Role of angiography and embolization for massive gastroduodenal hemorrhage

The role of mesenteric angiography and embolization for massive gastroduodenal bleeding is unclear. We reviewed the records of patients who underwent angiography for acute, nonmalignant, and nonvariceal gastric or duodenal hemorrhage that was documented but not controlled by endoscopy. Fifty patients were identified over a 7-year period ending in March 1998. Only 17 patients (34%) were originally admitted to the hospital with gastrointestinal bleeding. All required treatment in the intensive care unit (mean 15 days) with a mean APACHE III score of 79 (29% predicted hospital mortality), and 32 (64%) had organ failure. A mean of 2.1 endoscopies were performed to locate the source of acute duodenal bleeding in 37 (74%) and gastric bleeding in 13 (26%). An average of 24.3 units of packed red blood cells were transfused per patient. Twenty-five patients (50%) were found to have active bleeding at angiography; all were treated by embolization as were 22 who underwent empiric embolization. Twenty-six patients (52%) were successfully treated by embolization and thus spared imminent surgery. Multiple variables were compared between those who were successfully treated by embolization and those considered failures. Time to angiography was considerably shorter (2.5 vs. 5.8 days, P <0.017) and fewer total units of packed red blood cells were used (14.6 vs. 34, P <0.003) in those who were successfully treated. There was also a strong trend toward using fewer units of packed red blood cells for transfusion prior to angiography (11.2 vs. 17.1, P <0.08). No differences were found that could be attributed to gastric vs. duodenal sources, number of comorbid diseases, organ failure, APACHE score, age, or whether active bleeding was found at angiography. A total of 20 patients (40%) died including 9 of 17 patients operated on in an attempt to salvage angiographic failure. In summary, angiographic embolization should be performed early in the course of bleeding in otherwise critically ill patients.