Cardiovascular magnetic resonance imaging for accurate sizing of the left atrium: Predictability of pulmonary vein isolation success in patients with atrial fibrillation




To prospectively determine the most reproducible approach for left-atrial size assessment using cardiovascular magnetic resonance (CMR) imaging in patients with atrial fibrillation and its value for prediction of pulmonary vein isolation (PVI) treatment success.

Materials and Methods

Eighty patients underwent CMR imaging prior to PVI; the CMR examination included standard cine sequences, a multislice cine sequence in 4-chamber orientation with full left-atrial coverage, and a contrast-enhanced MR angiography of the left atrium. Left-atrial size was determined as: diameter, area, volume segmented from angiography, and diastolic/systolic volumes from cine imaging (Simpson's rule). All measurements were carried out by two independent observers and repeated by one observer to assess inter- and intrareader variability. Treatment success was defined as persisting sinus rhythm after PVI (follow-up period 12.6 ± 6.6 months).


All left-atrial measurements showed substantial intrareader agreement. Interreader agreement was substantial for diastolic/systolic left-atrial volumes only. Calculated bias was found to be minimal (0.1%–4.9%). Predictability of PVI treatment success was best using cine volumetric measurements (cutoff value for diastolic volume, 112 mL) yielding a sensitivity and specificity of 80% and 70%, respectively.


Left-atrial volumetry based on cine imaging represented the most reproducible approach to determine left-atrial size. PVI success was predicted best using cine volumetry. J. Magn. Reson. Imaging 2011;33:455–463. © 2011 Wiley-Liss, Inc.

ATRIAL FIBRILLATION represents the most common sustained arrhythmia and is associated with a limited quality of life, an increased mortality, and relevant healthcare costs (1, 2). During the last years, pulmonary vein isolation (PVI) by radiofrequency percutaneous catheter ablation has proved to be an important therapeutic alternative in patients with symptomatic atrial fibrillation (1, 3). The reported success rates of ablation procedures differ widely depending on the selection criteria of the study population (4, 5). The left-atrial (LA) size is known to play an important role for procedure planning, performance, and outcome (6, 7) and, thus, a standardized imaging approach for the accurate and reproducible assessment of LA size is needed. Currently, transthoracic echocardiography is the most commonly used method for initial determination of LA size and during follow-up examinations, as it is noninvasive and widely available. However, visualization of the left atrium by routinely performed transthoracic echocardiography is limited to one- or two-dimensional measurements, resulting in an insufficient estimate of LA size, particularly in case of asymmetric LA dilatation. Cardiovascular magnetic resonance (CMR) imaging is increasingly gaining importance since it allows 3D visualization of all cardiac cavities with a consistently high endocardial border delineation. However, the accurate determination of LA size, especially in patients with atrial fibrillation, is challenging for CMR imaging since varying RR interval lengths may lead to impaired image quality resulting from electrocardiographic mistriggering, thereby potentially decreasing reliability of measurements. At present, a standardized evaluation procedure of LA size in order to assist planning and prognostication of success of ablation procedures in patients with atrial fibrillation does not exist.

Thus, the objective of the present study was to determine the most reproducible measurement using CMR imaging for LA size assessment and to verify the value of the different LA measurements with regard to predictability of the treatment success of pulmonary vein isolation.



Eighty patients (51 men; mean age: 60 ± 10 years; range: 30–77 years) referred for pulmonary vein isolation were prospectively enrolled between June 2007 and March 2009 after written informed consent was obtained. Patients with symptomatic paroxysmal or nonparoxysmal atrial fibrillation were included if medical treatment had failed to obtain a stable sinus rhythm and the left atrium showed a diameter <5.5 cm on current transthoracic echocardiography. Patients with known contraindications to CMR imaging (eg, noncompatible biometallic implants, claustrophobia) were not considered for study inclusion. The study was approved by the Charité Institutional Review Board.

The CMR examination was performed within 7 days prior to the ablation procedure; cardiovascular risk factors were assessed and a resting electrocardiogram (ECG) was used to document the rhythm at the time of the CMR examination. A transesopheagal echocardiography was performed within 24 hours prior to the electrophysiological examination to exclude any intraatrial thrombus formation. During the 2D echocardiographic examination, LA end-diastolic diameter and area were measured using a 4-chamber orientation; LA diastolic and systolic volumes were calculated using the area–length method as previously recommended (8).

