- Email: [email protected]
A noninvasive index of atrial remodeling in patients with paroxysmal and persistent atrial fibrillation: a pilot study
Available online at www.sciencedirect.com
Journal of Electrocardiology 45 (2012) 109 – 115 www.jecgonline.com
A noninvasive index of atrial remodeling in patients with paroxysmal and persistent atrial fibrillation: a pilot study☆ Antonio Vincenti, MD, a Matteo Rota, b Monica Spinelli, MD, a Mariella Corciulo, MD, a Sergio De Ceglia, MD, a Giovanni Rovaris, MD, a Laura Antolini, PhD, b Simonetta Genovesi, MD c, d,⁎ a
b
Electrophysiology and Cardiac Pacing Unit, S. Gerardo Hospital, Monza, Italy Department of Clinical Medicine and Prevention, Centre of Biostatistics for Clinical Epidemiology, University of Milano-Bicocca, Monza, Italy c Department of Clinical Medicine and Prevention, University of Milano-Bicocca, Monza, Italy d Nephrology Unit, S. Gerardo Hospital, Monza, Italy Received 18 April 2011
Abstract
Purpose: This study aims to develop a noninvasive atrial remodeling index (RI) to separate patients presenting paroxysmal atrial fibrillation (ParAF) from those with sustained persistent atrial fibrillation (PerAF), that is, AF episodes interrupted 7 days or more after the onset. Methods: Signal-averaged P-wave duration (SAPWd) and left atrial anteroposterior diameter (LADd) were measured in 33 ParAF patients, in 26 sustained PerAF patients, and in 18 control subjects. By using SAPWd and LADd, a dichotomous (0/1) RI was created. A logistic regression model on the probability of having a sustained PerAF vs a ParAF episode was estimated, including the RI, sex, age, and cardiac comorbidities as covariates. Results: Signal-averaged P-wave duration was significantly longer in sustained PerAF (153 ± 15 milliseconds) than in ParAF patients (142 ± 13 milliseconds, P b .001) and in both ParAF and sustained PerAF groups vs control group (123 ± 7 milliseconds, P b .001). Left atrial anteroposterior diameter was larger both in sustained PerAF (43 ± 6 mm) vs ParAF patients (38 ± 5 mm, P = .002) and in sustained PerAF group vs control group (38 ± 2 mm, P = .004), but no differences were observed between ParAF patients and controls (P = .6). A 12-fold increase (odds ratio, 11.8; 95% confidence interval, 2.2-63.5) in the odds of having a sustained PerAF vs a ParAF episode was observed in patients with RI equal to 1. Conclusions: P-wave duration and left atrium diameter enabled to define a noninvasive atrial RI to separate patients with ParAF from those with sustained PerAF. This could be a useful tool to select a suitable strategy for AF treatment. © 2012 Elsevier Inc. All rights reserved.
Keywords:
Paroxysmal atrial fibrillation; Persistent atrial fibrillation; Signal-averaged P-wave duration; Atrial diameter; Atrial remodeling; Remodeling index
Introduction Atrial fibrillation (AF) is a frequent arrhythmia associated with almost all heart diseases. It can cause complications such as heart failure and thromboembolism, but it is also present in apparently healthy hearts. The incidence of AF is increasing, particularly in the elderly, with an impact on economic resources required for the management of patients with AF. 1,2 ☆
Conflict of interest: none declared. ⁎ Corresponding author. Dipartimento di Medicina Clinica e Prevenzione, Università di Milano-Bicocca, Via Cadore 48, 20052, Monza, Italy. E-mail address: [email protected] 0022-0736/$ – see front matter © 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.jelectrocard.2011.08.005
An AF episode is defined paroxysmal (ParAF) when it spontaneously stops within 7 days from the onset and persistent (PerAF) when pharmacologic or electrical interventions are needed for interruption. 3 Atrial fibrillation episodes are often interrupted with electrical or pharmacologic cardioversion within 7 days from the onset. In such a case, it is not possible to know the natural history of the arrhythmia. This raises the question on whether the episode in patients who underwent intervention within 7 days from the onset would have resolved spontaneously. The distinction between ParAF and PerAF is crucial because PerAF is normally associated to structural and electrophysiologic modifications that make the arrhythmia more sustained. A
110
A. Vincenti et al. / Journal of Electrocardiology 45 (2012) 109–115
dilatation of the atria and an increase of fibrous components are often present in PerAF. From the electrophysiologic point of view, a shortening of the effective refractory period and a reduction of the conduction velocity of the electrical impulse through the atrium are observed. These phenomena are known as structural and electrical atrial remodeling. 4,5 The P-wave duration, measured with the signal-averaged electrocardiogram (ECG) (SAPWd), is an expression of the intra-atrial conduction time, 6 whereas the left atrial anteroposterior diameter at end-diastole (LADd) is a measure of atrial dimension. The modifications of intra-atrial conduction time and of atrial dimensions are characteristics of atrial remodeling. The aim of the present pilot study is to create a noninvasive index using SAPWd and LADd to distinguish between patients presenting ParAF or sustained PerAF, the latter defined as AF episodes lasting more than 7 days from the onset and not self-terminating. Methods Groups of patients The study considered 95 subjects. Seventy-seven patients, presenting an AF episode in the previous month, were enrolled consecutively (from January 2009 to June 2010) in the Electrophysiology and Cardiac Pacing Unit of the San Gerardo Hospital (Monza, Italy). Patients were classified according to the type of AF episode as: ParAF group (n = 33): patients with at least 1 selfterminating, symptomatic, and/or asymptomatic episode of AF, documented by 12-lead electrocardiogram (ECG) or 24 hour ECG Holter. Sustained PerAF group (n = 26): patients with an AF episode interrupted by pharmacologic or electrical intervention after at least 7 days from the onset. Early terminated PerAF group (n = 18): patients with an AF episode interrupted by pharmacologic or electrical intervention within 7 days from the onset. The remaining subjects (n = 18) were controls (control group). These were subjects without any previous history of heart disease, hypertension, diabetes mellitus, and/or AF. In the AF groups, P-wave duration was evaluated after discontinuation of 1-C drugs for 5 half-lives, 1 month after the AF episode or the electrical cardioversion. Patients taking amiodarone and those who had recurrences of the arrhythmia within a month after cardioversion were not included in the study. The study was approved by an institutional review committee, and the subjects gave informed consent. Electrocardiographic analysis The ECG recording was performed by a digital Holter recorder (SpiderView; Sorin Group Company, Sorin CRM, Saluggia, Vercelli, Italy). Signal-averaged P-wave duration was obtained by the P-wave analysis software (Sorin CRM, Saluggia, Vercelli, Italy) derived from the work of
Stafford et al. 6 The P-wave analysis was based on ECG obtained with recorders that allow acquisition of 3 bipolar orthogonal leads (x, y, and z), sampled at 1000 Hz. The user had to choose a reference P wave to be used to compute the averaging template. The method consisted in automatically aligning different P waves with the reference one by looking for the position of maximal mutual correlation. P-wave complexes that did not match the reference P wave were automatically rejected. The selected P waves were then filtered by a 20- to 50Hz bandpass filter. For each orthogonal lead, averaged highresolution P-wave signal was obtained, and the vector magnitude of the 3 channels was calculated (x 2 + y 2 + z 2) 0.5. The duration of the P wave was obtained by automatic estimation of the beginning and the end of the P wave. Approximately, 150 to 200 beats were averaged. Cardiac ultrasound examination Echocardiograms were obtained in the standard precordial positions using digital echocardiography equipment (Aloka ProSound SSD Alpha 10; Aloka Co, Ltd, Tokyo, Japan) with 1- to 5-MHz transducers. The recommendations for standard measurements from M-mode echocardiograms were followed. 7 Instantaneous measurements were made during 3 cardiac cycles, and the average value of LADd from a parasternal long-axis view was obtained for each subject. Both ECG analysis and cardiac ultrasound examination were performed in the same day, 1 month after cardioversion (sustained PerAF group), or after episode documentation (ParAF group). The operator was blinded with respect to the patient AF classification group.
