Barrett’s esophagus (BE), a complication of gastroesophageal reflux disease, remains one the highest risk factors for esophageal adenocarcinoma.¹ Esophagogastroduodenoscopy with histopathology detects and establishes the diagnosis of BE.² Histologically, BE is characterized by lower esophageal squamous epithelium replacement by columnar epithelium.³ A high index of suspicion is crucial for prompt diagnosis of BE and initiation of treatment. Atypia in BE is classified as: low-grade dysplasia, indefinite for dysplasia, and high-risk for dysplasia. The management of BE with low-grade dysplasia includes endoscopic surveillance or endoscopic eradication therapy. However, BE with high-grade dysplasia requires definitive endoscopic treatment, such as endoscopic resection, radiofrequency ablation (RFA), argon plasma coagulation (APC), or cryotherapy.⁴

The current preferred therapy for ablation is RFA, but it is costly and has a high recurrence rate.^5,6 Another study reported a recurrence rate of 29.1% on long-term follow-up after RFA.⁷ APC is an alternative method for the management of BE. However, due to variable depth of ablation and increased risk of stricture formation, perforation, and buried glands, its use is limited to short-segment BE.

Hybrid-APC is a novel technique that combines APC with submucosal injection for the eradication of BE. Hybrid APC is performed by injecting aqueous solution submucosally with an endoscope needle prior to coagulation, which might decrease the risk of complications associated with classic APC, such as stricture formation.⁸ Hybrid APC is a promising method for the treatment of BE with low-grade and high-grade dysplasia.

However, data on the efficacy and safety of APC for BE are limited. Hence, this systematic review and meta-analysis was conducted to evaluate and compare the efficacy and safety of various types of APC in the management of BE.

1. Information sources and search strategy

A comprehensive search of all suitable studies was conducted using the databases of MEDLINE, EMBASE, and Scopus from January 2000 to November 2022. The keywords used were: Barrett AND Argon AND (Metaplasia OR Dysplasia). Manual searching of reference lists of the included studies was also undertaken. The study methodology was designed and executed to adhere to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.⁹

2. Study selection

The present analysis included prospective and retrospective studies fulfilling the following criteria: (1) study population–patients with BE with or without dysplasia; (2) intervention–APC; (3) outcomes–efficacy and safety. Two reviewers independently assessed each study’s title and abstract in line with the aforementioned selection criteria. A third reviewer resolved any differences. Studies with sequential use of APC after endoscopic resection, case series with fewer than five patients, and those with insufficient or irrelevant clinical data were also excluded.

3. Data extraction and quality assessment

The data extraction was done by two separate reviewers, and a third reviewer settled any disputes. Data were collected under the following headings: study author and year, country of study, study design, number of patients, age and sex distribution, details of the lesion, type of APC used, efficacy, and adverse events (AE). Two independent reviewers assessed the quality of the included studies, a scale modified from the Newcastle-Ottawa Scale for cohort studies.¹⁰ A third independent individual was consulted in case of any discrepancy.

4. Statistical analysis

Using a random-effects inverse-variance model, the pooled proportions were calculated. When the incidence of an outcome in a study was zero, a continuity adjustment of 0.5 was applied before statistical analysis. The heterogeneity of the studies was assessed using Cochran's Q test and I² statistics. Significant results were defined as either an I² value >75% or a p-value of Q test <0.1. Visual inspection of funnel plots was used for publication bias assessment. In order to examine each research's impact on the total effect-size estimate and identify influential studies, the sensitivity analysis was carried out using a leave-one-out meta-analysis, in which one study is eliminated at each analysis. Stata software version 17 (StataCorp., College Station, TX, USA) was used for statistical analysis.

