Ioannis A. Paraskevaidis, MD; Thomas Dodouras, MD; Dimitrios Tsiapras, MD; and Dimitrios T. Kremastinos, MD

Transcription

1 Prediction of Successful Cardioversion and Maintenance of Sinus Rhythm in Patients With Lone Atrial Fibrillation* Ioannis A. Paraskevaidis, MD; Thomas Dodouras, MD; Dimitrios Tsiapras, MD; and Dimitrios T. Kremastinos, MD Objective: We aimed to prospectively investigate the predictive value of echocardiographic parameters for the prediction of successful cardioversion and long-term sinus rhythm (SR) maintenance in patients who have experienced a lone episode of atrial fibrillation (AF). Measurements and results: Clinical and echocardiographic data, including mean left atrial appendage (LAA) peak flow velocity and mitral annulus motion, were analyzed in 78 consecutive patients (mean [ SD] age, years) with AF lasting > 48 h and < 6 months. Sixty-one patients (78%) underwent successful external electrical cardioversion, while the remaining remained in AF. At the 1-year follow-up, of the 61 patients who had successfully been converted to SR, 24 (39.3%) remained in SR. For predicting the success of the cardioversion, we used a model consisting of two variables. LAA flow velocity (> 20 cm/s) and left ventricular (LV) fractional shortening (> 30%) appear to be quite strong, yielding 83.3% correct results. For predicting the maintenance of SR, we used a model consisting of two variables. The absence of the early systolic abnormal mitral annulus motion and LAA flow velocity (> 20 cm/s) appears to be quite strong, yielding 84.6% correct results. LAA flow velocity only marginally enters the model, and, if removed, little predictive value is lost (dropping to 83.3%). Removing the early systolic abnormal mitral annulus motion variable, the prediction value drops significantly to 70.5%. Conclusion: LAA flow velocity combined with LV fractional shortening can predict the success of the conversion of AF to SR. Additionally, LAA flow velocity, combined with the analysis of mitral annulus motion before cardioversion, can predict the long-term maintenance of SR. (CHEST 2005; 127: ) Key words: atrial fibrillation; left atrial appendage flow; mitral annulus motion Abbreviations: AF atrial fibrillation; LAA left atrial appendage; LV left ventricular; SR sinus rhythm The duration of atrial fibrillation (AF), 1,2 the type of underlying disease, 3 patient age, 1 left atrial diameter, 2 4 left ventricular (LV) function, 2 and the continuation of therapy with antiarrhythmic medication 5,6 have been all proposed as predictors of successful cardioversion of AF and maintenance of sinus rhythm (SR) on a long-term basis. However, the predictive values of these parameters are far from optimal. 7 The long-term maintenance and conversion of AF *From the Second Department of Cardiology, Onassis Cardiac Surgery Center, Athens, Greece. Manuscript received January 13, 2004; revision accepted September 9, Reproduction of this article is prohibited without written permission from the American College of Chest Physicians ( permissions@chestnet.org). Correspondence to: Ioannis A. Paraskevaidis, MD, Second Department of Cardiology, 356 Syngrou Ave, Athens, Greece; iparas@otenet.gr to SR remains a challenge. In this regard, it has been shown 8 that the pulmonary venous flow velocity pattern 24 h after cardioversion might identify patients who will remain in SR 1 year later. It has also For editorial comment see page 424 been suggested that the restoration and maintenance of SR may be predicted by evaluating left atrial appendage (LAA) flow velocity before cardioversion. 9,10 However, this finding has been challenged by others, 11,12 and the conclusions are weakened by the heterogeneous patient population and retrospective design of the available studies Previous studies 13,14 in patients with AF have described an early systolic reverse wave in pulmonary venous flow, which is apparently due to an abnormal mitral annulus motion (not always present), with both phenomena disappearing following successful car- 488 Clinical Investigations

