Previous studies have noted a correlation between the presence of atrial fibrillation (AF) and elevated brain natriuretic peptide hormone level, although the exact nature of this association is unclear. Understanding the relation between AF and brain natriuretic peptide may enhance care for this patient population. The aim of this study was to establish the relationship between AF burden and N-terminal pro–brain natriuretic peptide (NT-proBNP) level. One hundred eighty-four patients who presented to the University of North Carolina electrophysiology clinic with AF underwent baseline questionnaires, laboratory testing (including NT-proBNP), echocardiography, and 1-week ambulatory rhythm monitoring. Multivariate linear regression was used to determine the association between AF burden and NT-proBNP level. Increased AF burden was associated with increased NT-proBNP level, and this association remained significant after adjusting for possible confounders. Compared with a 0% AF burden, those with an AF burden of 1% to 25% had a nearly 1.5-fold increase (p = 0.102), those with an AF burden of 26% to 99% had a nearly fourfold increase (p <0.001), and those with an AF burden of 100% had a nearly 4.5-fold increase (p <0.001). In conclusion, AF burden as assessed by continuous 1-week ambulatory rhythm monitoring is directly associated with NT-proBNP level. NT-proBNP may act as a useful surrogate for assessing AF burden.
Previous studies have noted a correlation between the presence of atrial fibrillation (AF) and elevated brain natriuretic peptide (BNP) level, although the exact nature of this relation is not entirely clear. Restoration of sinus rhythm by electrical cardioversion and radiofrequency ablation has also been demonstrated to affect BNP level. However, most previous studies have been small, have relied on patient-reported symptoms as markers for AF burden, and have not adequately controlled for known causes of fluctuating BNP level. In addition, previous studies have not considered the affect of AF burden as a continuous variable, which may prove useful in monitoring patients before and after therapy. To further evaluate this association, we performed a substudy of the Symptom Mitigation in Atrial Fibrillation (SMART) registry. We hypothesized that adjusted N-terminal pro-BNP (NT-proBNP) level may directly correlate with AF burden as assessed by ambulatory rhythm monitoring.
Methods
The SMART study is a prospective cohort outcomes study of stable outpatients recruited through the electrophysiology clinic at the University of North Carolina at Chapel Hill. Patients were eligible to participate if they had ≥1 episode of AF not attributable to a reversible cause documented by electrocardiography or ambulatory rhythm monitoring and had not undergone previous AF ablation. The protocol was approved by the appropriate institutional review boards, and all participants provided written informed consent.
From September 2008 to November 2011, all 184 subjects from the SMART study with a complete data set necessary for analysis were enrolled in this substudy. Entry data were obtained by chart review, questionnaire, physical exam, and laboratory studies. All patients underwent baseline 1-week ambulatory rhythm monitoring and transthoracic echocardiography. The pattern of AF was classified as persistent if the patient had 100% AF burden on baseline monitor or a personal history of cardioversion as assessed by questionnaire. Glomerular filtration rate was calculated using the Cockcroft-Gault equation. From the transthoracic echocardiographic images, left atrial diameter was calculated in 1 dimension from a standard parasternal long-axis view, and the left ventricular ejection fraction was calculated using the standard Simpson’s method of discs.
Our primary predictor variable was AF burden as measured by 1-week continuous ambulatory rhythm monitoring at the time of enrollment (ACT III, Lifewatch Corporation, Rosemont, Illinois; or Zio Patch, iRhythm Technologies, Inc., San Francisco, California). AF burden, which was calculated as the percentage of total time while wearing the monitor that the participant was in AF, excluding artifacts, was divided into 4 categories. Participants who had no documented AF by 1-week monitoring were grouped together (0% AF burden). Those who were persistent throughout the duration of the monitoring were grouped together (100% AF burden). The remaining patients were cut at the median of 25% AF burden into 2 additional categories, 1% to 25% and 26% to 99% AF burden.
Our primary outcome variable was baseline NT-proBNP, drawn at the time the 1-week monitor was placed. Nonfasting blood samples were drawn from the antecubital vein, and NT-proBNP was assayed using Vitros NT-proBNP (Ortho Clinical Diagnostics, Rochester, New York). Normal NT-proBNP levels were defined in accordance with the reference standard established at McLendon Clinical Laboratories, University of North Carolina Hospital.
