Impact of B-Type Natriuretic Peptide Level on Risk Stratification of Thromboembolism and Death in Patients With Nonvalvular Atrial Fibrillation　― The Hokuriku-Plus AF Registry ―

Abstract

Background: B-type natriuretic peptide (BNP) may be a predictor of stroke risk in patients with nonvalvular atrial fibrillation (NVAF); because heart failure is associated with the incidence of stroke in AF patients. However, limited data exist regarding the association between BNP at baseline and risks of thromboembolic events (TE) and death in NVAF patients.

Methods and Results: We prospectively studied 1,013 NVAF patients (725 men, 72.8±9.7 years old) from the Hokuriku-plus AF Registry to determine the relationship between BNP at baseline and prognosis among Japanese NVAF patients. During the follow-up period (median, 751 days); 31 patients experienced TE and there were 81 cases of TE/all-cause death. For each endpoint we constructed receiver-operating characteristic curves that gave cutoff points of BNP for TE (170 pg/mL) and TE/all-cause death (147 pg/mL). Multivariate analysis with the Cox-proportional hazards model indicated that high BNP was significantly associated with risks of TE (hazard ratio [HR] 3.86; 95% confidence interval [CI] 1.83–8.67; P=0.0003) and TE/all-cause death (HR 2.27; 95% CI 1.45–3.56; P=0.0003). Based on the C-index and net reclassification improvement, the addition of BNP to CHA₂DS₂-VASc statistically improved the prediction of TE.

Conclusions: In a real-world cohort of Japanese NVAF patients, high BNP was significantly associated with TE and death. Plasma BNP might be a useful biomarker for these adverse clinical events.

Atrial fibrillation (AF) is the most common sustained arrhythmia. AF occurs together with many clinical risk factors and the development of AF is also influenced by genetic background.^1,2 AF is a major risk factor for stroke and systemic embolism,³ and the prevention of stroke is crucial for AF patients, because those experiencing a cardioembolic stroke have miserable outcomes.⁴ According to the present Japan Circulation Society (JCS) guidelines, appropriate antithrombotic therapy is recommended for nonvalvular AF (NVAF) patients with a CHADS₂ score ≥2, and dabigatran or apixaban is recommended for those with a CHADS₂ score of 1.⁵ The current European Society of Cardiology (ESC) guidelines recommend oral anticoagulation therapy for NVAF patients with a CHA₂DS₂-VASc score of 2 or more in men, and 3 or more in women.⁶

Congestive heart failure (CHF) is a major component of both the CHADS₂ and CHA₂DS₂-VASc score and is also an independent risk factor for stroke in AF. CHF in CHADS₂ score was defined as recent CHF exacerbation,⁷ whereas the 2016 ESC guidelines showed that CHF indicates signs/symptoms of HF or objective evidence of reduced left ventricular ejection fraction (LVEF).⁶ The 2013 JCS guidelines commented that CHF should be evaluated by signs/symptoms/laboratory findings of HF or drug therapy for CHF.⁵ The difference in HF definition between the guidelines may result in conflicting results according to HF-related stroke risk. B-type natriuretic peptide (BNP) has emerged as a powerful diagnostic tool for detecting acute HF and LV systolic and/or diastolic dysfunction.^8,9 In previous retrospective studies, a high BNP level was associated with left atrial appendage (LAA) thrombi,¹⁰ and it might be a predictor of thromboembolism in patients with NVAF.^11–13 However, there is little data from prospective cohort studies.

The Hokuriku-Plus AF Registry was designed as a multicenter, prospective, and observational cohort study for cardiologists to investigate the present status of anticoagulation treatment, to evaluate the effectiveness and safety of such treatment, and to determine the potential risk factors for thromboembolic events (TE), major bleeding, and the composite endpoints in Japanese patients with NVAF in real-world clinical practice. Here, we assess whether BNP level is associated with future TE events and death in Japanese NVAF patients using a real-world prospective cohort study.

Methods

Study Population

This study observed the principles outlined in the Declaration of Helsinki and was approved by the Ethics Committee for Medical Research of Kanazawa University Graduate School of Medical Science. All participants provided written informed consent. The Hokuriku-plus AF Registry is a multicenter, population-based, prospective cohort study. We enrolled patients who had AF of all types on 12-lead ECG or Holter monitoring at any time. In brief, 1,492 subjects aged 30–94 years were recruited from a total of 19 institutions in Hokuriku and Yokohama area. The baseline enrollment was performed between January 2013 and May 2014, and follow-up examinations were conducted after 2 years.

Paroxysmal AF was defined as AF that lasted >30 s and terminated spontaneously. Persistent AF was defined as AF that lasted >7 days and required either pharmacological therapy or electrical cardioversion for termination. AF refractory to cardioversion or where cardioversion was not attempted was classified as permanent.