2 dioversion. 15 However, in patients with AF the presence or absence of an abnormal mitral annulus motion in predicting cardioversion and/or SR maintenance, alone or in combination with other echocardiographic variables, has never been investigated. Therefore, we aimed to prospectively investigate in a well-defined group of patients with lone AF, the predictive value of various echocardiographic parameters for prediction of successful cardioversion and long-term SR maintenance. Study Group Materials and Methods From November 1996 until June 2000, we prospectively screened 438 adults who were in AF and had been admitted to our hospital for treatment with external electrical cardioversion. Patients with known or suspected coronary artery disease, organic valvular heart disease, cardiomyopathies, pericarditis, arterial hypertension (ie, 140/90 mm Hg), myocarditis, conduction abnormalities, and thyroid disorders, or patients with uncertain duration of AF were excluded from the study. Patients with AF for 48 h and 6 months duration (data selected from previous ECGs in the patient s records or from their attending physician), which was defined as having experienced a lone episode of AF were included in the study. Finally, 78 consecutive patients (mean [ SD] age, years) with AF and 10 apparently healthy volunteers (mean age, years) were enrolled into the study. All patients were treated conventionally with therapeutic anticoagulation (3 weeks before and 4 weeks after cardioversion; target international normalized ratio, 2.0 to 3.0), and all were receiving therapy with oral antiarrhythmic agents (sotalol, 40 patients; amiodarone, 38 patients). Antiarrhythmic treatment was started (after having reached the target international normalized ratio) from the time of diagnosis (at least 7 to 8 days before the scheduled cardioversion) by the referring physician and was continued for the entire observation period. Before cardioversion and after an initial transthoracic echocardiographic study was performed, a transesophageal echocardiographic study was performed, according to our hospital protocol. If thrombus 16,17 or severe spontaneous echo contrast 16,18 was present, the cardioversion was postponed (ie, the patient was not enrolled into the study). After successful conversion, the patients were followed up for 12 months. An ECG and clinical examination were performed at 24 h, and 3, 6, 9, and 12 months following conversion. This study was conducted according to the Helsinki Declaration and was approved by the ethics committee of our hospital, and all patients gave informed consent. Echocardiography Conventional transthoracic echocardiography was performed before cardioversion (Sonos 2500; Hewlett-Packard; Palo Alto, CA; or Vivid 7.0 ultrasonographic system; GE Healthcare; Chicago, IL). The following variables were measured using a parasternal long-axis view from two-dimensional targeted M- mode tracings according to the recommendations of the American Society of Echocardiography 19 : left atrial systolic diameter; LV end-diastolic and end-systolic diameter; and hence LV fractional shortening ([LV end-diastolic diameter LV end-systolic diameter]/lv end-diastolic diameter 100), and LV septal and posterior end-diastolic wall thickness. The ejection fraction was calculated using the modified Simpson rule. Following local anesthesia of the hypopharynx with a 10% lidocaine spray and IV sedation with midazolam, a 5.0-MHz transducer, mounted on the tip of a modified flexible multiplane gastroscope, was introduced to a depth of 25 to 30 cm from the teeth. During transesophageal echocardiography, images were analyzed online for the presence of intracardiac thrombus. The mitral E-wave velocity and pulmonary vein flow velocity were recorded, and the S/D peak velocity ratio as well as the systolic fraction (ie, velocity time integral S/velocity time integral [S D]) were calculated. The LAA flow velocity profiles were obtained by pulsed-wave Doppler interrogation, placing the sample volume 1 cm within the orifice of the LAA. Recordings of the mitral annulus motion during the cardiac cycle were obtained with the M-mode cursor directed for the modified four-chamber transesophageal view, approximately 33 cm from the teeth. The cursor was oriented toward the bright septal margin of the annulus (ie, the fibrous triangle) and then toward the lateral margin. On each side, the beam was oriented so that it was