This single-center, prospective observational study involved consecutive patients with drug-refractory, symptomatic atrial fibrillation who received their first RF catheter ablation (RFCA) at Asan Medical Center (Seoul, Republic of Korea) between September 2018 and February 2020. Patients with previous surgical- or catheter-based AF ablation and those who did not undergo endoscopic evaluations after the index RF ablation for AF were excluded (Figure 1). Written informed consent for data collection was obtained from all patients. This study was approved by the Institutional Review Board of Asan Medical Center (2020-0464).

AADs were discontinued for at least 5 half-lives before the procedure. Transesophageal echocardiography (TEE) was performed to exclude left atrium (LA) thrombus in 88 patients. Majority of patients underwent contrast-enhanced magnetic resonance (MR) imaging (n=114) or computed tomography (CT) (n=26, due to claustrophobia, presence of pacemaker/defibrillator, etc.) for evaluation of the anatomy of the LA and PVs and exclusion of the LA appendage thrombus before the procedure. Remifentanil with/without midazolam or dexmedetomidine was used for sedation/analgesia. General anesthesia was used in none of patients. Uninterrupted oral anticoagulation was administered in all patients. Of the 71 patients who did not undergo TEE, 27 patients have received oral anticoagulation for more than 1 year. Mean duration of anti-coagulation of the remaining 44 patients were 166 days and all patients underwent uninterrupted anticoagulation during minimum 38 days. Supplementary unfractionated heparin was administered to maintain the activated clotting time (ACT) at above 300 seconds. Venous access and catheterization were performed as previously described.11) Transseptal catheterization was performed under the guidance of hemodynamic monitoring, fluoroscopy, and intracardiac echocardiography (SOUNDSTAR catheter; Biosense Webster, Diamond Bar, CA, USA). Two 8 Fr SL1 sheaths (St. Jude Medical, St. Paul, MN, USA) were advanced into LA to handle a mapping catheter and an ablation catheter, respectively. Three-dimensional electroanatomic geometry of the LA was created using a multipolar mapping catheter (PentaRay; Biosense Webster). Voltage and activation mappings of LA were performed during constant pacing from the right atrium or distal coronary sinus (CS) at a pacing cycle length of 500–600 ms. All patients underwent PV isolation (PVI) using a 3.5 mm irrigated tip contact-force sensing RF ablation catheter (Thermocool smarttouch SF bidirectional navigation catheter; Biosense Webster). The point-by-point RF applications were delivered using the power-controlled mode, a temperature limited to 42°C, and normal saline irrigation (15 mL/min). For each point, RF energy was delivered with a power output of 50 W for 5 seconds initially. Due to high recurrence rate within blanking period, we prolonged the RF time to 10 seconds at the anterior wall and 6 seconds at the posterior wall. Automated lesion tagging (VisiTag™; Biosense Webster) was used to mark the location of each lesion. The VisiTag setting was as follows: catheter stability range of motion 3 mm during 3 seconds and minimum contact force >5 g over time of 50%. A tag size of 2 was used and further ablation was delivered to close the visual gap (Figure 2). The contact force in the posterior wall was maintained between 5–10 g. Entrance block into the PVs was demonstrated by the absence of PV potentials distal to the ablation line during constant pacing from the right atrium or distal CS. And the entrance block was also visualized by performing post-PVI voltage mapping of LA and PVs with inferior cut-off voltage of 0.05 mV. Additional anatomical ablations (LA roof/posterior wall, mitral isthmus, cavotricuspid isthmus, and/or superior vena cava) were performed at the discretion of the operator. The endpoint of the linear lesions was bidirectional block as confirmed by conventional pacing maneuvers.12)13) Isoproterenol infusion (5–10 μg/min) with burst atrial pacing was carried out after the ablations to induce non-PV triggers or atrial tachyarrhythmia.

Endoscopic examination of the upper gastrointestinal tract was performed under conscious sedation by an experienced gastroenterologist on the day after the AF ablation procedure. Esophageal lesion was classified based on the following endoscopic appearance: 1) erythema/erosion (without disruption of mucosa), 2) superficial ulcer involving only the mucosa, or 3) deep ulcer involving up to the muscularis externa.14) All patients received proton pump inhibitor (PPI) for 4 weeks after the ablation. In the case of esophageal injury, the management plans and determining whether and when to perform follow-up endoscopy were made in consultation with gastroenterologists. The primary outcome of this study is the occurrence of thermal injury of the esophagus detected by endoscopic evaluation after AF ablation using HPSD RF energy. Before discharge, the patients were informed on the possible symptoms associated with AEF (e.g., fever, chest pain, any neurologic symptom, hematemesis) and possible actions that should be taken (e.g., visiting a doctor's office or emergency room, taking chest CT, avoiding endoscopy).

During the index hospitalization after RFCA, 12-lead electrocardiogram (ECG) and telemetry monitoring were performed daily and the recurrence of atrial arrhythmia was screened. The follow-up visits to the outpatient clinics were scheduled at 1, 3, 6, and 12 months after the RFCA and then every 6 months thereafter. At each visit, 12-lead ECG was performed; 24-hours Holter monitoring was carried out at discharge and at 3, 6 and 12 months post-RFCA. Recurrence was defined by an episode of AF or atrial tachycardia (AT) lasting at least 30 seconds after the 3-month blanking period. If the arrhythmic recurrence was suspected, further Holter or event recording test was performed. The use of AADs after index ablation was determined at the physician's discretion.