1. Baseline characteristics with quality assessment

A total of 658 studies were identified with the above search strategy, of which 563 were screened after the removal of duplicates. Finally, 38 studies with 1,427 patients were included in the meta-analysis. Fig. 1 shows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses flowchart for study identification and selection process. Table 1 summarizes the baseline characteristics of the included studies. Twenty studies included standard APC (ranging from 30 to 60 W),^11-30 eight studies included high-power APC (>60 W),^31-38 two included mixed wattage,^39,40 and eight studies included on hybrid APC.^41-48 Among studies reporting high-power APC, one used 65 to 70 W,¹¹ two used 70 W,^35,38 four used 90 W,^32,33,36,37 and one used 150 W.³⁴ Supplementary Table 1 shows the detailed study quality analysis using a modified Newcastle-Ottawa Scale. Thirteen studies were of high quality, 18 were of medium quality, and seven were of low quality.

Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses flowchart for the study’s identification, inclusion, and selection process. APC, argon plasma coagulation; EMR, endoscopic mucosal resection.

Table 1 . Baseline Characteristics of the Included Studies.

Author (year)	Country	Study design	No. of patients	Age, yr*	M/F	Mean length, cm*	Dysplasia	No. of sessions*	Follow-up, mo*
Standard APC (≤60 W)
Morris et al. (2001)11	UK	Prospective	55	54.2	-	6.06±3.15	9 HGD, 9 LGD	3.02±1.69	38.5±14.6
Basu et al. (2002)12	UK	Prospective	50	61.4±11.5	-	5.9±3.1	0	4 (1–8)	14
Kahaleh et al. (2002)13	Belgium	Prospective	39	63.6±8.7	30/9	4.7±2.2	7 LGD	3 (1–4)	36 (12-46)
Familiari et al. (2003)14	Italy	Prospective	35	-	-	-	0	Median 2	49.5 (24-60)
Morino et al. (2003)15	Italy	Prospective	23	27–78	12/11	3.8 (2–8)	2 LGD	Mean 3.1 (2–6)	31.9 (16–45)
Pagani et al. (2003)16	Italy	Prospective	94	51.4 (17–82)	68/26	2.52 (0.5–9)	0	Mean 3 (1–5)	26 (6–45.9)
Ackroyd et al. (2003)17	Australia	RCT	20	46.5 (36–69)	15/5	4 (2–13)	2 LGD	Median 3 (2–6)	12
Dulai et al. (2005)18	California	RCT	26	58±11	21/5	4±1.5	0	3.8±1.6	36
Ragunath et al. (2005)19	UK	RCT	13	58 (35–79)	21/5	5 (3–9)	23 LGD, 3 HGD	Mean 2	12
Sharma et al. (2006)20	USA	RCT	19	65 (32–84)		4 (2–6)	3 LGD	Median 3 (2–6)	24
Ferraris et al. (2007)21	Italy	Prospective	96	57.1 (21–79)	70/26	4 (2.5–11)	0	Median 3.2	36 (12–98)
Mörk et al. (2007)22	Germany	Prospective	25	55 (37–73)	18/7	3.8 (2–11)	2 LGD	Median 4 (1–12)	30
Migaczewski et al. (2009)23	Poland	prospective	30	54.9	19/11	3.06 (2–5)	3 LGD, 5 HGD	Mean 1.3	12
Bright et al. (2008)24	Australia	RCT	26	57 (41–70)	20/6	3 (2–13)	1 LGD	Median 2 (1–6)	12
Zhang et al. (2009)25	China	prospective	18	55	12/6	Median 2.1	1 LGD	Mean 1.3 (1–3)	11.8 (4–15)
Sie et al. (2013)26	Australia	RCT	64	63.5	64/0	4.3 (1–3)	LGD	Median 3 (1–6)	84
Castaño et al. (2014)27	Colombia	Prospective	33	62.4±5	22/11	3±2	0	2±1	18
Milashka et al. (2014)28	Belgium	Prospective	32	64 (46–76)	26/6	4.5 (3–11)	5 LGD	Mean 3 (1-5)	16 yr
Szachnowicz et al. (2016)29	Brazil	Retrospective	13	52 (32–72)	6/7	-	0	Median 3.5	9 (1–18) yr
Michopoulos et al. (2022)30	Greece	Prospective	22	56.3±12.9	-	4.41±2.91	14 LGD, 8 HGD	Mean 3 (1–12)	-
High-power APC (>60 W)
Pereira-Lima et al. (2000)31	Brazil	Prospective	33	55.2 (21–84)	21/12	4.05	14 LGD, 1 HGD	Mean 1.94 (1–4)	10.6 (6–18)
Schulz et al. (2000)32	Germany	Prospective	73	55 (28–77)	45/28	4.0 (1–12)	0	Median 2 (1–5)	12 (2–51)
Van Laethem et al. (2001)33	Belgium	Prospective	10	74.2 (50–88)	7/3	5.8±2.7	4 HGD,3 TIS	3.3±1.5	24 (12–36)
Tigges et al. (2001)34	Germany	Prospective	30	53.5 (31–77)	23/7	3 (1–10)	0	Median 2 (1–7)	12
Attwood et al. (2003)35	UK	Prospective	29	64 (43–85)	-	6 (1–12)	29 HGD	Median 2 (1–13)	37 (7–78)
Madisch et al. (2005)36	Germany	Prospective	73	55±12	45/28	4 (1–12)	0	Median 2	51 (9–85)
Pedrazzani et al. (2005)37	Italy	Prospective	25	61.7 (34–74)	-	3.4	0	Mean 1.6	26.3
Brasil et al. (2010)38	Brazil	Prospective	30	49.8 (45–60)	25/5	3.2 (1–10)	0	Mean 2 (1–6)	18 (1–60)
Combined
Dotti et al. (2009)39	Australia	Prospective	16	54±11.1	12/4	3.6±3.1	12 HGD, 4 ADC	-	20
Wronska et al. (2021)40	Poland	RCT	71	62 (51–72)	54/17	Median 4	0	Median 2 (1–2)	24
Hybrid APC
Kashin et al. (2016)41	Russia	Prospective	12	54 (40–68)	-	Median 2	12 LGD	Mean 2.5 (1–4)	4.5
Manner et al. (2016)42	Germany	Prospective	50	62.4±8.4	46/4	5±3	0	3.5±2.4	3
Linn et al. (2020)43	USA	Retrospective	27	66.5	22/5	Mean 2.1	8 LGD, 4 HGD	-	6
Trindade et al. (2020)44	USA	Retrospective	5	51–76	3/2	5–10	1 LGD, 1 HGD	2 (2–3)	6
Kroupa et al. (2021)45	Czech Republic	Prospective	24	60	19/5	All <5	-	Mean 1.5	-
Shimizu et al. (2021)46	USA	Retrospective	22	67.8	18/4	1–8	7 HGD, 4 LGD	Mean 1.2	4.5
Staudenmann et al. (2021)47	Australia	Prospective	11	68.2±8.0	8/3	4.5±4.0	5 LGD, 4 HGD, 2 TIS	2.7±1.1	28.8±4.4
Knabe et al. (2022)48	Multicentric	Prospective	154	64.2 (42–84)	133/21	4.41 (1–13)	26 LGD, 11 HGD	2.69 (1–5)	24