Statistical methods Continuous variables were described in terms of mean and standard deviation, and categorical variables, as percentages. Baseline characteristics of ParAF, sustained PerAF, and controls were compared by using the t test for continuous variables and the χ 2 test for categorical variables. Correlation between continuous variables was evaluated by the Pearson correlation coefficient. The probability of having a sustained PerAF vs a ParAF episode was modeled through a multivariate logistic regression analysis, including the noninvasive parameters of atrial remodeling (SAPWd and LADd) as covariates. A continuous remodeling index (CRI) was computed for each AF patient from the model-based linear predictor as a linear combination between the estimated regression coefficients β1 and β2 and the individual values of SAPWd and LADd as follows: α + β1·SAPWd + β2·LADd (where α was the model intercept). Atrial fibrillation patients presenting with greater values of the CRI were more likely to develop sustained PerAF instead of ParAF episodes. A cutpoint c was chosen to dichotomize the CRI result to define a criterion to divide AF episodes into sustained PerAF or ParAF. The cutpoint selection was performed by a systematic search of the central 80% of the observed values of the CRI with the aim of choosing the one associated with the minimum P value (or maximum χ 2 statistic). 8 The chosen cutpoint for dichotomization was
A. Vincenti et al. / Journal of Electrocardiology 45 (2012) 109–115
111
Table 1 Baseline characteristics of 95 subjects Characteristics
ParAF (n = 33)
Sustained PerAF (n = 26)
Early terminated PerAF (n = 18)
(ParAF vs sustained PerAF P value) Sex, males
18 (55%)
20 (77%)
(ParAF vs control P value) 11 (61%)
(.08) Age (y)
62 (9)
LADd (mm)
38 (5)
SAPWd (ms)
142 (13)
69 (10)
62 (12)
43 (6)
41 (5)
153 (15)
154 (16)
(.08) 38 (2)
(.6)
(.004) 21 (63%)
(.26) 61 (16)
(.73)
(.002)
13 (72%)
7 (27%)
0 (0%)
3 (12%)
2 (11%)
7 (27%)
0 (0%)
3 (12%)
1 (6%)
17 (65%)
6 (33%)
16 (62%)
3 (17%)
10 (38%)
6 (33%)
9 (35%)
5 (28%)
(b.0001) 123 (7)
(b.0001) 17 (65%)
(Sustained PerAF vs control P value)
11 (61%) (.65)
(.02)
Hypertension
Control (n = 18)
(b.0001)
(.89) Coronary artery disease
4 (12%) (.15)
Diabetes mellitus
5 (15%) (.69)
Heart failure
0 (0%) (.002)
History of stroke or TIA
1 (3%) (.2)
Other comorbidities
19 (57%) (.54)
OAT
8 (24%) (.004)
ACE-I
9 (27%) (.36)
ARB
11 (33%) (.92)
CHADS2 score
1.2 (1.2)
1.6 (1.2)
1.1 (0.9)
(.18) Values are expressed as expressed as number of patients (%) for categorical variables and mean (SD) for continuous variables. OAT indicates oral anticoagulation therapy; ACE-I, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptors blockers; CHADS2, congestive heart failure, hypertension, age > 75 years, diabetes, stroke, TIA/thrombo-embolism.
equivalent to the one obtained maximizing the difference between sensitivity and 1 minus specificity. To avoid the problem of inflation of type I error rate arising from multiple testing in the cutpoint selection, we applied the MillerSiegmund P value correction formula. 9 Following this approach, a dichotomous remodeling index (RI) was created. The criteria identified a sustained PerAF condition when the CRI was greater than or equal to c (RI = 1), whereas ParAF was diagnosed when the CRI was less than c (RI = 0). A logistic regression model on the probability of having a sustained PerAF vs a ParAF episode was estimated including the binary RI, sex, age, and other cardiac comorbidities as covariates. All variables were tested for interactions. Stepwise model reduction technique was used. Model performance was assessed using a receiver operating characteristic curve (ROC) for discrimination and using Hosmer-Lemeshow test for calibration. 10 The ROC curve shows the trade-off between sensitivity and specificity varying the cutoff on the linear predictor of the adjusted model. A model is considered to be highly discriminant when the area under the ROC curve (AUC) is close to 1 and not discriminant if the area is close to 0.5. The Hosmer-Lemeshow test relies on the contrast between the observed and model expected number of sustained PerAF episodes in risk groups defined by the deciles of the model linear predictor.
P b .05 was considered statistically significant. All performed tests were 2 sided. All analyses were performed with SAS 9.1.2 (SAS Institute, Cary, NC) and using R version 2.11.0 for graphics.
Results Table 1 shows the baseline characteristics of the 95 AF subjects enrolled in the study. Sustained PerAF were older than ParAF patients (69 ± 10 vs 62 ± 9 years, P = .02). Signalaveraged P-wave duration was significantly longer in sustained PerAF than in ParAF patients (153 ± 15 vs 142 ± 13 milliseconds, P = .004) and in both ParAF and sustained PerAF group vs control group (123 ± 7 milliseconds), with P b .001. The anatomical parameter LADd was larger both in sustained PerAF vs ParAF patients (43 ± 6 vs 38 ± 5 mm, P = .002) and in the sustained PerAF group vs the control group (38 ± 2 mm) with P b .001, but no differences were observed between ParAF patients and controls (P = .6). The AF groups of patients did not differ regarding the prevalence of hypertension, diabetes mellitus, coronary artery disease, and history of stroke or transient ischemic attack (TIA), whereas in the sustained PerAF group, a higher fraction of subjects showed heart failure (P = .002).
A. Vincenti et al. / Journal of Electrocardiology 45 (2012) 109–115
190
ParAF (n = 33)
SAPWd (msec)
150
110
130
r=0.05
170
190 170
r=0.44
110
SAPWd (msec)
B
Control (n = 18)
150
A
130
112
30
35
40
45
50
30
35
D
50
190
Early terminated PerAF (n = 18)
130
170
r=0.60
150
150
SAPWd (msec)
170
r=0.35
110
130
190
Sustained PerAF (n = 26)
110
SAPWd (msec)
45
LADd (mm)
LADd (mm)
C
40
30
35
40
45
50
30
35
LADd (mm)
40
45
50
LADd (mm)
Fig. 1. Scatter plot of the relationship between SAPWd and LADd measures within each group. A, Control group; B, ParAF group; C, sustained PerAF group; D, early terminated PerAF group. In all panels, r denotes the Pearson correlation coefficient.
