After JCOG1217: Toward Risk-Adapted Steroid Prophylaxis for Post-ESD Esophageal Strictures

Abstract

Endoscopic submucosal dissection (ESD) is an established treatment for superficial esophageal squamous cell carcinoma, enabling en bloc resection and precise histological assessment. However, post-ESD stricture remains a clinically important adverse event after extensive resection involving a large circumferential mucosal defect. Dysphagia requiring repeated endoscopic balloon dilation (EBD) can substantially impair quality of life and compromise the minimally invasive benefit of endoscopic treatment. Stricture prevention is therefore not merely a technical adjunct to esophageal ESD but an essential component of perioperative management.

Among available prophylactic strategies, corticosteroid-based approaches have been the most widely adopted because of their efficacy, accessibility, and relatively low cost. Before JCOG1217 [1], both local triamcinolone injection and oral prednisolone had shown favorable outcomes, but the supporting evidence was derived mainly from small, single-arm, or non-randomized comparative studies. In representative early studies, local triamcinolone injection reduced post-ESD stricture rates to 10%–19%, compared with 66%–75% in untreated or historical controls [2, 3]. Oral prednisolone, usually initiated at 30 mg/day and tapered over several weeks, reduced stricture rates to 5.3%–17.6%, compared with 31.8%–68.7% after preemptive EBD or no prophylaxis [4-6]; however, the 30 mg/day starting dose subsequently adopted in JCOG1217 was not established through dose-ranging evaluation, and neither routine dose reduction nor escalation can currently be recommended. Thus, both routes of steroid administration were clinically reasonable, but the available evidence was fragmented and insufficient to define the preferred approach after extensive esophageal ESD.

In this issue of Digestive Endoscopy, Tanaka and colleagues report the results of JCOG1217, a multicenter, open-label, randomized phase 3 trial comparing oral prednisolone with local triamcinolone injection for the prevention of stricture after extensive esophageal ESD [1]. The trial enrolled 281 patients from 45 institutions with superficial esophageal squamous cell carcinoma involving at least one-half, but not the entire circumference, of the esophageal lumen. Patients were randomized to receive either local triamcinolone injection, administered immediately after ESD at a total dose of 100 mg with avoidance of exposed muscle, or oral prednisolone, started at 30 mg/day on day 3 after ESD and tapered over 8 weeks, with prophylactic trimethoprim-sulfamethoxazole. The primary endpoint was stricture-free survival (SFS), and the trial was designed as a superiority trial testing whether oral prednisolone was superior to local triamcinolone injection, rather than establishing equivalence or noninferiority. Its importance lies in placing the two most widely used steroid-based strategies on the same multicenter randomized platform, thereby providing a clinically relevant benchmark for post-ESD stricture prophylaxis.

The principal finding is clear: Oral prednisolone yielded numerically better outcomes but did not demonstrate superiority over local triamcinolone injection. The 12-week SFS rate was 88.5% in the triamcinolone arm and 94.8% in the prednisolone arm (hazard ratio, 0.672; 95% confidence interval, 0.321–1.408; one-sided p = 0.1442), and the prespecified criterion for superiority was not met [1]. Secondary endpoints showed a similar numerical pattern without statistical significance: EBD within 12 weeks was performed 38 times in 14 patients in the triamcinolone arm and 26 times in 10 patients in the prednisolone arm (p = 0.38), and the proportion of patients with a dysphagia score of 0 or 1 at 12 weeks was 73.8% and 80.7%, respectively (p = 0.20) [1]. Thus, JCOG1217 does not justify simply replacing local triamcinolone injection with oral prednisolone in routine clinical practice. However, the directionally consistent results across SFS, EBD use, and dysphagia score remain clinically relevant and support further evaluation of oral prednisolone in patients with higher stricture risk.

The trial provides a pragmatic framework for treatment selection. For many patients undergoing extensive noncircumferential esophageal ESD, local triamcinolone injection remains a reference strategy because prophylaxis can be completed during the index ESD session, avoiding a scheduled oral steroid course, prolonged systemic exposure, antimicrobial prophylaxis, and post-discharge adherence. This workflow may be particularly advantageous when additional treatment must be considered after noncurative resection. This strategy, however, is technique-dependent: ESD should preserve a thin residual submucosal layer over the muscularis propria, and triamcinolone should be delivered uniformly into this layer while avoiding intramuscular injection. Oral prednisolone is a reasonable alternative when systemic steroid exposure is acceptable and local injection is technically difficult or potentially unsafe. Thus, route selection should be individualized according to expected efficacy, systemic risk, procedural feasibility, and treatment burden.

