Authors: Yuxin Ma, Liang Zhu, Shutao Chen
Categories: Systematic Review/Meta-analysis, corticosteroid therapy, endoscopic balloon dilatation, endoscopic submucosal dissection, esophageal stricture, network meta-analysis
Source: International Journal of Surgery (London, England)
Authors: Yuxin Ma, Liang Zhu, Shutao Chen
Endoscopic submucosal dissection (ESD) is a commonly used minimally invasive technique for treating early esophageal tumors. Postoperative esophageal stricture is a frequent complication. This study aimed to compare the efficacy of existing pharmacological therapies in preventing esophageal stricture after ESD and to determine the most effective regimen to inform clinical practice.
A systematic literature search of the Cochrane Library, Embase, PubMed, and Web of Science databases was performed from inception to 20 March 2025. The main search terms included “endoscopic mucosal resection” and “ESD.” The main outcome assessed was the incidence of esophageal stricture, while the secondary outcome involved the number of endoscopic balloon dilatation (EBD) sessions needed to manage severe postoperative strictures. This network meta-analysis was conducted using random-effects models. Mean differences (MDs) and relative risks (RRs) were estimated along with their 95% credible intervals (CrIs). Interventions were ranked based on the surface under the cumulative ranking curve (SUCRA) values. All statistical analyses were conducted using R software (version 4.5.0).
Thirty-one studies involving 3085 patients who underwent ESD were included. In terms of esophageal stricture incidence, oral hydrocortisone sodium succinate combined with aluminum phosphate gel demonstrated the greatest reduction in the risk of stricture compared with standard care (RR = 0.075, 95% CrI: 0.017, 0.24; SUCRA = 95.21%). Regarding the number of EBD sessions, the combination of polyglycolic acid (PGA) sheet plus fibrin glue resulted in the fewest sessions compared with standard care (MD = − 6.4, 95% CrI: − 8.4, − 4.3; SUCRA = 95.82%).
In patients undergoing ESD, oral hydrocortisone sodium succinate combined with aluminum phosphate gel was the most effective intervention for reducing the incidence of postoperative esophageal stricture. The combination of PGA sheet plus fibrin glue was most effective in minimizing the number of EBD sessions required.
Esophageal cancer (EC) is a significant global health issue, with an increasing burden of incidence and mortality. According to the GLOBOCAN 2022 data, EC ranks 11th in incidence and 7th in mortality worldwide^[1]^. Although the age-standardized rates have decreased, the absolute number of new cases and deaths continues to increase^[2]^. It is predicted that the disease burden will further increase by 2050.
Endoscopic submucosal dissection (ESD) is the preferred minimally invasive treatment method for early EC and precancerous lesions. It allows for complete resection of lesions^[3–5]^, preservation of function, and minimal tissue trauma^[,6]^. However, esophageal stenosis after ESD is the main complication affecting its clinical application^[7–9]^, especially when there is a large area of mucosal defect (such as a circumferential range of ≥3/4 or a length of ≥50 mm). Its incidence increases markedly under these conditions, seriously reducing the patient’s quality of life and imposing a greater medical burden. Studies have shown that when the mucosal defect exceeds 75% of the esophageal circumference, the incidence of stenosis is as high as 15%–70%^[9,10]^. Its mechanism is mainly related to the fibrotic process mediated by postoperative inflammatory factors^[11–13]^. Current preventive strategies, such as local hormone injection, balloon dilation, and stent placement, have certain effects. However, some limitations remain, including large heterogeneity of efficacy, inconsistent standards, and limited clinical application. Moreover, patients often present with dysphagia, malnutrition, and repeated hospitalizations, which not only increase the medical burden but also significantly reduce the quality of life^[8,14]^. Previous studies have mostly focused on a single therapy and lacked a systematic comparison of different intervention measures, especially combined treatment strategies. In addition, there are limitations such as small sample size, regional bias, and methodological heterogeneity. Thus, optimization of treatment strategies is needed.
This study adopted the network meta-analysis (NMA) method to comprehensively evaluate the preventive effects of 15 intervention measures on esophageal stenosis after ESD, aiming to provide higher-level evidence-based guidance for clinical practice, optimize treatment strategies, and improve patient prognosis. This study adhered to the requirements of the TITAN guideline^[15]^.
This study conducted an NMA in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, and was registered on PROSPERO. The risk of bias in non-randomized controlled trials (NRCTs) was independently assessed using the validated ROBINS-I tool, and the risk of bias in RCTs was assessed using the Cochrane Risk of Bias 2 (RoB 2) tool, AMSTAR (Assessing the Methodological Quality of Systematic Reviews) Guidelines^[16]^.
A systematic search of the Cochrane Library, Embase, PubMed, and Web of Science databases was performed, covering all records from database inception to 20 March 2025. The principal search terms included “endoscopic submucosal dissection,” “endoscopic mucosal resection,” “Submucosal Tunneling Endoscopic Resection,” and “ESD.” The full search strategy and additional terms are provided in Supplemental Digital Content Table S1, available http://links.lww.com/JS9/G635. No language restrictions were applied; however, non-English articles were required to provide an English abstract.
All articles retrieved were imported into EndNote X9 for management, and duplicates were removed. Currently, international mainstream guidelines, such as the American College of Gastroenterology (ACG) and the European Society for Gastrointestinal Endoscopy (ESGE), primarily recommend the prevention of esophageal stenosis after ESD based on early studies or indirect comparisons of single interventions, lacking high-level, direct comparative evidence for multiple combined schemes. This study strictly adhered to the reporting standards for systematic reviews and NMA (PRISMA-NMA)^[18]^. Two independent researchers thoroughly checked the titles and abstracts to select studies relevant to the research topic. Full-text versions of studies deemed potentially eligible were then acquired for further evaluation. Discrepancies in study selection were resolved by a third reviewer. Eligible studies enrolled patients who had undergone ESD. Interventions included steroids (oral and injectable corticosteroid drugs, as well as other steroid medications), oral administration of prednisone-class drugs (prednisone, prednisolone, and prednisone acetate), oral hydrocortisone sodium succinate combined with aluminum phosphate gel, oral budesonide, combination of prednisone-class drugs with EBD, polyglycolic acid (PGA) sheets combined with triamcinolone acetonide, oral combination of prednisone-class drugs and proton pump inhibitor (PPI), PGA sheets combined with oral prednisolone, long-term oral statin therapy, local injection of triamcinolone acetonide combined with oral prednisone, PGA sheet plus fibrin glue, local injection of botulinum toxin A, and PPI or EBD alone. Control groups received standard therapy or a placebo. Some studies reported comparisons between two or more intervention measures. The main outcome was the incidence of esophageal stricture [expressed as relative risk (RR)], and the secondary outcome was the total number of EBD sessions required postoperatively (expressed as mean difference (MD)]. Eligible study designs included randomized controlled trials (RCTs) and cohort studies. During screening, only studies that at least met the specified patient characteristics and intervention criteria were retained; all others were excluded.