Risk Factor Definitions

The CHADS₂ and CHA₂DS₂-VASc stroke risk scores were recorded as the baseline stroke risk. The components of the CHADS₂ score are CHF, hypertension, age ≥75 years, diabetes, and stroke (doubled); those of the CHADS₂-VASc score are CHF, hypertension, age ≥75 years (doubled); diabetes mellitus, stroke/transient ischemic attack (doubled); vascular disease, age from 65 to 74 years, and female sex. CHF was diagnosed if patients had history of hospitalization for HF, had symptoms of HF, or received treatment for HF. Hypertension was diagnosed if peripheral blood pressure was >140/90 mmHg or if the patient took antihypertensive medication. Diabetes was diagnosed if the fasting plasma glucose was >126 mg/dL, if random plasma glucose was >200 mg/dL, if HbA1c was >6.5%, or if the patient received treatment for diabetes mellitus. Vascular diseases were diagnosed if the patient had coronary artery disease, peripheral artery disease, or large vessel disease.

Regarding examination findings, creatinine clearance (CrCl) was estimated using the Cockcroft-Gault formula. Echocardiographic data were collected at the time of entry into the registry. LV end-diastolic and end-systolic dimensions were recorded from M-mode imaging obtained in the parasternal windows at the level of the mitral leaflet. LVEF was determined by the Teichholz method. Left atrial diameter was recorded in the parasternal windows.

Measurement of BNP at Baseline

The BNP levels were measured at individual institutions of the Hokuriku-plus AF Registry, and different approaches to measurement were used. Venous blood samples were collected from a peripheral vein at the time of the baseline survey. BNP levels were measured in an EDTA blood sample using a fluorescence enzyme immunoassay (E-test TOSOH II BNP, Tosoh Bioscience); a chemiluminescent immunoassay (ARCHITECT BNP-JP assay, Abbot Laboratories Diagnostics Division; ADOVIA Centaur BNP assay, Siemens Healthcare Diagnostics Inc.); or a chemiluminescent enzyme immunoassay (MI 02 Shionogi BNP, Shionogi and Co., Ltd; Lumipulse Prestol II, Fujirebio Inc.). The reference range for BNP measured by these assays was the same (≤18.4 pg/mL). Several studies have shown close correlation of the measured BNP concentration among the following 5 assays: E-test TOSOH II BNP, ARCHITECT BNP-JP assay, ADOVIA Centaur BNP assay, MI 02 Shionogi BNP, and Lumipulse Prestol II.^14,15

Outcomes

The primary endpoint of this analysis was TE, including ischemic stroke, transient ischemic attack, and systemic embolism. Stroke was defined as a sudden onset of focal deficit lasting >24 h and further categorized as ischemic or hemorrhagic. Systemic embolism was defined as an acute vascular occlusion outside the brain. Other clinical endpoints included cardiac death, all-cause death, and a composite endpoint of “TE and cardiac or all-cause death” during the follow-up period.

Statistical Analysis

Continuous variables are presented as mean±SD, and categorical variables are presented as percentage. Continuous variables were compared using Student’s t-test for paired data, and categorical variables were compared using Fisher’s exact test. The best cutoff points for BNP in this study were explored, and receiver-operating characteristic (ROC) curves were generated to test the predictive discrimination cutoff to identify association with adverse events during the follow-up. To investigate differences between groups in the cumulative ratio for cardiac events, the occurrence of cardiac events is presented using Kaplan-Meier cumulative survival curves and compared using the log-rank test. Adjusted hazard ratios (HR) and corresponding 95% confidence intervals (CI) of each variable associated with cardiac events were calculated by Cox-proportional hazard model. The differences in the area under the curve (AUC, C-statistic); and the net reclassification improvement (NRI, continuous method) were tested to compare the modified risk score, including high BNP with CHA₂DS₂-VASc score. All statistical analyses were performed using JMP Pro version 10 (SAS Institute, Cary, NC, USA) or R version 3.3.1 (The R Foundation, Vienna, Austria).

Results

Of the 1,492 patients with AF who were enrolled in the Hokuriku-plus AF Registry, 96 were excluded because of mitral stenosis and/or mechanical prosthetic valve. Of the remaining 1,396 patients with NVAF, 383 lacked baseline BNP data, so a total of 1,013 patients (71.6% men, mean age 72.8±9.7 years) with NVAF were included in this analysis (Table 1). The median CHADS₂ and CHA₂DS₂-VASc scores (interquartile range, IQR) were 2 (1–3) and 3 (2–5), respectively. Warfarin was prescribed to 570 patients (56.3%), and direct oral anticoagulants (DOACs) were prescribed to 316 patients (31.2%). The average value of the time in therapeutic range [TTR; international normalized ratio (INR) 2.0–3.0 for patients <70 years old and 1.6–2.6 for ≥70 years old] of warfarin-treated patients was 72.8±21.9%.

