Outcome and risk factors of ulcer healing after gastric endoscopic submucosal dissection: A systematic review and meta-analysis

Abstract

BACKGROUND

Endoscopic submucosal dissection (ESD) is widely utilized for the treatment of large adenomas, submucosal lesions, and early gastric cancer. A significant artificial ulcer typically forms after ESD. Delayed or incomplete healing of these ulcers can result in complications such as delayed bleeding and perforation. However, a comprehensive review of the outcomes and risk factors related to ulcer healing following ESD is currently lacking.

AIM

To assess ulcer healing outcomes and identify risk factors associated with delayed ulcer healing.

METHODS

Databases retrieved by computer include PubMed, EMBASE, The Cochrane Library, Web of Science, China National Knowledge Infrastructure, Wan Fang Data, and VIP. The study collects reports on ESD post-surgical ulcer healing outcomes and risk factors, using Stata 16.0 and RevMan 5.0 software for meta-analysis.

RESULTS

Our analysis included 12 studies, involving a total of 3430 patients. The meta-analysis revealed an overall healing rate of 65.55% for ulcers following ESD [odds ratio (OR) = 2.71; 95% confidence interval (CI): 2.45-3.00]. The healing rate within eight weeks was 48.32% (OR = 0.76; 95%CI: 0.35-1.66), while the rate beyond eight weeks was 88.32% (OR = 6.73; 95%CI: 3.82-11.87). Risk factors included Helicobacter pylori (H. pylori) infection (OR: = 5.32; 95%CI: 1.90-14.87; P = 0.001), ulcer size (OR = 2.08; 95%CI: 1.19-3.61; P = 0.01), lesion site (OR = 2.08; 95%CI: 1.19-3.11), and pathological type (OR = 1.64; 95%CI: 1.06-2.52). Diabetes (OR = 0.56; 95%CI: 0.05-5.80; P = 0.63) and duration of operation (OR = 1.00; 95%CI: 0.99-1.01; P = 0.96) were not significant factors.

CONCLUSION

The healing rate of ulcers following ESD is high after eight weeks. Risk factors affecting the healing process include H. pylori infection, ulcer size, lesion site, and pathological type.

Key Words: Endoscopic submucosal dissection; Gastric ulcer; Artificial ulcer; Outcome; Risk factors; Systematic review; Meta-analysis

Core Tip: The factors influencing the healing of artificial ulcers created by endoscopic submucosal dissection (ESD) remain unclear. This study examines ulcer healing outcomes and the risks associated with delayed healing. The healing rate of ulcers following ESD is high after eight weeks. Treatment for eight weeks or longer may support complete ulcer healing. A meta-analysis of risk factors, based on multivariate regression analysis, indicates that ulcer size, lesion site, Helicobacter pylori infection, and pathological type are key variables affecting the healing process, while diabetes and the duration of the operation do not have a significant impact.

INTRODUCTION

The endoscopic resection technique known as endoscopic submucosal dissection (ESD) was initially developed in the late 1990s and early 2000s[1]. ESD has been widely employed for large adenomas, submucosal lesions, and early stomach cancer due to ongoing research and procedural refinements[2,3]. This method allows for the complete removal of the lesion, increasing the resection rate. In general, ESD technology is effective and safe. However, complications such as bleeding and perforation can occur when lesions are excised with ESD. Unlike traditional ulcer formation mechanisms in the clinic, a significant artificial ulcer usually forms post-procedure. The risk of bleeding and perforation increases when these ulcers take longer to heal[4-6]. While it has been proposed that all artificial ulcers created by electrosurgical drainage (ESD) will heal within 8 weeks[7], recent research indicates that 5%-20% of individuals with post-ESD ulcers do not heal within this timeframe[8,9]. Consequently, further research on ESD ulcers and their healing is needed, as well as the development of a treatment plan for ESD ulcers. Relevant studies have investigated and examined the factors contributing to the slower healing of artificial ulcers after ESD[2,5,10-12]. However, the results obtained are inconsistent or conflicting due to varying standards across studies. It remains unclear what influences the healing of artificial ulcers created by ESD. Currently, there is no published systematic review or meta-analysis on the ulcer healing rate or the risk factors for delayed ulcer healing following endoscopic therapy. Our goals are to perform a meta-analysis to gain a thorough understanding of the ulcer healing rate, identify common risk factors for delayed ulcer healing, and provide clinicians with a foundation for management and treatment.