M, male; F, female; APC, argon plasma coagulation; RCT, randomized controlled trial; HGD, high-grade dysplasia; LGD, low-grade dysplasia; TIS, tumor in situ..

*Mean±SD or median (range)..

2. Complete response of intestinal metaplasia

A total of 34 studies (n=1,247) reported on clearance rate of intestinal metaplasia (CR-IM) with APC in BE. The pooled event rate for CR-IM with APC in BE was 86.8% (95% confidence interval [CI], 83.5% to 90.2%; I²=86.2%) (Fig. 2). On subgroup analysis, the CR-IM rate with standard, high-power, hybrid APCs were 78.5% (95% CI, 71.7% to 85.4%), 98.6% (95% CI, 96.1% to 100.0%), and 89.6% (95% CI, 83.8% to 95.5%), respectively. The CR-IM rate was significantly higher with both hybrid APC (p=0.016) and high-power APC (p=0.000) compared to standard APC. On comparing the CR-IM of standard APC in randomized controlled trials (RCT) versus observational studies, there was no difference (74.6% [95% CI, 66.0% to 83.2%] vs 80.6% [95% CI, 73.1% to 88.2%], p=0.304). On subgroup analysis of those undergoing standard APC, CR-IM was lower in the group with dysplasia compared to studies without any dysplasia (74.0% [95% CI, 68.0% to 88.0%] vs 88.0% [95% CI, 79.5% to 96.5%], p=0.009). However, no such difference was observed for those undergoing high-power or hybrid APC.