Studies must meet the following (1) studies enrolling patients aged 18 or older who received ESD for esophageal tumors; (2) interventions primarily consisting of pharmacological therapies, and interventions clearly defined without duplication; (3) availability of outcome data on drug efficacy for postoperative esophageal stricture, such as the incidence of esophageal stricture or the number of EBD sessions; and (4) study design limited to RCTs or cohort studies.
Exclusion criteria were as (1) studies involving non-adult populations or animal experiments; (2) patients who did not have EC and underwent ESD surgery; (3) interventions not clearly defined or only nonpharmacological; (4) studies without full-text access or lacking relevant data; and (5) study designs other than RCTs or cohort studies.
Two investigators independently reviewed the included articles and performed data extraction. Data collected included first author, country, publication year, patient age, sample size, intervention measures, control measures, number of patients in the experimental group and the control group, study design, NCT identifier (if available), adverse reactions, and outcome measures (esophageal stricture incidence, number of EBD sessions). Mean values and standard deviations (SDs) before and after intervention for both experimental and control groups were extracted as primary data. When SD was not provided, it was calculated from the standard error, 95% confidence interval (CI), range, or interquartile range (IQR). Specifically, when IQR was provided, the median was used as the mean, and the SD was estimated as IQR/1.135. If only minimum, maximum, and median values were reported, those data were excluded from statistical analysis. Because different outcome scales were used across studies, MD and RR were employed to pool continuous and dichotomous outcome data, respectively.
The risk of bias in non-RCTs (NRCTs) was independently assessed using the validated ROBINS-I tool. The evaluation covered seven bias due to confounding, selection of participants, classification of interventions, deviations from intended interventions, missing data, measurement of outcomes, and selection of the reported result. Each domain was graded as low, moderate, high, or critical risk of bias. According to the ROBINS-I guidelines, studies with a low risk in all domains must demonstrate a level of evidence comparable to that of high-quality randomized trials in the same field.
The risk of bias in RCTs was assessed using the Cochrane Risk of Bias 2 (RoB 2) tool. The assessment covered the following the randomization process, deviations from intended interventions, missing data, outcome measurement, and selection of the reported result. Each domain was rated as low risk, some concerns, or high risk. Two reviewers independently conducted the quality assessment, and any disagreements were resolved through consultation with a third reviewer.
This NMA was implemented using R version 4.5.0 and GeMTC software. All data analyses were conducted under the Bayesian random effects model supported by the R “gemtc” and “rjags” packages. The Bayesian approach provided the probability distribution of each effect estimate by simulating the posterior and prior distributions of parameters using the Markov chain Monte Carlo method. A total of 25 000 tuning iterations and 250 000 simulations were set up with a refinement factor of 10 to generate the Bayesian statistical model of this NMA. The inconsistency model was used to check the inconsistency. Since closed loops were found in the network graph, the node-splitting method was used to evaluate the local inconsistency between direct and indirect evidence. The convergence was evaluated using the Brooks–Gelman–Rubin diagnostic tool in combination with visual inspection of trace plots and density plots. The effect size was estimated using a Bayesian random-effects model and was presented as MDs, RRs, and their 95% credible intervals (CrIs). The comparison results among all interventions were shown in forest plots and ranking tables. When the 95% CrIs did not include the effect value of 0, the comparison was considered statistically significant. The probability distribution of the rankings for different drug efficacy results was generated, and the surface under the cumulative ranking curve (SUCRA) was calculated to stratify these interventions. The SUCRA values ranged from 0% to 100%, with higher SUCRA values indicating a greater probability of a treatment being superior. Sensitivity analyses were conducted by sequentially excluding individual studies (Tables 1 and 2). The funnel plot was used to assess publication bias, and Egger’s test and Begg’s test were used to confirm the predicted results, with P > 0.05 indicating no significant publication bias. To assess the robustness of the results, statistical analyses and sensitivity analyses at the level of the literature data were conducted. In terms of statistical analysis, comparisons were made between the results of outcome data analysis using the random-effects model. In terms of literature data, after screening and excluding studies with small sample sizes (n < 20 participants), the results before and after exclusion were compared to test the robustness of the results.Table 1Sensitivity analysis for the incidence of esophageal stricture.Random treatmentSURCAFixed treatmentSURCAIDDescription10.1330321410.132274551Placebo20.4080486620.407754912steroid30.4117254530.411336613PT40.952058484****0.951950894H + A50.816691075****0.817679915BS60.2569214360.256777686PPI70.0447750070.045331707EBD80.0370754580.037329028PT + EBD90.4880763490.489094209PGA sheet plus fibrin glue110.72854375110.7281883911PGA + TA120.8664625012****0.8648348212PT + PPI130.70176920130.7017156313PGA + prednisolone140.7655062514****0.7657491114Statin150.41149420150.4116446415steroid + prednisone160.47782009160.4783379516BTX-AETI: endoscopic injection of triamcinolide; H + A: oral administration of hydrocortisone sodium succinate in combination with aluminum phosphate gel; BS: oral budesonide; PPI: proton pump inhibitors; PGA: polyglycolic acid; BTX: botulinum toxin; EBD: balloon dilatation; TA: triamcinolone acetonide. The bolded parts are the indicators reported in the article. Table 2Sensitivity analysis for EBD frequency used for severe esophageal stricture.Random treatmentSURCAFixed treatmentSURCAIDDescription10.141477910.13736021Placebo20.84233732****0.85402052Steroid30.498091230.50791823PT40.605084540.60035574H + A50.498452150.49521485BS60.301458560.29665006PPI70.220537070.21799777EBD80.578419480.57179898PT + EBD90.95824099****0.96511939PGA sheet plus fibrin glue140.3293576140.327575014Statin150.7022373150.708028415****Steroid + Prednisone160.3243064160.317961416BTX-APT: prednisone-type drugs; H + A: oral administration of hydrocortisone sodium succinate in combination with aluminum phosphate gel; BS: oral budesonide; PPI: proton pump inhibitors; PGA: polyglycolic acid; BTX: botulinum toxin; EBD: balloon dilatation.
