Early identification of oesophageal lesions after atrial fibrillation ablation to prevent atrio-oesophageal fistula

This commentary refers to ‘A worldwide survey on incidence, management, and prognosis of oesophageal fistula formation following atrial fibrillation catheter ablation: the POTTER-AF study’, by R. R. Tilz et al., https://doi.org/10.1093/eurheartj/ehad250 and the discussion piece ‘Be aware and hit hard and early: oesophageal lesions after atrial fibrillation ablation’, by R. R. Tilz and C.-H. Heeger, https://doi.org/10.1093/eurheartj/ehad692. We have read the POTTER-AF1 study with interest and congratulate the authors for this contribution. Although uncommon, atrial-oesophageal fistula (AOF) after ablation of atrial fibrillation (AF) is a challenging condition with high mortality rates. Remarkably, despite POTTER-AF being a worldwide survey, it did not bring any information from Latin America centres, where AF ablation has been increasingly performed. The AOF was first described by our group2 and by Pappone et al.3 in 2004 through two simultaneous publications reporting three cases. These reports occurred after the extra-ostial wide circumferential pulmonary veins (PVs) isolation became the default ablation approach, suggesting that this strategy can help avoiding PV stenosis, but increases the risk of oesophageal lesions. And, as the initial symptoms are not cardiac in nature, but neurological or infectious instead, high levels of suspicion are required for the diagnosis. The purpose of this letter is to alert that AOF incidence may be higher than reported by the POTTER-AF registry, as a post-procedural detection routine has not been implemented in most centres, and many cases could go undetected and unreported. The fact that no AOF had been detected until 2003 could be explained, in part, by a recognition failure and lack of active screening. It was noteworthy that most (63.6%) of POTTER-AF centres have never identified an AOF, whereas one particular centre has reported an unexpectedly elevated incidence of 0.4%. These data denote the highly variable diagnostic accuracy. A Brazilian survey published in 2016,4 including 8863 procedures from 2003 to 2015, reported 10 AOF cases (incidence of 0.113%) with a median time to diagnosis of 7 ± 3 days. Only patients with early diagnosis and prompt surgical treatment survived. Seven (70%) patients died. Only one patient (10%) was discharged without neurological sequelae. In response to these observations, preventive measures were adopted by our group, such as oesophageal temperature monitoring. We also implemented an active screening routine, consisting in oesophageal endoscopy after 24–48 h of the procedure and clinical evaluation of all patients around the 30th day of follow-up, when the fistulas usually become symptomatic (Figure 1).2,3 Characteristics and outcomes of POTTER AF and the active endoscopic surveillance protocol. EGD, oesophagogastroduodenoscopy. Between 2016 and 2022, our group performed 2700 AF ablation procedures under this protocol. Most patients had favourable clinical evolution except for two, who evolved with deep ulcer and developed initial oesophageal perforation detected by contrast-enhanced oesophageal computed tomography in the third week. One patient has been treated by using the endoclip technique and the other by using endoscopic negative pressure therapy to the oesophagus for 5 days.5 Both patients had completed healing. The general incidence of AOF was 0.07% during the last 6 years, and patients at highest risk were immediately identified and effectively treated without surgery and no sequelae. After 20 years of the first AOF, the management of this complication remains a significant challenge for electrophysiologists. We expect that the hereby presented data encourage the implementation of active screening measures, improving the safety of patients undergoing AF catheter ablation, especially at this time, when new techniques and technologies have been increasingly introduced. We thank Dr Alberto Ferraz for his assistance in writing the manuscript. All authors declare no disclosure of interest for this contribution.