A successful ablation procedure was defined as the absence of symptomatic or documented atrial fibrillation following an initial 3-month blanking period and was verified by periodic ECGs at rest (at 1, 2, 3, 6, 9, and 12 months) and repetitive 24-hour Holter monitoring recorded at 1, 3, 6, and 12 months after PVI. In case symptomatic arrhythmias occurred, the patients were asked for an immediate presentation and additional ECG documentation was performed.

MR Study

CMR imaging was performed with the patient in the supine position using a 3.0 T MR scanner (Philips Achieva, Best, Netherlands) equipped with a Quasar Dual gradient system (40 mT/m; 200 mT/m/ms) based on Philips software release 2.6.1. A 6-element cardiac synergy coil was used for signal reception and cardiac synchronization was performed with a vector-ECG. After acquisition of a multistack, multislice localizer of the heart, cine imaging of the standard cardiac geometries was done: 1) short axis orientation (12 to 14 slices, no gap, full left-ventricular coverage), and 2) three standard long axis views (ie, 4-, 2-, and 3-chamber orientation). In addition, for full LA coverage 8 to 10 slices without a gap based on the 4-chamber orientation were acquired.

For all cine images, balanced turbo field echo sequences were used during multiple end-expiratory breathold maneuvers with retrospective gating (TR/TE/flip angle: 3.5 msec/1.8 msec/45°; 50 phases per cardiac cycle; measured spatial resolution: 1.8 × 1.8 × 8.0 mm). Finally, a contrast-enhanced MR angiography of the left atrium and the pulmonary veins was acquired during breatholding without ECG gating (TR/TE/flip angle: 4.3 msec/1.5 msec/20°; measured spatial resolution: 1.1 × 1.1 × 1.6 mm; 60 slices) using real-time bolus tracking with a bolus injection of 0.1 mmol/kg Gd-BOPTA (Multihance, injection rate 3.0 mL/s).

Pulmonary Vein Isolation

The ablation procedure was guided by CARTO Merge electroanatomical mapping. Three catheters were introduced via right femoral venous access under local anesthesia, supplemented by sedation if necessary. A four-pole catheter was placed in the right ventricular apex and a diagnostic catheter in the coronary sinus. An irrigated tip catheter and a mapping catheter were introduced in the left atrium after double transseptal puncture. Intravenous heparin was administrated to maintain an activated clotting time between 250 and 350 seconds. Bipolar electrograms were filtered at bandpass settings of 30–500 Hz and were recorded digitally (EPMed Systems, Lakeland, FL). Radiofrequency ablation was performed to encircle the right and left pulmonary veins in pairs at a distance of 5–10 mm to their ostia using a 3.5-mm irrigated tip catheter (17 mL/min, 30–35W, 43°C; NaviStar Thermocool, Biosense Webster, Diamond Bar, CA). Radiofrequency energy was delivered until the amplitude of the local bipolar atrial electrogram had been reduced by >80% or was <0.1 mV. The endpoint of the procedure was the electrical isolation of all pulmonary veins (entrance-block) as confirmed by abolishment of the PV potential.

CMR Image Analysis

CMR data were evaluated using Extended Workspace Software (Philips Medical Systems). For determination of interreader agreement, LA measurements were independently carried out by two experienced MR readers. For determination of intrareader agreement, one reader repeated the analysis after 3 months. The following measurements of the left atrium were performed.

Diameter (mm)

LA anterior posterior diameter was measured on the 3-chamber view based on echocardiographic standard criteria with LA diastole defined as the last phase image before opening of the mitral valve (Fig. 1a).

Figure 1.

Illustration of LA size CMR measurements. a: LA diameter was assessed in anterior–posterior direction using the 3-chamber view. b: LA area was determined from the LA diastolic endocardial contour using the 4-chamber view. c: LA volume was measured by segmentation of the left atrium (light gray) and pulmonary veins (dark gray) using contrast-enhanced, 3D MR angiography and was calculated pixel-wise. d: Calculation of LA diastolic and systolic volumes according to Simpson's rule: endocardial contours of the left atrium were traced on cine images in 4-chamber orientation with full gapless coverage of the left atrium resulting in the volume–time curve as shown.

Area (cm2)

LA area was measured in the 4-chamber view using the last phase image before mitral valve opening. A contour was manually drawn on the endocardial border of the left atrium (Fig. 1b).