Table 1. Baseline Characteristics and Comorbidities of the Study Patients With Nonvalvular AF

	Total (n=1,013)	TE (n=31)	No TE (n=982)	P value
Baseline characteristics
Age, years, mean±SD	72.8±9.7	75.9±6.8	72.7±9.8	0.076
Male (%)	725 (71.6)	21 (67.7)	704 (71.7)	0.69
Body mass index (kg/m2)	23.6±3.8	23.5±4.0	23.6±3.8	0.83
Systolic BP (mmHg)	124.1±17.2	127.4±13.5	124.0±17.3	0.28
Diastolic BP (mmHg)	72.7±12.0	74.5±10.9	72.7±12.0	0.40
Heart rate (beats/min)	74.8±14.9	74.2±16.3	74.8±14.8	0.80
Persistent or permanent AF (%)	677 (66.8)	28 (90.3)	649 (66.1)	0.0033
Comorbidities
HF (%)	386 (38.1)	18 (58.1)	368 (37.5)	0.024
Hypertension (%)	637 (62.9)	21 (67.7)	616 (62.7)	0.71
Diabetes mellitus (%)	296 (28.5)	10 (33.3)	279 (28.4)	0.69
Stroke/TIA (%)	143 (14.1)	12 (38.7)	131 (13.3)	0.0005
Vascular disease (%)	231 (22.8)	12 (38.7)	219 (22.3)	0.047
Cardiomyopathy (%)	113 (11.2)	7 (22.6)	106 (10.8)	0.072
CHADS2 score	2.0±1.3	2.9±1.7	2.0±1.3	0.0004
Median (IQR)	2 (1–3)	3 (2–4)	2 (1–3)
CHA2DS2-VASc score	3.4±1.7	4.5±2.1	3.4±1.7	0.0002
Median (IQR)	3 (2–5)	4 (3–6)	3 (2–5)
Examination findings
Hemoglobin (g/dL)	13.5±1.9	13.6±2.0	13.5±1.9	0.81
Serum Cr (mg/dL)	0.97±0.62	1.05±0.43	0.97±0.62	0.47
Median (IQR)	0.87 (0.73–1.03)	0.96 (0.76–1.19)	0.86 (0.73–1.02)
Calculated CrCl (mL/min)	66.5±28.0	56.8±27.0	66.8±28.0	0.048
Left atrial diameter (mm)	44.9±8.5	47.9±6.2	44.8±8.5	0.055
LVEF (%)	70.5±12.3	71.1±11.9	70.5±12.3	0.80
BNP (pg/mL)	169.0±252.0	239.6±176.6	166.7±253.7	0.11
Median (IQR)	104 (52.1–199.9)	227 (122–305.5)	102 (51.2–194.9)
OAC prescription at baseline
Warfarin (%)	570 (56.3)	25 (80.7)	545 (55.5)	0.0054
DOAC (%)	316 (31.2)	5 (16.1)	311 (31.7)	0.076
No OAC (%)	127 (12.5)	1 (3.2)	126 (12.8)	0.16

AF, atrial fibrillation; BNP, B-type natriuretic peptide; BP, blood pressure; CrCl, creatinine clearance; HF, heart failure; LVEF, left ventricular ejection fraction; DOAC, direct oral anticoagulant; OAC, oral anticoagulant; TE, thromboembolism; TIA, transient ischemic attack.

Baseline characteristics of patients with and without TE are listed in Table 1. Compared with patients without TE, those with TE had significantly persistent or permanent AF [66.1% (649/982 patients) vs. 90.3% (28/31 patients); P=0.0033], HF [37.5% (368/982 patients) vs. 58.1% (18/31 patients); P=0.024], previous stroke/transient ischemic attack [13.3% (131/982 patients) vs. 38.7% (12/31 patients); P=0.0005], and vascular disease [22.3% (219/982 patients) vs. 38.7% (12/31 patients); P=0.047]. Patients with TE had significantly higher mean CHADS₂ and CHA₂DS₂-VASc scores compared with patients without TE (CHADS₂ score, 2.9 vs. 2.0, P=0.0004; CHA₂DS₂-VASc score, 4.5 vs. 3.4, P=0.0002). Compared with patients without TE, those with TE were significantly more likely to be on warfarin therapy [55.5% (545/982 patients) vs. 80.7% (25/31 patients); P=0.0054]. The TTRs of patients with and without TE were comparable (75.3±19.0 and 72.7±22.0, P=0.68).

Median (IQR) BNP was 104 (52.1–199.9) pg/mL. There was no significant difference in baseline BNP between patients with TE and without TE (Table 1). For each endpoint we constructed ROC curves that gave a cutoff point of 170 pg/mL for the primary endpoint of TE (AUC, 0.70); and 147 pg/mL for TE/cardiovascular death (AUC, 0.72) and TE/all-cause death (AUC, 0.70; Figure 1). Comparison of the baseline characteristics of patients with low BNP (<170) and those with high BNP (≥170) is shown in Table 2. Patients with high BNP were older (P<0.0001), more often female (P=0.013); and had a lower body mass index (P=0.0008). The mean CHADS₂ and CHA₂DS₂-VASc scores were higher in those with high BNP (P<0.0001), and persistent or permanent AF, HF, and cardiomyopathy were more common in those with high BNP (all P<0.001). The mean hemoglobin, mean calculated CrCl, and mean LVEF were lower, and the mean LA diameter was larger in those with high BNP (all P<0.0001). Analysis of potential confounding variables revealed no interaction between high BNP (≥170) and cardiomyopathy or LVEF (P for trend, 0.77 and 0.69, respectively).

Figure 1.

