2.1. Study population
The study population consisted of 649 patients with drug-refractory AF who underwent cardiac CT before catheter ablation in the Heart and Vascular Center of Semmelweis University, Budapest, Hungary between 2014 and 2017. Exclusion criteria were age under 18 years, non-diagnostic image quality of CT, repeated ablation and patients in whom assessment of LAA flow velocity was not feasible. Those without data on the history of stroke/TIA were also excluded from the analysis. History of stroke/TIA was collected from patients’ chart reviews.
2.2. Cardiac CT imaging
Cardiac CT examinations were performed with a 256-slice scanner (Brilliance iCT 256, Philips Healthcare, Best, The Netherlands) with prospective ECG-triggered axial acquisition mode. For cardiac CT, 100-120 kV with 200-300 mAs tube current was used depending on patient anthropometrics. Image acquisition was performed with 128x0.625 mm detector collimation, and 270 msec gantry rotation time. For heart rate control, a maximum of 50-100 mg metoprolol was given orally and 5-20 mg intravenously, if necessary. Iomoprol contrast material (Iomeron 400, Bracco Ltd, Milan, Italy) was used with 85-95 ml contrast agent at a flow rate of 4.5-5.5 ml/sec through an antecubital vein via an 18-gauge catheter, using a four-phasic protocol. Bolus tracking in the left atrium was used to obtain proper scan timing. 0.8 mg sublingual nitroglycerin was given between the native and CT angiography acquisitions. CT data sets were reconstructed with 0.8 mm slice thickness and 0.4 mm increment.
After defining LA and LAA borders with caution to the orifices of the pulmonary veins and the level of the mitral valve, we measured LA and LAA volumes and determined LAA morphologies based on three-dimensional volume-rendered images using a semiautomated software package (EP Planning, Philips IntelliSpace Portal, Philips Healthcare, Best, The Netherlands). Since assessment of LAA morphology can be highly subjective, LAA morphologies were determined by consensus reading of three expert readers using rigorous definitions in order to minimize inter-observer variability. LAA morphologies were classified in four different types as previously described: 1.) Cauliflower, if LAA has limited length and the distal width exceed the proximal width; 2.) Windsock, if the primary structure is one dominant lobe with sufficient length; 3.) Chicken wing, if the dominant lobe has an obvious bend in the proximal or middle part; and 4.) Swan if LAA has a second sharp curve folding the dominant lobe back . Representative examples are provided in Figure 1.
2.3. LAA flow velocity measurement
Maximum 24h before ablation, all patients underwent TEE examination to exclude the presence of LAA thrombus. iE33 and Epiq 7C (Philips Medical System, Andover, MA) systems equipped with S5-1 phased array and X7-2t matrix TEE transducers were used. TEE was performed during conscious awake sedation. The LAA was imaged from 0°, 45°, 90° and 135° views to detect spontaneous echo contrast, sludge or thrombus. Subsequently, a sample volume was placed at the middle portion of the LAA and the peak velocity of the outflow of the LAA was measured.
2.4. Statistical analysis
Categorical variables are expressed as frequencies (percentages), and continuous values are expressed as mean±standard deviation (SD). Normality of continuous parameters was tested with Shapiro-Wilk test. Tests for significance were conducted using Mann-Whitney-Wilcoxon or Kruskal-Wallis tests for continuous variables and Pearson’s chi-square or Fisher exact tests (if 5 or less observations were included) for categorical variables. The odds ratio (OR) and 95% confidence interval (CI) values of stroke/TIA were computed using uni- and multivariate logistic regression analyses. In the multivariate model, adjustment was made for CHA2DS2-VASc-score risk factors, such as, heart failure, left ventricular systolic dysfunction (defined as left ventricular ejection fraction (LVEF) <50%), blood pressure >140/90 mmHg or antihypertensive therapy, age >65 years, diabetes mellitus, peripheral vasculopathy, obstructive coronary artery disease (CAD; defined as >50% luminal stenosis), and female sex. Moreover, obesity (defined as body mass index (BMI) >30kg/m2), hyperlipidemia, and renal dysfunction (defined as eGFR <60ml/min/1.73m2) were also included in the model beyond LA and LAA parameters. Receiver operating characteristic curve analysis was performed to determine optimal cut-off points of the LAA orifice area and flow velocity for stroke/TIA based on Youden-index. Differences between the various LAA morphology types were determined with ANOVA and post-hoc Tukey’s honest significant difference test. All tests were two-sided and a p-value <0.05 was considered statistically significant. All analyses were performed using statistical software R (version: 3.6.1) and its packages, namely ‘pROC’ (version: 1.15.3), ‘yarrr’ (version: 0.1.5).
The data underlying this article will be shared on reasonable request to the corresponding author.