Angiographic Volumetry (mL)

LA angiographically derived volume was assessed by segmentation of the contrast-enhanced MR scan. The mitral valve leaflets were used as landmarks to separate left atrium from left ventricle; pulmonary veins and the LA appendage were excluded (Fig. 1c).

Diastolic and Systolic Cine Volumetry (mL)

LA diastolic and systolic volumes were measured by applying Simpson's disc summation method using the cine 4-chamber view sequence with full LA coverage. The endocardial border of the left atrium was traced manually in all slices using the phase images showing the maximum LA extent and were subsequently propagated to all remaining phases using a semiautomatic contour detection algorithm and manually corrected if necessary. The resultant volume–time curves determined the LA diastolic volume as the maximal volume and the LA systolic volume as the minimal volume (Fig. 1d).

Statistical Analysis

Statistical analysis was performed using the SPSS software package release 17.0.0 (Chicago, IL). Pearson correlation was used to test for statistical correlation between LA measurements. Bland–Altman analysis was carried out to assess the inter- and intrareader reproducibility of LA measurements; the degree of agreement was determined as the mean absolute difference (bias), 95% confidence interval (95% CI) of the mean difference and mean relative difference (%bias, mean difference of two measurements divided by their mean value). In addition, Lin's concordance correlation coefficient was calculated with the following scale to describe the strength of agreement: >0.99 indicates almost perfect agreement; 0.95–0.99, substantial agreement; 0.90–0.95, moderate agreement; <0.90, poor agreement. Unpaired Student's t-test was used to examine differences between groups. Univariate analysis was done to identify clinical variables and LA size measurements related to the prediction of PVI treatment success. To determine the relationship between LA size and PVI treatment success, receiver-operating curve (ROC) analysis was performed and the area under the curve was calculated.

All tests were two-tailed; P < 0.05 was considered statistically significant.


Patient Characteristics

Paroxysmal atrial fibrillation was present in 49 patients (61%) and nonparoxysmal atrial fibrillation in 31 patients (39%). During the CMR examination 38 patients (48%) had regular sinus rhythm while 42 patients (52%) had atrial fibrillation; the CMR examination was completed successfully in all patients. Additional patient characteristics and left-ventricular function at rest as assessed by CMR imaging are summarized in Table 1.

Table 1. Patient Demographics and Clinical Data
 80 Patients
Patient characteristics
 Sex, M/F51/29
 Age, years60.2 ± 9.9
 Range30 – 77
 BMI, kg/m227.0 ± 3.4
Atrial fibrillation, n (%)
 Paroxysmal49 (61.3%)
 Persistent23 (28.7%)
 Long-term persistent8 (10.0%)
Cardiovascular risk factors, n (%)
 Arterial hypertension53 (66.3%)
 Diabetes mellitus10 (12.5%)
 Hyperlipoproteinemia48 (60.0%)
 Known CAD21 (26.3%)
Left ventricular function 
 LVEF, %58.6 ± 5.0
 LVEDV, mL135.0 ± 39.6
 LVESV, mL57.4 ± 20.5

Evaluation of LA Size

For all patients LA diameter ranged from 27.1–54.7 mm, LA area from 13.6–37.4 cm2, and LA angiographic volume from 50.4–247.7 mL. Diastolic and systolic LA cine volumes ranged from 55.6–183.2 mL and from 29.0–148.5 mL, respectively. LA diameter showed only a weak correlation with area, angiographic volume, and diastolic cine volume (r = 0.65, r = 0.43, and r = 0.53, respectively). LA area correlated weakly with angiographic volume (r = 0.49) but stronger with diastolic cine volume (r = 0.72). LA angiographic volume and diastolic cine volume showed a strong correlation (r = 0.71, P < 0.01). Figure 2 illustrates the scatterplots with corresponding 95% confidence bands of the regression lines for the different CMR approaches to determine LA size.

Figure 2.

Scatterplots illustrating the relationship between the different CMR approaches for LA size assessment. In each plot the black line indicates the regression line and the gray lines the corresponding 95% confidence band.