MATERIALS AND METHODS

This systematic review and meta-analysis adhered to the standards set forth by the preferred reporting items for systematic reviews and meta-analyses[13] and the meta-analysis of observational studies in epidemiology[14]. Additionally, the study was registered with the international prospective register of systematic reviews under the registration number No. CRD42024543288.

Literature sources and search strategy

The following search terms were utilized: (“ulcer” OR “scar” OR “mucosal healing” OR “ulcer healing”), (“endoscopic submucosal dissection” OR “ESD”), and (“risk factors” OR “influence factor”). We searched MEDLINE via PubMed, Web of Science, EMBASE, Cochrane Library, China National Knowledge Infrastructure (CNKI), Wan-fang database, and VIP database from the inception until April 17, 2024. All of the search terms used in each database are shown in Supplementary Table 1.

Inclusion and exclusion criteria

Studies were selected for inclusion based on the following criteria: (1) Study population: Individuals with post-ESD; (2) Outcome indicators: Ulcer healing rate and/or risk factors after ESD; and (3) Multivariate regression analysis of risk factors. The following studies were excluded: (1) The ulcer healing indicator is not in the scarring stage (S stage); (2) Duplicate publication, conference abstracts, reviews, comments; (3) Data missing or univariate analysis or not available odds ratio (OR), 95% confidence intervals (CIs) or not transformable raw data; (4) Theses and case reports; (5) Animal studies; and (6) Non-English or non-Chinese articles.

Data extraction and quality assessment

Various categories of data were extracted from all included studies: Study characteristics such as the first author’s name, publication year, country of origin, study design, and sample size. Outcome indicators: Ulcer healing rate, risk factors with prolonged ulcer healing, OR, and 95%CIs. The Cochrane Collaboration’s tool was employed to evaluate the quality of randomized controlled trials (RCTs) and their post hoc analyses. The Newcastle-Ottawa scale checklist[15] was used to assess the quality of case-control and cohort studies. Every item on this checklist was given a score of either 0 or 1. The classifications for low, medium, and high quality were 0-5, 6-7, and 8-9, respectively. The publications that satisfied the minimum score requirement of 7 to be considered for the meta-analysis process were incorporated into the data extraction and meta-analysis procedure. Two researchers conducted independent data extraction and assessed the quality of the eligible studies. Any disagreements were addressed through by consensus, with a third researcher resolving any further disagreements to reach a conclusion.

Evaluation criteria for ulcer healing

A six-stage framework was employed to categorize the phases of gastric ulcers: (1) A1 stage: Characterized by a mucus-coated ulcer exhibiting mild elevation due to edema; (2) A2 stage: Ulcer with diminished edema, still coated in mucus, and displaying more defined borders in comparison to the A1 stage; (3) H1 stage: Open wound with less than 50% epithelial regeneration, which may or may not feature converging folds; (4) H2 stage: An ulcer presenting a mucosal breach, predominantly covered by regenerating epithelium; (5) S1 stage: Absence of a mucosal break, revealing a red scar with irregular epithelialization; and (6) S2 stage: Complete re-epithelialization accompanied by the presence of a white scar. The process of ulcer healing was delineated through the scarring stages as per the Sakita-Fukutomi classification system[16-19]. The healing stage or ulcer shrinkage indicated that the ulcer had not yet achieved complete healing.

