Brain Natriuretic Peptide and the Risk of Cardiovascular Events and Death in Patients with Atrial Fibrillation

BNP levels during AF (both paroxysmal AF attack chronic AF) are 2-3 times higher compared to during SR, and BNP level AF is the sum of the BNP level from the ventricle (reflecting left ventricular function) and the atrium to atrial wall stress). Atrial Fibrillation-Basic Research and Clinical Applications is designed to provide a comprehensive review and to introduce outstanding and novel researches. This book contains 22 polished chapters and consists of five sections: 1. Basic mechanisms of initiation and maintenance of atrial fibrillation and its pathophysiology, 2. Mapping of atrial fibrillation and novel methods of signal detection. 3. Clinical prognostic predictors of atrial fibrillation and remodeling, 4. Systemic reviews of catheter-based/surgical treatment and novel targets for treatment of atrial fibrillation and 5. Atrial fibrillation in specific conditions and its complications. Each chapter updates the knowledge of atrial fibrillation, providing state-of-the art for not only scientists and clinicians who are interested in electrophysiology, but also general cardiologists.


Introduction
Brain natriuretic peptide (BNP) is a hormone that is secreted by the heart, especially from the ventricle (Sudoh et al., 1988;Yasue et al., 1994). The plasma BNP concentrations (BNP levels) are correlated positively with the left ventricular end-diastolic pressure and negatively with the left ventricular ejection fraction Maeda et al., 1998), so BNP levels should be measured to evaluate left ventricular function. BNP levels have proved to be good markers of congestive heart failure. In addition, BNP levels are useful in screening test for left ventricular dysfunction and also heart disease. Some studies have shown that BNP levels have a prognostic value of mortality and morbidity in patients with chronic heart failure (Maeda et al., 2000;Anand et al., 2003), in general populations (Wang et al., 2004) and in clinical practice (Tsuchida & Tanabe, 2008). However, the prognostic value of BNP levels in patients with atrial fibrillation (AF) is not well known. This study investigated the relations of BNP levels to cardiovascular events and death in patients with AF. Hohnloser et al. suggested that warfarin therapy was needed in patients with paroxysmal AF similarly as in those with sustained AF (Hohnloser et al., 2007). Another report showed that, if sinus rhythm was maintained with antiarrhythmic therapy, the prognosis of the patients with paroxsmal AF for ischemic stroke was better than those with permanent AF (Komatsu et al., 2004). We examined the necessity of warfarin therapy in patients with paroxysmal AF as in those with chronic AF. Furthermore, CHADS 2 score is known to be very useful to decide the indication of warfarin therapy in patients with AF (Gage et al., 2001). The patients with CHADS 2 score of 2 or more are recommended to take warfarin therapy, those with CHADS 2 score of 1 to take Warfarin or antiplatelet drugs, and those with CHADS 2 score of 0 need not any take warfarin. We investigated the usefulness of BNP as an aid in CHADS 2 score to decide the indication of warfarin therapy in patients with CHADS 2 score of 0 or 1. Patients with chronic AF had a mean CHADS 2 score of 1.76 compared with 1.23 in those with paroxysmal AF (p<0.0001). The reason for the higher mean CHADS 2 score in patients with chronic AF was probably because of older age, the presence of more structural heart disease (valvular disease, DCM, prior pacemaker operation and CHF) and prior stroke. The prior use of warfarin, wntiplatelet drugs, beta-blocker, thiazide and antialdosterone drugs was significantly higher in chronic AF. And the prior use of antiarrhythmic drugs was significantly higher in paroxysmal AF.

Kaplan-Meier curves for the endpoints and Incidence of death or cardiovascular events
Patients were stratified into two groups based on cut-off level of BNP (100 pg/ml), and a cumulative cardiovascular event-free curve was constructed according to Kaplan-Meier analysis. Cumulative cardiovascular event-free rate, as evaluated by Kaplan-Meier analysis, was significantly lower with a BNP level≥100 pg/ml (p<0.0001) (Figure 1). Similarly, in secondary analyses (cardiovascular mortality, all-cause mortality, heart failure, ischemic stroke, development of paroxysmal AF into chronic AF), cumulative survival rate (eventfree rate) was significantly lower with a BNP level≥100 pg/ml ( Figure 2, Figure 3, Figure4, Figure 5, Figure 6). But only with regard to coronary heart disease events, the cumulative event-free rate was not significantly associated with the BNP level.