Intra- and Interreader Reproducibility

Intrareader reproducibility was high with regard to all LA measurements, demonstrating a substantial agreement; the corresponding bias was found to be minimal (Table 2). Interreader reproducibility was poor for the measurement of diameter, area, and angiographic volume; however, the measurements of diastolic and systolic cine volumes yielded a substantial interreader agreement. The corresponding bias was minimal (Table 2). Bland–Altman plots illustrating intra- and interreader agreement of LA measurements are given in Fig. 3. Comparing patients with sinus rhythm to patients with atrial fibrillation during the CMR examination, the bias of all LA measurements was not significantly different (Table 3).

Figure 3.

a,b: Bland–Altman plots demonstrating intra- and interreader agreement (upper and bottom row, respectively) of LA size CMR measurements (diameter, area, angiographic volume, and cine volumes). In each plot the central horizontal line indicates the mean absolute difference, upper and lower lines represent the mean ± two standard deviations.

Table 2. Intra- and Interreader Agreement of Left-Atrial Size Measurements
 Read 1Read 2Bias [95%-CI]%BiasPearson, rLin's, ρc [95%-CI]
Intrareader variability
 Diameter [mm]37.7 ± 6.237.7 ± 6.10.03 [−0.26; 0.32]0.080.98*0.98 [0.97; 0.99]
 Area [cm2]26.7 ± 5.626.7 ± 5.3−0.05 [−0.43; 0.34]0.190.95*0.95 [0.93; 0.97]
 Angiographic volume [mL]105.2 ± 31.3106.1 ± 29.9−0.93 [−2.92; 1.06]0.880.96*0.96 [0.94; 0.98]
 Diastolic cine volume [mL]106.6 ± 26.0106.2 ± 24.80.40 [−1.12; 1.92]0.380.97*0.96 [0.95; 0.98]
 Systolic cine volume [mL]76.1 ± 27.476.0 ± 26.00.69 [−0.78; 2.17]0.910.97*0.97 [0.96; 0.98]
 Reader 1Reader 2Bias [95%-CI]%BiasPearson, rLin's, ρc [95%-CI]
Interreader variability
 Diameter [mm]37.7 ± 6.237.6 ± 6.00.14 [−0.54; 0.82]0.370.87*0.87 [0.82; 0.93]
 Area [cm2]26.7 ± 5.627.3 ± 5.8−0.59 [−1.27: 0.09]2.210.86*0.85 [0.79; 0.91]
 Angiographic volume [mL]105.2 ± 31.3105.8 ± 33.8−0.63 [−4.59; 3.33]0.600.85*0.85 [0.79; 0.91]
 Diastolic cine volume [mL]106.6 ± 26.0106.0 ± 28.10.54 [−1.46; 2.54]0.510.95*0.94 [0.92; 0.97]
 Systolic cine volume [mL]76.1 ± 27.472.3 ± 28.63.73 [1.89; 5.57]4.900.96*0.95 [0.93; 0.97]
Table 3. Mean Bias of Left-Atrial Size Measurements in Patients With and Without Atrial Fibrillation During CMR Imaging
 Sinus rhythm (n=38)Atrial fibrillation (n=42)P
Bias - intrareader variability
 Diameter [mm]−0.00 ± 1.320.05 ± 1.320.853
 Area [cm2]−0.12 ± 1.540.02 ± 1.880.717
 Angiographic Volume [mL]0.11 ± 8.29−1.87 ± 9.490.325
 Diastolic cine volume [mL]−0.59 ± 7.711.30 ± 5.870.218
 Systolic cine volume [mL]−0.64 ± 7.581.90 ± 5.420.087
Bias - interreader variability
 Diameter [mm]−0.36 ± 3.020.59 ± 3.050.167
 Area [cm2]−0.58 ± 2.66−0.60 ± 3.400.976
 Angiographic Volume [mL]−3.79 ± 18.862.23 ± 16.450.132
 Diastolic cine volume [mL]−0.91 ± 10.021.85 ± 7.790.172
 Systolic cine volume [mL]2.53 ± 9.124.81 ± 7.330.220

LA Size Measurements in Patients With Sinus Rhythm or Atrial Fibrillation During the CMR Examination

Comparing patients with sinus rhythm (n = 38) to patients with atrial fibrillation (n = 42) during the CMR examination, angiographic volumes (94.8 ± 27.5 mL vs. 114.6 ± 31.9 mL, P = 0.004) and systolic cine volumes (63.1 ± 25.7 mL vs. 87.8 ± 23.4 mL, P < 0.01) were significantly smaller; no significant group differences were found for diameter (36.4 ± 6.0 mm vs. 38.9 ± 6.2 mm, P = 0.08), area (25.8 ± 5.9 cm2 vs. 27.4 ± 5.3 cm2, P = 0.21), and diastolic cine volumes (101.0 ± 28.0 mL vs. 111.6 ± 23.2 mL, P = 0.07).

