Abstract
Atrio-oesophageal fistula (AOF) is a potentially lethal complication of atrial fibrillation (AF) ablation. Many studies have evaluated the presence and prevention of endoscopically-detected oesophageal lesions (EDOL) as a proxy measure for risk of AOF. This systematic review and meta-analysis sought to determine the prevalence of EDOL and effectiveness of general preventive measures during AF ablation.
We searched electronic databases for studies reporting prevalence or prevention of EDOL post-AF ablation. Pooled prevalence were reported with 95% confidence intervals (CI) while studies evaluating preventive measures including oesophageal temperature monitoring (OTM), esophageal manipulation and type of anaesthesia were analyzed descriptively or by random-effects modeling. Twenty-five studies were included in the analysis. Any and ulcerated EDOL pooled prevalence was 11% (95%CI, 6–15%) and 5% (95%CI, 3–7%), respectively. In six studies, there was no difference in EDOL with or without OTM (pooled OR 1.65, 95%CI, 0.22–12.55). There was no difference using a multi-sensor versus single-sensor OTM (one study) nor when using a deflectable probe (two studies). Oesophageal displacement was associated with significant instrumentation injury in one study. Two studies evaluating Oesophageal cooling showed conflicting results. General anaesthesia was associated with more EDOL than conscious sedation in two studies.
The pooled prevalence of any and ulcerated EDOL post-ablation was 11% and 5%, but varied between studies. Techniques such as OTM and oesophageal displacement or cooling have not conclusively demonstrated a reduction in EDEL, while general anaesthesia may be associated with higher EDOL risk. Further randomized data are critically needed to validate and develop measures to prevent EDOL and AOF.
Atrio-oesophageal fistula is a potentially lethal complication of atrial fibrillation ablation. The prevalence of esophageal injury and effectiveness of preventive measures are uncertain.
In this systematic review and meta-analysis, the pooled prevalence of any and ulcerated esophageal lesions diagnosed on endoscopy was 11% and 5%, respectively.
Current data regarding oesophageal temperature monitoring are conflicting while there remains sparse evidence to support oesophageal displacement or cooling.
General anaesthesia may be associated with more frequent oesophageal injury compared with conscious sedation.
Randomized data are critically needed to develop preventive measures for this serious post-ablation complication.
Introduction
Catheter ablation is being increasingly performed for the treatment of atrial fibrillation (AF).1 Although regarded as a minimally-invasive procedure, it is not without risk. Atrio-oesophageal fistula (AOF) is one of the most serious potential complications.2
Atrio-oesophageal fistula primarily develops from oesophageal injury during ablation, particularly when ablating the left atrium (LA) posterior wall. However, other possible contributing mechanisms include injury from pre-existing gastro-oesophageal reflux, pre-existing oesophageal dysmotility, and potential instrumentation injury from transoesophageal echocardiogram (TOE) or oesophageal temperature monitoring (OTM) probe.3 Given that AOF is rare,4 many studies have used thermal-induced oesophageal lesions as determined by post-procedure endoscopy (endoscopically-detected oesophageal lesions; EDOL) as a proxy measure for the likelihood of developing AOF. A recent study has demonstrated that oesophageal ulcer detected by post-ablation endoscopy is a significant predictor for oesophageal perforating complications.5
While restrictions to ablation temperature and power settings as well as various ablation catheters have been studied with variable success,6–8 the applicability of such measures are usually limited to the specific type of catheter ablation. As such, interest has grown in preventive techniques that are generalizable to any percutaneous ablation approach such as OTM, oesophageal manipulation, the type of anaesthesia used during the procedure, and prophylactic proton pump inhibitors (PPIs).9
In this systematic review and meta-analysis, we sought to determine the prevalence of oesophageal injury following AF ablation and evaluate the effectiveness of techniques to prevent EDOL during catheter ablation for AF.
Methods
Data sources and search strategy
A digital literature search was performed through MEDLINE and EMBASE databases for published articles up to August 2017. Key words using Medical Subject Heading (MeSH), where available, included ‘atrial fibrillation’, ‘catheter ablation’, ‘complications’, and ‘oesophageal injury’. The study protocol was prospectively registered with the PROSPERO international register (CRD42017077495). Example search strategies are provided in Supplementary material online, Tables S1 and S2. Reference lists of previous relevant articles were also reviewed for further articles.
Study objectives and eligibility criteria
The inclusion criteria for studies determining the prevalence of EDOL following percutaneous AF ablation were as follows: (i) studies in humans, (ii) all patients underwent upper gastrointestinal (GI) endoscopy follow-up <7 days of ablation procedure for evaluation of oesophageal injury, (iii) ≥50 patients, (iv) English language, and (v) fully published status. Studies using different ablation energy systems were included for comprehensive review and analysis, with sensitivity analyses of different ablation platforms performed. We excluded studies performed in animal subjects or in vitro, and studies that only performed endoscopy in patients whose OTM reached a certain cut-off temperature. Studies using surgical AF ablation or studies, which evaluated oesophageal injury through a modality other than upper GI endoscopy (e.g. rate of rise in oesophageal temperature, computed tomography findings) were also excluded.
In the assessment of techniques to prevent EDOL, we included studies that evaluated a generalizable preventive measure as their primary objective. In these studies, no minimum cut-off for the number of patients was required.
Data items and collection process
Data items to be collected were determined prior to the literature search. Two investigators (F.J.H. and H.H) conducted the literature search independently and extracted data for study design, baseline patient demographics, procedural characteristics including use of OTM, procedural and ablation time, and rate of EDOL. Extracted data were verified by the senior author (H.L) with any discrepancies resolved by consensus. Risk of bias was assessed using the Newcastle-Ottawa Scale (see Supplementary material online, Table S3).
Clinical endpoints
For the prevalence of EDOL following AF ablation, the clinical endpoints were the rate of any EDOL and the rate of ulcerated EDOL, as defined by the individual study. For the evaluation of preventive measures, the clinical endpoint was the rate of any EDOL. One study reported a mean severity of ulceration score after assigning numerical value according to lesion severity, where the minimum was one (normal) and the maximum was four (severe ulcer >3 mm).10
Statistical analyses
Data for rate of any EDOL and ulcerated EDOL were analysed by Freeman–Tukey double arcsine transformation with random-effects modelling. Pooled prevalence was reported as a summary estimate with 95% confidence interval (CI). Sensitivity analyses were performed to assess differences based on routine pre-procedure TOE and routine post-ablation PPI therapy. For the evaluation of OTM, only comparative studies were included in the analysis and data were analysed by random-effects modelling with summary statistics reported as a pooled odds ratio (OR) with 95% CI. Where multiple studies originated from the same institution and referred to the same cohort of patients, only data from the larger study or the most recently published data were included in quantitative analysis. Statistical heterogeneity was quantified with the I2 statistic, where an I2 >50% was considered significant heterogeneity. A two sided P-value of <0.05 was considered significant. Statistical analyses were performed using Stata MP 13.0 (Stata Corp LP, College Station, TX, USA) with the metaprop and the metan suite of commands.
Results
A total of 472 citations were reviewed and screened, with 56 studies identified for potential inclusion and further evaluation. Of these articles, 35 studies were excluded due to lack of endoscopic follow-up (n = 15) or fewer than 50 patients included in the assessment of EDOL prevalence (n = 12). Two studies were excluded from statistical analysis of EDOL prevalence as only a subgroup of patients whose luminal oesophageal temperature (LOT) reached >39°C underwent post-ablation endoscopy.11,12 Twenty-five studies were included in the final analysis. Full details of reasons for study exclusion are outlined in the PRISMA study flow chart (see Supplementary material online, Figure S4).