Statistical analysis

Data analyses were executed utilizing review manager 5.4 (Cochrane Collaboration, Oxford, United Kingdom) and STATA (version 15, Stata Corp LLC, College Station, TX, United States) for the meta-analysis encompassing all selected studies. An initial heterogeneity analysis was conducted on the included articles. If the results indicated I² value of less than 50%, a fixed effect model was employed for the meta-analysis; conversely, a random effect model was utilized. OR and 95%CIs were computed, and sensitivity analyses were performed to assess the stability of the findings. A P value of < 0.05 was considered statistically significant.

RESULTS

Study selection and characteristics

A sum of 2631 articles was collected: 454 from PubMed, 1172 from EMBASE, 445 from Web of Science, 92 from Cochrane Library, 154 from CNKI, 305 from Wan-fang database, and 9 from VIP database. After removing duplicates, the titles and abstracts of 2085 papers were screened, and 13 studies met our inclusion criteria. One study was excluded due to incorrect 95%CI data, leaving 12 full-text studies[9,16,18,20-28] for the meta-analysis. Figure 1 illustrates the flowchart outlining the search and selection process. Among the 12 included publications, 7 were cohort studies, 3 were case-control studies, 4 were RCTs, and 1 was mixed (cohort + RCT). All studies demonstrated a medium to high level of quality and collectively included a total of 3430 participants. Six risk factors were incorporated in the multifactorial regression analysis, with data extraction from at least two publications mentioning the same risk factor: (1) Helicobacter pylori (H. pylori) infection; (2) Ulcer size; (3) Lesion site; (4) Diabetes; (5) Pathology of the lesion; and (6) Duration of operation. The fundamental characteristics of the included studies are presented in Table 1, while the quality assessment is detailed in Supplementary Table 2 and Supplementary Table 3, as well as Supplementary Figure 1.

Figure 1  Flowchart of the search and selection process.

Table 1 Study selection and characteristics.

Ref.	Country	Total (n)	Study type	Outcomes: Ulcer healing rate		Risk factor
				< 8 weeks (%)	≥ 8weeks (%)
Gong et al[20]	China	196	Prospective cohort	79.59	-	2
Fei et al[21]	China	200	Case control	78.40	-	1,2,3
Huang et al[22]	China	112	Case control	-	-	3,4
Kobayashi et al[18]	Japan	170	Randomized controlled study	17.10	-	2
Lee et al[23]	Korea	192	Retrospective cohort	61.8	80.5	2
		95	Randomized controlled study	42.6	83.3
Lim et al[9]	Korea	1680	Retrospective cohort	-	94.3	2,4,5
Nakamura et al[16]	Korea	109	Randomized controlled study	13.8	84.4	2,3,5
Otsuka et al[24]	Japan	132	Prospective cohort	-	92.1	1,2,3,6
Shimozato et al[25]	Japan	73	Retrospective cohort	-	78.1	2,3,6
Shen et al[26]	China	152	Prospective cohort	69.09	-	2,3,5,6
Shao et al[27]	China	152	Case control	-	78.1	1
Yamasaki et al[28]	Japan	167	Retrospective cohort	24.5	-	2,3

¹Helicobacter pylori infection.

²Ulcer size.

³Lesion site.

⁴Diabetes.

⁵Pathology of lesion.

⁶Duration of operation.

Healing of ulcers was defined as the ulcer scarring stages.

Outcome of ulcer healing

The ulcer healing rate was used to determine the ulcer healing outcome. The pooled analysis revealed a total ulcer healing rate of 65.55 percent; the heterogeneity test showed (P < 0.00001; I² = 98%). A sensitivity analysis was performed by sequentially excluding individual studies and re-evaluating the remaining studies to determine if the findings could have been substantially influenced by any single study. The results indicated that heterogeneity remained consistent throughout the sensitivity analysis. A meta-analysis was performed utilizing the random effects model. P > 0.05, no statistical significance, (OR = 2.71; 95%CI: 2.45-3.00; P > 0.05), no statistical significance, Supplementary Figure 2A. Subgroup analysis included one subgroup < 8 weeks and another subgroup ≥ 8 weeks. The ulcer healing rate was 48.32% for the subgroup < 8 weeks, (OR = 0.76; 95%CI: 0.35-1.66; P > 0.05), Supplementary Figure 2B; for ≥ 8 weeks, the ulcer healing rate was 88.34%, (OR = 6.73; 95%CI: 3.82-11.87; P < 0.00001), Supplementary Figure 2C. The results showed a statistically significant variation in the rate of ulcer healing postoperatively, as assessed over an 8-week period (Table 2).

