Cardiac computed tomography angiography‐derived analysis of left atrial appendage morphology and left atrial dimensions for the prediction of atrial fibrillation recurrence after pulmonary vein isolation

Abstract Background Left atrial appendage (LAA) is a potential source of atrial fibrillation (AF) triggers. Hypothesis LAA morphology and dimensions are associated with AF recurrence after pulmonary vein isolation (PVI). Methods From cardiac computed tomography angiography (CCTA), left atrial (LA), pulmonary vein (PV), and LAA anatomy were assessed in cryoballoon ablation (CBA) patients. Results Among 1103 patients undergoing second‐generation CBA, 725 (65.7%) received CCTA with 473 (42.9%) qualifying for detailed LAA analysis (66.3 ± 9.5 years). Symptomatic AF reoccurred in 166 (35.1%) patients during a median follow‐up of 19 months. Independent predictors of recurrence were LA volume, female sex, and mitral regurgitation ≥°II. LAA volume and AF‐type were dependent predictors of recurrence due to their strong correlations with LA volume. LA volumes ≥122.7 ml (sensitivity 0.53, specificity 0.69, area under the curve [AUC] 0.63) and LAA volumes ≥11.25 ml (sensitivity 0.39, specificity 0.79, AUC 0.59) were associated with recurrence. LA volume was significantly smaller in females. LAA volumes showed no sex‐specific difference. LAA morphology, classified as windsock (51.4%), chicken‐wing (20.7%), cactus (12.5%), and cauliflower‐type (15.2%), did not predict successful PVI (log‐rank; p = 0.596). Conclusions LAA volume was strongly correlated to LA volume and was a dependent predictor of recurrence after CBA. Main independent predictors were LA volume, female sex, and mitral regurgitation ≥°II. Gender differences in LA volumes were observed. Individual LAA morphology was not associated with AF recurrence after cryo‐PVI. Our results indicate that preprocedural CCTA might be a useful imaging modality to evaluate ablation strategies for patients with recurrences despite successful PVI.


Funding information
A research agreement between the local institution and Abbott Medical exists, and the most recent clinical software of EnSite Precision™ (Abbott Medical GmbH, Eschborn, Germany) was provided for the CT reconstruction and post-processing of the CCTA images. female sex, and mitral regurgitation ≥ II. Gender differences in LA volumes were observed. Individual LAA morphology was not associated with AF recurrence after cryo-PVI. Our results indicate that preprocedural CCTA might be a useful imaging modality to evaluate ablation strategies for patients with recurrences despite successful PVI.

| INTRODUCTION
Left atrial appendage (LAA) morphology can be determined by cardiac computed tomography angiography (CCTA). However, the role of the LAA in the initiation and perpetuation of AF has to be defined further. Pulmonary vein electrical isolation (PVI) is the standard strategy at the initial ablation procedure for the treatment of paroxysmal and persistent atrial fibrillation (AF). 1 Despite continuing technical improvements, 2 a significant proportion of patients have recurrence after the first ablation procedure. 3 Several mechanisms causing recurrence have been identified. Electrical reconnection of the PVs is the main mechanism, and gap closure in the second procedure remains the standard of care. 1,4,5 With the introduction of next generation ablation catheters and smart ablation protocols, the majority of PVs show durable isolation. 4,6,7 Additional lesions beyond PV isolation (PVI) (e.g., empirical lines, complex fractionated electrograms, ganglionated plexi, triggers) have been studied, but until now, these measures have not demonstrated additional benefit as compared to PVI alone in the initial ablation procedure. 8 Clinical predictors of AF recurrence after catheter ablation are LA size, AF type, female sex, and in-hospital AF relapse, as well as comorbidities such as impaired cardiac and renal function. 9,1011 LAA itself might also be a source of extra-PV AF triggers 12 or might serve as the substrate of perpetuating AF, though it is unknown whether the LAA anatomy correlates with the recurrence rate after PVI. LAA is a complex anatomical structure with substantial variation in size and morphology; it is not possible to determine LAA morphology or LAA size through transthoracic echocardiography (TTE). However, CCTA is the best modality to determine LAA morphology and atrial dimensions. 13 Other benefits from preprocedural cardiac computed tomography angiography (CCTA) include ruling out possible thrombi, detecting underlying coronary artery disease (CAD), and gaining advanced information of individual PV anatomy. 1,14 The present study sought to investigate whether LAA morphology and detailed measurement of a variety of LAA and LA parameters, determined by CCTA, can predict atrial arrhythmia recurrence after initial cryoballoon PVI in symptomatic AF patients.

| Study design
This prospective single-centre registry study enrolled consecutive patients undergoing second-generation CBA between May 2012 and September 2016. Informed consent was obtained from all patients.
The study was approved by the regional ethics review board and was conducted in accordance with the Declaration of Helsinki.

| Study participants
Consecutive symptomatic patients scheduled for the initial AF ablation procedure aiming at PV isolation were treated with the second-generation cryoballoon (Arctic Front Advance™, Medtronic Inc., MN, USA) and were prospectively enrolled into the institutional observational registry. If a recent preprocedural CCTA of sufficient quality to assess the LAA anatomy was available, the patient was considered for the present blinded analysis. The clinical indications for CCTA were, firstly, exclusion of CAD, and secondly, determination of LA and PV anatomy prior to the CBA. No patient was excluded from cryoballoon ablation based on variations in LA, PV, LAA anatomy, or LA volume, as determined by CCTA. None of the consecutive patients were treated by means of point-by-point radiofrequency (RF) ablation for PVI. Standard exclusion criteria for AF ablation were applied. 15 Baseline characteristics were collected prospectively.

| Objectives and endpoints
The primary objective of the study was to evaluate the impact of LAA anatomy and morphology on the recurrence of AF after CBA. The secondary objective was the general determination of possible independent clinical risk factors for AF recurrence.