Prevalence of oesophageal injury
Fourteen studies with a total of 2948 patients evaluated the prevalence of any or ulcerated EDOL after AF ablation (see Supplementary material online, Table S5).5,10,13–24 Most were prospective studies (12/14) and the proportion of patients with paroxysmal AF ranged from 36% to 100%. Transoesophageal echocardiogram was routinely performed prior to ablation procedure in at least eight studies (not stated in six studies). Conventional radiofrequency (RF) ablation was the most common ablative approach (10 studies) with other types of catheter ablation including cryoballoon (two studies),14,19 a multi-electrode ablation catheter (nMARQ catheter)25 and a RF hot balloon system (designed to perform box isolation).10
For conventional RF ablation, all studies used pulmonary vein (PV) isolation with additional lesions (e.g. roof line, mitral isthmus line, complex fragmented atrial electrogram ablations) created depending on operators’ discretion. The maximum ablation temperature and power settings was heterogeneous ranging from 40 to 52°C and 20–50 W, respectively. Similarly, the maximum ablation time per lesion varied from 20 to 40 s. Most studies used an OTM probe (12/14) with a LOT cut-off point of 38.5–43°C. There were some differences in EDOL definitions between studies. Lesion location was specified in 9 out of 14 studies and was widely considered to be thermal-induced if located on the anterior wall of the mid-oesophagus or at least close to the region of the LA (see Supplementary material online, Table S6). While five studies made a distinction between any ulcer and necrotic/large ulcer, only two studies specified the definition of oesophageal ulcer detected on endoscopy (see Supplementary material online, Tables S6 and S7). No studies performed routine endoscopy prior to AF ablation for baseline EDOL measure.
The pooled prevalence of any and ulcerated EDOL across assessable studies was 10.7% (95% CI 6.3–15.0%, range 2.2–21%; Figure 1A) and 5.2% (95% CI 3.2–7.1%, range 1.5–11.1%; Figure 1B), respectively, with significant heterogeneity present (I2 = 93% and 80%, respectively). This heterogeneity was not explained by use of routine pre-procedure TOE (P = 0.89) or routine post-ablation PPI therapy (P = 0.92) on sensitivity analyses (see Supplementary material online, Figures S8 and S9). When stratified by ablation type, the few studies for certain ablation platforms (cryoballoon, nMARQ, surround flow irrigated RF) limited interpretation (see Supplementary material online, Figures S10 and S11). The pooled prevalence for any and ulcerated EDOL in conventional RF ablation studies was 10.9% (95% CI 6.0–15.8%, I2 = 94%) and 4.2% (95% CI 2.3–6.2%, I2 = 75%). Examples of EDOL are shown in Figure 2. Across the included studies, two patients developed fistula (pooled prevalence, 2/2948) and subsequently died.5 Both patients had undergone AF ablation with OTM where the maximum temperature recorded was 40.4°C and 41.7°C.
A forest plot of ES for the prevalence of any oesophageal lesions (A) and oesophageal ulcer (B). ES, summary estimates.
![Examples of endoscopically-detected oesophageal lesions post-ablation. (A) Necrotic ulceration and (B) mucosal erythema. Reprinted with permission from Martinek et al. [18] from Elsevier.](https://oup.silverchair-cdn.com/oup/backfile/Content_public/Journal/europace/21/1/10.1093_europace_euy121/3/m_euy121f2.jpeg?Expires=1747998490&Signature=q7coibJa7Pv92Ha9TjjfKxSiirTFmLf4nE0fGl3bZu7eS7S6uD9ZNXgwAhmKIrU9fyMYHtuprEs0iU6vU2O5~Sa6LfbzV~hsTbjkD3Cm~C8K2eR4D8mrQbuRsQfvhbnPyHlPLrYepkJ628Ijxdp1cHCe8Ctop1RCH8UmtgPWt5n6tktJ0WzWxbsSzMsRQ7vxif3EkE80I~x8IVERDPKkpgPdvSF3zv1e9S3VwOiO2n9PNQQ8O0~5ESuctpf71tUSoQIrh4LpN913WUUOmvFmpywXG-xQGTr4YLNEoyVhh-~XtdFOXJrmdx0qNsFK564bSJviuvEUl8vAF0Tav29PHQ__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Examples of endoscopically-detected oesophageal lesions post-ablation. (A) Necrotic ulceration and (B) mucosal erythema. Reprinted with permission from Martinek et al. [18] from Elsevier.
Oesophageal temperature monitoring
Six studies with a total of 636 patients evaluated the use of OTM during AF ablation with endoscopic follow-up (Table 1).12,15,22,25–27 All were prospective, non-randomized studies comparing patients undergoing AF ablation with or without OTM. Although RF ablation was used across the studies, four studies used conventional, irrigated-tip ablation catheters,15,22,26,27 one study used a system which allowed duty-cycled phased RF energy to be delivered via an over-the-wire catheter,12 and one study used a circular, multi-electrode nMARQ catheter designed to achieve near-circumferential antral ablation lesions around the PV.25 The preferred RF ablation strategy across the studies was PV isolation. All procedures were performed under conscious sedation (CS) and fluoroscopic guidance for OTM probe positioning. Four studies used a 7 Fr, multi-sensor (MS), non-deflectable OTM probe12,15,25,26 while an earlier study used a 9 Fr, single-sensor (SS), non-deflectable OTM probe22 and a recent study used an S-shaped OTM probe with insulated thermocouples (12 sensors).27 OTM-detected LOT cut-off ranged from 38.5 to 40°C.
Endoscopic studies evaluating the use of oesophageal temperature monitoring during AF ablation
| Authors | Year (study period) | Design | No. of patients | Paroxysmal AF (%) | Body mass index (kg/m2) | Ablation type | Ablation strategy | Max temp. (°C) | Max. power (W) | Max. ablation time per lesion (s) | Oesophageal temperature monitoring probe | Oesophageal cut-off temperature (°C) | Total procedural time (min) | Total ablation time (min) | EDOL | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| OTM | No OTM | P-value for difference | |||||||||||||||
| Singh et al.22 | 2008 (2006–2007) | P | 81 (67/14) | 42/50 | 31/28 | RF | PVI | 40 | 35 | NR | 9 Fr, SS | >38.5 | NR | 59 ± 24/67 ± 37 | 4/67 (6%) | 5/14 (36%) | <0.01 |
| Deneke et al.12 | 2011 (NR) | P | 90 (48/42) | 75/80 | NR | Duty- cycled RF | PVI | 60 | 10 | 60 | 7 Fr, MS | ≥40 | 107 ± 22/ 94 ± 26 | 26 ± 9/22 ± 5 | 5/48 (10%)a | 0/42 (0%) | <0.01 |
| Deneke et al.25 | 2015 (2013-2014) | P | 145 (103/42) | 58/41 | NR | RF with nMARQ | PVI | 45 | 25 (PW-20) | 60 | 7 Fr, MS | >39.5 | 112/93 | 19/15 | 29/103 (28%) | 0/41 (0%) | <0.0001 |
| Muller et al.26 | 2015 (NR) | P | 80 (40/40) | 65 | 27/28 | RF | Operators discretion (mostly PVI) | 43 | 35 (PW-25) | NR | 7 Fr, MS | >39.5 | 119 ± 27/ 127 ± 40 | 31 ± 11/30 ± 10 | 12/40 (30%) | 1/40 (2.5%) | <0.01 |
| Kiuchi et al.15 | 2016 (NR) | P | 160 (80/80) | 56/51 | 23/23 | RF | PVI ± LARA/ CFAEA/MIA | 43 | 30 (PW-20) | 40 (PW-20) | 7 Fr, MS | >39 | 264 ± 59/ 242 ± 91 | NR | 0/80 (0%) | 6/80 (7.5%) | 0.03 |
| Halbfass et al.27 | 2017 (NR) | P | 80 (40/40) | 27/46 | 30/30 | RF | PVI | 43 | 35 (PW-20) | 20 | MS, S-shaped | >39 | NR | NR | 3/40 (7.5%) | 4/40 (10%) | NS |
| Authors | Year (study period) | Design | No. of patients | Paroxysmal AF (%) | Body mass index (kg/m2) | Ablation type | Ablation strategy | Max temp. (°C) | Max. power (W) | Max. ablation time per lesion (s) | Oesophageal temperature monitoring probe | Oesophageal cut-off temperature (°C) | Total procedural time (min) | Total ablation time (min) | EDOL | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| OTM | No OTM | P-value for difference | |||||||||||||||
| Singh et al.22 | 2008 (2006–2007) | P | 81 (67/14) | 42/50 | 31/28 | RF | PVI | 40 | 35 | NR | 9 Fr, SS | >38.5 | NR | 59 ± 24/67 ± 37 | 4/67 (6%) | 5/14 (36%) | <0.01 |
| Deneke et al.12 | 2011 (NR) | P | 90 (48/42) | 75/80 | NR | Duty- cycled RF | PVI | 60 | 10 | 60 | 7 Fr, MS | ≥40 | 107 ± 22/ 94 ± 26 | 26 ± 9/22 ± 5 | 5/48 (10%)a | 0/42 (0%) | <0.01 |
| Deneke et al.25 | 2015 (2013-2014) | P | 145 (103/42) | 58/41 | NR | RF with nMARQ | PVI | 45 | 25 (PW-20) | 60 | 7 Fr, MS | >39.5 | 112/93 | 19/15 | 29/103 (28%) | 0/41 (0%) | <0.0001 |
| Muller et al.26 | 2015 (NR) | P | 80 (40/40) | 65 | 27/28 | RF | Operators discretion (mostly PVI) | 43 | 35 (PW-25) | NR | 7 Fr, MS | >39.5 | 119 ± 27/ 127 ± 40 | 31 ± 11/30 ± 10 | 12/40 (30%) | 1/40 (2.5%) | <0.01 |
| Kiuchi et al.15 | 2016 (NR) | P | 160 (80/80) | 56/51 | 23/23 | RF | PVI ± LARA/ CFAEA/MIA | 43 | 30 (PW-20) | 40 (PW-20) | 7 Fr, MS | >39 | 264 ± 59/ 242 ± 91 | NR | 0/80 (0%) | 6/80 (7.5%) | 0.03 |
| Halbfass et al.27 | 2017 (NR) | P | 80 (40/40) | 27/46 | 30/30 | RF | PVI | 43 | 35 (PW-20) | 20 | MS, S-shaped | >39 | NR | NR | 3/40 (7.5%) | 4/40 (10%) | NS |
Data are presented as OTM group/non-OTM group, unless otherwise stated.
