Effect of esophageal cooling on ablation lesion formation in the left atrium: Insights from Ablation Index data in the IMPACT trial and clinical outcomes

Abstract Introduction The IMPACT study established the role of controlled esophageal cooling in preventing esophageal thermal injury during radiofrequency (RF) ablation for atrial fibrillation (AF). The effect of esophageal cooling on ablation lesion delivery and procedural and patient outcomes had not been previously studied. The objective was to determine the effect of esophageal cooling on the formation of RF lesions, the ability to achieve procedural endpoints, and clinical outcomes. Methods Participants in the IMPACT trial underwent AF ablation guided by Ablation Index (30 W at 350–400 AI posteriorly, 40 W at ≥450 AI anteriorly). A blinded 1:1 randomization assigned patients to the use of the ensoETM® device to keep esophageal temperature at 4°C during ablation or standard practice using a single‐sensor temperature probe. Ablation parameters and clinical outcomes were analyzed. Results Procedural data from 188 patients were analyzed. Procedure and fluoroscopy times were similar, and all pulmonary veins were isolated. First‐pass pulmonary vein isolation and reconnection at the end of the waiting period were similar in both randomized groups (51/64 vs. 51/68; p = 0.54 and 5/64 vs. 7/68; p = 0.76, respectively). Posterior wall isolation was also similar: 24/33 versus 27/38; p = 0.88. Ablation effect on tissue, measured in impedance drop, was no different between the two randomized groups: 8.6Ω (IQR: 6–11.8) versus 8.76Ω (IQR: 6–12.2; p = 0.25). Arrhythmia recurrence was similar after 12 months (21.1% vs. 24.1%; 95% CI: 0.38–1.84; HR: 0.83; p = 0.66). Conclusions Esophageal cooling has been shown to be effective in reducing ablation‐related thermal injury during RF ablation. This protection does not compromise standard procedural endpoints or clinical success at 12 months.


| INTRODUCTION
Recent randomized data on controlled esophageal cooling suggests that it can prevent esophageal thermal injury, 1 an important contributor to serious complications of ablation for atrial fibrillation (AF). 2 Esophageal cooling during RF left atrial ablation involves application of thermal energy at both ends of the spectrum, in the esophageal luminal and left atrial endocardial walls, which are in close proximity to one another and may be at juxtaposed sites anatomically. The impact of esophageal cooling on ablation lesion formation is unknown.
Ablation Index (AI) was introduced in 2017 as an improved method to help standardize radiofrequency (RF) energy deliveries to ensure the creation of durable ablation lesions with irrigated, contactforce sensing catheters. [3][4][5] AI is a nonlinear weighted formula which incorporates time, contact force, and power to provide the operator with a real-time measure of ablation lesion formation. The AI system collects data about the characteristics of each RF delivery and the response to that delivery in electrical impedance, where a 5-10Ω drop has been viewed to be a marker of successful lesion formation. 6,7 AI has been validated in several studies and has been shown to improve both acute and long-term outcomes of catheter ablation in AF. [8][9][10] IMPACT was a randomized study that evaluated the ability of an esophageal temperature control device (ensoETM ® , Attune Medical) to reduce thermal injury during AI-guided AF ablation. 11 The device is a double-lumen silicone tube; it permits closed-loop water irrigation 2.4 L/min at a temperature as cold as 4°C, creating a large capacity to extract heat from the vicinity. 12,13 The IMPACT study endoscopy results showed significant evidence of protection, with a relative reduction of thermal injury of 83.4%. 11 In this further study of the IMPACT study cohort, we sought to determine if esophageal cooling affected the ability to achieve acute ablation procedural endpoints and clinical success.

| Trial design
The IMPACT study was an investigator-initiated single-center, prospective double-blind randomized trial of esophageal temperature control in adult patients undergoing AF ablation using AI technology (ClinicalTrials.gov NCT03819946). 11 The study was approved by the London-Stanmore Research Ethics Committee (IRAS ID 253844, NIHR CPMS ID 40619).

| Study purpose
The primary and secondary endpoints of the IMPACT study have been outlined. 11 In this further analysis, we sought to determine the effect of esophageal cooling on the characteristics of RF lesions associated with esophageal injury.

| Study population
All adult patients attending for RF ablation for AF under general anesthesia by participating electrophysiologists were screened for study eligibility during pre-assessment. Both first-time and redo AF ablations were included in the study. Table 1 illustrates the proportion of first-time and redo patients in each randomized group, with no significant differences between them. Indications for catheter ablation for AF were in keeping with international guidelines. Exclusion criteria were: Age <18 or >88 years; inability to consent for any reason and contra-indication for upper gastrointestinal endoscopy for any reason.
Enrolled participants were randomized 1:1 (via electronic randomization www.sealedenvelope.com) to either receive thermal protection with the ensoETM ® device or standard care consisting of the use of a single sensor temperature probe during ablation.
Participants were blinded to the treatment assignment.

