Cross-Sectional Imaging Obtained Immediately Following Radiofrequency Atrial Fibrillation Ablation Does Not Predict Endoscopic Evidence of Esophageal Injury

Background Radiofrequency atrial fibrillation ablation (AFA) is commonly performed in patients with atrial fibrillation. It is imperative to develop a strategy for the early detection of esophageal lesions secondary to AFA. The current protocol is to obtain cross-sectional imaging before and immediately after the procedure. If patients have evidence of esophageal inflammation, they undergo esophagogastroduodenoscopy (EGD). We hypothesized that esophageal abnormalities seen on imaging immediately post-ablation are a poor predictor of the damage seen during EGD. Methods Patients referred for EGD following AFA from 1/2009 to 11/2010 were included. Two endoscopists reviewed and scored the EGD images. Two radiologists reviewed the post-AFA imaging studies. For computed tomography (CT) scans, esophageal inflammation was scored from 0 to 2. For T2 and delayed magnetic resonance imaging (MRI) pictures, esophageal enhancement was scored from 0 to 2, with the circumference involved as 0, < 50%, or > 50%, and the length of esophageal enhancement in mm. Results In total, 76 patients were included; 22 patients had only endoscopic images and 54 had both endoscopic and radiologic images for review. Of the post-AFA imaging studies, 16 were CTs and 60 were MRIs. The kappa score for the inter-rater agreement of esophageal inflammation on EGD was 0.4584 (moderate). For MRIs, the kappa scores for T2 images were 0.1980 and 0.2857 for edema and circumference, respectively. For delayed images, the kappa scores were 0.2687 and 0.3101 for edema and circumference, respectively. The kappa scores were negative between EGD score by T2 edema (-0.2104) and circumference (-0.2212), and between EGD score and delayed edema (-0.0588) and circumference (-0.0446). When measures were treated as dichotomous, the overall agreement between CT measures and EGD scores was kappa = 0, for T2 measures and EGD kappa = -0.2963, 95% confidence interval (CI) (-0.5643, -0.0282), and between delayed measures kappa = -0.0244, 95% CI (-0.1420, -0.0932). Conclusions There was no agreement between immediate imaging and the endoscopic findings of esophageal inflammation after AFA. A longer period of time between AFA and obtaining an imaging study may be useful in detecting patients with significant esophageal injury who should undergo EGD to assess for complications of AFA. Further studies are needed in order to determine the best modalities and optimal timing to detect post-AFA esophageal damage in an attempt to prevent the formation of atrial-esophageal fistulas.