Comparison of CMR Imaging With Echocardiography

For all patients, echocardiographic measurements yielded an average LA diameter of 40.5 ± 5.7 mm, an area of 23.9 ± 5.1 cm2, and a diastolic and systolic volume of 112.7 ± 32.4 mL and 79.8 ± 26.1 mL, respectively. Echocardiographic assessments correlated weakly with CMR imaging regarding diameter and area (r = 0.35 and r = 0.65, respectively) but stronger for diastolic and systolic volume determination (r = 0.77 and r = 0.75, respectively; P < 0.01). Corresponding Bland–Altman plots are shown in Fig. 4. Comparing patients with sinus rhythm to patients with atrial fibrillation during the CMR examination, the bias of all LA measurements between echocardiography and CMR imaging did not differ significantly (Table 4).

Figure 4.

Bland–Altman plots illustrating the agreement of echocardiography and CMR imaging for the assessment of LA size (diameter, area, diastolic, and systolic volumes). In each plot the central horizontal line indicates the mean absolute difference, upper and lower lines represent the mean ± two standard deviations.

Table 4. Mean Bias of Left-Atrial Size Measurements in Patients With Sinus Rhythm or Atrial Fibrillation During CMR Imaging in Comparison to Echocardiography
 Sinus rhythm (n=38)Atrial fibrillation (n=42)P
Bias - CMR vs. echo
 Diameter [mm]−0.3 ± 5.2−2.4 ± 8.00.583
 Area [cm2]2.2 ± 4.73.3 ± 4.30.259
 Diastolic volume [mL]−8.0 ± 21.4−4.4 ± 19.90.436
 Systolic volume [mL]−7.8 ± 15.7−0.1 ± 20.70.066

Prediction of PVI Success

Seventy-four out of the 80 patients finished the ablation procedure; reasons for an incomplete or postponed ablation procedure included an insufficient registration of atrial electrograms due to extensive atrial fibrosis (n = 1), the detection of an intraatrial thrombus during prior transesophageal echocardiography (n = 1), an amiodarone-induced hyperthyreosis (n = 1), the need for prior coronary intervention (n = 2), and the decision to perform isthmus ablation due to documented atrial flutter (n = 1). Of the 74 patients with a completed pulmonary vein isolation procedure, 54 patients (73%) showed a constant sinus rhythm during a mean follow-up period of 12.6 ± 6.6 months (range, 6–25 months) and thus were classified as successfully treated; the remaining 20 patients (27%) had a documented relapse of atrial fibrillation following the 3-month blanking period and were classified as nonsuccessful.

Patients with successful PVI showed significant differences in diameter, angiographic volume, and cine volume measurements of LA size when compared to patients with relapsed atrial fibrillation during the follow-up period; measurement of LA area did not yield a significant difference between either patient group (Table 5).

Table 5. Comparison of Left-Atrial Size Measurements in Patients with Persistent Sinus Rhythm and Patients with Relapse of Atrial Fibrillation During Follow-up
 Successful PVI (n=54)Relapse of atrial fibrillation (n=20)P
Diameter [mm]36.7 ± 6.339.4 ± 5.80.033
Area [cm2]25.7 ± 5.128.2 ± 5.90.128
Angiographic volume [mL]97.6 ± 24.3123.9 ± 39.10.001
Diastolic cine volume [mL]99.0 ± 20.2122.5 ± 26.10.001
Systolic cine volume [mL]67.4 ± 22.593.8 ± 27.5<0.001

ROC analyses were performed in order to identify the cutoff value of each LA measurement approach for prediction of PVI treatment success (Table 6; Fig. 5). The best predictability was achieved with LA cine volumetry using a diastolic cutoff value of 112 mL (sensitivity and specificity of 80% and 70%, respectively). Clinical variables (ie, age, hypertension, and left-ventricular ejection fraction) and LA size measurements were tested regarding predictability of PVI treatment success: on univariate analysis only the presence of paroxysmal atrial fibrillation and LA diastolic volume derived from cine volumetry demonstrated a significant relationship (P = 0.026 and P < 0.001, respectively).