| Definitions
Technical success was defined as demonstrable isolation of all pulmonary veins (PVs). This was assessed by standard methods; we determined entrance and exit block, using multipolar mapping catheters either the Lasso or the Pentaray (Biosense Webster). The achievement of enduring first-pass isolation of the veins was a desirable outcome, defined as isolation that occurred on completion or before completion of the encircling lesion set for each pair of veins and endured for at least 20 min including adenosine provocation when performed. Adenosine provocation testing was not a substitute for the full waiting period but was permitted to be performed in this protocol during the waiting period. This option of either a full waiting period or a waiting period and adenosine testing was in line with standard ablation practice at this center. Proven block across all other lines created was also a desired endpoint. The duration of the procedure and duration of fluoroscopy were documented, as well as ablation delivery parameters including total RF ablation time, power, force, force-time integral, and AI. All the lesion-related data stored in the Carto ® system (Biosense Webster) were exported into Microsoft Excel for analysis ( Figure 1).
Clinical success was also measured as freedom from atrial arrhythmia at >3 months after ablation. Atrial arrhythmia recurrence was defined as a record of AF or related atrial tachycardia of >30 s from standard cardiac monitoring devices available at standard care such as 24-h ECG, implanted loop recorders, and so on. A major adverse cardiovascular or cerebrovascular event (MACCE) was defined as in-hospital death from any cause, acute myocardial infarction, or acute ischemic stroke and was screened for acutely up to 12-month follow-up.

| Method of RF ablation
All procedures were performed under general anesthesia. Patients continued anticoagulation as per local practice at the center.
Ultrasound-guided vascular access was available for all operators.
All ablation lesions were guided by AI with those in the anterior part of the left atrium created at 40 W with an AI target of ≥450; posterior lesions were at 30 W with an AI target of 350-400.
Anterior and posterior segmentation was in keeping with previously recognized definitions. 3 Point-by-point ablation was performed with interlesion distance <6 mm. The Visitag Surpoint TM was standardized at the recommended settings: minimum force of 5 g, force overtime of 25%, and lesion tag size of 3 mm. Respiratory adjustment was enabled.

| Protected group: Utilizing the ensoETM ® device
After transseptal puncture, the transesophageal echocardiography probe was withdrawn and an ensoETM ® probe was introduced in its place by the attending anesthetist and connected to a mobile console (Blanketrol III, Gentherm Medical). Before commencing ablation on the posterior part of the left atrium, the probe was set to cooling mode at 4°C for at least 10 min. Cooling continued until posterior wall ablation was complete. Body temperature was recorded throughout with a temperature probe placed in the axilla or nasopharynx.

| Postprocedural management
All participants were prescribed proton pump inhibitors, postprocedure for a duration of 6-8 weeks (Lansoprazole 30 mg od or an alternative to an equivalent dose).

| Endoscopy
Patients in both groups were invited to attend for esophageal endoscopy at 7 days after ablation by one of two senior endoscopists. The patient and the endoscopist were blinded to the treatment assignment of the patient and following a standardized protocol. The site of any physical or thermal injury to the esophagus was documented, and any abnormality of gastric or esophageal motility. These results have already been reported. 11

| Focus on ablation lesion data
Ablation data were analyzed from all IMPACT study participants, including those who did not have endoscopy performed. Ablation data was extracted from the Carto workstations, which included all the lesions applied in all of the cases included in this study. The additional information yielded includes RF time (sec), force (g), power (W), base impedance (ohms), impedance drop (ohms), FTI, and AI achieved. Due to X, Y, and Z coordinates for each lesion, consecutive interlesion distance could be measured. The maximum ablation catheter temperature tip (degrees) was also recorded per lesion.

| Statistical analysis
Categorical data were compared using the χ² test, post hoc test was used to detect the difference between individual groups. Student's t-test was used to compare normally distributed data and Mann-Whitney U-test to compare nonparametric data. Spearman rank correlation coefficient (ρ) was used to measure the ordinal association between nonparametric variables. Kaplan Meier curve was used to measure the fraction of patients with AF recurrence.
Analysis was performed with IBM SPSS statistical software (Version 26.0, IBM SPSS Statistics).

