EUS for the evaluation of esophageal injury after catheter ablation for atrial fibrillation.

Background and aims: Atrial fibrillation (AF) ablation is an increasingly used rhythm control strategy that can damage adjacent structures in the mediastinum including the esophagus. Atrioesophageal fistulas and esophagopericardial fistulas are life-threatening adverse events that are believed to progress from early esophageal mucosal injury (EI). EUS has been proposed as a superior method to EGD to survey EI and damage to deeper structures. We evaluated the safety of EUS in categorizing postablation EI and quantified EUS-detected lesions and their correlation with injury severity and clinical course.

Methods: We retrospectively reviewed 234 consecutive patients between 2006 and 2020 who underwent AF ablation followed by EUS for the purpose of EI screening. The Kansas City classification was used to classify EI (type 1, type 2a/b, or type 3a/b).

Results: EUS identified pleural effusions in 31.6% of patients, mediastinal adventitia changes in 22.2%, mediastinal lymphadenopathy in 14.1%, pulmonary vein changes in 10.6%, and esophageal wall changes in 7.7%. EGD revealed 175 patients (75%) without and 59 (25%) with EI. Patients with type 2a/b EI and no EI were compared with multivariate logistic regression, and the presence of esophageal wall abnormality on EUS (odds ratio [OR], 72.85; 95% confidence interval [CI], 13.9-380.7), female sex (OR, 3.97; 95% CI 1.3-12.3), and number of energy deliveries (OR, 1.01; 95% CI, 1.003-1.03) were associated with EI type 2a or 2b. Preablation use of proton pump inhibitors was not associated with a decreased risk of EI.

Conclusions: EUS safely assesses mediastinal damage after ablation for AF and may excel over EGD in evaluating mucosal lesions of uncertain significance, with a reduced risk of gas embolization in the setting of a full-thickness injury (enterovascular fistula). We propose an EUS-first guided approach to post-AF ablation examination, followed by EGD if it is safe to do so.