AF, atrial fibrillation; CFAEA, ablation targeting complex fractionated atrial electrograms; EDOL, endoscopically-detected oesophageal lesion; LARA, left atrium roof ablation; max., maximum; max temp., maximum temperature; MIA, mitral isthmus ablation; MS, multi-sensor; NR, not recorded; NS, non-significant; OTM, oesophageal temperature monitoring; P, prospective; PVI, pulmonary vein isolation; PW, posterior wall; RF, radiofrequency; SS, single-sensor.
Endoscopic evaluation was performed in patients with oesophageal temperature monitoring recording ≥39°C in the OTM group based on the assumption that patients <39°C would not have oesophageal lesions.
Endoscopic studies evaluating the use of oesophageal temperature monitoring during AF ablation
| Authors | Year (study period) | Design | No. of patients | Paroxysmal AF (%) | Body mass index (kg/m2) | Ablation type | Ablation strategy | Max temp. (°C) | Max. power (W) | Max. ablation time per lesion (s) | Oesophageal temperature monitoring probe | Oesophageal cut-off temperature (°C) | Total procedural time (min) | Total ablation time (min) | EDOL | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| OTM | No OTM | P-value for difference | |||||||||||||||
| Singh et al.22 | 2008 (2006–2007) | P | 81 (67/14) | 42/50 | 31/28 | RF | PVI | 40 | 35 | NR | 9 Fr, SS | >38.5 | NR | 59 ± 24/67 ± 37 | 4/67 (6%) | 5/14 (36%) | <0.01 |
| Deneke et al.12 | 2011 (NR) | P | 90 (48/42) | 75/80 | NR | Duty- cycled RF | PVI | 60 | 10 | 60 | 7 Fr, MS | ≥40 | 107 ± 22/ 94 ± 26 | 26 ± 9/22 ± 5 | 5/48 (10%)a | 0/42 (0%) | <0.01 |
| Deneke et al.25 | 2015 (2013-2014) | P | 145 (103/42) | 58/41 | NR | RF with nMARQ | PVI | 45 | 25 (PW-20) | 60 | 7 Fr, MS | >39.5 | 112/93 | 19/15 | 29/103 (28%) | 0/41 (0%) | <0.0001 |
| Muller et al.26 | 2015 (NR) | P | 80 (40/40) | 65 | 27/28 | RF | Operators discretion (mostly PVI) | 43 | 35 (PW-25) | NR | 7 Fr, MS | >39.5 | 119 ± 27/ 127 ± 40 | 31 ± 11/30 ± 10 | 12/40 (30%) | 1/40 (2.5%) | <0.01 |
| Kiuchi et al.15 | 2016 (NR) | P | 160 (80/80) | 56/51 | 23/23 | RF | PVI ± LARA/ CFAEA/MIA | 43 | 30 (PW-20) | 40 (PW-20) | 7 Fr, MS | >39 | 264 ± 59/ 242 ± 91 | NR | 0/80 (0%) | 6/80 (7.5%) | 0.03 |
| Halbfass et al.27 | 2017 (NR) | P | 80 (40/40) | 27/46 | 30/30 | RF | PVI | 43 | 35 (PW-20) | 20 | MS, S-shaped | >39 | NR | NR | 3/40 (7.5%) | 4/40 (10%) | NS |
| Authors | Year (study period) | Design | No. of patients | Paroxysmal AF (%) | Body mass index (kg/m2) | Ablation type | Ablation strategy | Max temp. (°C) | Max. power (W) | Max. ablation time per lesion (s) | Oesophageal temperature monitoring probe | Oesophageal cut-off temperature (°C) | Total procedural time (min) | Total ablation time (min) | EDOL | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| OTM | No OTM | P-value for difference | |||||||||||||||
| Singh et al.22 | 2008 (2006–2007) | P | 81 (67/14) | 42/50 | 31/28 | RF | PVI | 40 | 35 | NR | 9 Fr, SS | >38.5 | NR | 59 ± 24/67 ± 37 | 4/67 (6%) | 5/14 (36%) | <0.01 |
| Deneke et al.12 | 2011 (NR) | P | 90 (48/42) | 75/80 | NR | Duty- cycled RF | PVI | 60 | 10 | 60 | 7 Fr, MS | ≥40 | 107 ± 22/ 94 ± 26 | 26 ± 9/22 ± 5 | 5/48 (10%)a | 0/42 (0%) | <0.01 |
| Deneke et al.25 | 2015 (2013-2014) | P | 145 (103/42) | 58/41 | NR | RF with nMARQ | PVI | 45 | 25 (PW-20) | 60 | 7 Fr, MS | >39.5 | 112/93 | 19/15 | 29/103 (28%) | 0/41 (0%) | <0.0001 |
| Muller et al.26 | 2015 (NR) | P | 80 (40/40) | 65 | 27/28 | RF | Operators discretion (mostly PVI) | 43 | 35 (PW-25) | NR | 7 Fr, MS | >39.5 | 119 ± 27/ 127 ± 40 | 31 ± 11/30 ± 10 | 12/40 (30%) | 1/40 (2.5%) | <0.01 |
| Kiuchi et al.15 | 2016 (NR) | P | 160 (80/80) | 56/51 | 23/23 | RF | PVI ± LARA/ CFAEA/MIA | 43 | 30 (PW-20) | 40 (PW-20) | 7 Fr, MS | >39 | 264 ± 59/ 242 ± 91 | NR | 0/80 (0%) | 6/80 (7.5%) | 0.03 |
| Halbfass et al.27 | 2017 (NR) | P | 80 (40/40) | 27/46 | 30/30 | RF | PVI | 43 | 35 (PW-20) | 20 | MS, S-shaped | >39 | NR | NR | 3/40 (7.5%) | 4/40 (10%) | NS |
Data are presented as OTM group/non-OTM group, unless otherwise stated.