Our systematic search identified 2228 records, supplemented by eight additional studies from citation tracking. Following duplicate removal (1549 records) and initial screening, 624 records were excluded based on title and abstract review. Among 55 potentially eligible articles, 12 were inaccessible. After the full-text assessment, 11 studies were excluded (5 for population mismatch, 5 for non-RCT/cohort design, and 1 duplicate). The final analysis included 31 studies published prior to 25 February 2025, as detailed in the PRISMA flowchart (Fig. 1).HIGHLIGHTSOral hydrocortisone + aluminum phosphate gel best reduces stricture risk (RR = 0.093).Polyglycolic acid (PGA) sheets + triamcinolone minimize EBD sessions (MD = −8.6 vs. standard care).Network meta-analysis ranks 15 hydrocortisone-aluminum gel top for stricture prevention.Surface under the cumulative ranking Hydrocortisone-aluminum gel (95.17%) and PGA-TA (96.85%) were optimal.Geographic 97% studies of the research come from China/Japan; needs multicenter validation. Figure 1.Flowchart of the study selection process.
This study included 31 studies^[19–49]^, enrolling 3058 patients who received ESD. The basic characteristics of these studies are summarized in Table 3. The first authors were predominantly from China and Japan (14 from China, 16 from Japan, and 1 from Australia). The primary outcome was the incidence of esophageal stricture, and the secondary outcome was the number of EBD sessions required for severe postoperative strictures. Study designs included RCTs and cohort studies. Patient ages ranged from 50 to 81 years.Table 3Baseline information of the included studies.First authorYearDistrictSampleAge, years (mean ± SD)TreatmentResearch typeOutcomeSide-effectLei Chen2024China75 (25/25)68.69 ± 6.69Steroid vs. placeboRCT1, 2A small number of transient retrosternal pains.Dan Nie2019China27 (14/13)66.67 ± 9.20H + A vs. steroidRCT1, 2Reversible hypokalemia (66.7%).Farzan F. Bahin2015AUS104 (29/75)NABS vs. PPICS1, 2Not have.Gaosu Zhou2017China23 (13/10)66.22 ± 6.27Pred + PPI vs. PPICS1, 2Not have.Xiaojian He2017China67 (22/23)56.99 ± 6.64Steroid vs. placeboRCT1, 2A small number of cases of mild facial edema.Hiroki Sato2013JPN23 (10/13)70.45 ± 7.33Pred + EBD vs. EBDCS1, 2Not have.Toshiro Iizuka2017JPN60 (39/31)67.99 ± 8.71PGA + fibrin glue vs. steroidCS1, 2One case of postoperative bleeding occurred.Hajime Isomoto2011JPN7 (4/3)67.52 ± 5.13Pred vs. EBDCS1, 2Not have.Kaname Uno2012JPN31 (15/16)70.77 ± 7.29TN + EBD vs. EBDCS1, 2Not have.Mikinori Kataoka2015JPN33 (17/16)70.57 ± 6.75Pred vs. EBDCS1, 2Not have.Yudai Kawamura2021JPN13 (7/6)65.37 ± 7.17steroid vs. placeboCS1Not have.Noboru Hanaoka2012JPN59 (30/29)65.49 ± 6.96Steroid vs. placeboCS1, 2Not have.Yasuaki Nagami2017JPN150 (37/37)69.55 ± 6.97Steroid vs. placeboCS1, 2Not have.Ningli Chai2017China16 (8/8)72.57 ± 3.51Steroid + PGA vs. PGARCT1, 2Not have.Rong-Wei Ruan2019China49 (22/27)66.06 ± 7.67Steroid vs. placeboCS1, 2Not have.Yoshiki Sakaguchi2020JPN448 (52/30/349)68.16 ± 8.60Steroid vs. PGA vs. placeboCS1, 2One case of delayed perforation occurred during EBD treatment.Yoshiki Sakaguchi2023JPN699 (105/53/505)68.70 ± 9.00Steroid vs. PGA vs. placeboCS1, 2Not have.Wei Tian2024China45 (18/27)63.67 ± 8.18Pred + PGA vs. PredCS1Not have.Yanjuan Wang2023China225 (23/21/20/145)66.58 ± 6.85Statin vs. BTX-A vs. steroid vs. placeboCS1, 2Not have.Qingxia Wang2025China207 (53/67/87)67.13 ± 7.95Pred vs. steroid + Pred vs. PPICS1, 2One case of lung infection related to inhalation.Jing Wen2016China67 (33/34)60.93 ± 7.59BTX-A vs. placeboRCT1Not have.Hailu Wu2022China81 (28/23)62.09 ± 7.95Pred vs. placeboRCT1One case of lung infection related to inhalation.Naoyuki Yamaguchi2011JPN41 (19/22)68.02 ± 7.16Pred vs. EBDCS1, 2Not have.Yiyang Zhang2021China63 (32/31)64.70 ± 9.44H + A vs. steroid + PredRCT1, 2Infection: control 9.7% (3/31); experimental 31.3% (10/32), significantly higher than the control group (P = 0.034). And control group 19.4% (6/31); experimental 31.3% (10/32), there was no statistically significant difference (P = 0.278).Yoshiaki Ando2024JPN100 (52/48)66.55 ± 8.805Steroid vs. placeboCS1, 2One case of lung infection related to inhalation.Yuan Chu2018China70 (34/36)64.32 ± 7.57Steroid + Pred vs. placeboCS1, 2Not have.Yasuaki Nagami2015JPN56 (28/28)69.35 ± 8.07Steroid vs. placeboCS1, 2Not have.Tomohiro Kadota2016JPN115 (53/29/33)69.85 ± 1.77Steroid vs. placeboCS1One case of lung infection related to inhalation.Yoshiki Tsujii2017JPN38 (28/10)69.69 ± 7.42Steroid vs. placeboCS1, 2EBD-related perforation, one case in the control group (14%); five cases in the experimental group (45%).Satoru Hashimoto2011JPN41 (21/20)71.11 ± 9.45Steroid vs. placeboCS1, 2Not have.Yu Qiu2021China52 (15/20)64.00 ± 5.09Steroid vs. placeboCS1, 2Two cases of severe perforation in the control group.H + A: Oral administration of hydrocortisone sodium succinate in combination with aluminum phosphate gel; BS: oral budesonide; PPI: proton pump inhibitors; PGA: polyglycolic acid; BTX: botulinum toxin; EBD: balloon dilatation; TA: triamcinolone acetonide.Outcome: 1, esophageal stenosis; 2, the number of EBD treatments.
The evaluation by ROBINS-1 suggested that among 25 NRCTs, 18 were judged to be at high risk of bias, 6 at moderate risk, and 1 at low risk. High risk was primarily attributed to confounding bias, bias due to selection of participants, bias due to deviations from intended interventions, bias due to missing data, and bias in measurement of outcomes. Moderate risk was mainly related to confounding bias, bias in classification of interventions, bias due to selection of participants, bias in deviations from intended interventions, bias due to missing data, bias in measurement of outcomes, and bias in selection of the reported result (Figs 2 and 3).Figure 2.Point plot for ROBINS-1 for cohort studies. Figure 3.Bar plot for ROBINS-1 for cohort studies.
