Risk stratification for stricture formation after endoscopic submucosal dissection for esophageal dysplasia.

We aimed to evaluate the demographic, clinical, procedural, and histopathologic factors associated with stricture development following esophageal endoscopic submucosal dissection (ESD). We conducted a retrospective cohort study of patients undergoing ESD for esophageal lesions from 2019 to 2024 at St. Michael's Hospital, in Toronto, Canada. The primary outcome was stricture formation, defined as a symptomatic luminal narrowing at the ESD site confirmed on follow-up endoscopy, requiring intervention. Strictures requiring dilation developed in 24% of patients, 85% of which were impassable with a standard gastroscope (9.9 mm diameter). Stricture rates increased with defect circumferential involvement: <50% (7.7%), 50%-74% (11.5%), 75%-89% (23.1%), and ≥90% (57.7%). Intraprocedural local triamcinolone acetate (LTA) injection was administered in 40 of 108 patients (37%), with a mean defect circumferential size of 87.5%. Among patients receiving LTA, stricture rates varied based on defect size: for <50% circumferential defect involvement (n = 1) and 50%-74% (n = 3), no strictures developed; for 75%-90% (n = 17), 6 patients (35%) developed strictures, 5 of which were impassable; and for 90%-100% (n = 19), 11 patients (58%) developed strictures, all of which were impassable. Patients selectively discharged on prophylactic steroids demonstrated varied stricture rates depending on the steroid regimen: prednisone (61.5%), oral budesonide (26.9%), and combination therapy (7.7%). Independent predictors of stricture formation included defect circumferential involvement (OR 1.07, 95% CI 1.03-1.12, p < 0.001), length of hospitalization (OR 1.88, 95% CI 1.11-3.16, p = 0.018), and presence of deep mural injury (OR 6.28, 95% CI 1.10-35.88, p = 0.039). Stricture formation post-ESD is strongly associated with lesion and procedural characteristics, including defect circumferential involvement, deep mural injury, and length of hospitalization.