This retrospective study included 128 lesions in patients who underwent ESD for superficial esophageal neoplasms at Chonnam National University Hwasun Hospital between January 2012 and December 2023. All procedures were performed using standard ESD techniques. Prophylactic strategies were applied at the endoscopist’s discretion based on lesion characteristics. Local triamcinolone injection (typically 10–40 mg) was administered into the post-ESD mucosal defect margin, and oral prednisolone was prescribed for selected high-risk cases. Post-procedural stricture was defined as inability of standard upper endoscope passage at follow-up examination. First follow-up endoscopy was generally performed approximately 2 months after ESD (median 66 days; interquartile range, 49–81 days), with earlier endoscopy when dysphagia developed; however, follow-up timing was not fully standardized because of the retrospective design. Circumferential extent was defined according to the proportion of the esophageal circumference involved by the post-ESD mucosal defect and was classified as <50%, 50% to 75%, or >75% circumference. Tumor location was categorized as upper (≤25 cm), middle (26–32 cm), or lower (≥33 cm) esophagus according to the proximal lesion margin from the upper incisors. JES classification was defined according to the Japan Esophageal Society magnifying endoscopic classification for superficial esophageal squamous neoplasia [2,3]. Muscular injury was graded using the 5-grade classification proposed by Xu et al. [14] as follows: Grade 0, no obvious myofilament fracture; Grade 1, mild circular muscle fracture involving <50% of the circumference; Grade 2, mild fracture involving >50%; Grade 3, severe fracture involving <50%; and Grade 4, full-thickness injury with visible longitudinal muscle or severe fracture involving >50%. Baseline comparisons between stricture and non-stricture groups were performed using Mann–Whitney U test for continuous variables, Fisher’s exact test or chi-square test for categorical variables. Univariate logistic regression analyses were performed to identify factors associated with post-procedural stricture. Because tumor size, specimen size, and circumferential extent represent overlapping measures of lesion or resection extent, highly correlated variables were not entered simultaneously into the multivariate model. Given the limited number of stricture events, multivariate logistic regression included age and circumferential extent (>75% vs. ≤75%). The study was approved by the Institutional Review Board (IRB) of Chonnam National University Hwasun Hospital (IRB No: CNUHH-2026-026). Informed consent was waived by the IRB because of the retrospective design.

Baseline characteristics by stricture status are summarized in Table 1. Among 128 lesions analyzed, mean patient age was 67.0±10.1 years, and all 13 patients who developed stricture were male. Post-procedural stricture developed in 13 lesions (10.2%), with en bloc resection achieved in 125 lesions (97.7%). Stricture rates showed a strong association with circumferential extent of the post-ESD mucosal defect (p<0.001). Among 49 lesions involving <50% circumference, 1 (2.0%) developed stricture. For 50% to 75% circumference (n=54), stricture occurred in 1 case (1.9%). In contrast, 11 of 25 lesions (44.0%) involving >75% circumference developed stricture.

Prophylactic strategies varied according to lesion risk (Table 2). Overall, 52 lesions (40.6%) received no prophylaxis, 54 (42.2%) received local triamcinolone injection only, and 22 (17.2%) received combined triamcinolone injection plus oral prednisolone. Among patients receiving oral prednisolone (n=22), mean daily starting dose was 23.0±9.3 mg (range 10–30 mg) with mean duration of 55.1±20.0 days (median 43 days, range 19–88 days). Treatment selection was strongly associated with circumferential extent, with prophylaxis used more frequently for larger post-ESD mucosal defects. Overall stricture rates were 1.9% with no prophylaxis, 13.0% with triamcinolone injection only, and 22.7% with combined therapy, reflecting selection bias because higher-risk lesions preferentially received prophylaxis. In lesions with >75% circumference mucosal defects, stricture occurred in 1 of 3 lesions without prophylaxis, 6 of 12 treated with triamcinolone alone, and 4 of 10 treated with combined triamcinolone plus oral prednisolone. These subgroup results are presented descriptively only and should not be interpreted as comparative efficacy estimates.

The clinical burden of stricture was substantial (Table 3). Among 13 stricture cases, endoscopic balloon dilation was performed in 11 patients (84.6%), requiring mean 2.2±1.8 sessions (range 0–6). Any muscle layer injury during ESD occurred in 11 of 13 stricture cases (84.6%), and 69 of 115 non-stricture cases (60.0%) (Table 1). When analyzed using the 5-grade muscular injury classification, higher muscle injury grade was associated with post-ESD stricture in univariate logistic regression (odds ratio [OR] 1.604 per 1-grade increase, 95% confidence interval [CI] 1.003–2.566; p=0.049).