AF, atrial fibrillation; CFAEA, ablation targeting complex fractionated atrial electrograms; EDOL, endoscopically-detected oesophageal lesion; LARA, left atrium roof ablation; max., maximum; max temp., maximum temperature; MIA, mitral isthmus ablation; MS, multi-sensor; NR, not recorded; NS, non-significant; OTM, oesophageal temperature monitoring; P, prospective; PVI, pulmonary vein isolation; PW, posterior wall; RF, radiofrequency; SS, single-sensor.
Endoscopic evaluation was performed in patients with oesophageal temperature monitoring recording ≥39°C in the OTM group based on the assumption that patients <39°C would not have oesophageal lesions.
There was no significant difference in EDOL rate with the use of OTM (pooled OR 1.65, 95% CI 0.22–12.55; Figure 3). Significant heterogeneity was present between studies (I2 = 81%, P < 0.001) with two studies favouring OTM,15,22 three studies from one research group favouring no OTM12,25,26 and one study finding no significant difference.27 Across the six studies, one case of oesophageal fistula occurred in a patient who underwent AF ablation with nMARQ catheter.28
Risk estimates for oesophageal injury for studies performing AF ablation with OTM vs. without OTM. CI, confidence interval; OR, odds ratio; OTM, oesophageal temperature monitoring; RF, radiofrequency.
Type of oesophageal temperature monitoring probe
Four studies which evaluated a specific type of OTM probe during AF ablation were systematically reviewed with endoscopic follow-up (Table 2).29–32
Endoscopic studies evaluating types of oesophageal temperature monitoring probes
| Authors | Year (study period) | Design | No. of patients | Paroxysmal AF (%) | Ablation type | Ablation strategy | Target temperature (°C) | Maximum power (W) | Max. ablation time per lesion (s) | Comparison probe | Control probe | Oesophageal cut-off temperature | Total procedural time (min) | Total ablation time (min) | EDOL | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Comparison probe (%) | Control probe (%) | P-value for difference | |||||||||||||||
| Leite et al.30 | 2011 (2008–2009) | P | 45 | 47 | RF | PVI | 42 | 30 (PW-25) | 30 | 7 Fr, SS, deflectable, ICE-guided | NA | Temperature ↑ by >2°C | 255 | NR | 0/45 (0) | NA | NA |
| Multi-sensor vs. single-sensor probe | |||||||||||||||||
| Kuwahara et al.17,29 | 2014 | P | 100 (50/50) | 56/66 | RF | PVI + SVCI/ LARA/MIA/ CFAEA | 44 | 25-30 | 30 | 16 Fr, MS (5 sensors), non-deflectable | 16 Fr, SS, deflectable | ≥42°C | 156/159 | 27.5/27.8 | 10/50 (20) | 15/50 (30) | 0.25 |
| Deflectable vs. non-deflectable probe | |||||||||||||||||
| Tschabrunn et al.32 | 2015 (2012–2013) | P | 20 (10/10) | 50/40 | RF | PVI | NR | 40 (PW-20) | 40 (PW-20) | 1× 9 Fr, SS, non- deflectable; 1× 10 Fr, MS, deflectable | 1× non- deflectable, 9 Fr, SS | Temperature ↑ by ≥2°C | NR | PW-7.8/7.2 | 5/10 (50) | 4/10 (40) | NS |
| Miyazaki et al.31 | 2016 | P | 40 (27/13) | 100/100 | Cryo | PVI | NA | NA | 180 | 1× MS, non- deflectable; 1× 7 Fr, SS, deflectable | 2× MS, non-deflectable | 0°C | NR | NR | 5/27 (19) | 3/13 (23) | 0.81 |
| Authors | Year (study period) | Design | No. of patients | Paroxysmal AF (%) | Ablation type | Ablation strategy | Target temperature (°C) | Maximum power (W) | Max. ablation time per lesion (s) | Comparison probe | Control probe | Oesophageal cut-off temperature | Total procedural time (min) | Total ablation time (min) | EDOL | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Comparison probe (%) | Control probe (%) | P-value for difference | |||||||||||||||
| Leite et al.30 | 2011 (2008–2009) | P | 45 | 47 | RF | PVI | 42 | 30 (PW-25) | 30 | 7 Fr, SS, deflectable, ICE-guided | NA | Temperature ↑ by >2°C | 255 | NR | 0/45 (0) | NA | NA |
| Multi-sensor vs. single-sensor probe | |||||||||||||||||
| Kuwahara et al.17,29 | 2014 | P | 100 (50/50) | 56/66 | RF | PVI + SVCI/ LARA/MIA/ CFAEA | 44 | 25-30 | 30 | 16 Fr, MS (5 sensors), non-deflectable | 16 Fr, SS, deflectable | ≥42°C | 156/159 | 27.5/27.8 | 10/50 (20) | 15/50 (30) | 0.25 |
| Deflectable vs. non-deflectable probe | |||||||||||||||||
| Tschabrunn et al.32 | 2015 (2012–2013) | P | 20 (10/10) | 50/40 | RF | PVI | NR | 40 (PW-20) | 40 (PW-20) | 1× 9 Fr, SS, non- deflectable; 1× 10 Fr, MS, deflectable | 1× non- deflectable, 9 Fr, SS | Temperature ↑ by ≥2°C | NR | PW-7.8/7.2 | 5/10 (50) | 4/10 (40) | NS |
| Miyazaki et al.31 | 2016 | P | 40 (27/13) | 100/100 | Cryo | PVI | NA | NA | 180 | 1× MS, non- deflectable; 1× 7 Fr, SS, deflectable | 2× MS, non-deflectable | 0°C | NR | NR | 5/27 (19) | 3/13 (23) | 0.81 |
Data are presented as comparison probe group/control probe group, unless otherwise stated.
CFAEA, ablation targeting complex fractionated atrial electrograms; Cryo, cryoballoon; EDOL, endoscopically-diagnosed oesophageal lesion; ICE, intracardiac echocardiography; LARA, left atrium roof ablation; MIA, mitral isthmus ablation; MS, multi-sensor; NA, not applicable; NR, not recorded; NS, non-significant; P, prospective; PVI, pulmonary vein isolation; PW, posterior wall; RF, radiofrequency; SS, single-sensor; SVCI, superior vena cava isolation.