0
1
2
cutpoint c = −0.9 of the CRI was associated with a maximum χ 2 of 12.8 and an uncorrected P value of .0003. Using the Miller and Siegmund formula, the corrected P value was .01. Fig. 2 shows the distribution
Continuous Remodeling Index
The prevalence of anticoagulation therapy was higher (P = .004) in the sustained PerAF than in ParAF patients. Angiotensin-converting enzyme inhibitor use (P = .36) and angiotensin receptor blocker use (P = .92) did not differ among sustained PerAF and ParAF. No statistically significant differences were observed on the Congestive heart failure, Hypertension, Age > 75 years, Diabetes, Stroke, TIA/Thrombo-embolism (CHADS2) score in ParAF and sustained PerAF (P = .18). The relationship between SAPWd and LADd measures is depicted within each group in Fig. 1. These variables can be noticed to be modestly correlated in the control group (Fig. 1A; r = 0.44, P = .07), whereas this correlation completely disappeared in ParAF (Fig. 1B; r = 0.05, P = .78). In the sustained PerAF group, the 2 variables were still weakly correlated (Fig. 1C; r = 0.35, P = .07). In univariate logistic regression analyses, both LADd (odds ratio [OR], 1.17; 95% confidence interval [CI], 1.051.29) and SAPWd (OR, 1.06; 95% CI, 1.02-1.11) were significantly related to the probability of having a sustained PerAF vs a ParAF episode. The CRI, defined by the model linear predictor, was CRI = −12 + 0.13·LADd + 0.05·SAPWd. The optimal
ParAF (n = 33)
Sustained PerAF (n = 26)
Fig. 2. Box-plot representation of the CRI by AF episode. The box extends from the 25th to the 75th percentile. The inner line of the box marks the median, and whisker lines out of the box represent the minimum and the maximum observed values.
A. Vincenti et al. / Journal of Electrocardiology 45 (2012) 109–115
P .0009 .04 .05
0.8 0.6
95% CI 2.2-63.5 1.1-20.3 1.0-16.6
0.4
OR 11.8 4.8 4.0
Sensitivity
RI (1 vs 0)a Sex (male vs female) Age ≥69 y
1.0
Table 2 Results from multivariate logistic regression on the probability of having a sustained PerAF vs a ParAF episode
113
RI = 1 if −12 + 0.13·LADd (millimeters) + 0.05·SAPWd (milliseconds) ≥ −0.9, and RI = 0 otherwise.
0.0
AUC=0.81
0.0
0.2
0.4
0.6
0.8
1.0
1−Specificity
Fig. 3. Receiver operating characteristic curve of the logistic regression model predicting probability of having a sustained PerAF vs a ParAF episode as a function of RI, sex, and age.
probability of 60% ranging from 3% to 85% was estimated from the multivariate model. Discussion The present study shows that in a group of patients with AF, the P-wave duration was longer in sustained PerAF than
A
Age >= 69 years
0
0.2
0.4
0.6
P(Sustained PerAF)
0.8
1
male Age < 69 years
RI=0
RI=1
RI=0
RI=1
female Age < 69 years
Age >= 69 years
0
0.2
0.4
0.6
0.8
1
B
P(Sustained PerAF)
of the CRI in sustained PerAF and ParAF patients. One can note that patients with sustained PerAF episodes were associated with higher values of the CRI than ParAF patients. The binary RI, defined according to the cutpoint c = −0.9, classifies subjects as at risk for being sustained PerAF or ParAF depending on whether CRI is −0.9 or greater (RI = 1) or CRI is less than −0.9 (RI = 0). Results of the adjusted logistic multivariate model predicting a ParAF or sustained PerAF episode, including the binary RI, sex, and age, are shown in Table 2. Heart failure and oral anticoagulation therapy use were not statistically significant. Almost a 12-fold increase in the odds of having a sustained PerAF vs a ParAF episode was observed for patients with RI = 1 (OR, 11.8; 95% CI, 2.2-63.5). An increase of the risk was observed in males (OR, 4.8; 95% CI, 1.1-20.3) and in patients older than 69 years (OR, 4; 95% CI, 1.0-16.6). No significant interactions were found between these variables. The ROC curve of the adjusted model is shown in Fig. 3. The AUC was 0.81. Sensitivity and specificity of the point closest to the upper left corner of the graph was 0.92 (95% CI, 0.821.00) and 0.52 (95% CI, 0.34-0.69), respectively. This means that the RI yielded 2 (8%) of 26 false-negative ParAF diagnoses in the sustained PerAF group and 16 (48%) of 33 false-positive sustained PerAF diagnoses in the ParAF group. The Hosmer-Lemeshow test (P = .94) indicated that the observed number of patients having sustained PerAF episodes was not statistically significantly different from those predicted by the model and that there was good fit. Fig. 4 shows the model-based estimated probability along with 95% CI of having a sustained PerAF episode according to the RI (1 vs 0) by sex and age. The model predicted a probability of approximately 85% (95% CI, 59%-96%) of having a sustained PerAF episode for males aged 69 years or older presenting CRI of −0.9 or greater (RI = 1) (Fig. 4A), whereas for females younger than 69 years presenting CRI of less than −0.9 (RI = 0), the predicted probability was 3% (95% CI, 0.3%18%) (Fig. 4B). The adjusted model was used to classify the group of 18 early terminated PerAF patients whose AF type was unknown because they had been electrically or pharmacologically cardioverted to normal sinus rhythm within 7 days from the onset. The mean SAPWd of these patients was 154 ± 16 milliseconds, whereas mean LADd was 41 ± 5 mm. According to the model, 14 (78%) of these patients had a greater probability of having sustained PerAF. For these early terminated AF subjects, a median sustained PerAF
0.2
a
RI=0
RI=1
RI=0
RI=1
Fig. 4. Model-based estimated probability, with 95% CI, of having a PerAF episode. Results for male (A) and results for female (B), both according to age (≥69 vs b69 years) and RI (1 vs 0).