CHADS 2 Score and Incidence of Ischemic stroke
Based on Kaplan-Meier analysis of five groups stratified by CHADS 2 score (0, 1, 2, 3, 4-6) in Figure 10, it was found that as CHADS 2 score was higher, the cumulative event-free rate for ischemic stroke decreased significantly (p<0.0001). As detailed in Table 3, there was the number of prior use of warfarin and the incidence of ischemic stroke, stratified by CHADS 2 score.  In patients with CHADS 2 score of 0 or 1, a BNP level≧100 pg/ml was associated with a HR (95% CI) of 3.84 (1.18-12.47) for ischemic stroke compared with a BNP < 100 pg/ml (p=0.0254) (Figure 11).

BNP and the risk of cardiovascular events and death in patients with AF
BNP is a hormone that is secreted by the heart, especially from the ventricle (Sudoh et al., 1988;Yasue et al., 1994), and BNP levels are useful in diagnosis and screening for left ventricular dysfunction and heart failure. Furthermore, during AF, BNP was known to be secreted mainly from the atrium in response to atrial wall stretch (Inoue et al., 2000). Our previous study in outpatients with paroxysmal AF showed that BNP levels during AF attack were increased 2.4 times compared with BNP levels during sinus rhythm (SR) (Tsuchida & Tanabe, 2004). Another study on electric defibrillation in patients with chronic AF showed an increase of about three times during AF than during SR after electric defibrillation (Ohta et al., 2001). These studies revealed that BNP levels during AF (both paroxysmal AF attack and chronic AF) are 2-3 times higher compared to during SR, and therefore BNP level during AF is the sum of the BNP level from the ventricle (reflecting left ventricular function) and the atrium (due to atrial wall stress).
Some studies have shown that BNP levels have a prognostic value of mortality and morbidity in patients with chronic heart failure (Maeda et al., 2000;Anand et al., 2003), in general populations (Wang et al., 2004) and in clinical practice (Tsuchida & Tanabe, 2008). This study shows that BNP level in patients with AF is an important prognostic marker of cardiovascular events, cardiovascular mortality, all-cause mortality, heart failure, ischemic stroke and development of paroxysmal AF into chronic AF, by stratification into two groups based on routinely used cut-off levels of BNP (100pg/ml).
The 14 year follow-up study of paroxysmal AF (Kato T et al., 2004) revealed that paroxysmal AF eventually developed into chronic AF in 132 of 171 patients (77.2%, 5.5% of patients per year), despite changing the drugs as necessary, and the development ratio was significantly increased by aging, an enlarged left atrium, myocardial infarction and valvular disease. In this study (5.4 year follow-up), paroxysmal AF developed into chronic AF in 39 of 231 patients (16.9%, 3.1% of patients per year), and the development ratio was significantly higher in patients with a BNP≧100 pg/ml than a BNP<100 pg/ml. It is conceivable that the reason for lower development ratio in this study than in the former study, is because of less myocardial infarction (5% in this study versus 11% in the former study) and valvular disease (10% versus 20%).
In patients with coronary heart disease, BNP levels were definitely associated with acute phase and outcome of myocardial infarction (Morita et al., 1993;Bibbins-Domingo et al, 2003;Morrow et al., 2003;Suzuki et al., 2004). However, in this study, we did not find an association between BNP levels and the risk of coronary heart disease events in patients with AF, reflecting a similar finding in the report of the Framingham study in a communitybased population (Wang et al., 2004). Hohnloser et al. (2001) suggested that Warfarin therapy was needed in patients with paroxysmal AF, similarly as in those with sustained AF (Hohnloser et al., 2007). Meanwhile, another study showed that, if SR was maintained with antiarrhythmic therapy, the prognosis of the patients with paroxysmal AF for ischemic stroke was better than those with permanent AF (Komatsu et al., 2004). In this study, the incidence of ischemic stroke was significantly higher in patients with chronic AF than in those with paroxysmal AF. The reason for it was because of older age, the presence of more structural disease (valvular disease, DCM, prior pacemaker operation and CHF), more prior stroke and higher mean CHADS 2 score in patients with chronic AF. This study showed that the incidence of ischemic stroke was significantly higher with a BNP≧100 pg/ml than with a BNP<100 pg/ml. In addition, in patients with paroxysmal AF, the incidence of ischemic stroke was significantly higher with a BNP≧100 pg/ml than with a BNP<100 pg/ml, whereas in patients with chronic AF, it was not significantly associated with the BNP levels.

