Pulmonary vein antrum isolation using radiofrequency ablation has become an effective therapy in symptomatic patients with atrial fibrillation. Additional linear ablation in the left atrium is performed in persistent atrial fibrillation and atrial flutter. Non-lethal complications such as cardiac perforation, stroke and pulmonary vein stenosis have been reported to occur with rates of 1.3%, 0.2% and 1.3% respectively [1, 2]. A recently reported complication describes esophageal injury leading to left atrial esophageal fistula. Despite its low incidence (0.03-0.1%) this usually lethal complication is of tremendous clinical importance [3–7]. Other serious esophageal injuries include vagus nerve damage with acute pyloric spasm and gastro-paresis [8]. Different strategies are proposed to avoid esophageal injury. Power and temperature settings are limited when ablating at the posterior wall of the left atrium [9]. Visualization of the esophagus by barium swallow or tagging the course of the esophagus by electro-anatomical mapping has been proposed in order to modify ablation lines in areas of close proximity to the esophagus [10]. A pilot study using an irrigated intraesophageal cooling balloon resulted in a significant reduction of intraluminal esophageal temperature [11, 12].

Post ablation esophageal wall changes (erosion or ulceration) are reported to occur in up to 47% of patients [13]. Real time temperature monitoring can detect rapid esophageal heating during radiofrequency ablation [14].

The aim of this prospective study was to investigate the incidence of thermal esophageal lesions when limiting the intraluminal esophageal temperature in radiofrequency ablation of left atrial tachyarrhythmias.

Preablation endoscopy was performed in 23 of 184 patients. Eight patients showed mild reflux disease, three patients hemorrhagic gastritis, and four patients pyloric ulcers. All patients were treated with proton pump inhibitors. Ablation was postponed in all patients with hemorrhagic gastritis and pyloric ulcers until healing was confirmed. Eight patients had no pathologies on pre-ablation endoscopy.

The temperature threshold of 40°C was reached in 157 of 184 patients (85%). Despite stopping energy delivery further increase in esophageal temperature was observed. This temperature overshoot was registered in 146 of 157 patients (93%) in whom energy delivery was stopped due to esophageal temperature rise. The mean maximal recorded temperature was 40.8°C in all patients. The maximal recorded temperature was 45.3°C in one patient. Using a multiple linear regression analysis including box lesion, roof-line, mitral-isthmus-line, course of the esophagus, type of arrhythmia and index procedure, box lesion (Beta = 0.29, p = 0.004) remains a significant predictor variable of maximal esophageal temperature. Figure 2 illustrates the mean maximal temperature in patients with and without box lesion. Although the course of the esophagus turned out not being an independent predictor of max temperature, post hoc Bonferroni analysis showed a significantly higher max temperature in the LPV ostia group and the RPV ostia group as compared to the mid posterior left atrium group (table 2). No patient reported about thoracic discomfort such as dysphagia or odynophagia following ablation. Endoscopy showed thermal ulcerations at the anterior wall in the mid esophagus in three patients. This was the index pulmonary vein isolation procedure to treat paroxysmal atrial fibrillation in all three patients. Table 3 summarizes the procedural findings in patients with esophageal ulcerations.

An endoscopic picture of the esophageal ulceration in patient one is shown in figure 3. Lesion regression was documented after four to six days and complete lesion healing confirmed after two weeks in all three patients. No patient developed clinical signs of atrio-esophageal fistula on follow-up. Patient, disease and procedural characteristics had no influence on the incidence of esophageal ulcer formation (table 4).

Atrio-esophageal fistula is a rare but devastating complication of radiofrequency ablation in the left atrium. Thermal esophageal lesions are believed to be precursors of fistula formation. Little is known about the incidence of thermal esophageal lesions. In most studies less than 100 patients are included and different ablation protocols are used [13, 15, 16].