The evaluation by RoB 2 indicated that among six RCTs included, one was assessed as high risk of bias, and five as moderate risk. High risk was attributable to bias in the selection of the reported result, while moderate risk was related to the randomization process, outcome measurement, and selection of the reported result (Figs 4 and 5). Figure 4.Point plot for RoB 2 for randomized controlled trials. Figure 5.Bar plot for RoB 2 for randomized controlled trials. PT: prednisone-type drugs; H + A: oral administration of hydrocortisone sodium succinate in combination with aluminum phosphate gel; BS: oral budesonide; PPI: proton pump inhibitors; PGA: polyglycolic acid; BTX: botulinum toxin; EBD: balloon dilatation; TA: triamcinolone acetonide. The numbers represent the intervention Placebo; Steroid; PT; H + A; BS; PPI; EBD; PT + EBD; PGA + fibrin glue; PGA + TA; PT + PPI; PGA + prednisolone; Statin; TA + prednisone; BTX-A.
Thirty-one studies^[19–49]^ involving 3058 patients treated with ESD were included to evaluate the effects of various pharmacological and other interventions for preventing esophageal stricture. Fifteen interventions were steroids (oral and injectable corticosteroid drugs, as well as other steroid medications), oral administration of prednisone-class drugs (prednisone, prednisolone, and prednisone acetate), oral hydrocortisone sodium succinate combined with aluminum phosphate gel, oral budesonide, combination of prednisone-class drugs with EBD, PGA sheets combined with triamcinolone acetonide, oral combination of prednisone-class drugs and PPI, PGA sheets combined with oral prednisolone, long-term oral statin therapy, local injection of triamcinolone acetonide combined with oral prednisone, PGA sheet plus fibrin glue, local injection of botulinum toxin A, PPI or EBD alone, and placebo.
The NMA results are shown in Figure 6. Figure 6.Forest plot for the incidence of esophageal stricture.
Based on SUCRA values, oral hydrocortisone sodium succinate combined with aluminum phosphate gel was associated with the greatest reduction in the incidence of esophageal stricture compared with standard care (RR = 0.075, 95% CrI: 0.017, 0.24; SUCRA = 95.21%). The next most effective interventions were prednisone-type drugs plus PPI (RR = 0.12, 95% CrI: 0.020, 0.59; SUCRA = 86.65%), budesonide suspension (RR = 0.16, 95% CrI: 0.029, 0.69; SUCRA = 81.67%), and long-term oral administration of statin drugs (RR = 0.21, 95% CrI: 0.049, 0.71; SUCRA = 76.55%). Each of these interventions was associated with a significant reduction in the incidence of esophageal stricture. To assess the stability of the results, a convergence diagnosis of the model was conducted. The diagnostic efficacy results of the model showed that the CI index for esophageal stenosis was equal to 1, indicating that the diagnostic efficacy of the model was robust. Relevant details are provided in Supplemental Digital Content Figure S1, available http://links.lww.com/JS9/G635.
The inconsistency test showed that all comparisons among different interventions had P values greater than 0.05, indicating no significant inconsistency. Detailed results are presented in the forest plot for relative effects (Fig. 6) and league table (Table 4). The probability ranking chart is shown in Figure 7. The network analysis diagram is in Figure 8, and the inconsistency test results are provided in Supplemental Digital Content Figure S2, available http://links.lww.com/JS9/G635. The line graph is shown in Figure 9. Figure 7.Probability ranking for the incidence of esophageal stricture. Figure 8.Network diagram of the incidence of esophageal stenosis. Figure 9.Forest plot for EBD frequency used for severe esophageal stricture. Table 4League table for EBD frequency used for severe esophageal stricture.Placebo**−4.7 (−5.64, − 3.54)−2.52 (−3.78, − 1.27)−2.91 (−4.35, − 1.28)−2.43 (−4.27, − 0.58)−1.43 (−2.69, − 0.15)0.98 (−7.76, 9.75)−3.03 (−12.34, 6.36)−6.4 (−8.36, − 4.29)−0.03 (−9.64, 9.6)−3.33 (−4.22, − 2.41)−1.25 (−3.93, 1.41)Placebo4.7 (3.54, 5.64)Steroid2.18 (0.4, 3.66)**1.8 (−0.12, 3.56)2.27 (−0.01, 4.26)3.28 (1.48, 4.76)5.66 (−3.15, 14.48)1.65 (−7.75, 11.07)−1.71 (−3.44, 0.03)4.64 (−5.04, 14.27)1.37 (−0.14, 2.63)3.43 (0.51, 6.24)4.7 (3.54, 5.64)2.52 (1.27, 3.78)−2.18 (−3.66, − 0.4)**PT−0.39 (−1.17, 0.63)0.09 (−1.55, 1.75)**1.1 (0.15, 2.05)3.51 (−5.15, 12.2)−0.51 (−9.73, 8.8)−3.89 (−6.16, − 1.38)**2.48 (−7.21, 12.19)−0.81 (−1.71, 0.13)1.26 (−1.67, 4.21)2.52 (1.27, 3.78)****2.91 (1.28, 4.35)−1.8 (−3.56, 0.12)0.39 (−0.63, 1.17)H + A0.47 (−1.48, 2.26)**1.48 (0.07, 2.69)3.87 (−4.85, 12.59)−0.13 (−9.42, 9.21)−3.51 (−5.96, − 0.91)2.85 (−6.87, 12.58)−0.42 (−1.8, 0.77)1.64 (−1.46, 4.67)2.91 (1.28, 4.35)****2.43 (0.58, 4.27)−2.27 (−4.26, 0.01)−0.09 (−1.75, 1.55)−0.47 (−2.26, 1.48)BS1.01 (−0.34, 2.34)3.41 (−5.4, 12.26)−0.61 (−9.97, 8.85)−3.97 (−6.61, − 1.11)**2.39 (−7.38, 12.15)−0.9 (−2.51, 0.73)1.18 (−2.09, 4.41)**2.43 (0.58, 4.27)1.43 (0.15, 2.69)−3.28 (−4.76, − 1.48)−1.1 (−2.05, − 0.15)****−1.48 (−2.69, − 0.07)−1.01 (−2.34, 0.34)PPI2.4 (−6.31, 11.15)−1.61 (−10.89, 7.75)−4.98 (−7.26, − 2.48)1.39 (−8.28, 11.08)−1.9 (−2.81, − 0.98)0.17 (−2.78, 3.12)1.43 (0.15, 2.69)−0.98 (−9.75, 7.76)−5.66 (−14.48, 3.15)−3.51 (−12.2, 5.15)−3.87 (−12.59, 4.85)−3.41 (−12.26, 5.4)−2.4 (−11.15, 6.31)EBD−4 (−7.23, − 0.74)−7.37 (−16.39, 1.63)−1.04 (−13.86, 11.93)−4.31 (−13.04, 4.39)−2.24 (−11.37, 6.91)−0.98 (−9.75, 7.76)3.03 (−6.36, 12.34)−1.65 (−11.07, 7.75)0.51 (−8.8, 9.73)0.13 (−9.21, 9.42)0.61 (−8.85, 9.97)1.61 (−7.75, 10.89)**4 (0.74, 7.23)**PT + EBD−3.35 (−12.92, 6.21)2.97 (−10.26, 16.31)−0.3 (−9.65, 8.97)1.77 (−7.94, 11.49)3.03 (−6.36, 12.34)6.4 (4.29, 8.36)1.71 (−0.03, 3.44)3.89 (1.38, 6.16)****3.51 (0.91, 5.96)3.97 (1.11, 6.61)4.98 (2.48, 7.26)**7.37 (−1.63, 16.39)3.35 (−6.21, 12.92)PGA + fibrin glue6.36 (−3.48, 16.13)**3.08 (0.75, 5.21)5.14 (1.74, 8.43)6.4 (4.29, 8.36)**0.03 (−9.6, 9.64)−4.64 (−14.27, 5.04)−2.48 (−12.19, 7.21)−2.85 (−12.58, 6.87)−2.39 (−12.15, 7.38)−1.39 (−11.08, 8.28)1.04 (−11.93, 13.86)−2.97 (−16.31, 10.26)−6.36 (−16.13, 3.48)Statin−3.29 (−12.96, 6.35)−1.21 (−11.24, 8.8)0.03 (−9.6, 9.64)3.33 (2.41, 4.22)−1.37 (−2.63, 0.14)0.81 (−0.13, 1.71)0.42 (−0.77, 1.8)0.9 (−0.73, 2.51)**1.9 (0.98, 2.81)4.31 (−4.39, 13.04)0.3 (−8.97, 9.65)−3.08 (−5.21, − 0.75)3.29 (−6.35, 12.96)TA + prednisone2.07 (−0.75, 4.89)3.33 (2.41, 4.22)1.25 (−1.41, 3.93)−3.43 (−6.24, − 0.51)−1.26 (−4.21, 1.67)−1.64 (−4.67, 1.46)−1.18 (−4.41, 2.09)−0.17 (−3.12, 2.78)2.24 (−6.91, 11.37)−1.77 (−11.49, 7.94)−5.14 (−8.43, − 1.74)1.21 (−8.8, 11.24)−2.07 (−4.89, 0.75)BTX-A1.25 (−1.41, 3.93)Placebo−4.7 (−5.64, − 3.54)−2.52 (−3.78, − 1.27)−2.91 (−4.35, − 1.28)−2.43 (−4.27, − 0.58)−1.43 (−2.69, − 0.15)0.98 (−7.76, 9.75)−3.03 (−12.34, 6.36)−6.4 (−8.36, − 4.29)−0.03 (−9.64, 9.6)−3.33 (−4.22, − 2.41)−1.25 (−3.93, 1.41)PlaceboPT: prednisone-type drugs; H + A: oral administration of hydrocortisone sodium succinate in combination with aluminum phosphate gel; BS: oral budesonide; PPI: proton pump inhibitors; PGA: polyglycolic acid; BTX: botulinum toxin; EBD: balloon dilatation.
Eighteen studies^[19,21,23–25,27,29,31,35,36,38,39,42–46,48]^, including 1796 patients who had undergone ESD, were included to evaluate the effects of various interventions on reducing the number of postoperative EBD sessions. Twelve interventions were prednisone-type drugs; oral hydrocortisone sodium succinate combined with aluminum phosphate gel; budesonide suspension; PPI; EBD; prednisone-type drugs plus EBD; PGA sheet plus fibrin glue; steroid; long-term oral administration of statin drugs; combined triamcinolone acetonide and prednisone; and botulinum toxin A (BTX-A). The NMA results are shown in Figure 7.
Based on SUCRA values, the PGA sheet plus fibrin glue produced the greatest reduction in EBD sessions compared with standard care (MD = − 6.4, 95% CrI: − 8.4, − 4.7; SUCRA = 95.82%). The next most effective interventions were steroids (MD = − 4.8, 95% CrI: − 5.7, − 4.0; SUCRA = 84.23%) and combined triamcinolone acetonide and prednisone (MD = − 3.3, 95% CrI: − 3.8, − 2.8; SUCRA = 70.22%), which also reduced the number of EBD sessions.
