There Is Glory in Prevention!—Atrio-esophageal Fistula and Atrial Fibrillation Ablation

Thomas Deneke; Karin Nentwich; Philipp Halbfaß; Philipp Sommer


Europace. 2021;23(4):491-493. 

The incidence of major complications of atrial fibrillation (AF) ablation has decreased over the years and is inversely related to operator experience and training.[1–5] Atrio-oesophageal fistula (AEF) is an exception to this rule, as its incidence has remained quite stable over time (although at a low level) and the occurrence is not related to experience of the ablating centre or physician.[1,6–9] So far, no predictors of this rare but devastating complication have been identified. Strategies aiming to reduce collateral thermal damage during ablation of the left atrial posterior wall include oesophageal temperature monitoring, changes in amount or technique of energy delivery but have not been shown to resolve the risk of AEF.

Transmural thermal injury of the oesophageal wall as a direct consequence of ablation appears to be the offset of a cascade of events leading to progression towards oesophageal perforation and erosion of the left atrial wall. This creates a direct communication between oesophageal lumen and the left atrial cavity that may allow for gaseous embolization as the most severe complication. AEF is rare and appears to occur in between 0.03% and 0.2% of AF ablation cases but information about the accurate incidence is lacking.[9]

As a consequence, any information and data added to our current knowledge is helpful and may improve strategies for identification, treatment, and risk stratification of patients.

In this issue of Europace, Gandjbakhch et al.[6] published a survey among the French AF ablating centres summarizing AEF cases between January 2006 and June 2019. Responses to the survey came from 80% of all centres performing in total 93% of all registered AF ablations in France (encompassing nearly 130 000 AF ablations). A total of 29 cases of AEF and four cases of oesophago-pericardial fistula (the overall reported incidence of fistulous oesophageal events 0.026% and 0.022% reported incidence of AEF) are documented. Analysable data for 26 AEF and for four oesophago-pericardial fistula cases (for ease of discussion summarized as 33 AEF cases) is available and accounts for the largest number of presentations of de novo AEF. A total of 24 centres (29% of all participating centres) reported AEF cases, with all of them occurring in centres with at least 100 AF ablations per year. It is remarkable that no AEF was reported in smaller case-volume centres but only one-third of ablation centres doing less than 50 cases answered the survey. It is clear that AEF is most probably the one complication of AF ablation that is not influenced by operator experience and therefore it may happen in any AF ablation centre.

Keeping in mind the immanent drawbacks of surveys on severe complications of medical procedures like underreporting and underdetection, this survey in a meticulous effort that allows new insights into the reported incidence, presentation, diagnosis, and outcome of AEF.

Incidence and Timing of Atrio-oesophageal Fistula

Gandjbakhch et al. indicate a reported incidence of AEF (including oesophago-pericardial fistula) of 0.026% well in line with published reported incidences in surveys of 0.04% or 0.03%.[3,10] Reported incidences in surveys are to be considered with caution as they may underestimate the true incidence because severe complications may be underreported, and rare complications may be underdetected if no rigid follow-up exists. To illustrate this, a study in 1000 AF patients ablated in a high-volume centre with complete and consequent follow-up of every patient published a 'true' detected incidence of AEF of 0.2%.[11] Whether this difference by factor 10 is realistic for the current survey is unclear but reported incidences in surveys are lower than in the reality.[8] Alarming in the French survey is that there are no reported cases by smaller centres where AEF diagnosis may have been missed.

It is interesting to learn from the survey by Gandjbakhch et al.[6] that annual incidences of AEF are stable throughout the surveyed period. So far, it appears that no protective measures relevantly decreased AEF. It will be interesting to see the effects high-power short-duration ablation strategies or other ablation technologies like pulsed field ablation may have. It is clear that the discrepant incidence reported for cryo-balloon ablation (no AEF) in the current survey allows no conclusion as to the safety. AEF has been reported with almost any ablation technology currently available and the incidence for cryo-based pulmonary vein isolation may be comparable to radiofrequency ablation strategies.[12]

As indicated in recently published literature reports, AEF occurred in between 0 and 60 days after ablation.[7,13,14] The most common timeframe is 2–4 weeks post-ablation, and mean delay from ablation to AEF occurrence was 21 days in the current survey. Whereas very rare on the day of the procedure (only two reported cases with probably a different underlying pathomechanism), the occurrence within the first 2 months after ablation bears an important risk as the patient has left the ablating centre and may present to remote facilities not aware of the presenting symptoms and causal relationship to the ablation procedure. It is the responsibility of each ablating centre to assure adequate information to patients and medical facilities that may deal with these patients if AEF occurs.

Symptoms, Presentation, and Diagnostic in Patients With Atrio-oesophageal Fistula

In the current French survey[6] and also in prior reports of AEF cases,[7,8,13,14] the most common presenting symptoms of patients diagnosed with AEF are fever (63–73%), neurological symptoms (23–72%), and chest pain, odynophagia, or pain during swallowing (41–47%). This wide range of symptoms implements that patients with AEF may present to a wide variety of medical specialties and cardiac symptoms appear to be quite rare (most common AF recurrences). Different symptoms may relate to different stages of oesophageal perforation, for example, neurological symptoms usually result from (gaseous) emboli related to a fistulous connection into the left atrium and gastro-intestinal symptoms may relate to initial oesophageal perforation. Thereby it is clear that neurological symptoms related to AEF predict worse outcome.[7] Suspicion for AEF is warranted in any patient ablated within 2 months of admission presenting with symptoms such as chest discomfort, pain during swallowing—most probably as earliest symptoms—followed by fever and later by intermittent or sustained neurological deficits. Informing patients and anyone seeing this patient cohort in follow-up facilities is crucial.