ROC curves for BNP levels to predict TE and death show a cutoff point of 170 pg/mL for TE (area under the curve, 0.70) (A); 147 pg/mL for TE/cardiovascular death (area under the curve, 0.72) (B); and 147 pg/mL for TE/all-cause death (area under the curve, 0.70) (C). BNP, B-type natriuretic peptide; ROC, receiver-operating characteristic; TE, thromboembolic events.

Table 2. Baseline Characteristics and Comorbidities of NVAF Patients Categorized According to BNP Level

	Low BNP (<170pg/mL) (n=699)	High BNP (≥170pg/mL) (n=314)	P value
Baseline characteristics
Age, years, mean±SD	71.4±9.9	76.0±8.4	<0.0001
Male (%)	517 (74.0)	208 (66.2)	0.013
Body mass index (kg/m2)	23.9±3.8	23.0±3.8	0.0008
Systolic BP (mmHg)	124.6±16.7	123.2±18.3	0.22
Diastolic BP (mmHg)	72.9±11.7	72.2±12.6	0.39
Heart rate (beats/min)	74.0±14.1	76.7±16.2	0.0082
Persistent or permanent AF (%)	427 (61.1)	250 (79.6)	<0.0001
Comorbidities
HF (%)	192 (27.5)	194 (61.8)	<0.0001
Hypertension (%)	431 (61.7)	206 (65.6)	0.23
Diabetes mellitus (%)	194 (27.8)	95 (30.3)	0.45
Stroke/TIA (%)	92 (13.2)	51 (16.2)	0.21
Vascular disease (%)	156 (22.3)	75 (23.9)	0.63
Cardiomyopathy (%)	61 (8.7)	52 (16.6)	0.0005
CHADS2 score	1.8±1.3	2.5±1.3	<0.0001
Median (IQR)	2 (1–3)	2 (2–3)
CHA2DS2-VASc score	3.1±1.7	4.0±1.6	<0.0001
Median (IQR)	3 (2–4)	4 (3–5)
Examination findings
Hemoglobin (g/dL)	13.7±1.7	13.0±2.0	<0.0001
Serum Cr (mg/dL)	0.90±0.47	1.12±0.84	<0.0001
Median (IQR)	0.84 (0.70–0.99)	0.96 (0.80–1.18)
Calculated CrCl (mL/min)	72.2±28.2	53.9±22.9	<0.0001
Left atrial diameter (mm)	43.7±8.3	47.5±8.3	<0.0001
LVEF (%)	71.7±10.9	67.9±14.6	<0.0001
BNP (pg/mL)	75.8±43.9	376.4±372.0	<0.0001
Median (IQR)	71.9 (40.9–108.3)	277.0 (208.1–380.0)
OAC prescription at baseline
Warfarin (%)	380 (54.4)	190 (60.5)	0.075
DOAC (%)	212 (30.3)	104 (33.1)	0.38
No OAC (%)	107 (15.3)	20 (6.4)	<0.0001

NVAF, nonvalvular AF. Other abbreviations as in Table 1.

The cohort was followed for 2,133 person-years. During the median follow-up (IQR) of 751 (731–879) days, the incidence of major clinical events in the entire cohort was as follows: TE in 31 (1.5 per 100 person-years); TE/cardiovascular death in 55 (2.6 per 100 person-years); and TE/all-cause death in 81 (3.8 per 100 person-years; Table 3). During the follow-up period, TE significantly developed in patients with high BNP (≥170) compared with those with low BNP (<170) [21/314 (3.2 per 100 person-years) vs. 10/699 (0.7 per 100 person-years); P<0.0001] (Table 3). Additionally, both TE/cardiovascular death and TE/all-cause death significantly developed in patients with high BNP compared with those with low BNP [TE/cardiovascular death, 40/374 (5.3 per 100 person-years) vs. 15/639 (1.1 per 100 person-years); P<0.0001; TE/all-cause death, 54/374 (7.1 per 100 person-years) vs. 27/639 (2.0 per 100 person-years); P<0.0001] (Table 3). Additional analyses were performed to test whether or not the association between BNP level and the incidence of major clinical events was robust. During the follow-up period, the crude incidences of TE, TE/cardiovascular death, and TE/all-cause death were higher in patients with a BNP above the median value (104 pg/mL) than in those with a BNP at or below the median value (P=0.0018, P<0.0001, and P<0.0001, respectively; Table S1). The crude incidences of major clinical events among BNP quartiles are shown in Table S2. Among the BNP quartiles, the crude incidences of TE, TE/cardiovascular death, and TE/all-cause death differed, and increased, with the level of BNP (P=0.003, P<0.0001, and P<0.0001, respectively).

Table 3. Incidence of Major Clinical Events According to BNP Levels Dichotomized by the ROC Curve-Derived Cutoff Points

	Overall	Low BNP (BNP <170)	High BNP (BNP ≥170)	Low BNP (BNP <147)	High BNP (BNP ≥147)	P value
No. of patients	1,013	699	314	639	374
No. of patients with TE (per 100 person-years)	31 (1.5)	10 (0.7)	21 (3.2)	–	–	<0.0001
No. of patients with TE/cardiovascular death (per 100 person-years)	55 (2.6)	–	–	15 (1.1)	40 (5.3)	<0.0001
No. of patients with TE/all-cause death (per 100 person-years)	81 (3.8)	–	–	27 (2.0)	54 (7.1)	<0.0001

ROC, receiver-operating characteristic. Other abbreviations as in Table 1.