The inconsistency test indicated that all comparisons among different interventions had P values greater than 0.05, indicating no significant inconsistency. Heterogeneity analysis showed that the NMA met the homogeneity assumption. To assess the stability of the results, a convergence diagnosis of the model was conducted. The diagnostic efficacy results of the model showed that the CI index for the number of EBD sessions was under 1.05, indicating that the diagnostic efficacy of the model was robust. Relevant details can be found in Supplemental Digital Content Figure S3, available http://links.lww.com/JS9/G635. Detailed results are presented in the forest plot for relative effects (Fig. 10) and league table (Table 5). The probability ranking chart is shown in Figure 11. The network analysis diagram is in Figure 12, and the inconsistency test results are provided in Supplemental Digital Content Figure S4, available http://links.lww.com/JS9/G635. The line graph is shown in Figure 13. Figure 10.SUCRA line chart of esophageal stricture. Figure 11.Probability ranking for EBD frequency used for severe esophageal stricture. Figure 12.Network diagram of EBD frequency used for severe esophageal stricture. Figure 13.SUCRA line chart of EBD frequency used for severe esophageal stricture. Table 5League table with esophageal stricture as the outcome indicator.Placebo**0.39 (0.25, 0.55)****0.41 (0.24, 0.65)****0.07 (0.02, 0.24)****0.16 (0.03, 0.69)**0.47 (0.2, 1)1.98 (0.56, 8.45)1.9 (0.42, 10.12)**0.37 (0.17, 0.77)****0.23 (0.09, 0.56)****0.12 (0.02, 0.59)****0.23 (0.07, 0.75)****0.21 (0.05, 0.71)****0.4 (0.24, 0.65)****0.35 (0.13, 0.83)Placebo2.59 (1.8, 3.94)**Steroid1.05 (0.56, 1.95)**0.19 (0.04, 0.64)**0.41 (0.07, 1.93)1.21 (0.51, 2.91)**5.16 (1.4, 23.4)****4.95 (1.06, 28.17)**0.96 (0.42, 2.16)0.61 (0.22, 1.61)0.32 (0.05, 1.65)0.61 (0.17, 2.12)0.54 (0.12, 2.01)1.05 (0.57, 1.97)0.92 (0.33, 2.39)**2.59 (1.8, 3.94)****2.46 (1.55, 4.22)0.95 (0.51, 1.77)PT0.19 (0.04, 0.59)**0.39 (0.07, 1.7)1.15 (0.55, 2.41)**4.88 (1.57, 19.59)****4.68 (1.15, 23.87)**0.91 (0.36, 2.25)0.57 (0.2, 1.62)0.31 (0.05, 1.45)0.58 (0.18, 1.71)0.51 (0.11, 1.99)0.99 (0.57, 1.78)0.87 (0.3, 2.4)**2.46 (1.55, 4.22)****13.38 (4.14, 58.6)****5.15 (1.56, 22.23)****5.41 (1.69, 22.74)**H + A2.11 (0.29, 16.28)**6.26 (1.62, 30.77)****27.24 (5.14, 192.84)****26.08 (4.06, 219.01)****4.94 (1.22, 25.04)**3.13 (0.69, 17.49)1.67 (0.2, 13.61)3.14 (0.61, 18.87)2.76 (0.44, 19.21)**5.4 (1.56, 24.31)****4.75 (1.04, 25.71)****13.38 (4.14, 58.6)****6.39 (1.45, 34.88)**2.46 (0.52, 13.59)2.59 (0.59, 13.41)0.47 (0.06, 3.49)BS2.97 (0.83, 12.89)**12.92 (1.97, 106.88)****12.39 (1.57, 123.33)**2.35 (0.43, 14.68)1.49 (0.26, 9.94)0.8 (0.1, 5.92)1.49 (0.23, 10.62)1.32 (0.17, 10.83)2.58 (0.6, 13.44)2.27 (0.38, 14.81)**6.39 (1.45, 34.88)****2.13 (1, 4.89)**0.83 (0.34, 1.98)0.87 (0.41, 1.8)**0.16 (0.03, 0.62)0.34 (0.08, 1.21)PPI4.27 (1.1, 20.13)**4.09 (0.83, 24.06)0.79 (0.26, 2.36)0.5 (0.15, 1.66)0.27 (0.05, 1.05)0.5 (0.13, 1.84)0.44 (0.09, 1.98)0.87 (0.42, 1.82)0.76 (0.22, 2.49)**2.13 (1, 4.89)**0.51 (0.12, 1.78)**0.19 (0.04, 0.71)****0.2 (0.05, 0.64)****0.04 (0.01, 0.19)****0.08 (0.01, 0.51)****0.23 (0.05, 0.91)**EBD0.95 (0.42, 2.35)**0.18 (0.04, 0.79)****0.12 (0.02, 0.55)****0.06 (0.01, 0.43)****0.12 (0.02, 0.57)****0.1 (0.01, 0.62)****0.2 (0.05, 0.73)****0.18 (0.03, 0.82)**0.51 (0.12, 1.78)0.53 (0.1, 2.39)**0.2 (0.04, 0.94)****0.21 (0.04, 0.87)****0.04 (0, 0.25)****0.08 (0.01, 0.64)**0.24 (0.04, 1.2)1.06 (0.43, 2.4)PT + EBD0.19 (0.03, 1.02)**0.12 (0.02, 0.71)****0.06 (0.01, 0.53)****0.12 (0.02, 0.72)****0.11 (0.01, 0.77)****0.21 (0.04, 0.97)**0.18 (0.03, 1.04)0.53 (0.1, 2.39)**2.72 (1.3, 6.04)**1.05 (0.46, 2.4)1.1 (0.44, 2.74)**0.2 (0.04, 0.82)**0.43 (0.07, 2.31)1.27 (0.42, 3.81)**5.44 (1.26, 27.9)**5.2 (0.98, 33.3)PGA + fibrin glue0.63 (0.19, 2.08)0.34 (0.05, 1.97)0.63 (0.15, 2.62)0.56 (0.11, 2.48)1.1 (0.45, 2.75)0.96 (0.28, 3.08)**2.72 (1.3, 6.04)****4.28 (1.79, 11.16)**1.65 (0.62, 4.5)1.74 (0.62, 4.96)0.32 (0.06, 1.44)0.67 (0.1, 3.91)2.01 (0.6, 6.75)**8.61 (1.83, 47.87)****8.25 (1.41, 56.46)**1.58 (0.48, 5.27)PGA + TA0.53 (0.07, 3.32)1 (0.21, 4.53)0.88 (0.17, 4.27)1.73 (0.62, 4.99)1.52 (0.41, 5.46)**4.28 (1.79, 11.16)****8.03 (1.68, 50.58)**3.08 (0.61, 19.84)3.24 (0.69, 19.53)0.6 (0.07, 4.92)1.25 (0.17, 10.13)3.71 (0.95, 19.51)**16.32 (2.32, 149.11)****15.62 (1.87, 166.48)**2.96 (0.51, 20.93)1.88 (0.3, 14.26)PT + PPI1.88 (0.27, 15.1)1.65 (0.2, 15.11)3.24 (0.69, 19.73)2.84 (0.44, 21.22)**8.03 (1.68, 50.58)****4.28 (1.33, 15.35)**1.65 (0.47, 6.06)1.74 (0.59, 5.46)0.32 (0.05, 1.63)0.67 (0.09, 4.35)2 (0.54, 7.81)**8.61 (1.76, 50.64)****8.24 (1.39, 59.91)**1.58 (0.38, 6.83)1 (0.22, 4.68)0.53 (0.07, 3.68)PGA + prednisolone0.88 (0.14, 5.3)1.73 (0.51, 6.26)1.52 (0.33, 7)**4.28 (1.33, 15.35)****4.83 (1.4, 20.59)**1.86 (0.5, 8.18)1.97 (0.5, 8.81)0.36 (0.05, 2.26)0.76 (0.09, 6.04)2.27 (0.5, 11.46)**9.86 (1.62, 72.13)****9.42 (1.3, 82.28)**1.78 (0.4, 8.98)1.13 (0.23, 6.03)0.61 (0.07, 5.08)1.14 (0.19, 7.17)Statin1.96 (0.5, 8.9)1.71 (0.35, 9.05)**4.83 (1.4, 20.59)****2.47 (1.54, 4.18)**0.95 (0.51, 1.77)1.01 (0.56, 1.75)**0.19 (0.04, 0.64)**0.39 (0.07, 1.68)1.16 (0.55, 2.38)**4.91 (1.37, 21.75)****4.7 (1.03, 26.2)**0.91 (0.36, 2.24)0.58 (0.2, 1.61)0.31 (0.05, 1.45)0.58 (0.16, 1.95)0.51 (0.11, 1.99)TA + prednisone0.88 (0.3, 2.39)**2.47 (1.54, 4.18)****2.82 (1.2, 7.5)**1.09 (0.42, 3.03)1.15 (0.42, 3.37)**0.21 (0.04, 0.96)**0.44 (0.07, 2.66)1.32 (0.4, 4.54)**5.68 (1.23, 32.09)5.43 (0.96, 37.77)1.04 (0.32, 3.54)0.66 (0.18, 2.46)0.35 (0.05, 2.26)0.66 (0.14, 3.01)0.58 (0.11, 2.84)1.14 (0.42, 3.36)BTX-A2.82 (1.2, 7.5)Placebo0.39 (0.25, 0.55)****0.41 (0.24, 0.65)****0.07 (0.02, 0.24)****0.16 (0.03, 0.69)**0.47 (0.2, 1)1.98 (0.56, 8.45)1.9 (0.42, 10.12)**0.37 (0.17, 0.77)****0.23 (0.09, 0.56)****0.12 (0.02, 0.59)****0.23 (0.07, 0.75)****0.21 (0.05, 0.71)****0.4 (0.24, 0.65)****0.35 (0.13, 0.83)**PlaceboH + A: oral administration of hydrocortisone sodium succinate in combination with aluminum phosphate gel; BS: oral budesonide; PPI: proton pump inhibitors; PGA: polyglycolic acid; BTX: botulinum toxin; EBD: balloon dilatation.
