Incidence of acute thermal esophageal injury after atrial fibrillation ablation guided by prespecified ablation index

Ablation index (AI), a novel parameter defining energy application at single ablation lesions, calculated by integration of ablation time, energy, catheter stability, and contact force, has been documented to be associated with effective lesions and higher ablation efficacy. Using a prespecified target AI in addition to acute lesion efficacy may affect local collateral damage like esophageal thermal injury when used for guiding radiofrequency (RF) ablation at the posterior left atrial (LA) wall.

esophageal complications like atrioesophageal fistula. 2,3 The incidence of perforating esophageal complications reported may be lower than the true incidence because of underdetection and underreporting. According to a worldwide survey and a registry study, the incidence of atrioesophageal fistula after AF ablation was below 0.2%, whereas a recently published study with rigid postablation endoscopy found an incidence of atrioesophageal fistula and esophageal perforation of up to 0.6% in patients. 2,4,5 Limiting ablation time, contact force, and ablation energy at the LA posterior wall has been recommended in all AF ablations to decrease the risk of esophageal complications.
Whereas AI-guided ablation was designed to optimize individual ablation lesion quality (efficacy) to target prespecified values may be associated with thermal collateral damage (safety).
The purpose of this single-center study was to evaluate the incidence of esophageal thermal injury related to AI-guided ablations using postablation endoscopy in patients undergoing their first AF ablation.

| METHODS
All patients gave informed consent before entering the study. The study was approved by the local institutional review board.

| Study population
From July 2017 to November 2018, consecutive patients undergoing first-time AF ablation including pulmonary vein isolation using singletip RF force-sensing ablation catheters in conjunction with the AI module at our institution were included. All patients underwent esophageal endoscopy (EE) 1 to 3 days after ablation to document esophageal thermal damage (EDEL).

| Atrial fibrillation ablation procedure
LA thrombi were ruled out by transesophageal echocardiography performed on the day before the ablation procedure. All procedures were performed under continued oral anticoagulation. The procedures were performed using a three-dimensional electroanatomic mapping system (CARTO 3; Biosense Webster, Diamond Bar, CA) as previously published. In brief, exit and entrance block of each pulmonary vein (PV) as the procedural endpoint was confirmed by a lasso catheter (Biosense Webster). Patients were ablated using single-electrode RF ablation catheters with open catheter-tip irrigation at an irrigation flow rate of 30 mL/min (ThermoCool SmartTouch; Biosense Webster). Ablation power was limited to 25 W at the posterior wall and to 35 W at other LA ablation sites. Target contact force was between 5 and 15 g at the posterior wall and between 15 and 30 g at other ablation sites. Target ablation time was automatically defined by the AI algorithm incorporating ablation power, time, and contact force. A color change of the ablation lesion tag to dark red indicated that the targeted AI at the current ablation site was reached. The AI target value at posterior wall ablation sites was defined for each of the participating operators individually (AI at posterior wall sites between 300 and 350). Ablation procedures were performed by four experienced operators (each having performed more than 500 AF ablation procedures). According to our current standard approach, no luminal esophageal temperature probes were used.

| Follow-up
At 3 months follow-up any late-occurring esophageal complication was documented. Procedural efficacy at 3 months was evaluated by two Holter ECG recordings and clinical evaluation in relation to AF episodes.

| Statistical analysis
The data are expressed as mean ± SD for continuous variables or as numbers and percentages for categorical variables. For analysis of continuous variables the t test and for analysis of categorical variables contingency tables were used.
For analysis of possible risk factors for the occurrence of EDEL binary logistic regression was used. For logistic regression results were given as odds ratios and 95% confidence intervals and P values. P values less than .05 were considered statistically significant.
Statistical analyses were performed using SPSS. HALBFASS ET AL.
In all 211 patients, all PV were effectively isolated (100% acute success). Procedure duration, mean duration of RF at the posterior wall in the overall group, and mean AI at the posterior wall in overall group are displayed in Table 2. In univariate statistical evaluation only systolic EF (51 ± 10 in patients with EDEL vs 59 ± 9 in patients without EDEL, P < .01) was inversely associated with the incidence of EDEL (Table 1).

| Endoscopy and procedural results
Logistic regression analysis of parameters possibly influencing the incidence of EDEL (category 1 and category 2) did not identify any of the included variables to significantly influence the occurrence of EDEL (Table 2).