114
A. Vincenti et al. / Journal of Electrocardiology 45 (2012) 109–115
in ParAF. In both groups, the P-wave duration was prolonged compared with healthy subjects. Left atrium diameter was larger in sustained PerAF than in ParAF patients, as well as compared with control subjects, whereas ParAF patients and controls did not differ for this anatomical parameter. A prolonged P-wave duration in patients with a persistent form of AF compared with patients with paroxysmal form was described in a previous study in which atrial dimensions were not significantly different. 11 Moreover, Koide et al 12 showed that, in ParAF, a longer P-wave duration and a larger left atrial dimension were independent predictors of progression to PerAF. A recent European survey found that the progression from paroxysmal to more sustained forms of AF was nearly 15% per year. Age, heart failure, previous TIA or stroke, chronic obstructive lung disease, and arterial hypertension were identified as the independent predictors of this phenomenon. 13 In the present analysis, there is some correlation between P-wave duration and left atrium diameter in controls and sustained PerAF groups, but not in paroxysmal patients. A possible explanation could be that the left atrium of ParAF patients has a greater fibrous component compared with control subjects. This could determine a slower impulse conduction through the atrium. A further increase of the atrial fibrosis, together with a dilatation of atrial chambers, could be a mechanism leading to progression from the paroxysmal to the persistent form of AF. 4,5 The proposed index enables to classify an AF episode (paroxysmal or sustained persistent AF) at the onset. It can reach a sensitivity of 92% and a specificity of 52% when one focuses of the chance of identifying those who will become persistent. A high sensitivity enables to properly select patients with PerAF as candidates for an appropriate treatment. The low specificity, however, could lead to overestimation of the number of AF-persistent patients. Overall, the ability of the model to separate the paroxysmal from the persistent nature of AF episodes is satisfactory because the AUC is 0.81. The proposed index was applied to a group of patients (n = 18) who underwent pharmacologic cardioversion within 7 days from the arrhythmia onset. In these patients, it was not possible to know if the arrhythmia would have stopped spontaneously. The index, according to a cutpoint that privileges the chance of identifying PerAF, classified 14 (78%) of these 18 episodes as persistent. Our index could be relevant from a clinical point of view because it may help in deriving the therapeutic approach by indicating patients with higher probability of either forms of AF, when it is impossible to know the true natural history of the arrhythmia. The classic method of catheter ablation of AF, with antral isolation of pulmonary veins, is much more successful in the ParAF form, 14-16 whereas in the persistent form, additional linear lesions or the elimination of fragmented potentials are probably required. 17 Therapy with antiremodeling drugs, for example, angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers, is able to reduce the incidence of AF and its
recurrences in high-risk patients 18,19 but is ineffective in preventing relapses in PerAF. 20 Finally, when a single oral dose is considered to interrupt an episode of AF (pill in the pocket strategy), it is very useful to know if the arrhythmia is prone to stop spontaneously. 21 Limitations of this study are that the patient sample is relatively small and that atrial dimensions were evaluated only by the M-mode echocardiography. Furthermore, our index yielded as much as 48% of false-positive diagnoses of PerAF in the ParAF group, and in the group of patients used as testing group (early terminated group), it was not possible to use the follow-up to verify the correctness of the obtained classification. This was because of the lack of relapses probably caused by drugs used for rhythm control or the recurrence of early cardioversion in emergency department if new arrhythmic episodes occurred. Moreover, because of the cross-sectional nature of the study, the index cannot predict the chance of remaining in sinus rhythm. For these reasons, this has to be considered a pilot study. A prospective study with a validation process should be performed in other populations of AF patients before the index can be applied in clinical practice.
References 1. Stewart S, Hart CL, Hole DJ, McMurray JJ. Population prevalence, incidence, and predictors of atrial fibrillation in the Renfrew/Paisley study. Heart 2001;86:516. 2. Go AS, Hylek EM, Phillips KA, et al. Prevalence of diagnosed atrial fibrillation in adults: national implications for rhythm management and stroke prevention: the AnTicoagulation and Risk Factors in Atrial Fibrillation (ATRIA) Study. JAMA 2001;285:2370. 3. European Heart Rhythm Association, European Association for CardioThoracic Surgery, Camm AJ, Kirchhof P, Lip GY, et al. Guidelines for the management of atrial fibrillation: the Task Force for the Management of Atrial Fibrillation of the European Society of Cardiology (ESC). Eur Heart J 2010;31:2369. 4. Allessie M, Ausma J, Schotten U. Electrical, contractile and structural remodeling during atrial fibrillation. Cardiovasc Res 2002;54:230. 5. Nattel S, Burstein B, Dobrev D. Atrial remodeling and atrial fibrillation: mechanisms and implications. Circ Arrhythm Electrophysiol 2008;1:62. 6. Stafford P, Denbigh P, Vincent R. Frequency analysis of the P wave: comparative techniques. Pacing Clin Electrophysiol 1995;18:261. 7. Lang RM, Bierig M, Devereux RB, et al. Recommendations for chamber quantification: a report from the American Society of Echocardiography's Guidelines and Standards Committee and the Chamber Quantification Writing Group, developed in conjunction with the European Association of Echocardiography, a branch of the European Society of Cardiology. J Am Soc Echocardiogr 2005;18:1440. 8. Mazumdar R, Glassman JR. Categorizing a prognostic variable: review of methods, code for easy implementation and applications to decisionmaking about cancer treatments. Stat Med 2000;19:113. 9. Miller R, Siegmund D. Maximally selected chi square statistics. Biometrics 1982;38:1011. 10. Hosmer DW, Lemeshow S. Applied logistic regression. New York: John Wiley & Sons Inc; 1989. 11. Santoni-Rugiu F, Verma R, Mehta D, et al. Signal-averaged P-wave ECG discriminates between persistent and paroxysmal atrial fibrillation. J Electrocardiol 2001;34:189. 12. Koide Y, Yotsukura M, Sakata K, Yoshino H, Ishikawa K. Investigation of the predictors of transition to persistent atrial fibrillation in patients with paroxysmal atrial fibrillation. Clin Cardiol 2002;25:69. 13. de Vos CB, Pisters R, Nieuwlaat R, et al. Progression from paroxysmal to persistent atrial fibrillation clinical correlates and prognosis. J Am Coll Cardiol 2010;55:725.
A. Vincenti et al. / Journal of Electrocardiology 45 (2012) 109–115 14. Cappato R, Calkins H, Chen SA, et al. Worldwide survey on the methods, efficacy, and safety of catheter ablation for human atrial fibrillation. Circulation 2005;111:1100. 15. Fisher JD, Spinelli MA, Mookherjee D, Krumerman AK, Palma EC. Atrial fibrillation ablation: reaching the mainstream. Pacing Clin Electrophysiol 2006;29:523. 16. Shah AN, Mittal S, Sichrovsky TC, et al. Long-term outcome following successful pulmonary vein isolation: pattern and prediction of very late recurrence. J Cardiovasc Electrophysiol 2008;19:661. 17. Haïssaguerre M, Hocini M, Sanders P, et al. Catheter ablation of longlasting persistent atrial fibrillation: clinical outcome and mechanisms of subsequent arrhythmias. J Cardiovasc Electrophysiol 2005;16:1138.
115
18. Schneider MP, Hua TA, Böhm M, Wachtell K, Kjeldsen SE, Schmieder RE. Prevention of atrial fibrillation by renin-angiotensin system inhibition: a meta-analysis. J Am Coll Cardiol 2010;55:2299. 19. Jibrini MB, Molnar J, Arora RR. Prevention of atrial fibrillation by way of abrogation of the renin-angiotensin system: a systematic review and meta-analysis. Am J Ther 2008;15:36. 20. GISSI-AF Investigators, Disertori M, Latini R, et al. Valsartan for prevention of recurrent atrial fibrillation. N Engl J Med 2009;360: 1606. 21. Alboni P, Botto GL, Baldi N, et al. Outpatient treatment of recent-onset atrial fibrillation with the “pill-in-the-pocket” approach. N Engl J Med 2004;351:2384.
