BACKGROUND: Scant information is available regarding patients with upper gastrointestinal bleeding (UGIB) from tumours. AIM: To determine the presentation, endoscopic findings, treatment and outcomes in patients with UGIB from malignant tumours and identify risk factors associated with rebleeding. METHODS: Consecutive patients who were hospitalised with haematemesis, melena or haematochezia and underwent upper endoscopy were identified retrospectively by reviewing an endoscopy database. Patients with UGIB due to biopsy-proven malignant tumours were studied. RESULTS: Tumours were the source of bleeding in 106 (5%) of 2,166 patients with UGIB. Tumours were oesophageal in 17 (16%), gastric in 77 (73%) and duodenal in 12 (11%). At presentation, 84 (79%) did not have known cancer previously, and 79 (75%) had metastatic disease. Seventy-seven (73%) received transfusions at index hospitalisation. At endoscopy, 32 (30%) had active bleeding (31 oozing, 1 spurting). Among actively bleeding patients, haemostasis was achieved in 12 (86%) of 14 receiving endoscopic therapy and all 18 not receiving endoscopic treatment. Hospitalisation for rebleeding occurred in 50 (49%) of 103 at a median of 30 days (3-885). On multivariate analysis, age ≤60 years (OR = 2.49, 95% CI 1.06-5.81) and haemodynamic instability (OR = 2.42, 95% CI 1.08-5.46) were associated with rebleeding. CONCLUSIONS: Patients presenting with tumour-associated UGIB have substantial blood loss, with three-quarters requiring transfusion at presentation. Initial haemostasis occurs in almost all patients, with or without endoscopic therapy, but rebleeding requiring repeat hospitalisation occurs in approximately half the patients and is more common in patients who are ≤60 years of age and have haemodynamic instability at presentation.
BACKGROUND: Scant information is available regarding patients with upper gastrointestinal bleeding (UGIB) from tumours. AIM: To determine the presentation, endoscopic findings, treatment and outcomes in patients with UGIB from malignant tumours and identify risk factors associated with rebleeding. METHODS: Consecutive patients who were hospitalised with haematemesis, melena or haematochezia and underwent upper endoscopy were identified retrospectively by reviewing an endoscopy database. Patients with UGIB due to biopsy-proven malignant tumours were studied. RESULTS:Tumours were the source of bleeding in 106 (5%) of 2,166 patients with UGIB. Tumours were oesophageal in 17 (16%), gastric in 77 (73%) and duodenal in 12 (11%). At presentation, 84 (79%) did not have known cancer previously, and 79 (75%) had metastatic disease. Seventy-seven (73%) received transfusions at index hospitalisation. At endoscopy, 32 (30%) had active bleeding (31 oozing, 1 spurting). Among actively bleedingpatients, haemostasis was achieved in 12 (86%) of 14 receiving endoscopic therapy and all 18 not receiving endoscopic treatment. Hospitalisation for rebleeding occurred in 50 (49%) of 103 at a median of 30 days (3-885). On multivariate analysis, age ≤60 years (OR = 2.49, 95% CI 1.06-5.81) and haemodynamic instability (OR = 2.42, 95% CI 1.08-5.46) were associated with rebleeding. CONCLUSIONS:Patients presenting with tumour-associated UGIB have substantial blood loss, with three-quarters requiring transfusion at presentation. Initial haemostasis occurs in almost all patients, with or without endoscopic therapy, but rebleeding requiring repeat hospitalisation occurs in approximately half the patients and is more common in patients who are ≤60 years of age and have haemodynamic instability at presentation.