Risk factors for delayed bleeding after endoscopic resection of non-ampullary duodenal epithelial tumors and the effectiveness of complete mucosal closure in high-risk patients

Endoscopic resection (ER) of non-ampullary duodenal epithelial tumors (NADETs) is associated with a high incidence of delayed bleeding (DB). While previous reports have identified composite risk factors for delayed adverse events, including both DB and delayed perforation, the specific factors associated with DB remain unclear. This study aimed to identify factors associated with DB after ER of NADETs.

This retrospective study included 335 consecutive patients who underwent ER for a NADET between January 2004 and December 2023. Participants were divided into the DB and non-DB groups. We compared baseline characteristics and clinical outcomes between the two groups to identify factors associated with DB. ER included endoscopic submucosal dissection (ESD) and endoscopic mucosal resection (EMR) techniques. DB was defined as bleeding that required endoscopic hemostasis or transfusion after ER. Complete mucosal closure (CMC) was defined as the complete closure of the defect with no endoscopically visible exposure of the defect.

Overall, 128 ESD procedures and 207 non-ESD procedures were performed. DB occurred in 13 patients (3.9%) and its incidence was lower when CMC was achieved (1.1% vs. 13.9%, P < 0.001). Multivariate analysis revealed the use of antithrombotic agents (odds ratio (OR) 3.8; 95% confidence interval (CI): 1.0–14.6, P = 0.048) and a defect circumference ≥ 1/2 (OR 6.2; 95% CI 1.2–30.5, P = 0.029) as risk factors for DB, and CMC (OR 0.19, 95% CI 0.038–0.95, P = 0.043) as a protective factor. Among the 66 patients with risk factors, including the use of antithrombotic agents and/or a defect circumference ≥ 1/2, CMC reduced the incidence of DB (4.5% [2/44], vs. 31.8% [7/22], P = 0.005).

We identified significant factors associated with DB after duodenal ER. Particularly, in patients with risk factors for DB, CMC can reduce the incidence of DB.