Article Text

Download PDFPDF

Original research article
Natriuretic peptides for the detection of paroxysmal atrial fibrillation
  1. Joachim Seegers1,2,
  2. Markus Zabel1,
  3. Timo Grüter1,
  4. Antje Ammermann3,
  5. Mark Weber-Krüger1,
  6. Frank Edelmann1,4,5,6,
  7. Götz Gelbrich7,
  8. Lutz Binder8,
  9. Christoph Herrmann-Lingen4,9,
  10. Klaus Gröschel10,
  11. Gerd Hasenfuß1,4,
  12. Nicolas Feltgen3,
  13. Burkert Pieske5,6 and
  14. Rolf Wachter1,4
  1. 1Department of Cardiology and Pneumology, University of Göttingen, Göttingen, Germany
  2. 2Department of Internal Medicine II—Cardiology, University Hospital Regensburg, Regensburg, Germany
  3. 3Department of Ophthalmology, University of Göttingen, Göttingen, Germany
  4. 4German Cardiovascular Research Center (DZHK), Göttingen, Germany
  5. 5Department of Internal Medicine, Cardiology, Charité – Campus Virchow-Klinikum, Universitätsmedizin Berlin, Berlin, Germany
  6. 6German Cardiovascular Research Center (DZHK), partner site Berlin, Germany
  7. 7Institute for Clinical Epidemiology and Biometry, University of Würzburg, Würzburg, Germany
  8. 8Department of Clinical Chemistry, University of Göttingen, Göttingen, Germany
  9. 9Department of Psychosomatic Medicine and Psychotherapy, University of Göttingen, Göttingen, Germany
  10. 10Department of Neurology, Mainz University Medical Center, Mainz, Germany
  1. Correspondence to Dr Rolf Wachter; Wachter{at}


Background and purpose Silent atrial fibrillation (AF) and tachycardia (AT) are considered precursors of ischaemic stroke. Therefore, detection of paroxysmal atrial rhythm disorders is highly relevant, but is clinically challenging. We aimed to evaluate the diagnostic value of natriuretic peptide levels in the detection of paroxysmal AT/AF in a pilot study.

Methods Natriuretic peptide levels were analysed in two independent patient cohorts (162 patients with arterial hypertension or other cardiovascular risk factors and 82 patients with retinal vessel disease). N-terminal-pro-brain natriuretic peptide (NT-proBNP) and BNP were measured before the start of a 7-day Holter monitoring period carefully screened for AT/AF.

Results 244 patients were included; 16 had paroxysmal AT/AF. After excluding patients with a history of AT/AF (n=5), 14 patients had newly diagnosed AT/AF (5.8%) NT-proBNP and BNP levels were higher in patients with paroxysmal AT/AF in both cohorts: (1) 154.4 (IQR 41.7; 303.6) versus 52.8 (30.4; 178.0) pg/mL and 70.0 (31.9; 142.4) versus 43.9 (16.3; 95.2) and (2) 216.9 (201.4; 277.1) versus 90.8 (42.3–141.7) and 96.0 (54.7; 108.2) versus 29.1 (12.0; 58.1). For the detection of AT/AF episodes, NT-proBNP and BNP had an area under the curve in receiver operating characteristic analysis of 0.76 (95% CI, 0.64 to 0.88; p=0.002) and 0.75 (0.61 to 0.89; p=0.004), respectively.

Conclusions NT-proBNP and BNP levels are elevated in patients with silent AT/AF as compared with sinus rhythm. Thus, screening for undiagnosed paroxysmal AF using natriuretic peptide level initiated Holter monitoring may be a useful strategy in prevention of stroke or systemic embolism.

  • BNP brain-type natriuretic peptide

This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See:

Statistics from

Request Permissions

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.

Key messages

What is already known about this subject?

  • The early detection of paroxysmal atrial fibrillation (AF) is a very attractive strategy for primary stroke prevention. Extensive Holter monitoring up to 7 days is useful to detect AF, but it is not feasible to screen the whole population.

How might this impact on clinical practice?

  • A biomarker screening to decide whether or not to perform extended Holter monitoring for the first detection on AF may reduce the number of patients to be screened by Holter. Once AF is detected, >90% of patients need oral anticoagulants for primary stroke prevention. Biomarker level-guided indication of Holter ECG may be an attractive strategy for AF detection and thus primary stroke prevention.