perpendicular to the descending motion of the annular structure. Multiple M-mode recordings were made of the septal and lateral margins of the mitral annulus. The motion of these two margins was similar, and thus only the motion of the lateral margin was reported. The recordings of both LAA flow velocity and mitral annulus motion were analyzed offline by two experts who were unaware of clinical and other details about the patient. All measurements were averaged over a minimum of 10 consecutive cardiac cycles. Follow-up Patients were followed up regularly every 3 months for up to 1 year. Serial ECGs were recorded at each visit to document the maintenance of SR or the recurrence of AF. Suspected AF episodes between visits were confirmed by 24-h ECG monitoring (Holter monitor). Statistical Analysis Data were processed and analyzed using a statistical software package (SPSS 10.0 for Windows; SPSS; Chicago, IL). At this stage, we added categoric and grouping variables. As possible groupings we chose (1) whether the patient exhibited an abnormal early systolic mitral annulus motion before cardioversion, (2) whether the cardioversion was successful, and (3) whether the patient had maintained SR after 12 months. Each grouping was examined for underlying differences in demographics (ie, age and gender) and medical treatment. An independent samples t test and the Levene test for equality of variances were used for numeric variables, and the Kruskal-Wallis test was used for categoric variables. The compliance with normality was used to determine whether a paired t test should be used, or whether, in the case of noncompliance, a nonparametric test such as the Wilcoxon test should be used. Next, we utilized the logistic regression estimation method in order to construct a prediction model utilizing the variables that exhibited statistically significant differences among the groups. We applied a binary logistics regression statistical procedure on the full set of data, with preservation of SR after 12 months used as the dependent variable, and we utilized our findings to establish a set of independent variables as follows: (1) whether the patient exhibited an abnormal early systolic mitral annulus motion before cardioversion; (2) left atrial diameter; (3) LAA flow velocity; (4) LV end-systolic diameter; and (5) LV fractional shortening. Through the logit regression model, it is possible to describe the relationship between one or more continuous independent variables to the binary dependent variable group. In the logit CHEST / 127 / 2/ FEBRUARY,

3 regression model, the predicted values for the dependent variable will never be 0or 1, regardless of the values of the independent variables. Following this, we derived the model, using the stepwise forward conditional approach, and it explains most of the variability of the dependent variable. Results Successful vs Unsuccessful Cardioversion Of 78 consecutive patients, 61 patients (78%) underwent successful electrical conversion, while the remaining were in AF. There were no differences in age, gender, drug administration, and duration of AF between the unsuccessfully and successfully converted groups (Table 1). Therapy with antiarrhythmic drugs (ie, sotalol or amiodarone) was continued in all patients throughout the follow-up period. Prior to direct current cardioversion, the mean LAA peak flow velocity had been higher in the successfully converted group compared to that of the nonconverted group (Table 1). Additionally, a similar percentage of an abnormal mitral annulus motion (toward the left atrium instead of toward the LV during early systole) [Fig 1, bottom, c] had been recorded in both the unsuccessfully, and the successfully converted groups (Table 1). The time distances of the Table 1 Demographic, Echocardiographic, and Other Data on Patient Conversion of AF to SR* Variable Nonconverted (n 17) Converted (n 61) p Value Age, yr NS Duration of AF, mo NS Drugs Amiodarone 8 30 Sotalol 9 31 Mean LAA peak flow velocity, cm/s Presence of abnormal mitral 11 (64.7) 40 (65.6) R wave to abnormal mitral NS annulus motion, ms R-R interval, ms NS Left atrium, mm NS IVS, mm NS LPWTh, mm NS LVEDD, mm NS LVESD, mm NS FS, % NS EF, % NS Mitral E-wave velocity, cm/s NS Pulmonary