Univariate statistics were used to examine means, SDs, and shapes of distributions for continuous variables and frequencies for categorical variables. No imputation of missing data was performed, given the small overall number of missing variables. We then performed bivariate comparisons of the primary predictor variable in addition to other baseline characteristics, with NT-proBNP as a categorical outcome (normal vs elevated NT-proBNP) using Student’s t tests for continuous variables and chi-square analyses for categorical variables. Because NT-proBNP had a very skewed distribution, NT-proBNP was then log-transformed before further analyses. The correlation of ln NT-proBNP with AF burden was assessed by Pearson’s correlation analysis. We then analyzed the relation of our primary predictor variable, AF burden in 4 categories, with ln NT-proBNP as a continuous variable in bivariate and multivariate linear regression models. Potential confounders were identified on the basis of review of existing research and bivariate analyses as detailed previously. The partial F test was used to exclude potential confounders in analyses. For ease of presentation, ln NT-proBNP was then transformed back after analyses. Analyses were performed with Stata version 11 (StataCorp LP, College Station, Texas). Statistical tests were 2 tailed, with p values <0.05 considered significant.
Results
Baseline characteristics of our patient population are listed in Table 1 . Approximately half of the patients (91 of 184) were in AF when seen in the outpatient clinic at the time the NT-proBNP level was drawn and the 1-week monitor was placed. The mean recording time by continuous monitor was 7.7 ± 2.9 days. The mean baseline AF burden was 48.5%, although the distribution was relatively bimodal, with approximately 1/3 of participants having 0% AF burden and 1/3 having 100% AF burden during the monitoring period.
| Variable | Overall (n = 184) | Normal NT-proBNP (<178 pg/ml) (n = 50) | Elevated NT-proBNP (≥178 pg/ml) (n = 134) | p Value |
|---|---|---|---|---|
| Age (yrs) | 61.5 ± 13.2 | 55.5 ± 15.4 | 64.2 ± 11.4 | <0.001 |
| Men | 119 (65%) | 37 (74%) | 82 (61%) | 0.106 |
| White | 130 (71%) | 32 (64%) | 98 (73%) | 0.204 |
| Body mass index (kg/m 2 ) | 32.3 ± 8.1 | 31.7 ± 7.1 | 32.8 ± 8.5 | 0.414 |
| Hypertension | 109 (59%) | 26 (52%) | 83 (62%) | 0.047 |
| Diabetes mellitus | 33 (18%) | 7 (13%) | 26 (19%) | 0.422 |
| History of stable angina, myocardial infarction, or coronary revascularization | 30 (16%) | 7 (13%) | 23 (17%) | 0.518 |
| History of congestive heart failure | 27 (15%) | 2 (4%) | 25 (19%) | 0.029 |
| Current smokers | 18 (10%) | 6 (12%) | 12 (9%) | 0.690 |
| β blockers use | 108 (59%) | 24 (48%) | 84 (63%) | 0.074 |
| Nondihydropyridine calcium channel blockers use | 47 (26%) | 8 (16%) | 39 (29%) | 0.054 |
| Angiotensin-converting enzyme inhibitors or angiotensin II receptor blockers use | 74 (40%) | 16 (32%) | 58 (43%) | 0.108 |
| Statins use | 78 (42%) | 16 (32%) | 62 (46%) | 0.120 |
| Persistent AF | 113 (61%) | 15 (30%) | 98 (73%) | <0.001 |
| AF burden (%) | 49.9 ± 45.6 | 5.6 ± 15.9 | 67.3 ± 41.5 | <0.001 |
| Overall heart rate (beats/min) | 77.6 ± 16.5 | 79.0 ± 18.1 | 77.2 ± 16.0 | 0.604 |
| Ejection fraction (%) | 56.7 ± 8.8 | 59.7 ± 4.2 | 55.3 ± 9.8 | 0.003 |
| Left atrial diameter (cm) | 4.3 ± 0.8 | 3.9 ± 0.6 | 4.5 ± 0.8 | <0.001 |
| Hemoglobin (mg/dl) | 14.2 ± 1.9 | 14.7 ± 1.9 | 13.9 ± 1.9 | 0.019 |
| Glomerular filtration rate (ml/min) | 111.6 ± 47.1 | 119.9 ± 44.6 | 108.5 ± 48.3 | 0.155 |
In unadjusted analyses, ln NT-proBNP had a strong direct correlation with baseline AF burden (r = 0.63, p <0.001; Figure 1 ). Compared with a 0% AF burden by continuous monitoring, each increase in AF burden category was associated with an increase in NT-proBNP ( Figure 2 ). In multivariate analyses, using partial F tests to reduce the predictive model, this association of NT-proBNP with AF burden persisted after adjusting for potential confounding variables (age, gender, left atrial size, β-blocker use, and history of congestive heart failure). Compared with a 0% AF burden, those with an AF burden of 1% to 25% had a nearly 1.5-fold increase (p = 0.102), those with an AF burden of 26% to 99% had a nearly fourfold increase (p <0.001), and those with an AF burden of 100% had a nearly 4.5-fold increase (p <0.001) in NT-proBNP ( Table 2 ).