To comprehensively evaluate the relative effectiveness of each intervention measure, we plotted the cumulative ranking probability curve and calculated the SUCRA values. Figure 10 shows the SUCRA ranking probabilities of different intervention measures in reducing the incidence of esophageal stenosis. As shown in the figure, the curve of hydrocortisone sodium succinate combined with aluminum phosphate gel was consistently located at the top. Its SUCRA value (95.21%) was higher than that of any other interventions, indicating that it was likely to be the best intervention for preventing the occurrence of stenosis. The curves of prednisone drugs combined with PPI and budesonide suspension were the second highest, confirming their effectiveness as secondary options. Figure 13 shows the ranking of each intervention measure in reducing the number of EBD treatments after surgery. The PGA sheet combined with fibrin glue (SUCRA = 95.82%) was the most effective strategy for reducing the need for EBD. Notably, hydrocortisone sodium succinate combined with aluminum phosphate gel, which performed best in preventing the incidence of stenosis, did not show a prominent effect in reducing the need for EBD. This difference indicated that different intervention measures had different effects on preventing the occurrence of stenosis and reducing the severity of stenosis, providing a clear basis for the selection of treatment measures based on different treatment goals.
Publication bias among the included studies was assessed using the Begg test and Egger funnel plot. The outcomes of esophageal stenosis and the number of EBD procedures were analyzed for potential publication bias. For esophageal stenosis, the funnel plot results showed Egger’s P = 0.50 and Begg’s P = 0.42 (Fig. 14). For the number of EBD procedures, the funnel plot results showed Egger’s P = 0.15 and Begg’s P = 0.30 (Fig. 15). To assess the robustness of the results, statistical analyses and sensitivity analyses at the level of the literature data were conducted. In terms of the analysis methods, the random-effects model yielded consistent statistical analysis results for the outcomes included in the analysis. Therefore, the analysis methods and model construction in this study were relatively robust. In terms of the literature data, after excluding studies with small sample sizes, the results for esophageal stenosis remained consistent with those before exclusion. However, differences were observed for the number of EBD procedures. Before exclusion, based on the SUCRA values, league table, and relative effect forest plot results, the top three effective intervention measures were PGA + steroid, PGA sheet plus fibrin glue, and steroid. After excluding small-sample studies, based on the SUCRA, league table, and relative effect forest plot results, the top three effective intervention measures were PGA sheet plus fibrin glue, steroid, and combined steroid and prednisone. Figure 14.Funnel plot of esophageal stricture. PT: prednisone-type drugs; H + A: oral administration of hydrocortisone sodium succinate in combination with aluminum phosphate gel; BS: oral budesonide; PPI: proton pump inhibitors; PGA: polyglycolic acid; BTX: botulinum toxin; EBD: balloon dilatation. Figure 15.Funnel plot of EBD frequency used for severe esophageal stricture.
This study demonstrated that oral administration of sodium succinate hydrocortisone in conjunction with aluminum phosphate gel potentially lowered the rate of postoperative esophageal stricture (RR = 0.075, SUCRA = 95.21%). Additionally, the PGA sheet plus fibrin glue reduced the number of EBD sessions (MD = − 6.4, SUCRA = 95.82%). These findings are consistent with previous studies. For example, one study^[8]^ found that local steroid injections reduced the risk of stricture formation, while another study^[50]^ confirmed that the combination of PGA and corticosteroids decreased the need for EBD. However, some studies^[45]^ reported that BTX-A had limited efficacy, possibly due to its narrow pharmacological target or small sample sizes, indicating a potential need for combination therapies.
The postoperative stenosis rate is significantly associated with the circumferential mucosal defect rate after ESD. Previous studies have shown that when the mucosal defect exceeds 75% of the esophageal circumference, the stenosis rate can reach 15%–70%^[9,10]^. Another study reported that 81.8% of patients had a resection range of 75%–99% of the circumference, but the corresponding stenosis rate was not specified^[51]^. When the resection range is ≥1/2 of the circumference, the risk of stenosis significantly increases (40% vs. 8%, P = 0.039)^[22]^, and local hormone injection has a preventive effect on patients with < 3/4 and ≥1/2 circumference resection^[22]^. For resection range ≥90% of the circumference, the study mainly focused on the incidence of symptomatic stenosis at 3 months^[52]^. In addition, a mucosal defect width ≤5 mm (OR = 4.2) and resection length ≥50 mm (OR = 3.1) are independent predictors of stenosis^[52]^. The risk of stenosis in the cervical esophagus after ESD is higher, and tumor size (P = 0.026) and resection time (P = 0.028) are significantly associated with stenosis in patients who did not receive hormone injection^[22]^.