Journal of Electrocardiology 45 (2012) 109 – 115 www.jecgonline.com
A noninvasive index of atrial remodeling in patients with paroxysmal and persistent atrial fibrillation: a pilot study☆ Antonio Vincenti, MD, a Matteo Rota, b Monica Spinelli, MD, a Mariella Corciulo, MD, a Sergio De Ceglia, MD, a Giovanni Rovaris, MD, a Laura Antolini, PhD, b Simonetta Genovesi, MD c, d,⁎ a
b
Electrophysiology and Cardiac Pacing Unit, S. Gerardo Hospital, Monza, Italy Department of Clinical Medicine and Prevention, Centre of Biostatistics for Clinical Epidemiology, University of Milano-Bicocca, Monza, Italy c Department of Clinical Medicine and Prevention, University of Milano-Bicocca, Monza, Italy d Nephrology Unit, S. Gerardo Hospital, Monza, Italy Received 18 April 2011
Abstract
Purpose: This study aims to develop a noninvasive atrial remodeling index (RI) to separate patients presenting paroxysmal atrial fibrillation (ParAF) from those with sustained persistent atrial fibrillation (PerAF), that is, AF episodes interrupted 7 days or more after the onset. Methods: Signal-averaged P-wave duration (SAPWd) and left atrial anteroposterior diameter (LADd) were measured in 33 ParAF patients, in 26 sustained PerAF patients, and in 18 control subjects. By using SAPWd and LADd, a dichotomous (0/1) RI was created. A logistic regression model on the probability of having a sustained PerAF vs a ParAF episode was estimated, including the RI, sex, age, and cardiac comorbidities as covariates. Results: Signal-averaged P-wave duration was significantly longer in sustained PerAF (153 ± 15 milliseconds) than in ParAF patients (142 ± 13 milliseconds, P b .001) and in both ParAF and sustained PerAF groups vs control group (123 ± 7 milliseconds, P b .001). Left atrial anteroposterior diameter was larger both in sustained PerAF (43 ± 6 mm) vs ParAF patients (38 ± 5 mm, P = .002) and in sustained PerAF group vs control group (38 ± 2 mm, P = .004), but no differences were observed between ParAF patients and controls (P = .6). A 12-fold increase (odds ratio, 11.8; 95% confidence interval, 2.2-63.5) in the odds of having a sustained PerAF vs a ParAF episode was observed in patients with RI equal to 1. Conclusions: P-wave duration and left atrium diameter enabled to define a noninvasive atrial RI to separate patients with ParAF from those with sustained PerAF. This could be a useful tool to select a suitable strategy for AF treatment. © 2012 Elsevier Inc. All rights reserved.
Keywords:
Paroxysmal atrial fibrillation; Persistent atrial fibrillation; Signal-averaged P-wave duration; Atrial diameter; Atrial remodeling; Remodeling index
Introduction Atrial fibrillation (AF) is a frequent arrhythmia associated with almost all heart diseases. It can cause complications such as heart failure and thromboembolism, but it is also present in apparently healthy hearts. The incidence of AF is increasing, particularly in the elderly, with an impact on economic resources required for the management of patients with AF. 1,2 ☆
Conflict of interest: none declared. ⁎ Corresponding author. Dipartimento di Medicina Clinica e Prevenzione, Università di Milano-Bicocca, Via Cadore 48, 20052, Monza, Italy. E-mail address: [email protected] 0022-0736/$ – see front matter © 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.jelectrocard.2011.08.005
An AF episode is defined paroxysmal (ParAF) when it spontaneously stops within 7 days from the onset and persistent (PerAF) when pharmacologic or electrical interventions are needed for interruption. 3 Atrial fibrillation episodes are often interrupted with electrical or pharmacologic cardioversion within 7 days from the onset. In such a case, it is not possible to know the natural history of the arrhythmia. This raises the question on whether the episode in patients who underwent intervention within 7 days from the onset would have resolved spontaneously. The distinction between ParAF and PerAF is crucial because PerAF is normally associated to structural and electrophysiologic modifications that make the arrhythmia more sustained. A
110
A. Vincenti et al. / Journal of Electrocardiology 45 (2012) 109–115
dilatation of the atria and an increase of fibrous components are often present in PerAF. From the electrophysiologic point of view, a shortening of the effective refractory period and a reduction of the conduction velocity of the electrical impulse through the atrium are observed. These phenomena are known as structural and electrical atrial remodeling. 4,5 The P-wave duration, measured with the signal-averaged electrocardiogram (ECG) (SAPWd), is an expression of the intra-atrial conduction time, 6 whereas the left atrial anteroposterior diameter at end-diastole (LADd) is a measure of atrial dimension. The modifications of intra-atrial conduction time and of atrial dimensions are characteristics of atrial remodeling. The aim of the present pilot study is to create a noninvasive index using SAPWd and LADd to distinguish between patients presenting ParAF or sustained PerAF, the latter defined as AF episodes lasting more than 7 days from the onset and not self-terminating. Methods Groups of patients The study considered 95 subjects. Seventy-seven patients, presenting an AF episode in the previous month, were enrolled consecutively (from January 2009 to June 2010) in the Electrophysiology and Cardiac Pacing Unit of the San Gerardo Hospital (Monza, Italy). Patients were classified according to the type of AF episode as: ParAF group (n = 33): patients with at least 1 selfterminating, symptomatic, and/or asymptomatic episode of AF, documented by 12-lead electrocardiogram (ECG) or 24 hour ECG Holter. Sustained PerAF group (n = 26): patients with an AF episode interrupted by pharmacologic or electrical intervention after at least 7 days from the onset. Early terminated PerAF group (n = 18): patients with an AF episode interrupted by pharmacologic or electrical intervention within 7 days from the onset. The remaining subjects (n = 18) were controls (control group). These were subjects without any previous history of heart disease, hypertension, diabetes mellitus, and/or AF. In the AF groups, P-wave duration was evaluated after discontinuation of 1-C drugs for 5 half-lives, 1 month after the AF episode or the electrical cardioversion. Patients taking amiodarone and those who had recurrences of the arrhythmia within a month after cardioversion were not included in the study. The study was approved by an institutional review committee, and the subjects gave informed consent. Electrocardiographic analysis The ECG recording was performed by a digital Holter recorder (SpiderView; Sorin Group Company, Sorin CRM, Saluggia, Vercelli, Italy). Signal-averaged P-wave duration was obtained by the P-wave analysis software (Sorin CRM, Saluggia, Vercelli, Italy) derived from the work of
Stafford et al. 6 The P-wave analysis was based on ECG obtained with recorders that allow acquisition of 3 bipolar orthogonal leads (x, y, and z), sampled at 1000 Hz. The user had to choose a reference P wave to be used to compute the averaging template. The method consisted in automatically aligning different P waves with the reference one by looking for the position of maximal mutual correlation. P-wave complexes that did not match the reference P wave were automatically rejected. The selected P waves were then filtered by a 20- to 50Hz bandpass filter. For each orthogonal lead, averaged highresolution P-wave signal was obtained, and the vector magnitude of the 3 channels was calculated (x 2 + y 2 + z 2) 0.5. The duration of the P wave was obtained by automatic estimation of the beginning and the end of the P wave. Approximately, 150 to 200 beats were averaged. Cardiac ultrasound examination Echocardiograms were obtained in the standard precordial positions using digital echocardiography equipment (Aloka ProSound SSD Alpha 10; Aloka Co, Ltd, Tokyo, Japan) with 1- to 5-MHz transducers. The recommendations for standard measurements from M-mode echocardiograms were followed. 7 Instantaneous measurements were made during 3 cardiac cycles, and the average value of LADd from a parasternal long-axis view was obtained for each subject. Both ECG analysis and cardiac ultrasound examination were performed in the same day, 1 month after cardioversion (sustained PerAF group), or after episode documentation (ParAF group). The operator was blinded with respect to the patient AF classification group.
