We retrospectively analyzed 332 consecutive patients treated with NOAC who were referred to two centers between September 2013 and October 2016. All patients had documented symptomatic AF and underwent radiofrequency catheter ablation for AF with uninterrupted rivaroxaban. Baseline demographic and clinical characteristics, including age, gender, BMI, hypertension, diabetes mellitus, coronary artery disease, and congestive heart failure, were recorded for all patients. Data related to the diagnosis of AF, including AF pattern, oral anticoagulation, and previous interventions, was also recorded. Risk stratification for thromboembolism and bleeding was performed based on the CHA2DS2-VASc and HAS-BLED score, respectively. Creatinine clearance was estimated by Cockcroft-Gault eq. A 2D transthoracic echocardiogram was carried out on all patients to estimate left ventricular ejection fraction and any structural heart diseases.

This study was approved by the Institutional Review Board of Petrovsky National Research Centre of Surgery (IRB No. 130/1).

TEE was performed (iE33, Philips Medical Systems, MA, US) and interpreted by an experienced echocardiographer for the presence or absence of the LA thrombus. A thrombus was considered to be present if a mass detected in the LA appendage or other sites appeared to be distinct from the underlying endocardium in more than one imaging plane and was not caused by pectinate muscles.

ICE was performed by using a 10F AcuNav phased-array probe (Siemens AG, Germany) connected to a Vivid i system (General Electric, IL, US), and advanced through the left femoral vein. The catheter was initially positioned in the right atrium and then was rotated to visualize whole left atrium including the appendage. It was then curved and advanced into the right ventricle next to the pulmonic artery for a better visualization of the left atrial appendage. In some cases, the ICE probe was advanced into the coronary sinus, and LA appendage imaging was performed (Fig 1). Moreover, ICE was used during the procedure for the identification of procedure-related complications.

All patients received 20 mg of rivaroxaban 4 weeks prior to the procedure without discontinuation in the intra-procedural period. The LAA visualization approach was chosen by the operator, thus neither in the TEE nor the ICE group, we did not use both modalities. All ablation procedures were performed under deep sedation. A bolus of heparin (100 IU/L) was administered immediately after vascular access before double transseptal puncture. Anticoagulation therapy was continued throughout the procedure with a targeted activated clotting time (ACT) above 350 s. The ACT was repeated every 20–30 min, and additional bolus was administrated to maintain target ACT. Transseptal sheaths were continuously flushed with heparinized saline (2000 IU/L) 180 ml per hour. A 3D map of the LA geometry was created using the Carto 3 navigation system (Biosense Webster Inc., CA, US). Continuous point-by-point ablation using a 3.5 mm irrigated open-tip catheter (ThermoCool SmartTouch D-F curve, Biosense Webster Inc., CA, US) was used to create a pulmonary vein antrum lesion set in all patients. For patients with long-standing persistent AF were performed additional linear ablation across the LA roof, in the mitral valve isthmus or LA anteroseptal region, and in the cavotricuspid isthmus. The cavotricuspid line was applied in patients with previously reported or observed typical atrial flutter. In the case of spontaneous induction of atypical atrial flutter, an activation map had been constructed, and ablation of the zone of interest was done. A deflectable circular mapping catheter (Lasso 2515, Biosense Webster inc., CA, US) was used for verification of the pulmonary veins isolation and line conduction block. At the end of the procedure, if pulmonary vein isolation (PVI) alone or plus linear did not achieve sinus rhythm, electrical cardioversion was performed.

All complications within the first 48 h after the ablation procedure were included in this study. All patients were recommended to continue rivaroxaban for at least 3 months or long-life after discharge if the CHA2DS2-VASc score was 0 or ≥1, respectively.

Continuous variables were expressed as the mean ± standard deviation. Categorical variables were expressed as absolute numbers and percentages. Differences between means among groups were compared using the Student t test for continuous variables and with the nonparametric χ2 test or Fisher exact test when appropriate for categorical variables. A p value less than 0.05 was considered significant. SPSS software version 20.0 (IBM, NY, US) was used for statistical analysis.

A total of 115 patients (37 females, mean age 57.4 ± 11.5 years) underwent TEE prior to catheter ablation, and 217 patients (103 female, mean age 67.6 ± 8.3 years) underwent ICE during the procedure without prior TEE. Patients in the ICE group tended to be older, having higher BMI and LA size, and more often suffer from hypertension and diabetes mellitus, while the incidence of coronary artery disease was lower. The average CHA2DS2­VASc tended to be higher in the ICE group, but HAS-BLED scores were similar in both groups. However, the CHA2DS2­VASc score was ≥2 only in 64 (55.7%) patients in the TEE group versus 214 (98.6%) patients in the ICE group. Two hundred and sixty-seven patients (80.4%) had paroxysmal AF, 56 patients (16.9%) had persistent AF, and 9 patients (2.7%) had long-persistent AF. Thirty-eight patients underwent a redo procedure after the previous PVI. Three patients had a concomitant atypical atrial flutter before the procedure. In one patient, atypical atrial flutter was observed during the procedure.

The left atrial appendage was adequately visualized in all cases. None of the patients have an identifiable LA thrombus either in the TEE group or ICE group.

Baseline characteristics of the study population are summarized in Table 1.

One (0.3%) thromboembolic event occurred in a patient on uninterrupted 20 mg daily rivaroxaban. The patient had the persistent type of atypical atrial flutter after previous PVI and cavotricuspid isthmus ablation. His CHA2DS2-VASc score was equal to 2. He underwent the repeat procedure for PVI and linear ablation for right atrial upper loop re-entry flutter. The sinus rhythm was restored during linear ablation in the right atrium between the superior and inferior vena cava. The day after the procedure, the patient had an acute stroke with right temporal hemianopia. Magnetic resonance imaging revealed an acute infarction in the left parietal lobe.

A total of 6 (1.8%) patients had bleeding complications. Two (0.6%) major bleeding complications were pericardial effusion with cardiac tamponade, managed by percutaneous pericardial drainage with no sequelae. Four patients (1.2%) who had minor bleeding complications suffered a groin hematoma. No patient with groin hematoma required intervention.

The major limitation of our study is its retrospective nature. Thus, the approach of LA visualization was neither randomized nor systematic, but instead a reflection of the choices of the operator. Only rivaroxaban was included; therefore, it is difficult to generalize for the entire class of NOAC. The majority of patients in the ICE group had paroxysmal AF, in which the stroke risk is already minimized as compared to the non-paroxysmal population, that reflects the biased allocation of study participants. Also, ablation strategy and experience could influence thromboembolic events during the procedure regardless of TEE or ICE screening. Most importantly, this study was underpowered to provide statistically rigorous results, and was thus, exploratory in nature.