The safety profile further helps define the role of each strategy. Grade 3/4 adverse events occurred in 8.7% of patients in the triamcinolone arm and 6.7% in the prednisolone arm [1]. Although the overall incidence of severe adverse events was similar, the nature of risk differed between the two approaches. Oral prednisolone was associated with systemic corticosteroid-related adverse events, including hyperglycemia and infection; one serious adverse event, grade 4 hyperglycemia, occurred in this arm. Accordingly, patients receiving oral prednisolone require appropriate outpatient follow-up, including laboratory monitoring and clinical assessment for worsening glycemic control, infection, and psychiatric symptoms. In JCOG1217, prophylactic trimethoprim-sulfamethoxazole was also administered. By contrast, local triamcinolone injection avoids prolonged systemic steroid exposure but carries procedure-related risks, as delayed perforation with mediastinitis occurred in two patients in this arm [1].

The subgroup findings raise a practical clinical question: how to prevent strictures after near- or whole-circumferential post-ESD mucosal defects. For post-ESD mucosal defects involving three-fourths to less than seven-eighths of the circumference, the 12-week SFS rates were almost identical between the triamcinolone and prednisolone arms (96.0% vs. 96.5%). In contrast, among patients with post-ESD mucosal defects involving seven-eighths or more but not the entire circumference, the corresponding rates were 75.2% and 92.5%, respectively (hazard ratio, 0.545; 95% confidence interval, 0.198–1.502). For post-ESD whole-circumference mucosal defects, the rates were 50.0% and 75.0%, respectively (hazard ratio, 0.360; 95% confidence interval, 0.040–3.223) [1]. Although these subgroup estimates numerically favored prednisolone in more extensive defects, their main clinical implication is hypothesis-generating: Oral prednisolone may deserve further evaluation in risk-enriched noncircumferential defects, whereas stricture prevention remains insufficient as mucosal defects approach the entire circumference.

In a recent retrospective study of mucosal defects involving seven-eighths or more but not the entire circumference, local triamcinolone injection plus oral prednisolone showed numerically favorable outcomes compared with local injection alone: lower stricture incidence (26.9% vs. 42.9%), higher 12-week SFS (76.9% vs. 64.3%), and fewer EBD sessions among patients who developed strictures (median, 4 vs. 6), although none reached statistical significance [7]. In addition, muscular injury has emerged as an independent risk factor for severe stricture after extensive esophageal ESD [8, 9]. In 49 patients undergoing entire circumferential esophageal ESD with combined local and oral steroid prophylaxis, postoperative and refractory strictures occurred in 65% and 51%, respectively, and muscular injury independently predicted refractory stricture (odds ratio, 16.2; 95% confidence interval, 2.04–129.1) [9]. These findings underscore that effective prevention of refractory stricture in this subgroup may depend not only on intensified steroid prophylaxis, but also on high-quality ESD with meticulous preservation of the muscularis propria.

JCOG1217 and emerging data in higher-risk populations have two practical implications. Low- to intermediate-risk defects may be managed with well-performed local triamcinolone injection, avoiding unnecessary systemic steroid exposure. In contrast, near-circumferential and circumferential defects should not be treated as a single risk category: The former may be candidates for intensified steroid prophylaxis, including oral prednisolone-based or combination approaches, whereas the latter may require strategies beyond steroid intensification alone. Future studies should therefore move beyond circumferential extent alone and prospectively evaluate composite risk models incorporating longitudinal defect length, prior ESD or chemoradiotherapy, and intraoperative muscular injury [8-10]. Such stratification may help identify patients who require intensified prophylaxis, close follow-up, and early intervention before refractory stricture develops, and provide a basis for prospective trials integrating intensified prophylaxis with quality-controlled ESD aimed at minimizing muscular injury.

In conclusion, JCOG1217 provides the first randomized benchmark for steroid prophylaxis after extensive noncircumferential esophageal ESD and does not support routine replacement of local triamcinolone injection with oral prednisolone. Its findings instead point toward prospective, risk-defined studies to determine whether intensified perioperative strategies can prevent refractory stricture in selected high-risk patients.

The authors have nothing to report.

The authors declare no conflicts of interest.

This article is linked with Tanaka et al. http://doi.org/10.1111/den.70195.

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.