Unadjusted HRs in the Cox-proportional regression analysis for major clinical events are shown in Table 4. The CHA₂DS₂-VASc score, high BNP, persistent or permanent AF, and CrCl <30 mL/min were significantly associated with increased risk for TE, TE/cardiovascular death, and TE/all-cause death. In the multivariate analysis, after adjustment by CHA₂DS₂-VASc score, a high BNP level remained significantly associated with TE (adjusted HR, 3.86; 95% CI, 1.83–8.67; P=0.0003), TE/cardiovascular death (adjusted HR, 2.88; 95% CI, 1.67–5.09; P=0.0001), and TE/all-cause death (adjusted HR, 2.27; 95% CI, 1.45–3.56; P=0.0003) (Model 1 in Table 4). In model 2 of the multivariate analysis, after adjustment by CHA₂DS₂-VASc score, persistent or permanent AF, OAC prescription, and CrCl <30 mL/min, a high BNP level remained significantly associated with TE (adjusted HR, 3.11; 95% CI, 1.46–7.06; P=0.0031); TE/cardiovascular death (adjusted HR, 2.61; 95% CI, 1.48–4.69; P=0.0009); and TE/all-cause death (adjusted HR, 1.97; 95% CI, 1.24–3.14; P=0.0039) (Table 4). In the multivariate analysis, we sought to determine whether there is a strong correlation between CHA₂DS₂-VASc score and BNP level. There was no interaction between CHA₂DS₂-VASc score and high BNP in the models for the endpoints of TE (P for trend, 0.15); TE/cardiovascular death (P for trend, 0.14); and TE/all-cause death (P for trend, 0.15).

Table 4. Cox-Proportional Regression Model for the Endpoints of TE and Composite of Adverse Cardiovascular Events

Variable	Univariate analysis		Multivariate analysis
	HR (95% CI)	P value	Model 1		Model 2
			HR (95% CI)	P value	HR (95% CI)	P value
TE
CHA2DS2-VASc score	1.45 (1.17–1.80)	0.0007	1.38 (1.12–1.69)	0.0027	1.34 (1.08–1.65)	0.0076
High BNP (≥170 pg/mL)	4.92 (2.37–10.92)	<0.0001	3.86 (1.83–8.67)	0.0003	3.11 (1.46–7.06)	0.0031
Persistent or permanent AF	4.68 (1.66–19.59)	0.043			2.99 (1.03–12.71)	0.044
OAC prescription	4.26 (0.91–75.81)	0.069			2.09 (0.43–37.65)	0.42
CrCl <30 (mL/min)	2.79 (1.04–6.38)	0.043			1.68 (0.61–3.94)	0.29
TE/cardiovascular death
CHA2DS2-VASc score	1.47 (1.25–1.73)	<0.0001	1.46 (1.25–1.70)	<0.0001	1.43 (1.22–1.68)	<0.0001
High BNP (≥147 pg/mL)	4.88 (2.76–9.12)	<0.0001	2.88 (1.67–5.09)	0.0001	2.61 (1.48–4.69)	0.0009
Persistent or permanent AF	2.01 (1.08–4.09)	0.028			1.53 (0.79–3.21)	0.22
OAC prescription	1.15 (0.53–3.00)	0.74			0.69 (0.31–1.86)	0.44
CrCl <30 (mL/min)	2.87 (1.41–5.36)	0.0052			1.61 (0.77–3.08)	0.19
TE/all-cause death
CHA2DS2-VASc score	1.39 (1.22–1.59)	<0.0001	1.42 (1.25–1.61)	<0.0001	1.39 (1.22–1.58)	<0.0001
High BNP (≥147 pg/mL)	3.66 (2.33–5.89)	<0.0001	2.27 (1.45–3.56)	0.0003	1.97 (1.24–3.14)	0.0039
Persistent or permanent AF	1.90 (1.14–3.34)	0.013			1.51 (0.88–2.73)	0.14
OAC prescription	1.28 (0.66–2.89)	0.49			0.83 (0.41–1.92)	0.64
CrCl <30 (mL/min)	3.31 (1.90–5.47)	<0.0001			2.04 (1.15–3.45)	0.015

CI, confidence interval; HR, hazard ratio. Other abbreviations as in Table 1.

Kaplan-Meier curves for the incidence of TE, TE/cardiovascular death, and TE/all-cause death are shown in Figure 2. High BNP was significantly associated with a higher incidence of TE (P<0.001), higher incidence of TE/cardiovascular death (P<0.001), and higher incidence of TE/all-cause death (P<0.001) compared with low BNP. Kaplan-Meier analysis revealed that the cumulative incidence of TE, TE/cardiovascular death, and TE/all-cause death during the study follow-up was significantly higher in patients with a BNP above the median (104 pg/mL) than in those with a BNP at or below the median (P=0.0013, P<0.001, and P<0.001, respectively; Figure S1). The cumulative event rate of TE, TE/cardiovascular death, and TE/all-cause death during the study follow-up increased with increasing BNP quartiles (P=0.0016, P<0.001, and P<0.001, respectively, log-rank test; Figure S2).