| RESULTS-AI IN IMPACT
One Hundred eighty-eight patients were recruited from February 2019 to January 2020 and all underwent successful catheter ablation (Supporting Information: Figure 1). Because 36% of participants were unable or unwilling to return for endoscopy after ablation, recruitment was expanded to obtain the 120 with endoscopy required by our study design. Procedure characteristics and outcomes were recorded for all 188 patients; AI data were analyzed for 181 participants, with seven cases (four in the protected group, three

| Patient characteristics
Demographic data of the study groups were well-matched (Table 1). The mean age in the randomized protected and control groups was 65.1 and 65.2 years respectively and 58% and 63% were males.

| Procedure characteristics
PVI was achieved in all cases. The rates of success in creating block in the left atrial roof, the mitral isthmus, and the cavotricuspid isthmus were similar, and in those in whom isolation of the left atrial posterior wall was attempted, the rate of its success was similar in both groups ( Table 2). The total procedure, fluoroscopy, and RF time were similar in both randomized groups (p > 0.05; Table 2 ablation lesions analyzed. The QDot Micro was used in QMODE only, in all cases with the same AI technology and guidance as for STSF, therefore this data was included in the AI analysis, as both the practical use and the data extraction and analysis were the same.

| Enduring first-pass isolation
The achievement of first-pass isolation and the recurrence of

| Lesion parameters and stability
The overall values for impedance drop for all ablation points were similar in the protected group versus the control group (p = 0.25,

| Impedance drop
AI was found to have a highly significant but weak correlation with impedance drop that was similar in the protected group and the control group (ρ = 0.16-0.25, p < 0.0001; Figure 4).

| Lesion-to-lesion progression
There was a closer spatial relationship between consecutive lesions  (ρ = 0.14-0.25; p < 0.0001). A similar correlation was found between interlesion distance and impedance drop in the protected group and in the control group (ρ = 0.15-0.28, p < 0.0001).

| Esophageal injury
There was no case of esophageal perforation or atrioesophageal fistula. Endoscopic findings have been reported 9 : Mucosal injury was less common in the protected patients than in the control group (2/ 60 vs. 12/60; p = 0.008) with a trend toward lower incidence of gastroparesis in the protected group (2/60 vs. 6/60; p = 0.27).
The occurrence of mucosal injury was not associated with difference in RF duration, maximum power, or impedance drop (p = 0.08-0.69, Supporting Information:

| Complications
In the protected group there was one acute complication: A pericardial effusion that was conservatively managed without sequelae but was associated with a hospital stay of two nights. In the control group, there were two incidences of vascular accessrelated trauma requiring thrombin injection, one of pulmonary edema

| 12-month arrhythmia recurrence
The study patients had outpatient arrhythmia recurrence measured by standard of care methods, which involved intermittent, noninvasive monitoring in the form of a Holter for at least 24 h. Only those that had pre-existing implanted devices such as pacemakers or loop recorders had these devices interrogated to screen for arrhythmia recurrence. Short-term (>3-6 months) arrhythmia recurrence in the IMPACT study was interrupted by the covid-19 pandemic in that telephone clinics replaced face-to-face clinic appointments but with severe delay in objective assessment of arrhythmia recurrence. Therefore, short-term results were not calculated. By March 2021, 12-month outcomes were reliably measured, using these described methods in a standard care setting, either at the same hospital or in linked community hospitals. Kaplan Meier curve was used to measure the fraction of patients with AF recurrence: 12-month outcomes for arrhythmia recurrence showed no difference between both randomized groups. 21.1% versus 24.1%; 95% CI: 0.38-1.84; HR: 0.83; p = 0.66 ( Figure 6).

| Main findings
The IMPACT study was the first double-blind randomized controlled trial to identify an effective method for protecting the esophagus from thermal injury during ablation. 11 In this current study, we demonstrate that this effect was achieved without any evident disruption of our ability to create individual RF lesions with similar procedural endpoints reached and no discernible difference in AF recurrence rates at 12-month analysis. The study results, therefore, support the impression that controlled active cooling of the esophagus did not lead to a paradoxical effect on RF delivery and its effect on myocardial tissue.

| RF ablation guided by AI
Like previous studies, [3][4][5]9 our data show that AI methodology produces a high rate of enduring first-pass isolation of all PVs. As in previous studies, most points that require additional ablation after completing the first pass were located posteriorly, 9 mostly at the carina ( Figure 2). Our data shows that esophageal cooling does not make this difficult area more challenging to ablate.

| Energy delivered and lesion formed
Impedance drop is a surrogate marker of lesion depth, well known from impedance-guided RF ablation. 15,16 With effective application of RF energy, thermal energy disrupts the integrity of local tissue with irreversible cellular dysfunction and necrosis, which lowers the local electrical impedance. While AI integrates the characteristics that go into an effective RF delivery, impedance drop measures its outcome, providing a cross-check of effectiveness. Our data confirm this with an association between interlesion distance and impedance drop that is independent of AI or its components ( Table 2,