Table 2 Outcome of ulcer healing during different time periods.

Different durations	Studies	Patients (n)	OR	95%CI	P value	Heterogeneity
						χ2	df	P value	I2 (%)
Overall	11	3719	2.71	2.45-3.00	0.06	808.11	14	< 0.00001	98
< 8 weeks	7	1286	0.76	0.23-0.46	0.49	188.64	7	< 0.00001	96
≥ 8 weeks	6	2433	6.73	3.82-11.87	< 0.00001	74.58	6	< 0.00001	92

P value of < 0.05 was considered statistically significant. OR: Odds ratio; 95%CI: 95% confidence intervals; df: Degrees of freedom.

Risk factors

H. pylori infection: Three articles[21,24,27] reported the impact of H. pylori infection on the healing of ulcers following ESD. The test for heterogeneity showed (P = 0.005; I² = 81%). Subsequent analysis indicated that the primary contributor to heterogeneity was the study conducted by Fei et al[21], which featured a follow-up duration of less than 8 weeks. Except for this particular study, the other studies exhibited no significant statistical heterogeneity (P = 0.30; I² = 8%). Meta-analysis conducted utilizing a fixed-effects model indicated that H. pylori infection was a significant risk factor for ulcer healing, with an OR = 5.32, 95%CI: 1.90-14.87, P = 0.001 (Figure 2A).

Figure 2 Helicobacter pylori forest plots of various risk factors. A: Helicobacter pylori infection; B: Ulcer size; C: Lesion site; D: Pathology of lesion; E: Diabetes; F: Duration of operation. CI: Confidence interval.

Ulcer size: Ten articles[9,16,18,20,21,23-26,28] examined the impact of ulcer size on the healing process following ESD. The test for heterogeneity showed (P < 0.00001; I² = 92%). Using sensitivity analysis, the heterogeneity remained unchanged. Meta-analysis employing a random-effects model revealed that the size of the ulcer serves as a significant risk factor for the healing of ulcers postoperatively, with an OR = 1.13, 95%CI: 1.02-1.26, P = 0.02 (Figure 2B).

Lesion site: Seven articles[16,21,22,24-26,28] reported the effect of ulcer location (specifically, ulcers situated in the distal third of the stomach) on postoperative ulcer healing. The test for heterogeneity showed (P = 0.004; I² = 69%). Sensitivity analysis identified the study by Yamasaki et al[28], which had a follow-up period of 4 weeks and clinical selection bias, as the source of heterogeneity. Except for this study, no significant statistical heterogeneity was detected (P = 0.23; I² = 27%). Meta-analysis conducted utilizing a fixed-effects model indicated that ulcer location was a risk factor for postoperative ulcer healing, with an OR = 2.08, 95%CI: 1.19-3.61, P = 0.01 (Figure 2C).

Pathology of lesion: Three articles [9,16,26] examined the influence of the pathological characteristics of ulcer sites on the healing process following surgery. The test for heterogeneity showed (P = 0.18; I² = 42%). Meta-analysis conducted using a fixed-effects model indicated that the pathology of the ulcer was a significant risk factor for postoperative healing, with an OR = 1.64, 95%CI: 1.06-2.52, P = 0.03 (Figure 2D).

Diabetes: Two articles[9,22] reported the effect of diabetes mellitus on postoperative ulcer healing. The test for heterogeneity showed (P = 0.0003; I² = 92%). Meta-analysis was conducted utilizing a random effects model, which revealed that diabetes mellitus does not serve as a risk factor for the healing of postoperative ulcers, with an OR = 0.56, 95%CI: 0.05-5.80, P = 0.63 (Figure 2E).