Figure 5.

ROC analyses to determine the cutoff values for each LA size CMR measurement (diameter, area, angiographic volume, and cine volumes) being predictive of a constant sinus rhythm after PVI. Cine volumes predicted best the treatment success using a diastolic cutoff value of 112 mL (sensitivity and specificity of 80% and 70%, respectively). AUC indicates area under the curve.

Table 6. ROC Analysis for Prediction of PVI Success
 Cutoff valueSensitivity [%]Specificity [%]AUC
Diameter [mm]38.2 mm63.060.00.663
Area [cm2]28.0 cm268.555.00.616
Angiographic volume [mL]114.1 ml74.160.00.742
Diastolic cine volume [mL]112.3 ml79.670.00.754
Systolic cine volume [mL]84.0 ml72.265.00.765


PVI in patients with drug-resistant and symptomatic paroxysmal or persistent atrial fibrillation has become an accepted therapeutic standard. In the present study the success rate of PVI was within the range of previously published reports (62%–86% during the initial 6–12 month following PVI) (1). Known risk factors for early recurring atrial fibrillation after PVI are advanced age, arterial hypertension, increased LA size (>40 mm diameter as defined by echocardiography), and the preprocedural duration of atrial fibrillation (6). Yet it has been reported that these clinical risk factors are not sufficiently predictive of the treatment success of ablation procedures, which is suggestive of more complex interactions (7). In addition, several studies demonstrated that the echocardiographically measured increase in LA diameter was an insufficient predictor of postprocedural atrial fibrillation recurrence (5, 9). Most likely, diameter measurements of the left atrium do not accurately reflect the true extent of LA enlargement, particularly in case of geometrically complex LA remodeling. Consequently, direct volumetric measurement without geometric assumptions may help to overcome this limitation (10). Multislice computed tomography (MSCT) and CMR imaging offer the advantage of detailed, 3D visualization of the left atrium and provide important information regarding the anatomic characteristics of the pulmonary veins and anatomic relationships (11). Early MSCT studies yielded promising results regarding LA volumetry and the prediction of PVI success (7, 12). CMR imaging offers the additional advantage of functional evaluation of the left atrium allowing for the assessment of volume–time curves with accurate determination of diastolic and systolic volumes. However, a standard imaging procedure of LA size determination has not been defined yet.

In order to establish such an imaging strategy, two main requirements need to be fulfilled: first, reproducibility of LA size measurements must be high, and second, the prediction of ablation success should be possible. Considering this, CMR imaging is capable to allow direct comparison of diameter, area, angiographic volumetry, and cine volumetry in the same patient and during a single session examination.

The present study demonstrated that cine volumetry of the left atrium according to Simpson's rule achieved substantial inter- and intrareader agreement even in patients presenting with atrial fibrillation during the CMR examination. In addition, cine volumetry facilitated the most accurate prediction of PVI success among all parameters of LA size: the presence of a constant sinus rhythm during a mean follow-up period of 1 year after PVI (diastolic cutoff value of 112 mL) was predicted with a sensitivity and specificity of 80% and 70%, respectively. Thus, the high reproducibility of LA cine volumetry was accompanied by a reasonably high predictability of PVI success. Consequently, it may be considered the preferred parameter of LA sizing.

In conclusion, cine volumetry provided the most reproducible CMR approach for LA sizing and yielded high intra- and interreader agreement. Importantly, the presence of atrial fibrillation during the CMR examination did not affect the high reproducibility of LA measurements. In addition, preprocedural cine volumetry was highly predictive of the treatment success of pulmonary vein isolation. Thus, in patients with paroxysmal and nonparoxysmal atrial fibrillation, cine volumetry of the left atrium can be used to assist in procedure planning.

Study Limitations

The patients included in the current study showed a relatively narrow range of LA size due to the predefined echocardiographic exclusion criterion of a LA diameter >5.5 cm. Thus, the results do not reflect the entire range of LA sizes seen in patients with atrial fibrillation. In addition, patients with reduced left-ventricular ejection fraction or with uncontrolled arterial hypertension were not included and, consequently, the results may not be extrapolated to these patient groups. The time period of follow-up may be considered relatively short, with a mean of 12 months and the proportion of patients with relapse of atrial fibrillation at a later timepoint remains undetermined.