S/D peak velocity NS ratio Systolic fraction NS *Values given as mean SD or No. of patients (%), unless otherwise indicated. IVS interventricular septum; LPWTh LV posterior wall thickness in diastole; LVEDD LV end-diastolic diameter; LVESD LV end-diastolic diameter; FS LV fractional shortening; EF LV ejection fraction; NS not significant. Figure 1. Mitral annulus motion in a healthy individual (top, a), in a patient who remained in SR throughout the 1-year follow-up period (middle, b), and in a patient with an abnormal mitral annulus motion (note the early systolic notch) in whom AF recurred throughout the same observation period (bottom, c). early systolic abnormal mitral annulus motion from the R wave on the ECG and the previous R-R interval were similar between groups (Table 1). Maintenance of SR At the 1-year follow-up of the 61 patients who had successfully converted to SR, 24 (39.3%) remained in SR while AF recurred in the remaining 37 patients. There were no differences in age, gender, 490 Clinical Investigations

4 drug administration, and duration of AF between the two groups (Table 2). Prior to cardioversion, the mean LAA peak flow velocity had been higher in the SR group compared to that in the AF group (Table 2). In 5 of 24 patients (20.8%) in the SR group, an early systolic abnormal mitral annulus motion had been recorded, while in the AF group this had been present in 35 of 37 patients (94.6%). The R-R interval and the time distance of the R wave to abnormal mitral annulus motion were not different between the two groups (Table 2). Early Systolic Abnormal Mitral Annulus Motion Recordings of the abnormal mitral annulus motion were consistently obtainable as long as the beam was oriented perpendicularly to the descending motion of the annulus structure. Of the 61 successfully converted patients, 40 had presented with an abnormal early systolic mitral annulus motion (ie, anotch), while in the remaining 21 patients a pattern without a notch, which was similar to that of the normal volunteers, had been observed (Fig 1, top, a, and middle, b). Only 2 of 21 patients (9.5%) were in AF after 12 months of follow-up, while the remaining Table 2 Demographic, Echocardiographic, and Other Data in Patients who Presented with SR vs Those With AF After 12 Months* Variable SR (n 24) AF (n 37) p Value Age, yr NS Gender NS Male Female 9 14 Duration of AF, mo NS Drugs NS Amiodarone Sotalol Mean LAA peak flow velocity, cm/s Presence of abnormal mitral 5 (20.8) 35 (90) annulus notch R wave to abnormal mitral annulus motion, ms NS R-R interval, ms NS Left atrium, mm NS IVS, mm NS LPWTh, mm NS LVEDD, mm NS LVESD, mm NS FS, % EF, % NS Mitral E-wave velocity, cm/s NS Pulmonary S/D peak velocity NS ratio Systolic fraction NS *Values given as mean SD or No. of patients (%), unless otherwise indicated. See Table 1 for abbreviations not used in the text. were still in SR. On the other hand, of 40 patients who had presented with the abnormal early systolic mitral annulus motion, only 5 patients (12.5%) were in SR while the remaining had reverted to AF. Prediction Model According to the logit regression model, the dependent variable is predicted by the following function: y exp (b 0 b 1 * X 1... bn * Xn) {1 exp (b 0 b 1 * X 1... bn * Xn)} where X 1 is LAA flow velocity and Xn is LV fractional shortening, and b , b , and bn For predicting the success of the cardioversion procedure, a model consisting of two variables, LAA flow velocity and LV fractional shortening, appears to be quite strong, yielding 83.3% correct results. Although this model exhibits a high overall accuracy, it is highly skewed toward positive predictions (correct positive predictions, 95.1%; correct negative predictions, 41.2%). The model was applied over the whole sample, and although, as expected, LAA flow explains most of the variability of the data, it provides for a model with a strong positive bias (positive prediction, 98.4% correct; negative prediction, 29.4% correct). Further range analysis has indicated negative outcomes for values of mean LAA flow velocity of 20 cm/s or for values of LV fractional shortening of 0.3 (30%), and two gray zones in which (1) the product of mean LAA flow velocity LV fractional shortening is between 5.7 and 