What does this study add?

  • Brain natriuretic peptide and N-terminal-pro-BNP values in patients with paroxysmal AF were significantly higher and showed very good diagnostic properties for AF detection. Therefore, these markers may be very useful for identifying patients with otherwise undetected AF.


Atrial fibrillation (AF) is an established risk factor for ischaemic stroke.1 A meta-analysis indicates that oral anticoagulation therapy with warfarin reduces stroke risk by 64% on an intention to treat basis.2 Identifying paroxysmal AF may be challenging in patients presenting with sinus rhythm, especially as episodes of AF may not be symptom related,3 but the risk for thromboembolism is similar to that for permanent AF.4 Prolonged,5 ,6 or continuous,7 rhythm monitoring may enhance the detection of this clinically ‘silent’ paroxysmal AF. Although such extended diagnostics have been performed in a large number of patients,8 this seems to be not feasible for clinical routine as they are uncomfortable for the patient and time-consuming. Investigation on natriuretic peptides has shown their potential to predict AF within the general population,9 as well as in patients undergoing surgical procedures.10 Moreover, elevation of natriuretic peptide levels was shown in patients suffering from AF,11 ,12 which decreases after conversion to sinus rhythm.13 ,14 Recently, we have shown that the brain natriuretic peptide (BNP) is an independent predictor of paroxysmal AF detected by prolonged ECG monitoring in patients with cerebral ischaemia (secondary stroke prevention),15 Thus, we hypothesised that natriuretic peptide levels may also help to identify AF in patients with cardiovascular risk factors as a potential prevention strategy and aimed to evaluate the diagnostic value in a pilot study.

Materials and methods


Patients participating in this observational trial were recruited from the collective of the ongoing non-interventional diastolic congestive heart failure (Diast-CHF) trial, which is part of the nationwide German Competence Network Heart Failure. Inclusion and exclusion criteria have been described previously.16 Briefly, a total of 1735 patients with at least one risk factor for heart failure with preserved left ventricular function (defined as history of hypertension, diabetes mellitus, sleep apnoea syndrome or atherosclerotic disease) or established CHF were referred by a network of primary care physicians between 2004 and 2006. After the baseline visit which included N-terminal-pro-brain natriuretic peptide (NT-proBNP) analysis, NT-proBNP values were divided into quartiles. Only patients with NT-proBNP in the lowest (NT-proBNP<53 pg/mL) and in the highest NT-proBNP-quartile (NT-proBNP >226 pg/mL) at the baseline visit presenting with sinus rhythm in the 12-lead ECG and without an implanted cardiac device were selected and randomly asked to participate in this substudy which was part of the 1 year follow-up examination of the patients. Patient selection and all clinical examinations were performed by investigators blinded to NT-proBNP values and quartiles. As systolic dysfunction (ejection fraction<50%) is a known cause of elevated natriuretic peptide levels, these patients were excluded. Diast-CHF complies with the Declaration of Helsinki; the protocol was approved by the responsible ethics committee and all patients gave written informed consent. The recruitment process of the Diast-CHF substudy is shown in figure 1.

Figure 1

Flow chart of the recruitment process (diastolic congestive heart failure, Diast-CHF; NT-proBNP, N-terminal-pro-brain natriuretic peptide).

In addition, a second cohort was added using data from patients enrolled in the Find-AF eye study. In brief, Find-AF eye is a single centre study and consecutively enrolled 101 patients with retinal artery occlusion or venous retinal thrombosis. Only patients free of AF on admission ECG and with an evaluable 7-day Holter ECG were included in the analysis of this manuscript (n=82).

Data collection and clinical evaluation

All patients underwent medical history, physical examination, a 12-lead ECG and standard echocardiography at baseline.

Blood analysis

Blood samples were drawn immediately before Holter ECG was started and were stored at −80°C for later analysis. Plasma levels of NT-proBNP were measured using a sandwich immunoassay (Roche Diagnostics, Mannheim, Germany). BNP was measured by means of a sandwich immunoassay (Abbott GmbH & Co. KG, Wiesbaden, Germany). N-terminal pro atrial natriuretic peptide (NT-proANP) was measured by using a sandwich immunoassay (Biomedica Medizinprodukte, Vienna, Austria). Personnel responsible for the determination of natriuretic peptide levels were blinded to clinical patient data.