| 2259
using endoscopy already during the ablation procedure and most EDEL heal off within 7 days after the initial thermal injury. Timing of postablation endoscopy may be crucial for accurate documentation of esophageal thermal injury and for practical reasons endoscopy is usually done within the first 3 days after the procedure (days EE performed after ablation procedure: mean 2.4 ± 4.5).
In recent reports on patients ablated using our protocol thermal esophageal lesions were documented in 18% to 19% of all patients when using postablation endoscopy as a standard procedure and category 2 lesions comprised one-third of these lesions. 5,6 In a singlecenter registry analyzing 832 patients treated with different RF ablation catheters at our institution using prespecified ablation parameters when ablating at the posterior LA wall (maximum 25 W) resulted in an incidence of 18% EDEL. 5 The ablation approach used in these studies performed by the same operators resulted in slightly higher incidences of EDEL compared to individual AI-guided ablation although there is no head-to-head or randomized comparison available. The concept of AI as a parameter for local ablation lesion quality integrates ablation time, power, and contact force in a weighted formula at each ablation site. Our data indicate that AIguided ablation may also be safe in regard to creating low incidences of EDELs by potentially reducing local energy and ablation time to a minimum needed for adequate lesion formation but not more. AIguided ablation has been documented to improve the procedural and long-term outcome of AF ablation but randomized trials documenting efficacy on the one hand and safety on the other hand are lacking. 1 Of note, in our study, we used lower target AI values (compared to the CLOSE-protocol studies using 400 AI at posterior wall and 550 AI at anterior wall) at posterior LA wall which may affect the incidence and extent of collateral esophageal thermal injury. The incidence of EDEL was slightly higher after ablation with a target AI value of 350 compared to ablation with a target AI value of 300 supporting the concept of higher energy applications leading to a higher risk for esophageal thermal injury. However, this difference was statistically not significant.

| Efficacy
In all 211 patients, all PVs demonstrated exit and entrance block at the end of the procedure regardless of the targeted posterior wall AI. The short-term success rate after 3 months FU was 85%. However, long-term success was not evaluated in this study protocol and future comparative studies are needed. Optimal AI values to keep the incidence of EDEL as low as possible without decreasing ablation success need to be defined.

| Incidence of category 2 lesions
EDEL is a surrogate parameter for the risk of relevant esophageal complications including perforation and atrioesophageal fistula. The incidence of EDEL overestimates the number of patients progressing to atrioesophageal fistula and esophageal perforation but category 2 lesions have been documented to identify patients at a 1-out-of-10 risk to progress to esophageal perforation. 5 Therefore dedicated protocols for this patient cohort including repeat endoscopy and early esophageal stenting are being evaluated. In the present study, category 2 lesions (3%) were slightly lower than what has been documented for non AI-guided contact-force posterior wall ablation (5%) which may allude to a lower risk for perforating esophageal complications using posterior wall AIs ≤350. Of note, an ablation strategy resulting in an overall 3% rate of postprocedural esophageal ulcers still has to be used with caution and additional efforts to further reduce this rate of esophageal injury are required.

| LIMITATIONS
The major limitation of the current study is its observational noncomparative character. Patients were included consecutively and the incidence of EDEL in this study cohort was compared with the incidence of EDEL in previous studies performing postablation EE at our institution. We did not use luminal esophageal temperature probes as they seem not to be predictive for the occurrence of EDELs and have not been clearly related to reduced atrioesophageal fistula or EDEL rates. Conflicting data regarding the potential of temperature probes to reduce the risk of EDEL have been published so far. 5-7,16-18 A recent meta-analysis evaluating the prevalence of and strategies to prevent esophageal injury after atrial fibrillation ablation identified 56 studies for potential inclusion in the analysis.
In total 35 of 56 studies had to be excluded due to lack of endoscopic follow-up or an insufficient number of patients undergoing postablation endoscopy. 18 According to this study use of luminal esophageal temperature monitoring was not associated with a lower incidence of post ablation EDEL. Of note, endoscopically detected esophageal lesions are an important surrogate parameter for clinically relevant but rarely occurring esophageal complications like esophageal perforation or fistula.
Furthermore, a major limitation of currently used temperature probes is that esophageal tissue temperature rises might be detected too late to prevent thermal esophageal damage by interrupting RF energy application at LA posterior wall. A high-definition accurate temperature probe using infrared technology has been evaluated in an IDE-trial (HEAT-AF) and data in regard to identify patients at risk for EDEL are promising but further investigations are needed. 6

| CONCLUSION
In this noncomparative single-center study of patients undergoing AIguided AF ablation (with a target AI ≤350 at the posterior wall) the incidence of EDEL was slightly lower than expected from previous published trials. Different prespecified AI target values were not associated with significantly different incidences of EDEL. Larger prospective and randomized studies should evaluate the impact of AI on the incidence of collateral thermal injury and efficacy to define optimal AI values potentially further reducing the risk of esophageal complications.