Endoscopic studies evaluating types of oesophageal temperature monitoring probes
| Authors | Year (study period) | Design | No. of patients | Paroxysmal AF (%) | Ablation type | Ablation strategy | Target temperature (°C) | Maximum power (W) | Max. ablation time per lesion (s) | Comparison probe | Control probe | Oesophageal cut-off temperature | Total procedural time (min) | Total ablation time (min) | EDOL | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Comparison probe (%) | Control probe (%) | P-value for difference | |||||||||||||||
| Leite et al.30 | 2011 (2008–2009) | P | 45 | 47 | RF | PVI | 42 | 30 (PW-25) | 30 | 7 Fr, SS, deflectable, ICE-guided | NA | Temperature ↑ by >2°C | 255 | NR | 0/45 (0) | NA | NA |
| Multi-sensor vs. single-sensor probe | |||||||||||||||||
| Kuwahara et al.17,29 | 2014 | P | 100 (50/50) | 56/66 | RF | PVI + SVCI/ LARA/MIA/ CFAEA | 44 | 25-30 | 30 | 16 Fr, MS (5 sensors), non-deflectable | 16 Fr, SS, deflectable | ≥42°C | 156/159 | 27.5/27.8 | 10/50 (20) | 15/50 (30) | 0.25 |
| Deflectable vs. non-deflectable probe | |||||||||||||||||
| Tschabrunn et al.32 | 2015 (2012–2013) | P | 20 (10/10) | 50/40 | RF | PVI | NR | 40 (PW-20) | 40 (PW-20) | 1× 9 Fr, SS, non- deflectable; 1× 10 Fr, MS, deflectable | 1× non- deflectable, 9 Fr, SS | Temperature ↑ by ≥2°C | NR | PW-7.8/7.2 | 5/10 (50) | 4/10 (40) | NS |
| Miyazaki et al.31 | 2016 | P | 40 (27/13) | 100/100 | Cryo | PVI | NA | NA | 180 | 1× MS, non- deflectable; 1× 7 Fr, SS, deflectable | 2× MS, non-deflectable | 0°C | NR | NR | 5/27 (19) | 3/13 (23) | 0.81 |
| Authors | Year (study period) | Design | No. of patients | Paroxysmal AF (%) | Ablation type | Ablation strategy | Target temperature (°C) | Maximum power (W) | Max. ablation time per lesion (s) | Comparison probe | Control probe | Oesophageal cut-off temperature | Total procedural time (min) | Total ablation time (min) | EDOL | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Comparison probe (%) | Control probe (%) | P-value for difference | |||||||||||||||
| Leite et al.30 | 2011 (2008–2009) | P | 45 | 47 | RF | PVI | 42 | 30 (PW-25) | 30 | 7 Fr, SS, deflectable, ICE-guided | NA | Temperature ↑ by >2°C | 255 | NR | 0/45 (0) | NA | NA |
| Multi-sensor vs. single-sensor probe | |||||||||||||||||
| Kuwahara et al.17,29 | 2014 | P | 100 (50/50) | 56/66 | RF | PVI + SVCI/ LARA/MIA/ CFAEA | 44 | 25-30 | 30 | 16 Fr, MS (5 sensors), non-deflectable | 16 Fr, SS, deflectable | ≥42°C | 156/159 | 27.5/27.8 | 10/50 (20) | 15/50 (30) | 0.25 |
| Deflectable vs. non-deflectable probe | |||||||||||||||||
| Tschabrunn et al.32 | 2015 (2012–2013) | P | 20 (10/10) | 50/40 | RF | PVI | NR | 40 (PW-20) | 40 (PW-20) | 1× 9 Fr, SS, non- deflectable; 1× 10 Fr, MS, deflectable | 1× non- deflectable, 9 Fr, SS | Temperature ↑ by ≥2°C | NR | PW-7.8/7.2 | 5/10 (50) | 4/10 (40) | NS |
| Miyazaki et al.31 | 2016 | P | 40 (27/13) | 100/100 | Cryo | PVI | NA | NA | 180 | 1× MS, non- deflectable; 1× 7 Fr, SS, deflectable | 2× MS, non-deflectable | 0°C | NR | NR | 5/27 (19) | 3/13 (23) | 0.81 |
Data are presented as comparison probe group/control probe group, unless otherwise stated.
CFAEA, ablation targeting complex fractionated atrial electrograms; Cryo, cryoballoon; EDOL, endoscopically-diagnosed oesophageal lesion; ICE, intracardiac echocardiography; LARA, left atrium roof ablation; MIA, mitral isthmus ablation; MS, multi-sensor; NA, not applicable; NR, not recorded; NS, non-significant; P, prospective; PVI, pulmonary vein isolation; PW, posterior wall; RF, radiofrequency; SS, single-sensor; SVCI, superior vena cava isolation.
Multi-sensor vs. single-sensor oesophageal temperature monitoring probe
One prospective, non-randomized study compared a MS-OTM probe with a SS-OTM probe during RF ablation using a pulmonary vein isolation (PVI) strategy.29 With a LOT cut-off temperature of 42°C, maximum power of 25–30 W and 30 s limit at LA posterior wall, Kuwahara et al. reported similar rates of oesophageal injury between the groups (20% vs. 30%, respectively; P = 0.25).
Deflectable vs. non-deflectable oesophageal temperature monitoring probe
Two prospective studies evaluated the use of a deflectable OTM probe against a non-deflectable OTM probe using different ablation techniques (RF and cryoballoon).31,32 Both studies simultaneously used two OTM probes (deflectable and non-deflectable) in their comparison group while their control group consisted of either one or two non-deflectable OTM probes inserted simultaneously. There was no significant difference in the rate of oesophageal injury between the comparison or control group in either study. One further study assessed the use of intracardiac echocardiography (ICE) to guide positioning of a deflectable OTM probe.30 In this single-arm, pilot study of 45 patients undergoing RF ablation for AF with a 7 Fr, deflectable SS-OTM probe inserted under ICE guidance, no EDOL were found.
Oesophageal displacement during ablative procedure
Mechanical displacement of the oesophagus involves insertion of an instrument to deviate the oesophagus away from the LA posterior wall during AF ablation. This can be achieved by inflation of an intra-pericardial balloon within the oblique sinus,33 or through instruments inserted directly into the oesophagus such as a TOE probe34 or an endoscope.35 A recent study used a novel stylet inserted through a gastric tube which automatically changes its curvature based on body temperature.36 The only study with post-ablation endoscopic follow-up involved an orogastric tube to displace the oesophagus in 20 patients during AF ablation under general anaesthesia (GA) (Table 3).37 A 9-Fr, SS-OTM was also inserted (LOT cut-off, 38.5°C) with a target RF temperature of 40°C and maximum power of 35 W. On endoscopic assessment, one patient (5%) was found to have a thermal-induced EDOL (located on the anterior wall of the middle third of the oesophagus). However, injury secondary to oesophageal displacement was common (12/19; 63%) although most were of mild severity (9/12; moderate severity in 3/12).
Endoscopic studies evaluating oesophageal displacement or cooling during AF ablation
| Authors | Year | Design | No. of patients | Par-oxysmal AF (%) | Ablation type | Ablation strategy | Target temperature (°C) | Maximum power (W) | Maximum ablation time per lesion (s) | OTM (cut-off temperature (°C) | GA | Oesophageal displacement | EDOL (%) | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Oesophageal displacement | |||||||||||||||
| Koruth et al.37 | 2012 | P | 20 | 50 | RF (n = 16) or laser balloon (n = 4) | PVI | 40 | 35 | 20–30 | Y (38.5) | Y | 32 Fr flexible PVC thoracic catheter inserted with 14 Fr stylet inside to allow displacement | Thermal injury: 1/19 (5) instrumentation injury: 12/19 (63) | ||
| Oesophageal cooling | Oesophageal cooling (%) | No oesophageal cooling (%) | P-value for difference | ||||||||||||
| Kuwahara et al.17,29a | 2014 | P | 100 (50/50) | 76/84 | RF | PVI ± LARA/ MIA/CFAE | 44 | 25–30 | 30 | Y (42) | Y | Ice water (5 mL) injected before RF energy delivery or if oesophageal temperature >42°C | 10/50 (20) | 11/50 (22) | NS |
| Sohara et al.10b | 2014 (2007–2011) | P | 318 (128 168/22) | 52/60/73 | RF hot balloon system | Box isolation | Balloon temperature 70–75 | 200 | NA | Y (39 or 43) | Y | Saline mixed with contrast medium (10–20 mL) injected if oesophageal temperature >39°C or 43°C | Mean severity of ulceration scorec | ||
| Cooling at 43°C: 1.20 ± 0.6Cooling at 39°C: 1.05 ± 0.21 | 1.73 ± 1.12 | P < 0.001 for cooling initiating at 39°C | |||||||||||||
| Authors | Year | Design | No. of patients | Par-oxysmal AF (%) | Ablation type | Ablation strategy | Target temperature (°C) | Maximum power (W) | Maximum ablation time per lesion (s) | OTM (cut-off temperature (°C) | GA | Oesophageal displacement | EDOL (%) | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Oesophageal displacement | |||||||||||||||
| Koruth et al.37 | 2012 | P | 20 | 50 | RF (n = 16) or laser balloon (n = 4) | PVI | 40 | 35 | 20–30 | Y (38.5) | Y | 32 Fr flexible PVC thoracic catheter inserted with 14 Fr stylet inside to allow displacement | Thermal injury: 1/19 (5) instrumentation injury: 12/19 (63) | ||
| Oesophageal cooling | Oesophageal cooling (%) | No oesophageal cooling (%) | P-value for difference | ||||||||||||
| Kuwahara et al.17,29a | 2014 | P | 100 (50/50) | 76/84 | RF | PVI ± LARA/ MIA/CFAE | 44 | 25–30 | 30 | Y (42) | Y | Ice water (5 mL) injected before RF energy delivery or if oesophageal temperature >42°C | 10/50 (20) | 11/50 (22) | NS |
| Sohara et al.10b | 2014 (2007–2011) | P | 318 (128 168/22) | 52/60/73 | RF hot balloon system | Box isolation | Balloon temperature 70–75 | 200 | NA | Y (39 or 43) | Y | Saline mixed with contrast medium (10–20 mL) injected if oesophageal temperature >39°C or 43°C | Mean severity of ulceration scorec | ||
| Cooling at 43°C: 1.20 ± 0.6Cooling at 39°C: 1.05 ± 0.21 | 1.73 ± 1.12 | P < 0.001 for cooling initiating at 39°C | |||||||||||||
AF, atrial fibrillation; CFAE, ablations targeting complex fractionated atrial electrogram; EDOL, endoscopically-detected oesophageal lesion; GA, general anaesthesia; LARA, left atrium roof isolation; MIA, mitral isthmus ablation; NA, not applicable; NS, non-significant; OTM, oesophageal temperature monitoring; P, prospective; PVC, peripheral venous catheter; PVI, pulmonary vein isolation; RF, radiofrequency; SS, single-sensor; Y, yes.