Holter ECG

A dual-channel Holter ECG was recorded with digital portable recorders (Diast-CHF: Lifecard CF, Del Mar Reynolds Medical Ltd, Hertford, UK, and Find-AF eye: getemed, Teltow, Germany) for seven consecutive days. Analysis was performed with Pathfinder digital (Software Version V8.602, Del Mar Reynolds Medical Ltd) or Cardioday (getemed). Episodes with sudden rapid changes in heart rate or sudden onset of arrhythmia in the manual review of the Holter ECG were carefully analysed for AF and atrial flutter (atrial tachycardia, AT). AT/AF was diagnosed if episodes lasted more than 30 s. Equivocal findings were cograded by an additional physician blinded to the data.

Definition of primary end point

The primary hypothesis was that higher natriuretic peptide levels can discriminate patients with paroxysmal AT/AF from those without AT/AF. We measured both NT-proBNP and BNP (which is measured in many institutions instead of NT-proBNP).

Statistical analyses

Continuous data are shown as mean±SD. Values of natriuretic peptides were not normally distributed and are therefore given as median and IQR of 25th and 75th centile. Categorical variables are given as absolute numbers (percentage). Receiver operating characteristic (ROC) curves were used to describe test characteristics. Statistical tests were performed with SPSS Statistics 21.0 (IBM, Chicago, Illinois, USA). A p value below 0.05 was considered to be significant.


Study population

Of the patients included in the Diast-CHF trial, a total of 615 patients were eligible in principle; the number of Holter recorder systems and the analysis time available was the most relevant restriction to patient recruitment. Of the 200 patients asked to participate, 167 provided written informed consent. Five patients were excluded for technical problems during the Holter recording (figure 1). Of the remaining 162 patients, 156 (96%) had 7 days of recording, whereas the other 6 patients had at least 4 days of recording. The Find-AF eye cohort consisted of 82 patients with an evaluable 7 day Holter ECG.

The baseline characteristics of the 162 patients of the Diast-CHF substudy and the additional 82 patients taken from the Find-AF eye study are displayed in table 1. Five (2%) patients with a history of AF in the Diast-CHF cohort were excluded for further analysis. In both cohorts combined, 14 (5.8%) patients had AT/AF for at least 30 s during Holter monitoring (table 2). Median duration (IQR) of AT/AF was 24.5 (3.3–1046.8) min, and eight patients had an episode of at least 6 min (ASSERT criteria).17

Table 1

Patient characteristics

Table 2

AT/AF burden, prior antithrombotic therapy and CHA2DS2-VASc Score28 of the 14 patients with AT/AF detected by prolonged Holter monitoring

Natriuretic peptides and paroxysmal AF

Table 3 shows the median and IQR of BNP and NT-proBNP levels. NT-proANP was only measured in Find-AF eye. Plasma levels of natriuretic peptides were numerically higher in paroxysmal AT/AF patients in both cohorts.

Table 3

Natriuretic peptide plasma levels in patients with and without paroxysmal AT/AF (5 patients with a history of AF were excluded from the analysis)

Diagnostic utility of natriuretic peptides

For ROC analysis, data of both cohorts were pooled to determine the diagnostic utility of natriuretic peptides for the detection of paroxysmal AT/AF. Figure 2 shows the ROC curves for BNP and NT-proBNP; area under curve was 0.75 (95% CI, 0.61 to 0.89; p=0.004) and 0.76 (95% CI, 0.64 to 0.88; p=0.002), respectively.

Figure 2

Receiver operating characteristics curve of N-terminal pro-atrial natriuretic peptide plasma levels (dark line) and brain natriuretic peptide (dashed line) for the detection of paroxysmal atrial tachycardia and atrial fibrillation.

Clinical relevance

The finding of paroxysmal AF in our study was of clinical relevance. In 13 of the 14 patients, the detection of paroxysmal AF would have led to a change in antithrombotic therapy according to recent guidelines18 (from no therapy to oral anticoagulation in 5 patients and from aspirin to oral anticoagulation in 8 patients).


The major finding of this study is that natriuretic peptide plasma levels are higher in patients suffering from paroxysmal AT/AF compared with those without AT/AF in two different independent cohorts. Although natriuretic peptides are mainly used for the diagnosis of heart failure,19 they are also elevated in a number of different diseases (eg, asymptomatic systolic and diastolic dysfunction,20 pulmonary embolism21 and other diseases).