Recently Halm reported of 185 patients undergoing left atrial ablation using an open irrigated tip ablation catheter with 30 Watts at the posterior wall. Esophageal temperature monitoring without temperature limitation was utilized. Post-procedural endoscopy of the esophagus revealed ulcer like thermal lesions in 14.6% of patients [17]. Our series of 184 consecutive patients showed an incidence of 1.6% thermal esophageal injury. This is the lowest rate of esophageal lesion formation described so far. No progression to atrio-esophageal fistula occurred. An intraesophageal temperature limit of 40°C appears to be an effective means of reducing the incidence of thermal esophageal damage. Halm did not use a temperature limit and consequently the measured mean maximal temperature was higher than in our series (42.6 ± 1.7°C vs 41.6 ± 1.5°C in patients with ulcers and 41.4 ± 1.7 vs 40.8 ± 1.8°C in patients without ulcers). In accordance to their findings we did not detect ulcers on endoscopy when the maximal temperature stayed below 41°C.

We decided to use the esotherm temperature probe, because it consists of three independent thermocouples with rapid temperature detection capacities. Singh used a different temperature probe with a single thermocouple in 67 patients. When limiting the temperature to 38.5°C they observed 6% ulcerations as compared to 36% without temperature monitoring [15]. It is of great importance to ensure a minimal distance between the temperature probe and the ablation catheter. Using a probe with three thermocouples significantly reduces the chance of missing a temperature rise. In addition the NavX-Ensite system continuously displays the position of the thermocouples in relation to the ablation catheter without the need of fluoroscopy. Ablation directly adjacent to the presumed course of the esophagus was avoided if possible in our study. In spite of applying a lower esophageal temperature limit the higher incidence of ulcerations in the series by Singh could be explained by these differences. However, Singh also used a different ablation protocol with 35 Watts at the posterior wall of the left atrium. This also increases the likelihood of esophageal lesion formation.

Martinek found a total of 2.9% esophageal lesions in 175 patients. Using a maximal power of 25 Watt without esophageal temperature monitoring, additional ablation lines especially an inferior ablation line parallel to the coronary sinus were variables favouring the incidence of esophageal lesions [18]. We could not find a significant correlation of ulcer formation with additional lines or other variables.

The tissue separating the posterior left atrium from the anterior esophageal wall is less than 5 millimetres at its thinnest location [19]. A rapid heat transfer is described [14]. Even though we applied a temperature limit of 40°C a temperature overshoot was observed frequently. The mean maximal esophageal temperature was higher in patients with ulcer formation than in patients without ulcers (41.6°C ± 1.5°C versus 40.8°C ± 1.8°C). However, this difference was not significant, possibly due to the low incidence of lesions. Using a multiple linear regression analysis including roof line, box lesion, mitral isthmus line, type of arrhythmia, redo procedure and course of the esophagus, creating a box-lesion predicted a significant higher esophageal temperature. The box lesion implies an inferior-posterior line connecting the anterior encircling of the right and left pulmonary veins. In this area the esophagus usually shows the closest distance to the left atrial posterior wall.

As stated before the course of the esophagus along the left atrial posterior wall is not an independent predictor of max temperature. However, post hoc Bonferroni analysis revealed a significant higher temperature of 41.2°C in patients with close relationship to either the right or left pulmonary venous ostia as compared to 40.1°C when the esophagus took a course along the mid posterior left atrium.

Although the temperature limit could not be adhered to in most cases due to rapid heat transfer it might be speculated that without this limit even more energy had been applied resulting in more or greater esophageal lesions and eventually an increasing risk of atrio-esophageal fistula formation.

Limitation of esophageal temperature in radiofrequency ablation of left atrial tachyarrhythmias is associated with the lowest incidence of esophageal lesion formation described until today. Temperature overshoot after cessation of energy delivery is frequent. A posterior box lesion significantly increases the maximal temperatures in the esophagus. Follow-up visits including 7 day holter recordings showed no evidence of reduced efficacy.

Adding temperature limitation in the esophagus to a standard protocol when ablating at the posterior wall of the left atrium may contribute to increase the safety profile of the procedure.