CT and Surgery Best Diagnose and Manage Esophageal Fistula

Daniel M. Keller, PhD

April 18, 2023

Esophageal fistula (EF) is a rare but serious complication of catheter ablation for atrial fibrillation (AF), with high mortality. Partly because of its rarity, numbers on its incidence, management, and outcomes have been hard to come by.

The very large Prognosis Following Oesophageal Fistula Formation in Patients Undergoing Catheter Ablation for AF (POTTER AF) study has provided some of those data, documenting a 65.8% mortality rate for cases, as well as significant delays in diagnosis. The method of management is critical for survival, results suggest.

Presenting the study at the European Heart Rhythm Association (EHRA) 2023 congress in Barcelona, Prof Roland Tilz, MD, of the University Heart Center in Lübeck, Germany, said, "I promise you that you will have a very clear plan how to diagnose and treat patients with esophageal fistula once you have seen this presentation."

This multinational (214 centers in 35 countries), retrospective study amassed data on 553,729 anonymized procedures. Investigators identified 138 patients with atrial-esophageal or esophageal-pericardial fistula or esophageal perforation; the incidence was 0.025%, or approximately 25 per 100,000 patients.

There were no exclusion criteria for study entry. Data were available for 118 cases and were collected online on an individual patient level using a standardized, uniform questionnaire.

Delays in Diagnosis

The time to first symptoms was a median of 18 days (95% CI, 7.75 – 25 days; range, 0 – 60 days). Diagnosis took a median of 21 days (95% CI, 15 – 29.5 days; range, 2 – 63 days) after the ablation procedure. The most common initial symptom was fever (59.3%), followed by chest pain/odynophagia (54.2%) and neurologic signs (44.1%).

The most common diagnostic modalities were chest CT (80.2%), cerebral CT/MRI (34.5%), echocardiography (25%), and endoscopy (20.7%). The most common treatment method was surgery (47.4%).

Death occurred in 65.8% of cases, major/minor sequelae in 18.8%, and no sequelae in 15.4%. Complications consisted of severe sepsis or septic shock (57.9%), coma (46.7%), stroke/cerebral hemorrhage (23.4%), cardiac arrest (18.7%), gastrointestinal bleeding (16.8%), cardiac tamponade (11.2%), and others.

Among the treatment strategies, surgery was associated with the lowest mortality rate, reducing the risk of death by about 68% (mortality odds ratio, 0.321), but even with surgery, 51.9% of patients died.

Table. Esophageal Fistula Treatment Outcomes

Treatment Mortality odds ratio (95% CI) P value Death (%) Major sequelae (%) Minor sequelae (%) No sequelae (%)
Surgery (47.4%) 0.321 (0.144 – 0.714) .005 51.9 16.7 16.7 14.8
Endoscopic treatment (19.8%) 0.629 (0.248 – 1.598) .330 56.5 4.3 8.7 30.4
Conservative management (32.8%) 7.463 (2.414 – 23.072) < .001 89.5 2.6 7.9


Tilz noted that a limitation of the study is that it was retrospective, incorporating information on patients with existing periprocedural EF. Therefore, risk factors for EF could not be discerned. Furthermore, if some occurrences of EF were not reported, the true incidence could not be known and may be even higher.

Also, there were no data on procedural variables, such as esophageal temperature, and there were no subgroup analyses based on type of catheter used, lesion parameters, lesion size, and ablation index.

He said he conducted a small Twitter survey of colleagues to get some more real-world data, asking, "What is your first-line therapy for a patient with confirmed atrial-esophageal fistula presenting with fever but without a neurologic deficit, so at an early stage?"

While the majority recommended surgery, "I was really surprised to hear that almost 30% said they recommend a conservative treatment," he said. "If you treat the patient conservatively as recommended...in the Twitter survey, you have a 90% mortality rate."

The diagnostic modality of choice was chest CT, which was diagnostic more than 80% of the time, he said. "So, with a chest CT and or cerebral CT or MRI, you could almost identify it in the majority or almost all patients."

Half a Million Cases

As the appointed discussant, Stylianos Tzeis, MD, PhD, head of the Cardiology Clinic and Electrophysiology and Pacing Department at Mitera Hospital in Athens, commented on the POTTER AF study in the context of the overall field, saying that the more than half a million cases enrolled "are more than the number of the population in all the previous surveys."

He highlighted the fact that the incidence of EF, "despite our technological developments, remains stable without a significant decrease in trend, is something about 0.02 or 0.03%, almost three cases every 10,000 ablations," with extremely high mortality.

Tzeis attributed the complication to use of RF energy, with "significantly lower" risk with cryo energy as compared to RF ablation. He noted that the risk of these complications is not associated with the volume of procedures of the center or of operator experience.

Because the study is a survey, he raised some of the usual caveats and limitations, two being the retrospective design and voluntary participation in the survey, so the sample may not be representative of physicians performing ablations. Even if there was underreporting and thus an underestimate of the incidence of EF, "it is the best data that we have in our hands," Tzeis said.

Tilz has received research grants from Abbott Medical, Medtronic, Lifetech, Boston Scientific, and Biosense Webster; travel grants from Biosense Webster, Abbott Medical, Medtronic, and Boston Scientific; speaker’s bureau honoraria from Biosense Webster, Medtronic, Lifetech, Boston Scientific, Abbott Medical, Biotronik, Pfizer, Bristol-Myers Squibb, Bayer, and Sanofi-Aventis; and publication fees from Boston Scientific. He has served as a consultant to Biosense Webster, Biotronik, Boston Scientific, and Abbott Medical. Tzeis reported no relevant financial relationships.

European Heart Rhythm Association (EHRA) 2023: Presented April 16, 2023.

For more news, follow Medscape on Facebook, Twitter, Instagram, and YouTube.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.