Data are presented as oesophageal cooling group/no oesophageal cooling group, unless otherwise stated.
Data are presented as oesophageal cooling at 43°C/oesophageal cooling with 39°C/no oesophageal cooling group, unless otherwise stated.
Severity of ulceration score based on numerical value assigned to severity of lesion (severe ulcer >3 mm, 4; mild ulcer <3 mm with red spot, 3; erosion, 2; normal, 1).
Endoscopic studies evaluating oesophageal displacement or cooling during AF ablation
| Authors | Year | Design | No. of patients | Par-oxysmal AF (%) | Ablation type | Ablation strategy | Target temperature (°C) | Maximum power (W) | Maximum ablation time per lesion (s) | OTM (cut-off temperature (°C) | GA | Oesophageal displacement | EDOL (%) | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Oesophageal displacement | |||||||||||||||
| Koruth et al.37 | 2012 | P | 20 | 50 | RF (n = 16) or laser balloon (n = 4) | PVI | 40 | 35 | 20–30 | Y (38.5) | Y | 32 Fr flexible PVC thoracic catheter inserted with 14 Fr stylet inside to allow displacement | Thermal injury: 1/19 (5) instrumentation injury: 12/19 (63) | ||
| Oesophageal cooling | Oesophageal cooling (%) | No oesophageal cooling (%) | P-value for difference | ||||||||||||
| Kuwahara et al.17,29a | 2014 | P | 100 (50/50) | 76/84 | RF | PVI ± LARA/ MIA/CFAE | 44 | 25–30 | 30 | Y (42) | Y | Ice water (5 mL) injected before RF energy delivery or if oesophageal temperature >42°C | 10/50 (20) | 11/50 (22) | NS |
| Sohara et al.10b | 2014 (2007–2011) | P | 318 (128 168/22) | 52/60/73 | RF hot balloon system | Box isolation | Balloon temperature 70–75 | 200 | NA | Y (39 or 43) | Y | Saline mixed with contrast medium (10–20 mL) injected if oesophageal temperature >39°C or 43°C | Mean severity of ulceration scorec | ||
| Cooling at 43°C: 1.20 ± 0.6Cooling at 39°C: 1.05 ± 0.21 | 1.73 ± 1.12 | P < 0.001 for cooling initiating at 39°C | |||||||||||||
| Authors | Year | Design | No. of patients | Par-oxysmal AF (%) | Ablation type | Ablation strategy | Target temperature (°C) | Maximum power (W) | Maximum ablation time per lesion (s) | OTM (cut-off temperature (°C) | GA | Oesophageal displacement | EDOL (%) | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Oesophageal displacement | |||||||||||||||
| Koruth et al.37 | 2012 | P | 20 | 50 | RF (n = 16) or laser balloon (n = 4) | PVI | 40 | 35 | 20–30 | Y (38.5) | Y | 32 Fr flexible PVC thoracic catheter inserted with 14 Fr stylet inside to allow displacement | Thermal injury: 1/19 (5) instrumentation injury: 12/19 (63) | ||
| Oesophageal cooling | Oesophageal cooling (%) | No oesophageal cooling (%) | P-value for difference | ||||||||||||
| Kuwahara et al.17,29a | 2014 | P | 100 (50/50) | 76/84 | RF | PVI ± LARA/ MIA/CFAE | 44 | 25–30 | 30 | Y (42) | Y | Ice water (5 mL) injected before RF energy delivery or if oesophageal temperature >42°C | 10/50 (20) | 11/50 (22) | NS |
| Sohara et al.10b | 2014 (2007–2011) | P | 318 (128 168/22) | 52/60/73 | RF hot balloon system | Box isolation | Balloon temperature 70–75 | 200 | NA | Y (39 or 43) | Y | Saline mixed with contrast medium (10–20 mL) injected if oesophageal temperature >39°C or 43°C | Mean severity of ulceration scorec | ||
| Cooling at 43°C: 1.20 ± 0.6Cooling at 39°C: 1.05 ± 0.21 | 1.73 ± 1.12 | P < 0.001 for cooling initiating at 39°C | |||||||||||||
AF, atrial fibrillation; CFAE, ablations targeting complex fractionated atrial electrogram; EDOL, endoscopically-detected oesophageal lesion; GA, general anaesthesia; LARA, left atrium roof isolation; MIA, mitral isthmus ablation; NA, not applicable; NS, non-significant; OTM, oesophageal temperature monitoring; P, prospective; PVC, peripheral venous catheter; PVI, pulmonary vein isolation; RF, radiofrequency; SS, single-sensor; Y, yes.
Data are presented as oesophageal cooling group/no oesophageal cooling group, unless otherwise stated.
Data are presented as oesophageal cooling at 43°C/oesophageal cooling with 39°C/no oesophageal cooling group, unless otherwise stated.
Severity of ulceration score based on numerical value assigned to severity of lesion (severe ulcer >3 mm, 4; mild ulcer <3 mm with red spot, 3; erosion, 2; normal, 1).
Oesophageal cooling during ablative procedure
Oesophageal cooling is usually performed using a gastric tube placed within the oesophagus at the level of the LA posterior wall with a concomitant OTM probe. A solution is injected on to the oesophageal surface based on a LOT threshold to theoretically reduce the oesophageal temperature and thus any thermal-induced injury. Two studies which evaluated the effect of oesophageal cooling during AF ablation were systematically reviewed with endoscopic follow-up (Table 3).10,17 In a randomized study of 100 patients, Kuwahara et al. used ice water (0°C) for the cooling solution which was injected prior to RF energy delivery and again if the LOT reached 42°C, and found no difference in EDOL rate with or without oesophageal cooling (20% vs. 22%, respectively).17 Conversely, in a cohort of 318 consecutive patients, Sohara et al.10 used normal saline (10°C) mixed with contrast medium for the cooling solution and found reduced lesion severity when initiating cooling at an LOT threshold of 39°C compared with injecting at an LOT threshold of 43°C, or compared with no oesophageal cooling.