10.5, and (2) mean LAA flow velocity is 20 cm/s but LV fractional shortening is 0.3 (30%). In the case of gray zone 1, the model is positively biased, resulting in inaccurate positive predictions, whereas in the case of gray zone 2, the model is negatively biased, resulting in inaccurate negative predictions. For predicting the maintenance of the cardioversion procedure, a model consisting of two variables, the absence of the early systolic abnormal mitral annulus motion (X 1 ) and LAA flow (Xn) velocity, appears to be quite strong, yielding 84.6% correct results. By removing the abnormal early systolic mitral annulus motion variable, the prediction value drops significantly to 70.5%. The mean LAA flow velocity ( 20 cm/s) only marginally enters the model, and, if removed, little predictive value is lost (dropping to 83.3%). Such a model can give us a very straightforward indication based on the existence of the abnormal mitral valve annulus motion before cardioversion (the existence of abnormal mitral an- CHEST / 127 / 2/ FEBRUARY,

5 nulus motion indicates a negative outcome with regard to the preservation of SR after 12 months). This model is slightly biased toward negative results, predicting negative results correctly 85.2% of the time, and positive results 79.2% of time. The inclusion of mean LAA flow velocity results in a model that is more skewed, providing accurate negative predictions in 88.9% of cases, but accurate positive predictions in only 75.0% of cases. Discussion The results of this study show that transesophageal echocardiography gives important information regarding the success of cardioversion and the maintenance of SR for up to 1 year in patients who have experienced a lone episode of AF. Specifically, LAA flow velocity combined with LV fractional shortening before cardioversion can predict the success of the conversion of AF to SR. Moreover, LAA flow velocity combined with the analysis of mitral annulus motion before cardioversion can predict the maintenance of SR on a long-term basis, with the latter being the most powerful predictor. LAA Flow Velocity The use of LAA areas and flow velocities in patients with AF has been proposed in predicting the immediate success of cardioversion and in the longterm preservation of SR. 20 However, due to conflicting data among several reports 11,12,21 23 its unique contribution remains uncertain. Indeed, different cutoff points for LAA flow velocity have been proposed in order to predict the maintenance of SR or the identification of patients with an increased risk of thrombus formation. 11,23 Accordingly, it has been suggested 24 that although high LAA flow velocity identifies patients with a greater likelihood of remaining in SR for 1 year after successful cardioversion, low LAA velocity is of limited value. In the present study, we found that the mean LAA peak flow velocity per se is not well able to predict the success of cardioversion or the maintenance of SR. However, when LAA flow velocity of 20 cm/s and LV fractional shortening of 30% are combined, the success of conversion of AF to SR can be predicted with a high overall accuracy. Additionally, the absence of the early systolic abnormal mitral annulus motion variable, combined with LAA flow velocity before cardioversion can predict the maintenance of SR on a long-term basis. However, LAA flow velocity only marginally enters the model, suggesting the importance of the analysis of mitral annulus motion in predicting the maintenance of SR. Analysis of Mitral Valve Annulus Motion In the present study, as expected, the mitral annulus motion during atrial contraction is absent, and the early-diastolic mitral annulus motion during ventricular relaxation is normally present. Interestingly, although LV performance was similar between the group in which SR was maintained and the group that experienced AF, systolic mitral annulus descent due to ventricular contraction might be different in patients who will remain in SR over the long term. In fact, the presence of a notch at the onset of the descending phase of the mitral annulus (ie, early systole) might identify the patients who will revert to AF during a 12-month follow-up period. The abnormal mitral annulus motion that was observed in our study is difficult to explain; however, it remains a phenomenon that occurs during early systole and coincides with the time of atrial relaxation, which, theoretically at least, is absent or different in patients with AF. In patients with chronic nonvalvular AF, timerelated structural and histologic remodeling, such as the loss of myofibrils, collagen formation, the development of interstitial fibrosis, the accumulation of glycogen, and chamber dilation, develops in the atrial myocardium. 