Figure 2.

Cumulative incidence of TE and death according to BNP levels dichotomized by the ROC curve-derived cutoff points. Kaplan-Meier curves for the incidence of TE (A), TE/cardiovascular death (B) and TE/all-cause death (C) during the follow-up period according to BNP levels dichotomized by the exclusion cutoff points. BNP, B-type natriuretic peptide; ROC, receiver-operating characteristic; TE, thromboembolic events.

ROC analyses showed that the C-indices for prediction of TE were 0.65 (95% CI, 0.56–0.75) for CHA₂DS₂-VASc score and 0.75 (95% CI, 0.67–0.83) for CHA₂DS₂-VASc score+high BNP (Figure 3, Table 5). The CHA₂DS₂-VASc score+high BNP was superior to the CHA₂DS₂-VASc score in predicting TE in NVAF patients (P=0.017; Table 5). Similarly, the NRI analysis showed significantly improved reclassification when BNP was added to the CHA₂DS₂-VASc score for TE (P=0.00001; Table 5).

Figure 3.

Comparison of ROC analyses for prediction of TE between CHA₂DS₂-VASc and CHA₂DS₂-VASc+high BNP in patients with NVAF. BNP, B-type natriuretic peptide; NVAF, nonvalvular atrial fibrillation; ROC, receiver-operating characteristic; TE, thromboembolic events.

Table 5. Evaluating Increased Predictive Ability of BNP Added to CHA₂DS₂-VASc Score for Detecting TE Using C-Statistic, and NRI Index

Risk score	ROC curve analysis			NRI analysis
	C-index (95% CI)	Z score	P value	NRI (95% CI)	P value
CHA2DS2-VASc	0.65 (0.56–0.75)	−2.39	0.017	0.76 (0.42–1.09)	0.00001
CHA2DS2-VASc+high BNP (≥170)	0.75 (0.67–0.83)

NRI, net reclassification index; ROC, receiver-operating characteristic. Other abbreviations as in Table 1.

Discussion

In this prospective real-world cohort study of NVAF patients, we demonstrated that TE significantly occurred in NVAF patients with high BNP (≥170 pg/mL) compared with those with low BNP. Additionally, NVAF patients with high BNP (≥147 pg/mL) developed TE/cardiovascular death and TE/all-cause death compared with those with low BNP. We also showed that high BNP was an independent predictor of TE and a composite of adverse cardiovascular events in NVAF patients. Moreover, an addition of high BNP to the CHA₂DS₂-VASc score statistically improved prediction for endpoints.

Patients with HF and AF are at an increased risk of TE because abnormal blood flow, frequently in the LAA, and endothelial dysfunction secondary to dilatation and fibrotic changes in the left atrium contribute to a prothrombotic or hypercoagulable state.¹⁶ It has been reported that HF is an independent predictor of stroke/systemic embolism,¹⁷ and there are many scoring systems to predict stroke risk in AF, including the CHADS₂ and CHA₂DS₂-VASc scores, which both have CHF as a major component. The issue of concern is that the criteria of CHF differ in the various guidelines and studies. Therefore, LV systolic dysfunction has been proposed as an objective determinant of HF. A recent systemic review reported that HF defined as impaired LV dysfunction on echocardiography increased the risk of stroke/systemic embolism by 1.7–2.6-fold.¹⁷

BNP is a well-established biomarker for objectively determining HF. It is a neurohormone secreted from cardiac myocytes, mainly in response to cardiac wall stress such as volume or pressure overload. Elevated levels of BNP have been reported in patients with LV hypertrophy, ventricular dilatation, HF, acute coronary syndrome, and AF.¹⁸ Kara et al showed elevated levels of BNP associated with a significant excess of incident AF, independent of traditional AF risk factors, in a general population,¹⁹ and a recent retrospective and single-center study showed that BNP ≥173 pg/mL was the only independent predictor of LAA thrombus in anticoagulated Japanese NVAF patients undergoing preprocedural transesophageal echocardiography.¹⁰ Sadanaga et al¹² reported that 9 (1.8%/year) cases of TE occurred during an average follow-up time of 762±220 days in patients with AF (n=261) treated with warfarin. They showed that a high BNP level (≥200 pg/mL) was a significant predictor of subsequent TE in these patients, important findings that were similar to our findings. However, the number of patients and events in that study were small,¹² whereas the Hokuriku-plus AF Registry is a multicenter and prospective study, and has 4-fold the number of patients, which we believe overcomes the limitations discussed above.