General anaesthesia vs. conscious sedation during ablative procedure
Two prospective studies which investigated the impact of GA vs. CS during AF ablation were systematically reviewed for risk of EDOL using either capsule endoscopy or standard upper GI endoscopy (Table 4).38,39 Both demonstrated increased EDOL in the GA group compared with patients under CS. Of note, a higher maximum LOT was reached with GA compared with CS (40.6 ± 1°C and 39.6 ± 0.8°C, respectively; P < 0.001) and time to return to baseline temperature was longer (29 ± 3 s and 18 ± 2 s, respectively; P < 0.001) in the randomized controlled trial.38
Endoscopic studies evaluating general anaesthesia vs. conscious sedation during AF ablation
| Authors | Year (study period) | Design | No. of patients | Paroxysmal AF (%) | Ablation type | Ablation strategy | Target temperature (°C) | Maximum power (W) | Maximum ablation time per lesion (s) | GA approach | CS approach | OTM | Oesophageal cut-off temperature (°C) | EDOL | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GA | CS | P-value for difference | ||||||||||||||
| Di Biase et al.38 | 2009 | P | 50 (25/25) | 100/100 | RF | PVI | 41 | 45 | 20 | Induction anaesthesia with 1 mg/kg propofol followed by inhalational agents through ETT | IV admin. of fentanyl and midazolam | 9 Fr, SS, non-deflectable | 39 | 12/25 (48)a | 1/25 (4)a | <0.001 |
| Martinek et al.39 | 2009 | P | 175 (28/140) | 64/56 | RF | PVI | 43 | 30 (PW 15–25) | 20–30 | NR | NR | NA | NA | 3/35 (9) | 2/138 (1.5) | 0.055 |
| Authors | Year (study period) | Design | No. of patients | Paroxysmal AF (%) | Ablation type | Ablation strategy | Target temperature (°C) | Maximum power (W) | Maximum ablation time per lesion (s) | GA approach | CS approach | OTM | Oesophageal cut-off temperature (°C) | EDOL | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GA | CS | P-value for difference | ||||||||||||||
| Di Biase et al.38 | 2009 | P | 50 (25/25) | 100/100 | RF | PVI | 41 | 45 | 20 | Induction anaesthesia with 1 mg/kg propofol followed by inhalational agents through ETT | IV admin. of fentanyl and midazolam | 9 Fr, SS, non-deflectable | 39 | 12/25 (48)a | 1/25 (4)a | <0.001 |
| Martinek et al.39 | 2009 | P | 175 (28/140) | 64/56 | RF | PVI | 43 | 30 (PW 15–25) | 20–30 | NR | NR | NA | NA | 3/35 (9) | 2/138 (1.5) | 0.055 |
AF, atrial fibrillation; CS, conscious sedation; EDOL, endoscopically-detected oesophageal lesion; GA, general anaesthesia; IV, intravenous; P, prospective; PW, posterior wall; PVI, pulmonary vein isolation; RF, radiofrequency; SS, single-sensor.
Data are presented as general anaesthesia group/conscious sedation group, unless otherwise stated.
Oesophageal lesions evaluated through capsule endoscopy.
Endoscopic studies evaluating general anaesthesia vs. conscious sedation during AF ablation
| Authors | Year (study period) | Design | No. of patients | Paroxysmal AF (%) | Ablation type | Ablation strategy | Target temperature (°C) | Maximum power (W) | Maximum ablation time per lesion (s) | GA approach | CS approach | OTM | Oesophageal cut-off temperature (°C) | EDOL | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GA | CS | P-value for difference | ||||||||||||||
| Di Biase et al.38 | 2009 | P | 50 (25/25) | 100/100 | RF | PVI | 41 | 45 | 20 | Induction anaesthesia with 1 mg/kg propofol followed by inhalational agents through ETT | IV admin. of fentanyl and midazolam | 9 Fr, SS, non-deflectable | 39 | 12/25 (48)a | 1/25 (4)a | <0.001 |
| Martinek et al.39 | 2009 | P | 175 (28/140) | 64/56 | RF | PVI | 43 | 30 (PW 15–25) | 20–30 | NR | NR | NA | NA | 3/35 (9) | 2/138 (1.5) | 0.055 |
| Authors | Year (study period) | Design | No. of patients | Paroxysmal AF (%) | Ablation type | Ablation strategy | Target temperature (°C) | Maximum power (W) | Maximum ablation time per lesion (s) | GA approach | CS approach | OTM | Oesophageal cut-off temperature (°C) | EDOL | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GA | CS | P-value for difference | ||||||||||||||
| Di Biase et al.38 | 2009 | P | 50 (25/25) | 100/100 | RF | PVI | 41 | 45 | 20 | Induction anaesthesia with 1 mg/kg propofol followed by inhalational agents through ETT | IV admin. of fentanyl and midazolam | 9 Fr, SS, non-deflectable | 39 | 12/25 (48)a | 1/25 (4)a | <0.001 |
| Martinek et al.39 | 2009 | P | 175 (28/140) | 64/56 | RF | PVI | 43 | 30 (PW 15–25) | 20–30 | NR | NR | NA | NA | 3/35 (9) | 2/138 (1.5) | 0.055 |
AF, atrial fibrillation; CS, conscious sedation; EDOL, endoscopically-detected oesophageal lesion; GA, general anaesthesia; IV, intravenous; P, prospective; PW, posterior wall; PVI, pulmonary vein isolation; RF, radiofrequency; SS, single-sensor.
Data are presented as general anaesthesia group/conscious sedation group, unless otherwise stated.
Oesophageal lesions evaluated through capsule endoscopy.
Discussion
We systematically evaluated the prevalence of oesophageal injury after AF ablation and challenges with current preventive techniques. The main findings are that (i) the pooled prevalence of any and ulcerated EDOL after AF ablation is 11% and 5%, respectively, albeit with significant heterogeneity between studies, (ii) the pooled prevalence of fistula formation was approximately 1 in 1500, (iii) comparative studies evaluating the impact of OTM report conflicting findings, (iv) there is currently no definitive evidence to support oesophageal manipulation either through displacement or cooling, and (v) GA may be associated with increased EDOL rate compared with CS.
Prevalence of oesophageal injury
While the overall prevalence of any EDOL was 11%, significant heterogeneity was present with prevalence ranging from 2% to 21%. Various patient and procedural characteristics likely contribute to the inter-study variability. These include, but are not limited to, patient body mass index, AF duration, ablation approach, maximum power/temperature settings, and lesion time.18 One small single-centre study found that the pre-ablation TOE probe could contribute to oesophageal injury,40 although a large registry of patients who underwent endoscopy post-AF ablation did not specifically report any TOE-related injury.5 Currently, TOE appears safe with instrumentation-related injury being uncommon and able to be distinguished from thermal injury.
We sought to distinguish the presence of ulcerated EDOL from any oesophageal lesion to stratify the injury level and thus potential AOF risk. Importantly, oesophageal ulcers detected by endoscopy are suggested to be a significant predictor of oesophageal perforation, with up to 1 in 10 lesions eventually perforating despite routine PPI use as shown in one study.5 In our study, the pooled prevalence of ulcerated EDOL was 5%. The heterogeneity may in part be explained by different definitions where some studies included any ulcer while other studies included only large or necrotic ulcers. There is a need for uniform endoscopic reporting (e.g. standard size/depth of lesion to stratify severity) for thermal-induced EDOL post-ablation before more precise risk stratification can be offered to patients.