Figure 2. Forest plot for the pooled event rate for the clearance rate of intestinal metaplasia with argon plasma coagulation (APC) with subgroup analysis. DL, DerSimonian and Laird method; CI, confidence interval.

3. Adverse events

A total of 27 studies (n=984) reported on the incidence of AE with APC in BE. The pooled event rate for AE with APC in BE was 22.5% (95% CI, 15.3% to 29.7%; I²=93.9%). On subgroup analysis, the incidence of AE with standard, high-power, hybrid APCs were 17.2% (95% CI, 9.3% to 25.1%), 25.5% (95% CI, 9.7% to 41.3%), and 13.6% (95% CI, 2.6% to 24.7%), respectively. However, there was no significant difference in overall AE between standard APC and high-power APC (p=0.359) or hybrid APC (p=0.605). The pooled incidence of serious AE with APC in BE was 0.4% (95% CI, 0.0% to 1.0%; I²=0.0%), with no significant difference between the groups.

1) Stricture

All the included studies (n=1,238) reported on the development of stricture with APC in BE on follow-up. The pooled event rate for development of stricture was 1.7% (95% CI, 0.9% to 2.6%; I²=0.0%). On subgroup analysis, the incidence of stricture with standard, high-power, hybrid APCs were 1.0% (95% CI, 0.0% to 2.1%), 3.0% (95% CI, 0.9% to 5.1%), and 2.6% (95% CI, 0.7% to 4.6%), respectively, without any significant difference between the groups.

2) Pain

A total of 20 studies with 761 patients reported the incidence of pain following APC. The pooled incidence of pain after APC for BE was 25.5% (95% CI, 15.7% to 35.3%; I²=92.6%). The pooled rate was comparable between standard (20.7%; 95% CI, 15.4% to 26.0%), high-power (29.7%; 95% CI, 5.9% to 53.5%), and hybrid APC (21.4%; 95% CI, 15.7% to 35.3%).

4. Recurrence

Overall, 26 studies with 861 patients reported on the recurrence of IM after initial CR-IM. The pooled event rate for recurrence was 16.1% (95% CI, 10.7% to 21.6%; I²=88.8%). On subgroup analysis, the incidence of stricture with standard, high-power, hybrid APCs were 21.2% (95% CI, 12.2% to 30.2%), 7.3% (95% CI, 0.4% to 14.2%), and 14.7% (95% CI, 0.0% to 30.0%), respectively. The recurrence rate with standard APC was significantly higher compared to high-power APC (p=0.016) but comparable to hybrid APC (p=0.475). Table 2 summarizes the pooled event rates of various outcomes along with subgroup analysis based on the type of APC.

Table 2 . Summary of the Findings for Various Outcomes with APC for Barrett’s Esophagus with Subgroup Analysis.