Statistical methods Continuous variables were described in terms of mean and standard deviation, and categorical variables, as percentages. Baseline characteristics of ParAF, sustained PerAF, and controls were compared by using the t test for continuous variables and the χ 2 test for categorical variables. Correlation between continuous variables was evaluated by the Pearson correlation coefficient. The probability of having a sustained PerAF vs a ParAF episode was modeled through a multivariate logistic regression analysis, including the noninvasive parameters of atrial remodeling (SAPWd and LADd) as covariates. A continuous remodeling index (CRI) was computed for each AF patient from the model-based linear predictor as a linear combination between the estimated regression coefficients β1 and β2 and the individual values of SAPWd and LADd as follows: α + β1·SAPWd + β2·LADd (where α was the model intercept). Atrial fibrillation patients presenting with greater values of the CRI were more likely to develop sustained PerAF instead of ParAF episodes. A cutpoint c was chosen to dichotomize the CRI result to define a criterion to divide AF episodes into sustained PerAF or ParAF. The cutpoint selection was performed by a systematic search of the central 80% of the observed values of the CRI with the aim of choosing the one associated with the minimum P value (or maximum χ 2 statistic). 8 The chosen cutpoint for dichotomization was
A. Vincenti et al. / Journal of Electrocardiology 45 (2012) 109–115
111
Table 1 Baseline characteristics of 95 subjects Characteristics
ParAF (n = 33)
Sustained PerAF (n = 26)
Early terminated PerAF (n = 18)
(ParAF vs sustained PerAF P value) Sex, males
18 (55%)
20 (77%)
(ParAF vs control P value) 11 (61%)
(.08) Age (y)
62 (9)
LADd (mm)
38 (5)
SAPWd (ms)
142 (13)
69 (10)
62 (12)
43 (6)
41 (5)
153 (15)
154 (16)
(.08) 38 (2)
(.6)
(.004) 21 (63%)
(.26) 61 (16)
(.73)
(.002)
13 (72%)
7 (27%)
0 (0%)
3 (12%)
2 (11%)
7 (27%)
0 (0%)
3 (12%)
1 (6%)
17 (65%)
6 (33%)
16 (62%)
3 (17%)
10 (38%)
6 (33%)
9 (35%)
5 (28%)
(b.0001) 123 (7)
(b.0001) 17 (65%)
(Sustained PerAF vs control P value)
11 (61%) (.65)
(.02)
Hypertension
Control (n = 18)
(b.0001)
(.89) Coronary artery disease
4 (12%) (.15)
Diabetes mellitus
5 (15%) (.69)
Heart failure
0 (0%) (.002)
History of stroke or TIA
1 (3%) (.2)
Other comorbidities
19 (57%) (.54)
OAT
8 (24%) (.004)
ACE-I
9 (27%) (.36)
ARB
11 (33%) (.92)
CHADS2 score
1.2 (1.2)
1.6 (1.2)
1.1 (0.9)
(.18) Values are expressed as expressed as number of patients (%) for categorical variables and mean (SD) for continuous variables. OAT indicates oral anticoagulation therapy; ACE-I, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptors blockers; CHADS2, congestive heart failure, hypertension, age > 75 years, diabetes, stroke, TIA/thrombo-embolism.
equivalent to the one obtained maximizing the difference between sensitivity and 1 minus specificity. To avoid the problem of inflation of type I error rate arising from multiple testing in the cutpoint selection, we applied the MillerSiegmund P value correction formula. 9 Following this approach, a dichotomous remodeling index (RI) was created. The criteria identified a sustained PerAF condition when the CRI was greater than or equal to c (RI = 1), whereas ParAF was diagnosed when the CRI was less than c (RI = 0). A logistic regression model on the probability of having a sustained PerAF vs a ParAF episode was estimated including the binary RI, sex, age, and other cardiac comorbidities as covariates. All variables were tested for interactions. Stepwise model reduction technique was used. Model performance was assessed using a receiver operating characteristic curve (ROC) for discrimination and using Hosmer-Lemeshow test for calibration. 10 The ROC curve shows the trade-off between sensitivity and specificity varying the cutoff on the linear predictor of the adjusted model. A model is considered to be highly discriminant when the area under the ROC curve (AUC) is close to 1 and not discriminant if the area is close to 0.5. The Hosmer-Lemeshow test relies on the contrast between the observed and model expected number of sustained PerAF episodes in risk groups defined by the deciles of the model linear predictor.
P b .05 was considered statistically significant. All performed tests were 2 sided. All analyses were performed with SAS 9.1.2 (SAS Institute, Cary, NC) and using R version 2.11.0 for graphics.
Results Table 1 shows the baseline characteristics of the 95 AF subjects enrolled in the study. Sustained PerAF were older than ParAF patients (69 ± 10 vs 62 ± 9 years, P = .02). Signalaveraged P-wave duration was significantly longer in sustained PerAF than in ParAF patients (153 ± 15 vs 142 ± 13 milliseconds, P = .004) and in both ParAF and sustained PerAF group vs control group (123 ± 7 milliseconds), with P b .001. The anatomical parameter LADd was larger both in sustained PerAF vs ParAF patients (43 ± 6 vs 38 ± 5 mm, P = .002) and in the sustained PerAF group vs the control group (38 ± 2 mm) with P b .001, but no differences were observed between ParAF patients and controls (P = .6). The AF groups of patients did not differ regarding the prevalence of hypertension, diabetes mellitus, coronary artery disease, and history of stroke or transient ischemic attack (TIA), whereas in the sustained PerAF group, a higher fraction of subjects showed heart failure (P = .002).
A. Vincenti et al. / Journal of Electrocardiology 45 (2012) 109–115
190
ParAF (n = 33)
SAPWd (msec)
150
110
130
r=0.05
170
190 170
r=0.44
110
SAPWd (msec)
B
Control (n = 18)
150
A
130
112
30
35
40
45
50
30
35
D
50
190
Early terminated PerAF (n = 18)
130
170
r=0.60
150
150
SAPWd (msec)
170
r=0.35
110
130
190
Sustained PerAF (n = 26)
110
SAPWd (msec)
45
LADd (mm)
LADd (mm)
C
40
30
35
40
45
50
30
35
LADd (mm)
40
45
50
LADd (mm)
Fig. 1. Scatter plot of the relationship between SAPWd and LADd measures within each group. A, Control group; B, ParAF group; C, sustained PerAF group; D, early terminated PerAF group. In all panels, r denotes the Pearson correlation coefficient.