Figures 4 and 5).
Similar AI values occurred in both study and control groups and the impedance drop achieved in response to that RF delivery was similar. The fact that esophageal protection does not reduce the impedance drop produced by a lesion with a given AI value is strong evidence that lesion formation is not impeded ( Figure 5).

| Catheter tip temperature
Catheter tip temperature sensing and its measurements during a procedure are recorded in the ablation data set under "maximum catheter tip temperature." The median of this temperature parameter did not differ between the two randomized groups in the IMPACT study ablations. This result amongst the rest of the similar ablation lesion application parameters is supportive of the fact that esophageal temperature lowering did not extend significantly to the left atrial tissue myocardium to negatively attenuate or weaken the RF applications. Although further temperature information at the catheter tip level would have been more useful, this provided reassurance alongside other procedural evidence. The biophysics of RF ablations is well described in literature. 17

| Enduring first-pass isolation
The achievement of first-pass isolation and the persistence of that isolation to the end of the waiting period and through adenosine challenge predict greater long-term success. 19 From this endpoint, it appears that esophageal protection has not hindered the success of ablation.

| Lesion-to-lesion progress
The data on the distance between consecutive lesions suggest that the presence of esophageal protection permitted the operators to operate more comfortably in constructing lesion sets in the posterior left atrium. They appear to show a steady progression from one site to the next, adhering in most cases to the desired 4-6 mm between consecutive lesions compared to the control group in which juddering progress was evident with more instances of movement across a distance of more than 15 mm within the posterior region. We interpret this as evidence of hesitancy arising from the occurrence of temperature rises in the esophagus, or from fear of their occurrence.
This may be purposeful, as the so-called "skip" strategy is used by some operators to avoid the heat stacking phenomenon. Juddering movement from one lesion to the next may also reflect intracardiac or intra-thoracic movement which leads to catheter tip instability.
The impact of contact force variation, spatial movement per lesion application as well as the sequential lesion placement on arrhythmia recurrence were explored in the study by Jankelson et al. 20 Here, the study implies that nonsequential ablation lesions were associated with increased risk of arrhythmia recurrence. This was thought to be due to catheter instability leading to contact force variation and spatial movement of the catheter causing shorter ablation applications and therefore leading to nonsequential lesion placement as a

| Clinical results versus mathematical modeling
This study provides real-life confirmation of effects predicted by a mathematical model of heat transfer at the interface between the heart and the esophagus. 21 The model used average values for the thermal and electrical properties for left atrial myocardium, esophagus, and pericardium and a variety of different values for ablation settings, set temperature of the device, and tissue thickness. It predicted that esophageal cooling, even with the device set at body temperature (37°C) would reduce esophageal thermal injury and that cooling to 5°would not impact on temperature or lesion formation in the myocardium during RF application for 20 s at 10-50 W.

| Secondary endpoints: Procedure metrics and complications
Overall procedural workflow was not compromised by esophageal cooling as demonstrated by the similar procedural time and fluoroscopy duration in both groups (Table 2). Acute complication rates were similarly low in both groups, as were MACCE up to 12 months postprocedure.

| Long-term follow-up
Introduction of new technology or changes to procedural workflow as part of research necessitates scrutiny on long-term outcomes in addition to monitoring of the procedural and acute parameters as outlined so far. 12-month follow-up from the IMPACT study patients show that there was no difference in arrhythmia recurrence between the two randomized groups. The method in which AF recurrence was measured was using noninvasive, intermittent Holter monitoring as per standard care. The similar rate of recurrence measured at 12 months reassuringly supports the hypothesis that controlled esophageal cooling for esophageal protection is not at the expense of the efficacy of the RF ablation procedure when analyzed by lesion application, procedure workflow, and its clinical or therapeutic effect on patients.

| Limitations
This is a single-center study and although further experience with this device and technique in esophageal protection has since been reported, 22,23 a multicenter study is required to further investigate and confirm these findings. A multicenter randomized study is currently being set up to build on this work.
Both the follow-up methods and the frequency of Holter monitoring in both randomized groups fell in line with standard care.
Although this was favorable from a logistical perspective, this was also a limitation, as the evidence becomes clear that continuous ECG monitoring improves the objective measurement of AF recurrence and so true recurrence for both randomized groups may be higher.
Short-term recurrence rates (>3-6 months) are not presented as this was limited by the first wave of the covid-19 pandemic, where ECG diagnostics were severely rationalized as part of the pandemic protocol.

| CONCLUSION
Esophageal cooling has been shown to be effective in reducing ablation-related thermal injury during RF ablation. This protection does not compromise standard procedural endpoints or clinical success at 12 months.