All continuous variables were compared by Student's t-test and are reported as mean±standard deviation. Categorical variables were compared with the χ² test or Fisher's exact test as appropriate and are expressed as percentages. The cumulative probability and survival curves were constructed from Kaplan-Meier estimates and compared by the log-rank test. The p values <0.05 were considered statistically significant. All statistical analysis was performed using IBM SPSS Statistics for Windows, version 20.0 (IBM Corp., Armonk, NY, USA).

During the study period, 196 patients with atrial fibrillation underwent their first RFCA using 50 W of RF energy at our hospital. After excluding 37 patients who did not undergo upper GI endoscopy evaluation after the AF ablation, 159 patients were included in the study (Figure 1). The baseline clinical characteristics of the entire population are shown in Table 1. The mean age was 63±9 years, and 68.6% were male. Paroxysmal, persistent, and long-standing persistent AF were observed in 72 (45.3%), 44 (27.7%) and 43 (27.0%) patients, respectively. The mean LA anteroposterior diameter measured by echocardiography was 44.4±6.7 mm.

PVI was successfully achieved by anatomical PV antral ablation in all patients. Additional RF ablations beyond PVI were performed in 120 (75.5%) patients (Table 2). Linear ablations of the LA roof, left lateral mitral isthmus, cavotricuspid isthmus, and/or LA posterior wall isolation was performed in 62 (39.0%), 49 (30.8%), 61 (38.4%) and/or 40 patients (25.2%), respectively. Of the 49 patients who underwent ablation for the left lateral mitral isthmus, epicardial ablation inside the CS was required in 34 (69.4%) patients to achieve bidirectional block of the mitral isthmus. Three patients underwent slow pathway ablation due to the induction of typical atrioventricular nodal tachycardia during the index procedure.

Acute PV reconnection during a 20-minute waiting period occurred in 50/159 (31.4%) patients with the initially successful PVI. Acute reconnection of left, right PVs and bilateral PVs were observed in 33, 10, and 7 patients, respectively. Eleven patients (11/50, 22.0%) showed PV reconnections at multiple PVs. Patients underwent PVI with 5 seconds ablation protocol showed significantly higher rate of acute reconnection than the patients with 6–10 seconds ablation protocol (14/26 [53.8%] vs 36/133 [27.1%], p=0.007). Isoproterenol infusion test was done in 109 patients of all patients and PV reconnection was observed in 11 patients (10.1%). Higher rate of reconnection after isoproterenol infusion was also observed in patients with 5 seconds ablation protocol (7/26 [28.0%] vs. 4/84 [4.8%], p=0.003). All reconnected PVs were re-isolated finally.

Out of the 159 patients, 4 (2.5%) had esophageal thermal injury and their characteristics are summarized in Table 3. Two (1.3%) patients had superficial esophageal ulcer and another 2 (1.3%) patients had erythematous erosion of the esophagus. None of these patients had gastrointestinal symptoms including indigestion, dysphagia, and chest pain. One of the patients with esophageal ulcer had a longitudinal ulcer of 1 cm in length at 27 cm from the upper incisor (Figure 3). Chest CT did not show mural thickening, defect along the esophagus, or evidence of pneumomediastinum. Follow-up endoscopy at 1 month after the initial endoscopy showed healed state of the esophageal ulcer. Patients with esophageal injury were older (75±6 vs. 63±9, p=0.015) and had higher CHA₂DS₂-VASc scores (3.8±2.1 vs. 1.9±1.3, p=0.007) than those without thermal lesions of the esophagus. Other baseline or procedural characteristics were not significantly different according to the presence of thermal injury. No significant difference of RF delivery time was observed between patients with and without esophageal injury (esophageal injury [+]: 36.9±13.4 minutes vs. esophageal injury [−]: 34.2±14.4 minutes, p=0.712). And other procedural characteristics, including procedural time, ablation lesion set and applied RF protocol (5 seconds vs. 6–10 seconds), were not different between groups. Among the 4 patients with esophageal injury, 1 patient (patient #1) underwent PV isolation using initial RF protocol (5 seconds) and the other 3 patients (patient #2–4) underwent PVI using prolonged RF protocol (6–10 seconds).

Food stasis, a suggestive finding of gastroparesis, was observed in 25 (15.7%) patients; among them, 3 patients had clinical symptoms of indigestion, one of whom required conservative treatment with a prokinetic agent. All 3 patients with indigestion showed symptom improvement within a month after the AF ablation procedure. There was no statistically significant predictor for the occurrence of food stasis in this study population.

During the periprocedural period, there were no major procedure-related complications such as cardiac tamponade, phrenic nerve palsy, stroke, or death. All patients completed the outpatient clinic visits as scheduled with a median follow-up duration of 6.2 months (interquartile range [IQR], 3.4–8.7). No patients reported symptoms suggestive of AEF beyond 3 months after the index procedure. None of the patients who did not undergo endoscopy had symptoms or signs suggestive of AEF.

Among the total patient cohort, 133 patients (paroxysmal AF, n=57 [42.9%]; persistent AF, n=39 [29.3%]; long-standing persistent AF, n=37 [27.8%]) were followed-up over the blanking period of 3 months after the index procedure. During a median follow-up of 193 days (IQR, 117–283), 35 (26.3%) patients showed recurrence of AF (n=23 [65.7%]) or AT (n=12 [34.3%]) after a single RF ablation.