Parameter	Overall	Standard APC	High-power APC	Hybrid APC
CR-IM
% (95% CI)	86.8 (83.5–90.2)	78.5 (71.7–85.4)	98.6 (96.1–100.0)	89.6 (83.8–95.5)
I2, %	86.2	90.9	49.2	50.9
Adverse events
% (95% CI)	22.5 (15.3–29.7)	17.2 (9.3–25.1)	25.5 (9.7–41.3)	13.6 (2.6–24.7)
I2, %	93.9	90.7	91.6	85.1
Serious adverse events
% (95% CI)	0.4 (0.0–1.0)	0.1 (0.0–0.9)	0.4 (0.0–1.6)	1.1 (0.0–2.5)
I2, %	0.0	0.0	0.0	0.0
Stricture
% (95% CI)	1.7 (0.9–2.6)	1.0 (0.0–2.1)	3.0 (0.9–5.1)	2.6 (0.7–4.6)
I2, %	0.0	0.0	0.0	0.0
Recurrence
% (95% CI)	16.1 (10.7–21.6)	21.2 (12.2–30.2)	7.3 (0.4–14.2)	14.7 (0.0–30.0)
I2, %	88.8	88.3	74.3	87.1

APC, argon plasma coagulation; CR-IM, clearance rate for intestinal metaplasia; CI, confidence interval..

5. Publication bias, sensitivity analysis, and meta-regression

There was significant publication bias (Supplementary Fig. 1) and small-study effect (Supplementary Table 2) for the outcomes of CR-IM, all-cause AE, and serious AE. Leave-one-out analysis did not show a significant change in the overall pooled event rates of various outcomes (Supplementary Figs 2-4). Meta-regression analysis showed that the median duration of follow-up, but not the length of BE, was a significant contributor to the heterogeneity for the outcome of IM recurrence (p=0.0483). Meta-regression also showed that the length of the BE was associated with a higher recurrence with standard APC (p=0.0230) (Fig. 3) but not with high-power or hybrid APC.

Figure 3. Bubble plot showing a positive association of the mean length of the Barrett’s esophagus with recurrence. CI, confidence interval.

BE is a significant risk factor contributing to the development of esophageal adenocarcinoma. As the incidence of esophageal adenocarcinoma continues to rise, it is important to understand the treatment modalities of BE and identify those therapies which will effectively and safely treat this condition.⁴⁹ Ablation techniques, including RFA and APC have the advantages of being a simpler outpatient operation with lower costs and lower risks compared to resection techniques. APC continues to be the most used ablation technique due to its widespread availability. Hence, the present meta-analysis was conducted to investigate the role of APC with respect to response rates, complications, and recurrence rates, along with subgroup analysis based on the type of APC used.

In the present analysis, the pooled rate of CR-IM with APC was 86.8%. Patients who received high-power APC had the highest rates of response (98.6%), followed by hybrid-APC (89.6%). Both high-power APC and hybrid APC had significantly higher CR-IM than standard APC (78.5%). Similarly, high-power APC had the lowest recurrence rate of 7.3%, as compared to those who received hybrid-APC (14.7%) and standard APC (21.2%). Thus, high-powered APC was found to have the highest efficacy with the lowest recurrence rates. Certainly, the increased wattage used in high-power APC could be advantageous to abnormal mucosal destruction and the treatment of BE. In a meta-analysis on the efficacy of RFA for BE, the overall pooled rate of CE-IM was 88.17 % (95 % CI, 88.1% to 88.2%).⁵ Similarly, the pooled CR-IM with rate with cryotherapy for BE was 64.2% (95% CI, 52.9% to 74.8%).⁵⁰ Thus, hybrid APC may lead to at least similar efficacy results with those achieved using RFA and may be superior to cryotherapy. However, due to heterogeneity of the population, further studies are required comparing APC with RFA in BE.

The pooled event rate for AE with APC in BE was 22.5%. Those who received high-power APC showed the highest occurrence of AE (25.5%) compared to standard-APC (17.2%) and hybrid-APC (13.6%). Although the rates varied, the adverse effect profile was very similar for high-power, standard, and hybrid-APC. These side effects most commonly included retrosternal pain, dysphagia, odynophagia and sore throat with mild fever. Pain was most commonly mentioned as a side effect (25.5%) but was transient and resolved within 1 to 2 weeks regardless of the type of APC used. The reported incidence of AE with RFA and cryotherapy for BE were 8.8% (95% CI, 6.5% to 11.9%) and 12.2%, respectively.^50,51 However, this higher incidence of AE with APC is primarily due to a higher incidence of self-limited pain. The pooled rate of post-procedure pain with RFA and cryotherapy were 3.8% (95% CI, 1.9% to 7.8%) and 2.7%, respectively.^50,51 Thus, the pooled incidence of AE with APC is higher compared to those reported with RFA and cryotherapy. However, the majority of these were minor AEs. Serious AE were rare but did occur in 1.1% of those who received hybrid-APC, 0.4% of those who received high-APC and 0.1% of those who received standard-APC. Amongst high-powered APC and standard-APC, serious adverse included pneumomediastinum without perforation, esophageal perforation and ulcer formation resulting in hemorrhage and need for endoscopic hemostasis.^11,37