The N-terminal pro-BNP (NT-proBNP) is also known as a biomarker of mortality and morbidity in CHF and ischemic heart disease.⁸ A substudy of RE-LY (Randomized Evaluation of Long-Term Anticoagulation Therapy: n=6,189) showed that elevation of NT-proBNP is common in patients with AF and independently related to increased risk of stroke and death.²⁰ Roldan et al studied 1,172 patients with permanent AF and reported that the NT-proBNP level provided complementary prognostic information to the CHA₂DS₂-VASc score for the prediction of stroke/systemic embolism and all-cause death.²¹ Both NT-proBNP and BNP are produced from proBNP, and their physiologic features are similar. Elevated BNP and NT-proBNP levels correlate with increased risk of incident TE/all-cause death; however, the cutoffs and management based on elevated levels are unclear. Measurement of BNP is useful for establishing diagnosis, prognosis, or disease severity in CHF according to the 2010 JCS guideline for the treatment of HF (Class I recommendation). Additionally, measurement of BNP as well as NT-proBNP level is useful for establishing prognosis or disease severity in chronic HF according to the 2017 ACC/AHA/HFSA Focused Update of the 2013 ACCF/AHA Guideline for the Management of Heart Failure (Class I recommendation and Level A evidence).²² This present study’s results clearly demonstrated that high BNP (≥170 pg/mL) at baseline could predict future TE.

Study Limitations

First, this study included patients on different anticoagulant therapies: 570 patients were on warfarin, 316 patients were taking DOACs, and 127 patients not taking any OAC. TE occurred in 25/570 (4.4%) of those on warfarin, 5/316 (1.6%) with DOACs, and in 1/127 (0.8%) without OAC. However, patients on warfarin therapy were well controlled with high TTR (72.8±21.9%). Second, the BNP values were measured at the individual Hokuriku-plus AF Registry institutions using different methods. Third, several patients lacked baseline BNP data. Patients without this information were immediately excluded from this study, but some selection bias may exist. Finally, it was a prospective and multicenter study with a modest sample size. However, our results showed a strong association of BNP with TE, TE/cardiovascular death and TE/all-cause death, emphasizing the predictive value of BNP.

Conclusions

In this real-world prospective cohort study of Japanese NVAF patients, high plasma BNP level was associated with TE, TE/cardiovascular death, and TE/all-cause death. BNP might be a useful biomarker, either used alone or in combination with established scoring systems, for identifying patients at risk of adverse clinical events, including TE/cardiovascular death/all-cause death. BNP-guided anticoagulant therapy for AF could reduce these events and improve the outcomes of NVAF patients.

Disclosures

M.Y. received lecture fees from Daiichi-Sankyo and Boehringer Ingelheim; M.Y. received research grants from Boehringer Ingelheim and Bayer Healthcare. All other authors report no relationships relevant to the content of this paper. The Hokuriku-plus AF Registry was partially supported by the Practical Research Project for Life-Style related Diseases including Cardiovascular Diseases and Diabetes Mellitus from the Japan Agency for Medical Research and Development, AMED under Grant number JP17ek0210082.

Appendix

The following is a list of the institutions participating in the Hokuriku-plus AF Registry.

Kanazawa University Hospital (Yamagishi M, Fujino N, Nohara A, Kawashiri M, Hayashi K, Sakata K, Yoshimuta T, Konno T, Tsuda T); Ishikawa Prefectural Central Hospital (Matsubara T, Inoue M, Yasuda T, Miwa K); Kanazawa Cardiovascular Hospital (Namura M, Horita Y, Ikeda M, Terai H, Kimura R); Komatsu Municipal Hospital (Ino H, Kaneda T, Takata M); Kaga Medical Center (Inoue T, Tagawa S); Wajima Municipal Hospital (Kita Y); Suzu General Hospital (Koizumi J); KKR Hokuriku Hospital (Itoh Y); Saiseikai Kanazawa Hospital (Araki T, Oe K); JCHO Kanazawa Hospital (Minamoto M, Tanaka Y); Houju Memorial Hospital (Mori K); Toyama Red Cross Hospital (Kaku B, Taguchi T, Katsuta S); Takaoka City Hospital (Hirase H); Kouseiren Takaoka Hospital (Okeie K, Kiyama M, Fujita T, Oota M); Hokuriku Central Hospital (Todo Y); Fukui Prefectural Hospital (Aoyama T, Yamaguchi M, Noji Y); Fukui Cardiovascular Center (Mizuno S, Ohsato K, Misawa K); Yokohama Sakae Kyosai Hospital (Michishita I, Iwaki T, Nozue T, Kato H); and Ishikawa Health Service Association Clinic (Yamagishi M).

Supplementary Files

Supplementary File 1

Figure S1. Cumulative incidence of TE and death according to BNP levels dichotomized by the median values.

Figure S2. Cumulative incidence of TE and death among BNP quartiles.