In addition, natriuretic peptides have also been shown to be elevated in patients with AF as compared to patients with sinus rhythm,14 Moreover, natriuretic peptides have been shown to predict future AF in several clinical settings,9 ,22 Pathophysiological, increased haemodynamic load causes atrial stretch, which is well known as a reason for development of AT/AF.23 Natriuretic peptides correlate with haemodynamic load;19 thus, their indicative value for diagnosis of unknown AT/AF may be equal or superior to clinical risk factors of AF.

Recently, we showed that natriuretic peptides are elevated in patients with stroke with undiagnosed paroxysmal AF as compared with sinus rhythm.15 We now extend these findings to two different clinical settings: patients with risk factors for heart failure and patients with retinal vessel occlusion.

The clinical relevance of early detection of the so-called ‘silent AT/AF’ was recently underlined by the ASSERT study which revealed that patients with short episodes of AF (>6 min within 3 months) have a 2.5-fold increase in the risk of stroke and systemic embolism.17 The early detection of subclinical AF in patients at risk for AF is a major challenge and potentially a very effective stroke prevention strategy.17 ,24 Recently, a stepwise screening approach was reported in elderly patients to identify (asymptomatic) paroxysmal AF. During a 2-week recording of 20–30 s two times per day using a hand-held ECG in 75-year-old patients without a history of AF, paroxysmal AT/AF was identified in only 4%, indicating that a large number of patients have to be screened to diagnose one patient with paroxysmal AT/AF.8 Natriuretic peptides were not used in this study for preselecting study participants. In daily life, risk of AT/AF development are commonly determined first by clinical risk factors (eg, age) followed by a standard 12-lead ECG. However, these workups will identify a small number of patients with ‘silent’ AT/AF.8 ,25 Thus, if AT/AF is assumed but not diagnosed, identifying those patients with an increased risk of paroxysmal AT/AF using natriuretic peptides seems highly attractive.

Strengths and limitations

Our study is strengthened by its use and detailed analysis of a continuous 7-day Holter ECG for the evaluation of paroxysmal AF/AT allowed detecting a high number of affected patients, which is higher than that diagnosed by periodic ECGs and cardiac event loop recorders,26 and similar findings in two different independent cohorts. Our findings are in line with our prior study in a third cohort.15 However, the sample size was too small to perform a detailed analysis on the independence of the indicative value of natriuretic peptide plasma levels for paroxysmal AT/AF, and the indicative value of BNP found in the present analysis has to be validated in larger trials. Furthermore, we cannot exclude that more intensive monitoring (eg, by an implantable loop recorder) would have detected a larger number of patients with paroxysmal AF. However, the initial selection of patients with Diast-CHF, which was performed by the lowest and highest quartiles of the baseline NT-proBNP plasma level of the Diast-CHF trial during the baseline visit, may cause an overestimation of the discriminatory value.

The amount of paroxysmal AF which leads to an increase in stroke risk is not known. Although there is now some evidence that patients with episodes lasting longer than 6 min are at increased risk (ASSERT),17 the risk of stroke may already be elevated in patients with excessive supraventricular ectopic activity.27 Otherwise, it is not known whether the risk in these patient populations is similar to that in patients with permanent AF and whether these patients benefit similarly from oral anticoagulation.


In two independent patient cohorts, natriuretic peptide levels were higher in patients with AF/AT as compared to patients with sinus rhythm. It may therefore be worthwhile to further evaluate the utility of natriuretic peptides in identifying patients at risk for AF/AT to be chosen for further workup (eg, prolonged Holter monitoring) as a tool for stroke prevention.



  • Contributors JS and RW wrote the paper. TG collected the data. MZ, AA, MW-K, FE, LB, LB, CH-L, KG, GH, NF and BP provided significant input and improvement of the whole paper. GG performed statistical analysis.

  • Funding This study was supported by an unrestricted grant from Roche Diagnostics. This work was supported by grants from the German Federal Ministry of Education and Research (German Heart Failure Network, TP 7 (FKZ 01GI0205)).

  • Competing interests None declared.

  • Ethics approval Ethik-Kommission der Universitätsmedizin Göttingen.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Consent for publication of raw data not obtained at time of study start (2004).