0
1
2
cutpoint c = −0.9 of the CRI was associated with a maximum χ 2 of 12.8 and an uncorrected P value of .0003. Using the Miller and Siegmund formula, the corrected P value was .01. Fig. 2 shows the distribution
Continuous Remodeling Index
The prevalence of anticoagulation therapy was higher (P = .004) in the sustained PerAF than in ParAF patients. Angiotensin-converting enzyme inhibitor use (P = .36) and angiotensin receptor blocker use (P = .92) did not differ among sustained PerAF and ParAF. No statistically significant differences were observed on the Congestive heart failure, Hypertension, Age > 75 years, Diabetes, Stroke, TIA/Thrombo-embolism (CHADS2) score in ParAF and sustained PerAF (P = .18). The relationship between SAPWd and LADd measures is depicted within each group in Fig. 1. These variables can be noticed to be modestly correlated in the control group (Fig. 1A; r = 0.44, P = .07), whereas this correlation completely disappeared in ParAF (Fig. 1B; r = 0.05, P = .78). In the sustained PerAF group, the 2 variables were still weakly correlated (Fig. 1C; r = 0.35, P = .07). In univariate logistic regression analyses, both LADd (odds ratio [OR], 1.17; 95% confidence interval [CI], 1.051.29) and SAPWd (OR, 1.06; 95% CI, 1.02-1.11) were significantly related to the probability of having a sustained PerAF vs a ParAF episode. The CRI, defined by the model linear predictor, was CRI = −12 + 0.13·LADd + 0.05·SAPWd. The optimal
ParAF (n = 33)
Sustained PerAF (n = 26)
Fig. 2. Box-plot representation of the CRI by AF episode. The box extends from the 25th to the 75th percentile. The inner line of the box marks the median, and whisker lines out of the box represent the minimum and the maximum observed values.
A. Vincenti et al. / Journal of Electrocardiology 45 (2012) 109–115
P .0009 .04 .05
0.8 0.6
95% CI 2.2-63.5 1.1-20.3 1.0-16.6
0.4
OR 11.8 4.8 4.0
Sensitivity
RI (1 vs 0)a Sex (male vs female) Age ≥69 y
1.0
Table 2 Results from multivariate logistic regression on the probability of having a sustained PerAF vs a ParAF episode
113
RI = 1 if −12 + 0.13·LADd (millimeters) + 0.05·SAPWd (milliseconds) ≥ −0.9, and RI = 0 otherwise.
0.0
AUC=0.81
0.0
0.2
0.4
0.6
0.8
1.0
1−Specificity
Fig. 3. Receiver operating characteristic curve of the logistic regression model predicting probability of having a sustained PerAF vs a ParAF episode as a function of RI, sex, and age.
probability of 60% ranging from 3% to 85% was estimated from the multivariate model. Discussion The present study shows that in a group of patients with AF, the P-wave duration was longer in sustained PerAF than
A
Age >= 69 years
0
0.2
0.4
0.6
P(Sustained PerAF)
0.8
1
male Age < 69 years
RI=0
RI=1
RI=0
RI=1
female Age < 69 years
Age >= 69 years
0
0.2
0.4
0.6
0.8
1
B
P(Sustained PerAF)
of the CRI in sustained PerAF and ParAF patients. One can note that patients with sustained PerAF episodes were associated with higher values of the CRI than ParAF patients. The binary RI, defined according to the cutpoint c = −0.9, classifies subjects as at risk for being sustained PerAF or ParAF depending on whether CRI is −0.9 or greater (RI = 1) or CRI is less than −0.9 (RI = 0). Results of the adjusted logistic multivariate model predicting a ParAF or sustained PerAF episode, including the binary RI, sex, and age, are shown in Table 2. Heart failure and oral anticoagulation therapy use were not statistically significant. Almost a 12-fold increase in the odds of having a sustained PerAF vs a ParAF episode was observed for patients with RI = 1 (OR, 11.8; 95% CI, 2.2-63.5). An increase of the risk was observed in males (OR, 4.8; 95% CI, 1.1-20.3) and in patients older than 69 years (OR, 4; 95% CI, 1.0-16.6). No significant interactions were found between these variables. The ROC curve of the adjusted model is shown in Fig. 3. The AUC was 0.81. Sensitivity and specificity of the point closest to the upper left corner of the graph was 0.92 (95% CI, 0.821.00) and 0.52 (95% CI, 0.34-0.69), respectively. This means that the RI yielded 2 (8%) of 26 false-negative ParAF diagnoses in the sustained PerAF group and 16 (48%) of 33 false-positive sustained PerAF diagnoses in the ParAF group. The Hosmer-Lemeshow test (P = .94) indicated that the observed number of patients having sustained PerAF episodes was not statistically significantly different from those predicted by the model and that there was good fit. Fig. 4 shows the model-based estimated probability along with 95% CI of having a sustained PerAF episode according to the RI (1 vs 0) by sex and age. The model predicted a probability of approximately 85% (95% CI, 59%-96%) of having a sustained PerAF episode for males aged 69 years or older presenting CRI of −0.9 or greater (RI = 1) (Fig. 4A), whereas for females younger than 69 years presenting CRI of less than −0.9 (RI = 0), the predicted probability was 3% (95% CI, 0.3%18%) (Fig. 4B). The adjusted model was used to classify the group of 18 early terminated PerAF patients whose AF type was unknown because they had been electrically or pharmacologically cardioverted to normal sinus rhythm within 7 days from the onset. The mean SAPWd of these patients was 154 ± 16 milliseconds, whereas mean LADd was 41 ± 5 mm. According to the model, 14 (78%) of these patients had a greater probability of having sustained PerAF. For these early terminated AF subjects, a median sustained PerAF
0.2
a
RI=0
RI=1
RI=0
RI=1
Fig. 4. Model-based estimated probability, with 95% CI, of having a PerAF episode. Results for male (A) and results for female (B), both according to age (≥69 vs b69 years) and RI (1 vs 0).