25,26 These degenerative changes may have an impact on atrial systolic and diastolic function, as well as on atrial electrophysiologic properties precipitating the recurrence of AF. 27 Thus, we hypothesized that this abnormal motion may represent a retraction of the mitral annulus (toward the base of the heart) due to a stiff atrial myocardium, which cannot be passively distended, at least during the onset of ventricular contraction. Moreover, it can be inferred that the greater the extent and severity of structural abnormalities 25,26 resulting in a stiffer atrium (probably indicated by a marked early-systolic notch), the higher the recurrence rate of AF. Limitations A potential limitation of the study is the approximate estimation of the duration of arrhythmia, which is inherently reliant on the extended length of time elapsed between regular follow-ups (3 months), and this may underestimate the true incidence of AF recurrence. Additionally, although short episodes of sinus bradycardia or paroxysmal supraventricular tachycardia were not recorded during regular follow-up visits, we cannot rule out that such episodes might have been missed that resulted in an underestimation of some differences. LAA areas have not been assessed in our study; however, these measurements were subjected to substantial interobserver variability during both data registration and off-line 492 Clinical Investigations

6 analysis, possibly due to the complex three-dimensional anatomy of the LAA. 22,28 In contrast, the assessment of LAA function by Doppler echocardiography is easily performed, reproducible, and clinically highly relevant. 22,29 Finally, further studies are required to validate the model that was applied in this study, since this study has been carried out in a very specific group of patients. Presumably, the incidence of the abnormal mitral annulus motion might be higher in patients with AF and LV wall motion abnormalities or rheumatic mitral valve disease, suggesting, probably, the higher recurrence rate of AF. In conclusion, LAA flow velocity ( 20 cm/s) combined with LV fractional shortening ( 30%) before cardioversion can predict the success of the cardioversion of AF to SR. Moreover, LAA flow velocity ( 20 cm/s) combined with the absence of the abnormal early systolic mitral annulus motion variable can predict the maintenance of SR in the long term, with the latter being the most powerful predictor of SR maintenance over a 12-month follow-up period. ACKNOWLEDGMENT: The authors are indebted to George Tentis consultant statistician, Eleni Binou for meticulous secretarial support, and Polymnia Anthopoulou for excellent nursing assistance. References 1 Waris E, Kreus KE, Salokannel J. Factors influencing persistence of sinus rhythm after DC shock treatment of atrial fibrillation. Acta Med Scand 1971; 189: Flugelman MY, Hasin Y, Katznelson N, et al. Restoration and maintenance of sinus rhythm after mitral valve surgery for mitral stenosis. Am J Cardiol 1984; 54: Ewy GA, Ulfers L, Hager D, et al. Response of atrial fibrillation to therapy: role of etiology and left atrial diameter. J Electrocardiol 1980; 13: Henry WL, Morganroth J, Pearlman AS, et al. Relation between echocardiographically determined left atrial size and atrial fibrillation. Circulation 1976; 53: Gold RL, Haffajee CI, Charos K, et al. Amiodarone for refractory atrial fibrillation. Am J Cardiol 1986; 57: Karlson BW, Torstensson I, Abjorn C, et al. Disopyramide in the maintenance of sinus rhythm after electroconversion of atrial fibrillation: a placebo controlled one-year follow-up study. Eur Heart J 1988; 9: Palinkas A, Antonielli E, Picano E, et al. Clinical value of left atrial appendage flow velocity for predicting of cardioversion success in patients