevalence, incidence, prognosis, and predisposing conditions for atrial fibrillation: population‐based estimates. Am J Cardiol 1998; 82: 2N–9N. [DOI] [PubMed] [Google Scholar]
- 3. Parén P, Schaufelberger M, Björck L, Lappas G, Fu M, Rosengren A. Trends in prevalence from 1990 to 2007 of patients hospitalized with heart failure in Sweden. Eur J Heart Fail 2014; 16: 737–742. [DOI] [PubMed] [Google Scholar]
- 4. Schnabel RB, Yin X, Gona P, Larson MG, Beiser AS, McManus DD, Newton‐Cheh C, Lubitz SA, Magnani JW, Ellinor PT. 50 year trends in atrial fibrillation prevalence, incidence, risk factors, and mortality in the Framingham Heart Study: a cohort study. Lancet 2015; 386: 154–162. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. McMurray JJ, Adamopoulos S, Anker SD, Auricchio A, Böhm M, Dickstein K, Falk V, Filippatos G, Fonseca C, Gomez‐Sanchez MA. ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure 2012. Eur J Heart Fail 2012; 14: 803–869. [DOI] [PubMed] [Google Scholar]
- 6. Camm AJ, Kirchhof P, Lip GY, Schotten U, Savelieva I, Ernst S, Van Gelder IC, Al‐Attar N, Hindricks G, Prendergast B, Heidbuchel H, Alfieri O, Angelini A, Atar D, Colonna P, De Caterina R, De Sutter J, Goette A, Gorenek B, Heldal M, Hohloser SH, Kolh P, Le Heuzey J‐Y, Ponikowski P, Rutten FH. Guidelines for the management of atrial fibrillation. Eur Heart J 2010: 31: 2369–2429. [DOI] [PubMed] [Google Scholar]
- 7. Heist EK, Ruskin JN. Atrial fibrillation and congestive heart failure: risk factors, mechanisms, and treatment. Prog Cardiovasc Dis 2006; 48: 256–269. [DOI] [PubMed] [Google Scholar]
- 8. Wang TJ, Larson MG, Levy D, Vasan RS, Leip EP, Wolf PA, D'Agostino RB, Murabito JM, Kannel WB, Benjamin EJ. Temporal relations of atrial fibrillation and congestive heart failure and their joint influence on mortality: the Framingham Heart Study. Circulation 2003; 107: 2920–2925. [DOI] [PubMed] [Google Scholar]
- 9. Dries D, Exner D, Gersh B, Domanski M, Waclawiw M, Stevenson L. Atrial fibrillation is associated with an increased risk for mortality and heart failure progression in patients with asymptomatic and symptomatic left ventricular systolic dysfunction: a retrospective analysis of the SOLVD trials. J Am Coll Cardiol 1998; 32: 695–703. [DOI] [PubMed] [Google Scholar]
- 10. Dickinson O, Chen LY, Francis GS. Atrial fibrillation and heart failure: intersecting populations, morbidities, and mortality. Heart Fail Rev 2014; 19: 285–293. [DOI] [PubMed] [Google Scholar]
- 11. Marsiliani D, Buccelletti F, Carroccia A, Gilardi E, Silveri NG, Franceschi F. Natriuretic peptides and atrial fibrillation. Eur Rev Med Pharmacol Sci 2010; 14: 855–860. [PubMed] [Google Scholar]
- 12. Silvet H, Young‐Xu Y, Walleigh D, Ravid S. Brain natriuretic peptide is elevated in outpatients with atrial fibrillation. Am J Cardiol 2003; 92: 1124–1127. [DOI] [PubMed] [Google Scholar]
- 13. Corell P, Gustafsson F, Kistorp C, Madsen LH, Schou M, Hildebrandt P. Effect of atrial fibrillation on plasma NT‐proBNP in chronic heart failure. Int J Cardiol 2007; 117: 395–402. [DOI] [PubMed] [Google Scholar]
- 14. Balion C, Don‐Wauchope A, Hill S, Santaguida PL, Booth R, Brown JA, Oremus M, Ali U, Bustamam A, Sohel N, McKelvie R, Raina P. Use of natriuretic peptide measurement in the management of heart failure. Report No.: 13(14)-EHC118-EF. Rockville (MD): Agency for Healthcare Research and Quality (US), November 2013. [PubMed]
- 15. Panagopoulou V, Deftereos S, Kossyvakis C, Raisakis K, Giannopoulos G, Bouras G, Pyrgakis V, W Cleman M. NTproBNP: an important biomarker in cardiac diseases. Curr Top Med Chem 2013; 13: 82–94. [DOI] [PubMed] [Google Scholar]
- 16. Di Angelantonio E, Chowdhury R, Sarwar N, Ray KK, Gobin R, Saleheen D, Thompson A, Gudnason V, Sattar N, Danesh J. B‐type natriuretic peptides and cardiovascular risk systematic review and meta‐analysis of 40 prospective studies. Circulation 2009; 120: 2177–2187. [DOI] [PubMed] [Google Scholar]
- 17. Oremus M, Don‐Wauchope A, McKelvie R, Santaguida PL, Hill S, Balion C, Booth R, Brown JA, Ali U, Bustamam A. BNP and NT‐proBNP as prognostic markers in persons with chronic stable heart failure. Heart Fail Rev 2014; 19: 471–505. [DOI] [PubMed] [Google Scholar]
- 18. Rossi A, Enriquez‐Sarano M, Burnett JC, Lerman A, Abel MD, Seward JB. Natriuretic peptide levels in atrial fibrillation: a prospective hormonal and Doppler‐echocardiographic study. J Am Coll Cardiol 2000; 35: 1256–1262. [DOI] [PubMed] [Google Scholar]