Table S1. Incidence of major clinical events according to BNP levels dichotomized by the median values

Table S2. Incidence of major clinical events among BNP quartiles

Please find supplementary file(s);

http://dx.doi.org/10.1253/circj.CJ-17-1085

References

• 1.   Hayashi K, Konno T, Tada H, Tani S, Liu L, Fujino N, et al. Functional characterization of rare variants implicated in susceptibility to lone atrial fibrillation. Circ Arrhythm Electrophysiol 2015; 8: 1095–1104.
• 2.   Tada H, Shiffman D, Smith JG, Sjogren M, Lubitz SA, Ellinor PT, et al. Twelve-single nucleotide polymorphism genetic risk score identifies individuals at increased risk for future atrial fibrillation and stroke. Stroke 2014; 45: 2856–2862.
• 3.   Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: The Framingham Study. Stroke 1991; 22: 983–988.
• 4.   Frost L, Andersen LV, Vestergaard P, Husted S, Mortensen LS. Trend in mortality after stroke with atrial fibrillation. Am J Med 2007; 120: 47–53.
• 5.   JCS Joint Working Group. Guidelines for pharmacotherapy of atrial fibrillation (JCS 2013): Digest version. Circ J 2014; 78: 1997–2021.
• 6.   Kirchhof P, Benussi S, Kotecha D, Ahlsson A, Atar D, Casadei B, et al. 2016 ESC Guidelines for the management of atrial fibrillation developed in collaboration with EACTS. Eur Heart J 2016; 37: 2893–2962.
• 7.   Gage BF, Waterman AD, Shannon W, Boechler M, Rich MW, Radford MJ. Validation of clinical classification schemes for predicting stroke: Results from the National Registry of Atrial Fibrillation. JAMA 2001; 285: 2864–2870.
• 8.   Boomsma F, van den Meiracker AH. Plasma A- and B-type natriuretic peptides: Physiology, methodology and clinical use. Cardiovasc Res 2001; 51: 442–449.
• 9.   Maalouf R, Bailey S. A review on B-type natriuretic peptide monitoring: Assays and biosensors. Heart Fail Rev 2016; 21: 567–578.
• 10.   Kawabata M, Goya M, Sasaki T, Maeda S, Shirai Y, Nishimura T, et al. Left atrial appendage thrombi formation in Japanese non-valvular atrial fibrillation patients during anticoagulation therapy: Warfarin vs. direct oral anticoagulants. Circ J 2017; 81: 645–651.
• 11.   Shimizu H, Murakami Y, Inoue S, Ohta Y, Nakamura K, Katoh H, et al. High plasma brain natriuretic polypeptide level as a marker of risk for thromboembolism in patients with nonvalvular atrial fibrillation. Stroke 2002; 33: 1005–1010.
• 12.   Sadanaga T, Kohsaka S, Mitamura H, Ogawa S. Elevated B-type natriuretic peptide level as a marker of subsequent thromboembolic events in patients with atrial fibrillation. Heart Vessels 2011; 26: 530–535.
• 13.   Nakamura M, Koeda Y, Tanaka F, Onoda T, Itai K, Ohsawa M, et al. Plasma B-type natriuretic peptide as a predictor of cardiovascular events in subjects with atrial fibrillation: A community-based study. PLoS One 2013; 8: e81243.
• 14.   Ito Y, Namioka Y. Fundamental evaluation of two BNP assay reagents by chemiluminescent enzyme immunoassay using Lumipulse f and Lumipulse Presto II. Jpn J Med Pharm Sci 2010; 64: 931–939.
• 15.   Yamada M, Kuroda N, Arai T, Kajikawa T, Fujita J, Taminato T. Evaluation of BNP assay “Chemilumi BNP” by chemiluminescent immunoassay. Jpn J Med Pharm Sci 2010; 63: 913–918.
• 16.   Abraham JM, Connolly SJ. Atrial fibrillation in heart failure: Stroke risk stratification and anticoagulation. Heart Fail Rev 2014; 19: 305–313.
• 17.   Agarwal M, Apostolakis S, Lane DA, Lip GY. The impact of heart failure and left ventricular dysfunction in predicting stroke, thromboembolism, and mortality in atrial fibrillation patients: A systematic review. Clin Ther 2014; 36: 1135–1144.
• 18.   Hijazi Z, Oldgren J, Siegbahn A, Granger CB, Wallentin L. Biomarkers in atrial fibrillation: A clinical review. Eur Heart J 2013; 34: 1475–1480.
• 19.   Kara K, Geisel MH, Mohlenkamp S, Lehmann N, Kalsch H, Bauer M, et al. B-type natriuretic peptide for incident atrial fibrillation: The Heinz Nixdorf Recall Study. J Cardiol 2015; 65: 453–458.
• 20.   Hijazi Z, Oldgren J, Andersson U, Connolly SJ, Ezekowitz MD, Hohnloser SH, et al. Cardiac biomarkers are associated with an increased risk of stroke and death in patients with atrial fibrillation: A Randomized Evaluation of Long-term Anticoagulation Therapy (RE-LY) substudy. Circulation 2012; 125: 1605–1616.
• 21.   Roldan V, Vilchez JA, Manzano-Fernandez S, Jover E, Galvez J, Puche CM, et al. Usefulness of N-terminal pro-B-type natriuretic peptide levels for stroke risk prediction in anticoagulated patients with atrial fibrillation. Stroke 2014; 45: 696–701.
• 22.   Yancy CW, Jessup M, Bozkurt B, Butler J, Casey DE Jr, Colvin MM, et al. 2017 ACC/AHA/HFSA Focused Update of the 2013 ACCF/AHA Guideline for the Management of Heart Failure: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines and the Heart Failure Society of America. Circulation 2017; 136: e137–e161.