with non-valvular atrial fibrillation. Eur Heart J 2001; 22: Paraskevaidis IA, Theodorakis GN, Katritsis DG, et al. Pulmonary vein flow analysis by transesophageal echocardiography in patients with chronic atrial fibrillation: one year follow-up after cardioversion. Eur Heart J 1999; 20: Mitusch R, Garbe M, Schmucker G, et al. Relation of left atrial appendage function to the duration and reversibility of nonvalvular atrial fibrillation. Am J Cardiol 1995; 75: Tabara T, Oki T, Iuchi A, et al. Evaluation of left atrial appendage function by measurement of changes in flow velocity patterns after electrical cardioversion in patients with isolated atrial fibrillation. Am J Cardiol 1997; 79: Perez Y, Duval AM, Carville C, et al. Is left atrial appendage flow a predictor for outcome of cardioversion of nonvalvular atrial fibrillation? A transthoracic and transesophageal echocardiographic study. Am Heart J 1997; 134: Verhorst PM, Kamp O, Welling RC, et al. Transesophageal echocardiographic predictors for maintenance of sinus rhythm after electrical cardioversion of atrial fibrillation. Am J Cardiol 1997; 79: Pasierski JT, Alton AM, Pearson CA. Transesophageal echocardiography characterization of pulmonary vein flow not due to atrial contraction or mitral regurgitation. Am J Cardiol 1991; 68: Bartzokis T, Lee R, Yeoh KT, et al. Transesophageal echo Doppler echocardiographic assessment of pulmonary venous flow patterns. J Am Soc Echocardiogr 1991; 4: Paraskevaidis IA, Kremastinos D Th, Matsakas E, et al. Transesophageal detection of early systolic reverse pulmonary venous flow in atrial fibrillation. Am J Cardiol 1994; 73: Irani WN, Grayburn PA, Afridi I. Prevalence of thrombus, spontaneous echo contrast, and atrial stunning in patients undergoing cardioversion of atrial flutter: a prospective study using transesophageal echocardiography. Circulation 1997; 95: Fatkin D, Kelly RP, Feneley MP. Relations between left atrial appendage blood flow velocity, spontaneous echocardiographic contrast and thrombo-embolic risk in vivo. J Am Coll Cardiol 1994; 23: Fatkin D, Kuchar DL, Trorburn CW, et al. Transesphageal echocardiography before and during direct current cardioversion of atrial fibrillation: evidence for atrial stunning as a mechanism of thromboembolic complications. J Am Coll Cardiol 1994; 23: Sahn DJ, DeMaria A Kisslo J, et al. Recommendations regarding quantitation in M-mode echocardiography: results of a survey of echocardiographic measurements. Circulation 1978; 58: Grimn RA, Stewart WJ, Black IW, et al. Should all patients undergo transesophageal echocardiography before electrical cardioversion of atrial fibrillation? J Am Coll Cardiol 1994; 23: Omran H, Jung W, Schimpf R, et al. Echocardiographic parameters for predicting maintenance of sinus rhythm after internal atrial defibrillation. Am J Cardiol 1998; 81: Agmon Y, Khandheria BK, Gentile F, et al. Echocardiographic assessment of the left atrial appendage. J Am Coll Cardiol 1999; 34: Manabe K, Oki T, Tabata T, et al. Transesophageal echocardiographic prediction of initially successful electrical cardioversion of isolated atrial fibrillation: effects of left atrial appendage function. Jpn Heart J 1997; 38: Antonielli E, Pizzuti A, Palinkas A, et al. Clinical value of left atrial appendage flow for prediction of long-term sinus rhythm maintenance in patients with nonvalvular atrial fibrillation. J Am Coll Cardiol 2002; 39: Falk RH. Etiology and complications of atrial fibrillation: insights from pathology studies. Am J Cardiol 1998; 82: 10N 7N CHEST / 127 / 2/ FEBRUARY,

7 26 Aime-Sempe C, Folliguet T, Rucker-Martin C, et al. Myocardial cell death in fibrillating and dilated human right atria. J Am Coll Cardiol 1999; 34: Wijffels MC, Kirchhof CJ, Dorland R, et al. Atrial fibrillation begets atrial fibrillation: a study in awake chronically instrumented goats. Circulation 1995; 92: Pollick C, Taylor D. Assessment of left atrial appendage function by transesophageal echocardiography: implications for the development of thrombus. Circulation 1991; 84: Kamp O, Verhorst PM, Welling RC, et al. Importance of left atrial appendage flow as a predictor of thromboembolic events in patients with atrial fibrillation. Eur Heart J 1999; 20: Clinical Investigations