- 19. Rienstra M, Van Gelder IC, Van den Berg MP, Boomsma F, Van Veldhuisen DJ. Natriuretic peptides in patients with atrial fibrillation and advanced chronic heart failure: determinants and prognostic value of (NT‐)ANP and (NT‐pro)BNP. Europace 2006; 8: 482–487. [DOI] [PubMed] [Google Scholar]
- 20. Jug B, Šebeštjen M, Šabovič M, Pohar M, Keber I. Atrial fibrillation is an independent determinant of increased NT‐proBNP levels in outpatients with signs and symptoms of heart failure. Wien Klin Wochenschr 2009; 121: 700–706. [DOI] [PubMed] [Google Scholar]
- 21. Dini FL, Gabutti A, Passino C, Fontanive P, Emdin M, De Tommasi SM. Atrial fibrillation and amino‐terminal pro‐brain natriuretic peptide as independent predictors of prognosis in systolic heart failure. Int J Cardiol 2010; 140: 344–350. [DOI] [PubMed] [Google Scholar]
- 22. Linssen G, Rienstra M, Jaarsma T, Voors AA, Gelder IC, Hillege HL, Veldhuisen DJ. Clinical and prognostic effects of atrial fibrillation in heart failure patients with reduced and preserved left ventricular ejection fraction. Eur J Heart Fail 2011; 13: 1111–1120. [DOI] [PubMed] [Google Scholar]
- 23. Clerico A, Carlo Zucchelli G, Pilo A, Passino C, Emdin M. Clinical relevance of biological variation: the lesson of brain natriuretic peptide (BNP) and NT‐proBNP assay. Clinical Chemical Laboratory Medicine 2006; 44: 366–378. [DOI] [PubMed] [Google Scholar]
- 24. Wu AH, Smith A. Biological variation of the natriuretic peptides and their role in monitoring patients with heart failure. Eur J Heart Fail 2004; 6: 355–358. [DOI] [PubMed] [Google Scholar]
- 25. Wedel H, McMurray JJ, Lindberg M, Wikstrand J, Cleland JG, Cornel JH, Dunselman P, Hjalmarson Å, Kjekshus J, Komajda M. Predictors of fatal and non‐fatal outcomes in the Controlled Rosuvastatin Multinational Trial in Heart Failure (CORONA): incremental value of apolipoprotein A‐1, high‐sensitivity C‐reactive peptide and N‐terminal pro B‐type natriuretic peptide. Eur J Heart Fail 2009; 11: 281–291. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Pocock SJ, Wang D, Pfeffer MA, Yusuf S, McMurray JJ, Swedberg KB, Östergren J, Michelson EL, Pieper KS, Granger CB. Predictors of mortality and morbidity in patients with chronic heart failure. Eur Heart J 2006; 27: 65–75. [DOI] [PubMed] [Google Scholar]
- 27. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Ann Intern Med 1999; 130: 461–470. [DOI] [PubMed] [Google Scholar]
- 28. Youden WJ. Index for rating diagnostic tests. Cancer 1950; 3: 32–35. [DOI] [PubMed] [Google Scholar]
- 29. Shelton RJ, Clark AL, Goode K, Rigby AS, Cleland JG. The diagnostic utility of N‐terminal pro‐B‐type natriuretic peptide for the detection of major structural heart disease in patients with atrial fibrillation. Eur Heart J 2006; 27: 2353–2361. [DOI] [PubMed] [Google Scholar]
- 30. Balion C, McKelvie R, Don‐Wauchope AC, Santaguida PL, Oremus M, Keshavarz H, Hill SA, Booth RA, Ali U, Brown JA. B‐type natriuretic peptide‐guided therapy: a systematic review. Heart Fail Rev 2014; 19: 553–564. [DOI] [PubMed] [Google Scholar]
- 31. De Vecchis R, Esposito C, Di Biase G, Ariano C, Giasi A, Cioppa C. B‐type natriuretic peptide‐guided versus symptom‐guided therapy in outpatients with chronic heart failure: a systematic review with meta‐analysis. J Cardiovasc Med 2014; 15: 122–134. [DOI] [PubMed] [Google Scholar]
- 32. Pfisterer M, Buser P, Rickli H, Gutmann M, Erne P, Rickenbacher P, Vuillomenet A, Jeker U, Dubach P, Beer H. BNP‐guided vs symptom‐guided heart failure therapy: the Trial of Intensified vs Standard Medical Therapy in Elderly Patients With Congestive Heart Failure (TIME‐CHF) randomized trial. JAMA 2009; 301: 383–392. [DOI] [PubMed] [Google Scholar]
- 33. Jaekel K, Schley P, Sause A, Müller M, Fueth R, Scheffold T, Guelker H, Horlitz M. Plasma levels of NT‐pro‐BNP in patients with atrial fibrillation before and after electrical cardioversion. Z Kardiol 2005; 94: 795–800. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1. Median NT‐proBNP levels (ng/l) with respect to LVEF and heart rhythm in the matched and general sample, respectively.
Table S2. Baseline characteristics of the general patient sample and of CHF patients matched with respect to NT‐proBNP, eGFR and LVEF (stratified by rhythm).
Table S3. Baseline characteristics of the general CHF patient sample and of CHF patients matched with respect to NT‐proBNP, age, sex and eGFR (stratified by rhythm).