While APC is used widely for the treatment of BE, its usage is restricted by the risk of stricture development and perforation. Avoiding a high-power setting (reducing the depth of coagulation) may reduce the incidence of AE theoretically, but also reduces the efficacy of the treatment. An ex-vivo research, which controlled wattage for both standard and hybrid APC, reported that coagulation depth was reduced by half when a protective submucosal fluid was injected for the hybrid approach.⁸ The authors concluded that less thermal injury should result in fewer strictures when using the hybrid technique. However, the pooled rate of stricture formation with standard APC (1.0%) was comparable with those receiving high-power APC (3%) and hybrid APC (2.6%) in the present analysis. The pooled rate of stricture development with RFA has been reported as 5.6% (95% CI, 4.2% to 7.4%),⁵¹ while that with cryotherapy has been reported as 7.3%.⁵⁰ In a study comparing ablation after endoscopic resection, Knabe et al.⁵² reported a stricture rate of 2% with hybrid APC compared to 13% with RFA. Thus, irrespective of the type of APC, the overall incidence of strictures with APC remains low compared to other ablative therapies.

To the best of our knowledge, this is the first systematic review and meta-analysis to analyze and compare the safety and efficacy of various types of APC for the management of BE. This analysis provides valuable insight into the established and emerging techniques of APC, along with gaps in the present knowledge. Despite this, there are multiple limitations to the present analysis warranting discussion. First, most of the studies were retrospective, leading to selection and reporting bias. Second, there was significant heterogeneity for all the outcomes, which may have been due to significant differences in the study design, lesion type, length of the lesion, intervention, and follow-up duration. Third, there was a significant publication bias in the analysis, indicating that only studies that had statistically significant positive results might have been published, leaving out the statistically insignificant or negative studies. Fourth, the amount of “high-powered” wattage varied in several studies; 150 W versus 90 W versus 65 to 70 W. Therefore, it may be useful to investigate further what cut-off of wattage used in APC may be needed to produce efficacious results so that harmful consequences of high-powered wattage can be minimized. Similarly, there was a variation in the timing of the histologic evaluation of CR-IM. This varied from 1 to 6 months after complete endoscopic ablation. This has to be standardized for the proper evaluation of novel treatment modalities.

This present meta-analysis demonstrates that APC for BE with or without dysplasia can achieve CE-IM rates similar to those reported with RFA. High-power and hybrid APC have a higher success rate with a lower risk of recurrence compared to standard APC. APC is associated with a high incidence of post-procedural self-limited pain but with a low rate of stricture formation. Further randomized studies comparing hybrid APC with standard APC and RFA are required to compare the efficacy and AE between these modalities and decide the optimal therapy for BE.

No potential conflict of interest relevant to this article was reported.

Study concept and design: M.K., S.G. Data acquisition: L.K., M.T., S.G. Data analysis and interpretation: M.K., K.S., M.T., S.G. Drafting of the manuscript: M.K., M.W., S.G. Critical revision of the manuscript for important intellectual content: M.K., L.K., K.S., M.T., M.W., S.G. Statistical analysis: L.K., S.G. Administrative, technical, or material support: M.K., L.K., K.S., M.T., M.W., S.G. Study supervision: M.K., S.G. Approval of final manuscript: all authors.

Supplementary materials can be accessed at https://doi.org/10.5009/gnl230094.