Oesophageal temperature monitoring
The benefit of OTM during AF ablation in reducing EDOL remains controversial. A previous meta-analysis reported no difference using OTM with significant heterogeneity.41 From the current analysis, several studies conducted by one research group generally favoured no OTM12,25,26 while studies performed by other groups favoured the use of OTM.15,22 A limitation with OTM is that it detects delayed temperature rises which may not accurately reflect the intramural temperature and thus ongoing, real-time oesophageal injury. Additionally, while an ablation cut-off temperature (usually 39°C) based on OTM reduces mean maximal LOT, this does not necessarily translate to fewer EDOL. There remains no current consensus on an optimal LOT cut-off as AOF has occurred at a recorded LOT as low as 36.4°C which highlights the importance of probe positioning to detect temperature changes.42
In view of inadequate OTM positioning, MS-OTM probes were designed to cover a larger oesophageal surface area compared with standard SS-OTM probes. Kuwahara et al.29 did not detect a significant difference between MS-OTM compared with a SS-OTM probe, although the small cohort, few sensors (five thermocouples), and high LOT cut-off temperature (42°C) limits generalizability. One large prospective study comparing MS-OTM and SS-OTM was excluded from systematic analysis as only patients with a LOT ≥39°C underwent upper GI endoscopy.43 Their results of increased EDOL in the MS-OTM group are difficult to interpret given that patients with an SS-OTM probe may not have reached the specified oesophageal temperature and thus would not undergo endoscopy according to the study protocol. This is reflected in the significantly higher number of patients whose LOT reached ≥39°C in the MS-OTM group vs. SS-OTM group (75% vs. 39%, respectively). Furthermore, even with improved probe positioning, the latency between real-time and detected oesophageal temperature remains unresolved. There could be a role for novel infrared thermography techniques to more accurately assess dynamic temperature changes across a broad oesophageal region.44
The deflectable component of the OTM probe, which is designed to place the probe tip adjacent to the ablation site, has yet to definitively reduce EDOL prevalence. Only a single-arm study has demonstrated its efficacy when manoeuvred under ICE guidance.30 Amongst comparative studies, there was no difference in rate of EDOL compared with a non-deflectable probe.31,32 Of note, there are no randomized studies assessing the overall utility of OTM in EDOL prevention. Such data are needed before the value of any specific components of OTM, be it the number of thermocouples or the deflectable aspect, can be judged to be useful for routine clinical use.
Oesophageal displacement
Given that the proximity of the oesophagus to the LA is an independent risk factor for oesophageal injury,18 mechanical displacement of the oesophagus appears feasible. However, evidence to support this technique is currently sparse. A prospective study showed that the oesophagus could be displaced up to 6 cm laterally, enabling safer RF delivery in 97% of patients with reduced maximal LOT.34 However, the only prospective study with endoscopic follow-up showed that mechanical oesophageal injury from the displacing instrument itself was extremely common (63%).37 Other considerations are the impact on LA geometry, which can be altered by oesophageal displacement, difficulties in displacement depending on laxity of adjacent anatomical structures,35 logistical and practicality issues for institutions that do not routinely use oesophageal displacement, and the need for GA which carries its own risks.
Type of anaesthesia
Use of GA appears to increase the risk of EDOL in two prospective studies compared with CS.38,39 The proposed risk is due to reduced oesophageal motility combined with lack of patient swallowing during the procedure. Taken together, GA may lead to prolonged thermal energy exposure to a single location of the oesophagus.38 However, the few studies currently available limits the clinical implications of these findings.
Oesophageal cooling
Oesophageal cooling previously involved the use of a cooled intra-oesophageal balloon45 while more recent reports inject a cooling solution when the LOT reaches a certain temperature. Neither of the included studies of oesophageal cooling definitively reduced EDOL prevalence. However, there was a trend towards decreased ulcer severity when initiating oesophageal cooling at a lower LOT (39°C instead of 43°C),10 and evidence to suggest that the maximum LOT reached is lower.17 The different ablative systems, overall ablation times and approaches to measuring EDOL limits comparability. However, oesophageal cooling could mitigate the severity of oesophageal injury by reducing the thermal effects of the ablation energy source during the procedure.
Other preventive techniques
Some preventive techniques did not meet the scope of this systematic review due to lack of endoscopic follow-up data. Prophylactic PPIs may reduce the size of thermal-induced ulcers and relieve exacerbating factors such as gastro-oesophageal reflux.46 Despite its widespread use, AOF has been reported despite high-dose PPI therapy.47 On sensitivity analysis, there was no significant difference between studies that adopted routine post-ablation PPI and those that did not. Randomized trials are not possible due to the rarity of AOF and the extent of risk reduction achieved by PPIs post-ablation remains uncertain.
Localization of the oesophagus through imaging has also been considered in oesophageal injury prevention. Real-time imaging is important as the oesophageal course is dynamic and computed tomography performed even <24 h before ablation showed discordance with periprocedural electroanatomic mapping of the oesophagus.48 Yet traditional electroanatomic oesophageal imaging under fluoroscopic guidance still under-estimates the oesophagus location by >1 cm in up to one-half of patients.49 Intracardiac echocardiography could overcome this limitation by enabling real-time oesophageal localization, improving OTM positioning and revealing temperature increases through the detection of micro-bubbles.38,39 Controlled studies are warranted to evaluate the effectiveness of oesophageal localization techniques in routine practice.
Routine endoscopy post-AF ablation has been proposed as a potential screening measure, however, there are some limitations with this approach. First, not all patients with AOF present with an oesophageal ulcer on initial endoscopy. Other findings include erythema, thrombus, minor oesophageal defects, and no abnormalities in certain cases,42,50 although in these reports, endoscopy was mostly performed later and during the presentation of AOF (see Supplementary material online, Table S12). Second, the optimal timing for performing post-operative endoscopy with high sensitivity for detecting clinically significant EDOL remains uncertain. Third, routine endoscopy is not without its own procedural risks and should be avoided in patients who develop clinical features suspicious of AOF such as fever, chest pain, or neurological symptoms.2 An emergent contrast-enhanced computed tomography chest is warranted in such cases.51
We focused on general measures to prevent AOF applicable to various catheter ablation techniques in this review. Nevertheless, specific ablation techniques are important in addressing AOF risk. The few studies evaluating prevalence of post-ablation EDOL in second-generation cryoballoon have shown mixed results ranging from 3% to 12%.14,19 Power and lesion time parameters can be adjusted although no definitive reduction in EDOL has been shown while maintaining ablation efficacy.39 Current guidelines recommend reduced power when ablating the LA posterior wall,9 although one study demonstrated no significant difference in power settings between patients with or without oesophageal ulceration using conventional RF ablation.18 There may be a role for low-flow irrigation RF catheters that can produce lesions limited to the endocardial surface with epicardial sparing compared with high-flow catheters.52 Additionally, contact force (CF) sensing provides insight into lesion transmurality and ablation efficacy. However, AOF has been reported in both commercially-available CF platforms, and there is a possibility that improved contact could increase risk of AOF.53 The incidence of EDOL was reported to be low in one single-arm retrospective study which performed point-by-point PVI using CF without the use of OTM.54
Limitations
There are several inherent limitations to our analysis that should be considered. First, only studies with endoscopic follow-up were included to enable a standard of comparability between studies. Second, there was significant heterogeneity between the included studies with regards to ablation protocols including type of system, maximum temperature/power settings, lesion time, and OTM type (if used). While this reflects the varied approaches to AF ablation worldwide, certain uniform procedural aspects are needed for comparability and to further research into this complication. Similarly, the definition of EDOL varied between individual studies (e.g. accounting for iatrogenic injury and different stratification of oesophageal lesion severity) and standard definitions of reporting are needed for accurate estimation of EDOL prevalence. Third, while all studies were prospectively-designed, only two studies were randomized and the overall quality of studies according to the Newcastle–Ottawa Scale was variable. There is a critical need for high quality randomized data to achieve clarity in the prevention of AOF.
Conclusion
Oesophageal injury following AF ablation as detected by endoscopy is relatively common, with the pooled prevalence of EDOL and ulcerated EDOL at 11% and 5%, respectively. The benefit of OTM remains unclear and randomized controlled data are needed before evaluation of specific OTM components. Although other techniques including oesophageal displacement and cooling have been trialled, none have conclusively demonstrated an ability to prevent EDOL at present. General anaesthesia may be associated with an increased risk of EDOL compared with CS. Further research is critically needed to develop preventive measures for this serious post-ablation complication.
Conflict of interest: H.S.L. reports having received research funding from St Jude Medical. P.S. reports having served on the advisory board of Biosense-Webster, Medtronic, St Jude Medical, Sanofi-Aventis and Merck, Sharpe and Dohme. P.S. reports having received lecture fees from Biosense-Webster, Medtronic, St Jude Medical, Boston Scientific, Merck, Sharpe and Dohme, Biotronik and Sanofi-Aventis. P.S. reports having received research funding from Medtronic, St Jude Medical, Boston Scientific, Biotronik and Sorin.