114
A. Vincenti et al. / Journal of Electrocardiology 45 (2012) 109–115
in ParAF. In both groups, the P-wave duration was prolonged compared with healthy subjects. Left atrium diameter was larger in sustained PerAF than in ParAF patients, as well as compared with control subjects, whereas ParAF patients and controls did not differ for this anatomical parameter. A prolonged P-wave duration in patients with a persistent form of AF compared with patients with paroxysmal form was described in a previous study in which atrial dimensions were not significantly different. 11 Moreover, Koide et al 12 showed that, in ParAF, a longer P-wave duration and a larger left atrial dimension were independent predictors of progression to PerAF. A recent European survey found that the progression from paroxysmal to more sustained forms of AF was nearly 15% per year. Age, heart failure, previous TIA or stroke, chronic obstructive lung disease, and arterial hypertension were identified as the independent predictors of this phenomenon. 13 In the present analysis, there is some correlation between P-wave duration and left atrium diameter in controls and sustained PerAF groups, but not in paroxysmal patients. A possible explanation could be that the left atrium of ParAF patients has a greater fibrous component compared with control subjects. This could determine a slower impulse conduction through the atrium. A further increase of the atrial fibrosis, together with a dilatation of atrial chambers, could be a mechanism leading to progression from the paroxysmal to the persistent form of AF. 4,5 The proposed index enables to classify an AF episode (paroxysmal or sustained persistent AF) at the onset. It can reach a sensitivity of 92% and a specificity of 52% when one focuses of the chance of identifying those who will become persistent. A high sensitivity enables to properly select patients with PerAF as candidates for an appropriate treatment. The low specificity, however, could lead to overestimation of the number of AF-persistent patients. Overall, the ability of the model to separate the paroxysmal from the persistent nature of AF episodes is satisfactory because the AUC is 0.81. The proposed index was applied to a group of patients (n = 18) who underwent pharmacologic cardioversion within 7 days from the arrhythmia onset. In these patients, it was not possible to know if the arrhythmia would have stopped spontaneously. The index, according to a cutpoint that privileges the chance of identifying PerAF, classified 14 (78%) of these 18 episodes as persistent. Our index could be relevant from a clinical point of view because it may help in deriving the therapeutic approach by indicating patients with higher probability of either forms of AF, when it is impossible to know the true natural history of the arrhythmia. The classic method of catheter ablation of AF, with antral isolation of pulmonary veins, is much more successful in the ParAF form, 14-16 whereas in the persistent form, additional linear lesions or the elimination of fragmented potentials are probably required. 17 Therapy with antiremodeling drugs, for example, angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers, is able to reduce the incidence of AF and its
recurrences in high-risk patients 18,19 but is ineffective in preventing relapses in PerAF. 20 Finally, when a single oral dose is considered to interrupt an episode of AF (pill in the pocket strategy), it is very useful to know if the arrhythmia is prone to stop spontaneously. 21 Limitations of this study are that the patient sample is relatively small and that atrial dimensions were evaluated only by the M-mode echocardiography. Furthermore, our index yielded as much as 48% of false-positive diagnoses of PerAF in the ParAF group, and in the group of patients used as testing group (early terminated group), it was not possible to use the follow-up to verify the correctness of the obtained classification. This was because of the lack of relapses probably caused by drugs used for rhythm control or the recurrence of early cardioversion in emergency department if new arrhythmic episodes occurred. Moreover, because of the cross-sectional nature of the study, the index cannot predict the chance of remaining in sinus rhythm. For these reasons, this has to be considered a pilot study. A prospective study with a validation process should be performed in other populations of AF patients before the index can be applied in clinical practice.
References 1. Stewart S, Hart CL, Hole DJ, McMurray JJ. Population prevalence, incidence, and predictors of atrial fibrillation in the Renfrew/Paisley study. Heart 2001;86:516. 2. Go AS, Hylek EM, Phillips KA, et al. Prevalence of diagnosed atrial fibrillation in adults: national implications for rhythm management and stroke prevention: the AnTicoagulation and Risk Factors in Atrial Fibrillation (ATRIA) Study. JAMA 2001;285:2370. 3. European Heart Rhythm Association, European Association for CardioThoracic Surgery, Camm AJ, Kirchhof P, Lip GY, et al. Guidelines for the management of atrial fibrillation: the Task Force for the Management of Atrial Fibrillation of the European Society of Cardiology (ESC). Eur Heart J 2010;31:2369. 4. Allessie M, Ausma J, Schotten U. Electrical, contractile and structural remodeling during atrial fibrillation. Cardiovasc Res 2002;54:230. 5. Nattel S, Burstein B, Dobrev D. Atrial remodeling and atrial fibrillation: mechanisms and implications. Circ Arrhythm Electrophysiol 2008;1:62. 6. Stafford P, Denbigh P, Vincent R. Frequency analysis of the P wave: comparative techniques. Pacing Clin Electrophysiol 1995;18:261. 7. Lang RM, Bierig M, Devereux RB, et al. Recommendations for chamber quantification: a report from the American Society of Echocardiography's Guidelines and Standards Committee and the Chamber Quantification Writing Group, developed in conjunction with the European Association of Echocardiography, a branch of the European Society of Cardiology. J Am Soc Echocardiogr 2005;18:1440. 8. Mazumdar R, Glassman JR. Categorizing a prognostic variable: review of methods, code for easy implementation and applications to decisionmaking about cancer treatments. Stat Med 2000;19:113. 9. Miller R, Siegmund D. Maximally selected chi square statistics. Biometrics 1982;38:1011. 10. Hosmer DW, Lemeshow S. Applied logistic regression. New York: John Wiley & Sons Inc; 1989. 11. Santoni-Rugiu F, Verma R, Mehta D, et al. Signal-averaged P-wave ECG discriminates between persistent and paroxysmal atrial fibrillation. J Electrocardiol 2001;34:189. 12. Koide Y, Yotsukura M, Sakata K, Yoshino H, Ishikawa K. Investigation of the predictors of transition to persistent atrial fibrillation in patients with paroxysmal atrial fibrillation. Clin Cardiol 2002;25:69. 13. de Vos CB, Pisters R, Nieuwlaat R, et al. Progression from paroxysmal to persistent atrial fibrillation clinical correlates and prognosis. J Am Coll Cardiol 2010;55:725.
A. Vincenti et al. / Journal of Electrocardiology 45 (2012) 109–115 14. Cappato R, Calkins H, Chen SA, et al. Worldwide survey on the methods, efficacy, and safety of catheter ablation for human atrial fibrillation. Circulation 2005;111:1100. 15. Fisher JD, Spinelli MA, Mookherjee D, Krumerman AK, Palma EC. Atrial fibrillation ablation: reaching the mainstream. Pacing Clin Electrophysiol 2006;29:523. 16. Shah AN, Mittal S, Sichrovsky TC, et al. Long-term outcome following successful pulmonary vein isolation: pattern and prediction of very late recurrence. J Cardiovasc Electrophysiol 2008;19:661. 17. Haïssaguerre M, Hocini M, Sanders P, et al. Catheter ablation of longlasting persistent atrial fibrillation: clinical outcome and mechanisms of subsequent arrhythmias. J Cardiovasc Electrophysiol 2005;16:1138.
115
18. Schneider MP, Hua TA, Böhm M, Wachtell K, Kjeldsen SE, Schmieder RE. Prevention of atrial fibrillation by renin-angiotensin system inhibition: a meta-analysis. J Am Coll Cardiol 2010;55:2299. 19. Jibrini MB, Molnar J, Arora RR. Prevention of atrial fibrillation by way of abrogation of the renin-angiotensin system: a systematic review and meta-analysis. Am J Ther 2008;15:36. 20. GISSI-AF Investigators, Disertori M, Latini R, et al. Valsartan for prevention of recurrent atrial fibrillation. N Engl J Med 2009;360: 1606. 21. Alboni P, Botto GL, Baldi N, et al. Outpatient treatment of recent-onset atrial fibrillation with the “pill-in-the-pocket” approach. N Engl J Med 2004;351:2384.