In addition to the inflammatory response mentioned earlier, the formation of esophageal stenosis involves molecular mechanisms, including fibroproliferative remodeling during wound healing. Pathological activation of fibroblasts and disruption of collagen metabolism are central mechanisms in the progression of esophageal stricture. TGF-β1 not only promotes the transdifferentiation of fibroblasts into myofibroblasts but also upregulates collagen synthesis genes such as COL1A1 and COL3A1^[13,53]^. Concurrently, the overexpression of tissue inhibitor of metalloproteinase-1 (TIMP-1) impairs collagen degradation, leading to an imbalance in ECM metabolism. Spatial transcriptomics analyses further revealed that myofibroblast clusters in ulcer bed regions expressed high levels of α-smooth muscle actin and fibronectin, which are closely associated with the histopathological features of luminal narrowing. This dynamic imbalance in ECM remodeling alters the biomechanical properties of the esophageal wall and accelerates tissue hardening.
The pathogenesis of fibrotic strictures involves aberrant wound healing characterized by excessive extracellular matrix deposition during late-stage tissue remodeling. In the proliferative phase, large numbers of fibroblasts are recruited, and collagen cross-linking leads to the formation of dense, disorganized scar tissue. Hypoxia-inducible factor-1α–mediated pathological angiogenesis exacerbates this process by supplying nutrients to fibrotic regions. Notably, three-dimensional reconstruction of post-ESD lesions has shown that collagen fibers align concentrically in circles perpendicular to the esophageal lumen, directly causing circumferential narrowing. This structural abnormality creates a positive feedback loop with the continuous mechanical stress generated by esophageal peristalsis, ultimately leading to clinically significant dysphagia.
The therapeutic combination of hydrocortisone sodium succinate and aluminum phosphate gel showed marked preventive effects against post-procedural esophageal strictures in our investigation. Glucocorticoids (such as hydrocortisone) can inhibit inflammation and fibrosis. Aluminum phosphate gel acts as a physical barrier over the wound, thereby reducing acid irritation and mechanical friction and shortening the drug’s lag time. Although EBD can rapidly relieve strictures, it requires repeated procedures and carries a risk of perforation. Some agents may cause steroid-related adverse effects (such as immunosuppression and metabolic disturbances) or local irritation. The core objective of this study was to balance efficacy and safety to identify the regimen that provides the greatest patient benefit. This combined approach not only suppresses fibrosis at the molecular level but also promotes wound healing through mucosal protection, thereby significantly reducing the risk of stricture.
Combined treatment with the PGA sheet plus fibrin glue had the most prominent effect in reducing the number of EBD sessions. PGA, as a biodegradable scaffold material, provided mechanical support to the wound^[54,55]^, promoted adhesion formation, reinforced the surgical site^[56,57]^, and prevented early luminal narrowing. The fibrin gel enhances the effect of PGA and provides a good protective adhesion effect on the surgical wound. This combined intervention acted synergistically, and it effectively reduced the need for repeated EBD treatments. Although EBD can rapidly relieve stricture symptoms, repeated procedures increase the risk of perforation^[58]^. Therefore, minimizing the number of EBD sessions is critical for patient safety. The findings of this research strongly endorse the use of combined treatment approaches.
Compared to earlier studies, this research presents multiple benefits. First, this study represents the first systematic comparison of 14 interventions, including local injections, systemic medications, and combination therapies. Thus, our findings may provide multidimensional evidence for clinical practice. Second, methodological the analysis adhered to PRISMA-NMA guidelines and employed a Bayesian framework with SUCRA ranking, enhancing the reliability of results. Third, dual-outcome it considered both the incidence of esophageal stricture and the number of EBD sessions, optimizing the basis for clinical decision-making. Nevertheless, this study has certain limitations. First, our research has a significant geographical limitation. Specifically, 97% of the included studies were conducted among the East Asian population (China and Japan). This may raise concerns about the external validity of our research results for non-Asian populations (particularly residents of Western countries). Additionally, squamous cell carcinoma (ESCC) is the main histological type of EC in the East Asian region, while in Western countries, adenocarcinoma (EAC) is more common. These two subtypes may exhibit different biological behaviors and responses to anti-fibrotic therapies, which may affect the efficacy of the preventive intervention measures evaluated in this analysis. Therefore, although our research results clearly indicate that the combination of oral hydrocortisone succinate and aluminum phosphate gel is the most promising strategy in the Asian population, whether it is applicable to European or North American patients remains uncertain and needs to be further corroborated. In the future, it is necessary to conduct multi-center, international RCTs to confirm the efficacy of these intervention measures in different racial groups and clinical environments. Second, heterogeneity of study a high proportion of cohort studies may introduce selection bias. Third, differences in dosage and treatment variable drug dosages among studies may affect the comparability of effect sizes. Fourth, our article collected information on adverse reactions and drug safety. However, we only covered limited adverse reactions following drug use, because the majority of the included articles did not mention this aspect. Additionally, our study focuses on drug treatments for preventing esophageal stenosis after ESD. Nonetheless, many postoperative complications, such as perforation, infection, systemic side effects, and mainly surgery-induced complications, are not caused by drug treatments. Only one report suggested that hormone use might increase the risk of perforation^[41]^, and another report indicated an increase in the risk of infection^[48]^. The remaining studies did not report any drug-related adverse reactions. However, the small sample size in most studies (<50) and certain incomplete reporting limited the reliability of the results. Despite these limitations, the novelty of this study lies in its first systematic comparison of multiple pharmacological combination regimens and identification of the optimal treatment combination. Our findings may provide direction for future clinical research and practice. Future research could further explore the optimization of dosage and administration timing as well as the potential of novel biologics in preventing esophageal stricture.
This study analyzed and evaluated different drug regimens for preventing esophageal stenosis after ESD. The results showed that the oral hydrocortisone sodium succinate combined with aluminum phosphate gel (RR = 0.075) and PGA sheet plus fibrin glue (MD = −6.4 sessions) achieved the best outcomes. Based on the results of this study, future clinical research should focus on the following directions. First, systematically comparing the preventive effects and safety of different combined medication regimens (such as combinations of glucocorticoids and mucosal protectants, and anti-fibrotic drugs) in high-risk patients to establish individualized treatment strategies. Second, developing new local sustained-release delivery systems (such as hydrogels and nanocarriers) to improve drug targeting and retention time, and reduce systemic side effects. Third, exploring molecular targeted therapy directed at key fibrotic signaling pathways (such as TGF-β and IL-6/STAT3) to expand treatment options for non-hormonal patients. Fourth, combining molecular markers and clinical characteristics to construct a stenosis risk prediction model to achieve precise stratification and early intervention. Fifth, conducting long-term follow-up studies to comprehensively evaluate the long-term recurrence rate of stenosis, functional prognosis, and safety of effective regimens, thereby strengthening the evidence base for clinical practice.