Incidence of stroke in patients with paroxysmal AF and chronic AF
In our previous study (Tsuchida & Tanabe, 2004), BNP levels during AF attack in patients with paroxysmal AF are 2.4 times higher (due to atrial wall stretch) compared to during SR, and even an asymptomatic AF attack also showed substantial and significant BNP elevation (median BNP levels: 31 pg/ml during SR, 71 pg/ml during AF attack). These findings suggest that BNP elevation of unknown origin could be attributed to the occurrence of asymptomatic AF attack, and the incidence of AF attack may be higher in paroxysmal AF patients with a BNP≧100 pg/ml, besides the degree of left ventricular dysfunction.
Furthermore, in this study, the incidence of development of paroxysmal AF into chronic AF was significantly higher with a BNP≧100 pg/ml, and so paroxysmal AF with a BNP≧100 pg/ml may be going to develop close to chronic AF.
As to the mechanism of association between elevation of BNP levels and development of ischemic stroke in AF patients, there are a few reports. The recent studies demonstrated that BNP levels correlated negatively with left atrial appendage flow velocity in chronic AF patient, and suggested that pathological changes (such as hypertrophy, fibrosis and inflammation) in the atrial myocardium may also be underlying factors in elevated BNP secretion in patients with poor left atrial appendage function, and so BNP as a reflection of left atrial appendage function may be a useful marker to predict vulnerability to thromboembolism in AF patients (Frustaci et al., 1997;Shimizu et al., 2002). Further study needs to be performed as to the mechanism of association between elevation of BNP levels and development of ischemic stroke in AF patients.

4.3
Warfarin therapy in patients with CHADS 2 score of 0 or 1 CHADS 2 score is known to be very useful to decide the indication of warfarin therapy in patients with AF (Gage et al., 2001). The patients with CHADS 2 score of 2 or more are recommended to take warfarin therapy, those with CHADS 2 score of 1 to take warfarin or antiplatelet drugs, and those with CHADS 2 score of 0 need not take any warfarin. Based on Kaplan-Meier analysis of five groups stratified by CHADS 2 score (0, 1, 2, 3, 4-6), it was found that as CHADS 2 score was higher, the cumulative event-free rate for ischemic stroke decreased significantly. Furthermore, in the patients with CHADS 2 score of 0 or 1, the incidence of ischemic stroke was significantly higher with a BNP≥100 pg/ml than with a BNP < 100 pg/ml. So in the patients with CHADS 2 score of 0 or 1, BNP may be useful as an aid in CHADS 2 score to decide the indication of warfarin therapy for prevention to ischemic stroke.

Study limitations
The study population consisted of 371 outpatients of one local clinic in Japan. Although they were treated according to the accepted guidelines, it was unavoidable that this study showed a certain amount of bias in relation to patient background, diagnosis and treatment.

Conclusion
In patients with AF, BNP levels predicted the risk of cardiovascular events and death, except for coronary heart disease. Patients with chronic AF had a higher risk of ischemic stroke than patients with paroxysmal AF. BNP may be useful as an aid in CHADS 2 score to decide the indication of warfarin therapy in patients with paroxysmal AF and in patients with CHADS 2 score of 0 or 1.

Acknowledgment
We thank our patients; and Nobuo Shirahashi (Clinical Epidemiology, Osaka City Graduate School), for statistical support; Joukichi Suzuki (Suzuki Clinic) and Takashi Tomidokoro