Clinical outcomes of transcatheter arterial embolization in patients with angiography-negative duodenal ulcer bleeding after failed endoscopic hemostasis

Background Duodenal ulcer bleeding is a common cause of non-variceal upper gastrointestinal bleeding. Although endoscopic hemostasis is recommended as the first-line treatment, a subset of patients experience failure of endoscopic therapy and require further intervention, for which transcatheter arterial embolization(TAE) has become a commonly used salvage treatment. In clinical practice, however, angiography may fail to demonstrate definite active bleeding, posing challenges for subsequent therapeutic decision-making. Under such circumstances, empiric embolization is frequently performed, yet its clinical outcomes have not been systematically evaluated. Methods This multicenter retrospective observational study included 31 patients who underwent TAE for duodenal ulcer bleeding after failed endoscopic hemostasis between January 2017 and November 2024 and showed no definite evidence of active bleeding on angiography. Clinical characteristics, endoscopic findings, interventional procedures, and clinical outcomes were retrospectively reviewed. The primary outcome was initial clinical hemostasis following TAE, while secondary outcomes included early rebleeding, procedure-related complications, and short-term mortality. Results Most patients achieved initial clinical hemostasis after embolization, while early rebleeding occurred in a minority of cases. No major embolization-related complications, such as bowel ischemia or pancreatitis, were observed. Overall, TAE demonstrated acceptable safety and effectiveness in this high-risk patient population. Conclusions For patients with angiography-negative duodenal ulcer bleeding after failed endoscopic hemostasis, TAE is a feasible and safe salvage treatment. In the absence of definite angiographic evidence of active bleeding, empiric embolization can still achieve effective bleeding control and has important clinical applicability.