Duration of operation: Three articles[24-26] examined the impact of surgical duration on the healing of postoperative ulcers. The test for heterogeneity showed (P = 0.27; I² = 23%). Meta-analysis was conducted utilizing a fixed-effect model, which demonstrated that the length of surgical procedures did not serve as a risk factor for the healing of postoperative ulcers, with an OR = 1.00, 95%CI: 0.99-1.01, P = 0.96 (Figure 2F).

Bias assessment

Using the effect indicator (OR) as the horizontal axis and the standard error as the vertical axis, a bias funnel plot was drawn (see Supplementary Figure 3). Except for the risk factors H. pylori infection and diabetes, which were included in fewer studies, making the Egger test unable to provide further bias evaluation, the remaining studies did not exhibit publication bias Table 3.

Table 3 Bias assessment: Egger's test.

	Outcome of ulcer healing			Risk factors
	Overall	< 8 weeks	≥ 8 weeks	Ulcer size	Lesion site	Pathology of lesion	Duration of operation
t value	-1.06	-0.46	-2.21	2.05	0.71	7.87	3.03
P value	0.311	0.661	0.078	0.074	0.520	0.080	0.203

DISCUSSION

This systematic review and meta-analysis incorporated findings from 12 studies concerning ulcer healing rates and the risk factors associated with delayed healing following ESD. The findings indicated that there was no statistically significant difference in the rates of ulcer healing over an 8-week period; however, a statistically significant difference was observed over 8 weeks. Of the six risk factors examined, H. pylori infection, ulcer size, lesion site, and the pathological characteristics of the lesion were linked to a prolonged healing process of the ulcer., while diabetes and duration of operation were not.

As the mucosal defect after ESD often reaches the submucosal layer, the resulting artificial ulcers are considered similar to peptic ulcers. The healing process of these ulcers is classified into phases A, H, and S, referencing the criteria for peptic ulcers[18]. Nevertheless, the ideal duration of treatment for ulcers induced by ESD remains undetermined. Most ulcers achieve complete healing within 4-8 weeks[29,30]. Park et al[31] proposed that a treatment duration of 4 weeks is adequate for the management of ulcers induced by ESD, as there was no significant difference observed in the healing of ulcers between 4-weeks and 8-weeks treatment periods. A prospective multicenter observational study demonstrated that both six-week and eight-week intervals were equally effective in promoting the healing of artificial ulcers following ESD[32]. However, 5%-20% of patients with post-ESD ulcers did not heal after more than eight weeks. Kakushima et al[8] reported that eight weeks of treatment were necessary for ESD-induced macroscopic ulcers, as four weeks of proton pump inhibitor treatment were insufficient[8,18,23]. The absence of a well-defined criterion could be the cause of these contentious outcomes. Potential affecting factors include pathology, lesion site, size, and H. pylori infection. The findings from the meta-analysis indicate a notable enhancement in the healing rate of ulcers over 8 weeks, as well as a statistically significant difference between the healing rate during this timeframe and the rate of non-healing. This suggests that extending the treatment of post-ESD ulcer patients to more than 8 weeks contributes to the complete healing of postoperative ulcerations of ESD.

Infection with H. pylori has been associated with the healing process of peptic ulcers. Nevertheless, there remains a contentious discussion regarding the influence of H. pylori infection on the healing of ulcers induced by medical interventions. Several studies have indicated that the eradication of H. pylori facilitates the healing process of ulcers[2,33]. Conversely, a multi-center trial has proposed that H. pylori infection does not influence the healing of ulcers following ESD[10,34]. Chronic infection with H. pylori was not found to be associated with a prolonged healing process of ulcers[9]. According to this meta-analysis, H. pylori infection is a risk factor for ulcers following ESD, and its eradication aids in ulcer healing. H. pylori eradication may improve inflammation and atrophy of the mucosa and promote ulcer recovery, however, further investigation is required to substantiate these claims.