Risk factor analysis revealed several significant associations (Table 4). In univariate logistic regression, age (OR 1.072 per year, p=0.046), tumor size (long axis; OR 1.082 per mm, p<0.001), specimen size (long axis; OR 1.076 per mm, p<0.001), JES classification (OR 1.444 per grade, p=0.020), procedure time (OR 1.025 per minute, p=0.001), muscular injury grade (OR 1.604 per 1-grade increase, p=0.049), and circumferential extent (>75% vs. <50%; OR 37.714, p<0.001) were associated with stricture. Tumor location in the upper esophagus showed a trend toward increased stricture risk compared with the lower esophagus (OR 4.889, 95% CI 0.846–28.236; p=0.076), although this did not reach statistical significance. Because of the small number of events and the potential for multicollinearity among lesion extent variables, the multivariate model included age and circumferential extent (>75% vs. ≤75%) only. In this model, >75% circumferential mucosal defect remained an independent risk factor (adjusted OR [aOR] 34.773, 95% CI 6.880–175.747; p<0.001), whereas age was not (aOR 1.046, 95% CI 0.967–1.132; p=0.261).

This 12-year experience demonstrates that circumferential extent of the post-ESD mucosal defect is the major risk factor for post-ESD stricture. The 10.2% overall stricture rate is consistent with previous reports, though lower than some series likely reflecting the predominance of smaller mucosal defects in our cohort [4-6]. Stricture risk increased markedly when the post-ESD mucosal defect exceeded three quarters of the circumference, reaching 44.0% in that subgroup. This finding is consistent with prior studies showing that circumferential extent is the strongest predictor of post-ESD stricture [4-6,8,15]. Notably, all patients who developed stricture in our cohort were male, reflecting the well-known male predominance of esophageal squamous cell neoplasms in Korea rather than a sex-specific risk for stricture development.

Our analysis of prophylactic strategies should be interpreted cautiously because treatment allocation was non-randomized and strongly influenced by lesion extent. The higher observed stricture rates in prophylaxis-treated groups likely reflect preferential use of steroid therapy in larger and higher-risk post- ESD mucosal defects rather than inferior treatment efficacy. For transparency, we present descriptive stricture rates according to prophylactic regimen both overall and within the >75% subgroup. In lesions with >75% circumferential mucosal defects, stricture occurred in 6 of 12 lesions treated with triamcinolone alone and 4 of 10 lesions treated with combined triamcinolone plus oral prednisolone, while 1 of 3 lesions without prophylaxis developed stricture. Given the small subgroup sizes and treatment selection bias, these findings should be interpreted as descriptive and hypothesis-generating only.

When muscle injury was evaluated using the 5-grade system, higher injury grade was associated with stricture on exploratory univariate analysis (OR 1.604 per 1-grade increase, 95% CI 1.003–2.566; p=0.049). This finding is consistent with recent reports that deeper or more extensive muscular injury increases post-ESD stricture risk [14,16]. Muscular injury grade was not entered into the multivariate model because it was strongly correlated with circumferential extent in our data and likely represents an intermediate step on the causal pathway from extensive resection to stricture. The association between longer procedure time and stricture in univariate analysis likely reflects technical difficulty and extent of tissue manipulation, consistent with recent risk stratification models [13]. These observations underscore the importance of operator experience and refined technique in minimizing procedural trauma during extensive resections.

Tumor size, specimen size, and circumferential extent showed significant associations with stricture in univariate analysis. Because these variables are closely related, they were not included together in the multivariate model. We therefore selected circumferential extent, based on the post-ESD mucosal defect, as the representative extent-related variable. After adjustment for age, a >75% circumferential mucosal defect remained an independent risk factor, whereas age did not. This approach was used to avoid overfitting and multicollinearity given the small number of stricture events.

The clinical burden of stricture in our series was substantial, as evidenced by the need for repeated endoscopic balloon dilation (mean 2.2 sessions, up to 6 sessions), emphasizing the importance of effective prophylactic strategies and careful patient selection for ESD versus alternative treatment approaches in very high-risk lesions. This is consistent with contemporary guidelines suggesting that when the predicted stricture risk is extremely high, alternative approaches including esophagectomy or definitive chemoradiotherapy should be considered, particularly in surgical candidates [1-3].

Several limitations should be acknowledged. First, this was a retrospective single-center study, which limits generalizability and introduces the possibility of selection bias. Second, prophylactic treatment was not randomly assigned but rather determined according to clinical judgment and lesion characteristics, making comparative assessment of prophylactic efficacy difficult. Third, the number of stricture events was relatively small, which limited the robustness of subgroup analyses and the extent of multivariable adjustment. Fourth, followup timing and steroid regimens were not fully standardized, and clinical practice patterns may have changed over the 12-year study period. Finally, circumferential extent of the post-ESD mucosal defect and muscle injury grade were assessed retrospectively, which may have introduced some degree of measurement bias. Future prospective studies with standardized risk assessment and prophylactic protocols are needed.

In conclusion, circumferential extent of the post-ESD mucosal defect was the strongest predictor of post-ESD stricture, with a marked increase in risk when the defect involved >75% of the esophageal circumference. These findings highlight the importance of careful risk stratification after esophageal ESD and suggest that lesions expected to result in extensive circumferential mucosal defects require particular attention with respect to prophylactic planning and treatment selection.