| Preprocedural investigations
Prior to the ablation procedure, all patients underwent transthoracic and transesophageal echocardiography to exclude possible LA thrombus formation. Additionally, individual left atrial anatomy was revealed using a 64-slice CT scanner (Brilliance 64, Philips Medical Systems, Cleveland, OH, USA) with retrospective electrocardiography (ECG) gating and 3D reconstruction prior to the procedure.
Scanning was performed at 120 kVp, with an effective tube current of 600 mAs. The slice collimation was 64 Â 0.625 mm, with a gan-

| Data acquisition, management and quality control
The main 2D parameters included the maximal length, height, and depth of the LAA. The volume of the LAA and LA was computed automatically after anatomical segmentation. Major parameters analyzed three-dimensionally were the LAA length, roof top and bottom lines, distance from the mitral valve annulus to the middle of the LA roof, depth of the LA, and septum orifice distance (see Figure S3). cactus, and cauliflower. 16,17 After initial classification, there was a reassessment of all images among a team of four physicians for objective validation. CBA was performed as previously described 15,18 (Appendix).
All ablation procedures were performed by experienced interventional electrophysiologists (EH, UD, FS, MW). Following the intervention, all patients were monitored with continuous ECG for at least 24-48 h. TTE was performed to exclude pericardial effusion. In the event of symptoms, additional ECG and Holter studies were continued for up to 7 days. Holter recordings after 1, 3, 6, and at least 12 months were organized to screen for symptomatic or asymptomatic atrial arrhythmias.
Follow-up was also ensured in cooperation with referring physicians, and detailed questionnaires (15 questions) were administered for each case via mail and telephone calls. If there was any suspicion of recurrence, the referring physician was contacted to validate the diagnosis. Only recurrences outside of the 90-day blanking period were categorized as failures.   Figure S1 represents a flow diagram of the study. The mean age of patients who underwent PVI was 66.2 ± 9.5 years, of whom, 189 (40%) were females. The majority of patients (58.6%) had symptomatic paroxysmal AF. Table 1 illustrates the baseline characteristics of the study population in terms of recurrence and non-recurrence.

| Statistical analysis
All patients underwent CBA, and all PVs were successfully iso-

| LAA morphology and outcome
The distribution of the LAA morphological types is shown in Figure 1.
Among the 166 recurrence events, chicken-wing morphology had the highest chance of recurrence at 37.8%, followed by windsock at 36.5%, cauliflower at 32.0%, and cactus at 28.8%. None of these categories was found to have a statistically significant impact on the AF recurrence rate (log-rank; p = 0.596). The corresponding Kaplan-  Figure S2). p < 0.001). All significant results are listed in Table 2 and Table S1.

| Outcome predictors: Correlation analysis
To avoid problems of multicollinearity in multivariate models, correlation analysis was performed for LA volume and all baseline parameters and measurement data according to the Pearson and Spearman tests (see Table S2, Table S3). Among the baseline parameters, mitral regurgitation, AF type, sex, age, structural heart disease, and hypertensive heart disease were significantly correlated to LA volume (the correlation matrix is provided in the Table S2). All CCTA results that related logically to the LA volume, such as the septum orifice distance, depth of the LA, distance of the mitral valve annulus to the LA roof, and trapezoid area of the posterior left atrial wall, showed significant positive correlations, as did the LAA volume and its companion parameters (e.g., perimeter of the LAA ostium or area of the LAA ostium). Details are provided in the Table S3. LA volume was the best CCTA-derived measurement parameter for predicting AF recurrence and was, therefore, included in multivariate regression analysis.

| Outcome predictors: Multivariate analysis
After precise analysis of intervariable correlation, a stepwise multivari-  Table 3. All independent risk factors were evaluated using Kaplan-Meier Survival curves for AF recurrence during follow-up (see Figure 2).  Determined by transthoracic echocardiography. b Univariate analysis of baseline characteristics and measurement data according freedom of AF after cryoballoon ablation provided multiple highly significant parameters. To prevent multicollinearity in the multivariate analysis, CCTA measurements were reduced to LA volume as it represents the most significant univariate parameter and showed highly significant correlation to all other measurement parameters. After stepwise multivariate regression with bidirectional elimination, three parameters could be identified as independent risk factors: female sex, mitral regurgitation ≥ II and LA volume. The statistical significance of the multivariate Cox regression model was p < 0.000001.

| DISCUSSION
To the best of our knowledge, this is the largest trial evaluating the impact of LAA anatomy and morphology, determined by preprocedural CCTA, on the recurrence of atrial arrhythmias after PVI.
LAA morphology was not associated with AF recurrence after cryoballoon PVI. This finding is in contrast to that of Kocyigit et al.
who identified a relationship between cauliflower-type LAA morphology and recurrences after CBA. 19 In this sense, therefore, our study must be considered negative. Nevertheless, secondary and exploratory analyses strongly suggest that LAA size has a role in AF recur- Female sex has already been described as a clinical predictor of recurrence, 23 but meta-analyses have shown ambiguous evidence for sex-related differences in outcomes with AF ablation. 1 Our results indicated that female sex was an independent risk factor for AF recurrence after CBA. Female patients were older and underrepresented compared with male patients. These results are consistent with those of all AF ablation trials. However, age was not an independent predictor for AF recurrence after CBA, which is in line with the recently published data. 24,25 Interestingly, LA volume was significantly smaller in females, while the LAA volume in females was similar to that in males.
Hypothetically, sex-related differences in the presence of fibrotic atrial myopathy or additional extra-PV trigger sites may explain our observations. Ultimately, the answer to these questions may lead to specific, sex-tailored AF ablation strategies.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.