Emergency chest computed tomography (CT) must be the first line diagnostic to identify or rule out if AEF is suspected.[9,15] An initial sensitivity of 80% is indicated in the current survey, and 21% of initial CTs were reported to be normal.[6] The reason for a 'non-detected AEF' at initial CT may be multifactorial and can be a true negative (no signs of AEF visible despite true AEF) or a false negative (signs of AEF but not detected by the one scrutinizing the CT) test. In two cases only repeat chest CT was found to show AEF, and again the reason may be a false negative initial CT or the occurrence of AEF during the waiting period for the second scan. Non-contrast CT may identify air in the mediastinum or in the arterial blood system as a direct sign of oesophageal perforation or infiltration into the left atrium but direct visualization using per-os contrast (water soluble) or intra-venous contrast should warrant higher sensitivity to AEF or perforating oesophageal complications. It is important to recognize, that normal chest CT does not exclude oesophageal perforation and i.v. and p.o. contrast and repeat CT should be performed.

Other diagnostic measures like trans-thoracic echo (negative in 95%) and cerebral imaging (negative in 95%) are not helpful, whereas endoscopy and transoesophageal echo are contraindicated if AEF is not excluded. The recommended workflow if AEF is suspected is (i) chest CT with per-os and intra-venous contrast and (ii) endoscopy (using CO2) only if AEF is ruled out.

Treatment, Mortality, and Prognosis of AEF

AEF is the most lethal complication of AF ablation and mortality has been published to rank in between 83% and 100% without surgical repair and 33% in those undergoing surgery.[7,8,10,13,14] The current survey[6] adds more information to this topic with a reported mortality of 31% in patients undergoing corrective surgery vs. 93% mortality in those without surgery. In total, 12 patients were not treated for AEF (mostly because of the critical clinical situation of the patients) and 11 of them died (with only one surviving with severe neurological impairment). Two patients received isolated oesophageal stenting for AEF and both died. Twelve patients underwent corrective surgery (with or without oesophageal stenting) and eight survived (67%). Overall mortality was documented to be 65% in AEF vs. 0% in the four cases of oesophago-pericardial fistulas (all underwent surgical repair).

It is notable, that surgical correction appears to be the only appropriate treatment option in manifest AEF even though this procedure must be considered a highly complex procedure with a fatality of around one-third of patients. The surgical procedure usually consists of a combined left atrial and oesophageal repair via thoracotomy and requires high surgical skills of cardiothoracic and abdominal surgeons. As this is an emergency procedure (and only half of the patients may be in a clinical situation to undergo such a surgery), organization is complex and adequate management maybe available only in very few centres worldwide.

In contrast, oesophago-pericardial fistula appears to have a rather good prognosis if undergoing surgical repair. In these cases, a less invasive approach with oesophageal stenting and long-term antibiotic therapy may be worth exploring even though so far data are minimal.

In summary, all data from surveys and summaries of published reports indicate the high mortality of AEF with conservative non-surgical management either because of the desperate critical clinical situation of the patient or because less invasive measures were preferred, or surgery was not available. From all data available a surgical approach should be performed when true AEF is detected. It can only be stressed that early diagnosis is key for the prognosis of these patients and a high amount of suspicion and awareness is needed to identify and adequately diagnose patients with warning signs.

We have learned from reports like the French survey by Gandjbakhch et al.[6] that AEF is the most lethal complication of AF ablation and it may occur in any patient cohort including redo-cases (7% of the current reported cases), radiofrequency with or without contact force measurement or cryo-balloon cases, independent of the use of oesophageal temperature probes (used in 20% of current reported cases), and independent of using general anaesthesia (27% of current cases without general anaesthesia)—but direct comparison cannot be obtained from surveys. AEF may occur more often as suggested from reported incidences in surveys and it is key to educate patients for early warning signs like fever, pain during swallowing and thoracic pain during the first 2 months after an AF ablation procedure. In these cases, emergent CT imaging using intravenous and per-os contrast to rule out or diagnose AEF or oesophageal perforation is recommended (no manipulation on the oesophagus like transoesophageal echo or endoscopy without AEF being ruled out). It appears critical to identify oesophageal perforation before fistulizing into the left atrium as prognosis of AEF still remains poor and surgical repair is the only treatment option with a fatality of around 30%.

Early identification of patients with oesophageal complications is key to organize appropriate diagnosis and treatment. Organization of effective surgical treatment is limited to very few centres worldwide. In an effort to distribute knowledge from experts to centres coming along a detected or suspected AEF case the German Working Group on Electrophysiology (AG EP) has established the first world-wide web-based hotline. This hotline is open to anyone needing help in identifying, diagnosing or treating these patients and immediate reply is guaranteed. Go to 'Kollegialer Austausch zum Thema Ösophagus-Verletzungen nach Vorhofflimmer-Ablation'.

Gandjbakhch et al.[6] are to be congratulated for their meticulous work allowing more insights into AEF management and treatment in a real-world scenario. We need to work on identifying risk factors for the occurrence of AEF, and some centres have used post-ablation endoscopy to scrutinize for oesophageal thermal injury (EDEL) as a standard procedure after ablations at the left atrial posterior wall. The finding of EDEL may be used as a surrogate parameter for the potential risk of perforating oesophageal complications (as severe oesophageal thermal ulcers mark the risk of perforation) and may allow comparison of transformed AEF risk for different ablation strategies and techniques.[15,16] There is glory in prevention and reducing incidence of EDEL that will most probably relate to lower incidences of AEF. Testing new approaches to AF ablation therefore should include assessing oesophageal thermal injury.