In the present investigation concerning the determinants influencing the healing of stomach ulcers following ESD, most of the research suggests that the size of the ulcer serves as an independent risk factor[1,16,35]. Post-ESD ulcer healing is similar to peptic ulcers, with a larger area of mucous membrane defect prolonging the mucosal reconstruction process[25]. The findings of this meta-analysis further corroborate this conclusion. However, the included studies may have a publication bias, suggesting that negative results are difficult to publish. Further investigation is required to ascertain the validity of these results. The meta-analysis revealed the presence of heterogeneity, which may be due to differences in research methods, design, ulcer evaluation criteria, and sample size, requiring more randomized studies for validation.

Kobayashi et al[18] found that antral lesions exhibited a superior marginal healing rate compared to lesions located in the body of the stomach. Additionally, a multi-center randomized study indicated that the anatomical positioning of gastric tumors, specifically in the upper and middle thirds of the stomach, may have a comparable effect on the healing process of ulcers resulting from ESD when contrasted with those located in the lower third. However, some studies indicates that the positioning of the lesion within the lower third of the stomach serves as an independent risk factor for the delayed healing of ulcers following ESD[16,25,36], which was supported by our meta-analysis. The differences in findings may be related to inconsistencies in multifactor regression analysis adjustment and ulcer cure criteria.

Although the three included studies suggested that ulcer healing was not related to the pathological type of the ulcer, our meta-analysis indicates that ulcer pathology does affect healing. Chang et al[37] identified that histologically confirmed high-grade dysplasia or early-stage gastric cancer serve as independent risk factors for the delayed healing of ulcers following ESD. Early-stage stomach cancer, particularly when the degree of differentiation is high, is more likely to delay healing, which may be linked to the depth and extent of infiltration[38]. However, owing to the restricted volume of literature included, it is not feasible to conduct further subgroup analyses, thereby necessitating additional research.

Lim et al[9] identified that diabetes mellitus serves as an independent risk factor for the delayed healing of ulcers following ESD. However, the findings of this meta-analysis indicate that diabetes does not constitute a significant risk factor for the delayed healing of ulcers. The differences in results may be due to varying study designs and sample sizes, as well as the lack of specific details on blood sugar control and the duration of diabetes in the included articles. Since only two studies were included, the reliability of this conclusion needs to be confirmed by further research.

Despite ongoing debate regarding the impact of procedure duration on the healing of ulcers, long ESD operations may affect ulcer healing due to an increased risk of musculoskeletal damage or bleeding[24,25]. However, our meta-analysis suggests that procedure duration does not affect ulcer healing.

The findings of this meta-analysis may offer valuable insights for the personalized treatment and management of endoscopic removal ulcers within clinical settings, but several limitations exist: Firstly, the results were based on observational studies. Although we strictly defined the outcome indicators and included results of multi-factor analysis, the details of potentially mixed factors remain unclear due to the inherent bias risk in observational descriptive studies. Secondly, individual risk factors were included in fewer studies, preventing subgroup analysis, which may impact the meta-analysis results. Thirdly, the risk factors included in each study varied, and the same risk factor was stratified differently. Factors such as musculoskeletal damage, mucous membrane shrinkage, bleeding, and drug effects were studied less frequently and could not undergo meta-analysis. Finally, individual risk factors demonstrate evidence of publication bias, which could potentially undermine the reliability of the results. Consequently, it is imperative to conduct high-quality studies to validate our findings.

CONCLUSION

The findings of our meta-analysis indicate a notable enhancement in the healing rate of ulcers over an eight-week period. Prolonging the treatment duration for patients with post-ESD ulcers beyond eight weeks is associated with improved outcomes in the complete resolution of postoperative ulcerations resulting from ESD. Furthermore, factors such as H. pylori infection, ulcer size, lesion location, and the pathological characteristics of the lesion have been identified as significant risk factors contributing to delayed ulcer healing. More large-sample, high-quality research is needed to further validate these risk factors for delayed post-ESD ulcers.

ACKNOWLEDGEMENTS

Huang FX provided guidance in the prospective register of systematic reviews.