BACKGROUNDGastrointestinal bleeding in patients with continuous flow left ventricular assist devices (LVADs) causes significant morbidity. Arteriovenous malformations (AVMs) cause 30–60% of bleeds, yet the efficacy of endoscopic interventions and risk factors for rebleeding have not been studied. PATIENTS AND METHODSThe charts of all LVAD patients undergoing endoscopy for gastrointestinal bleeding at Stanford between January 2010 and December 2017 were reviewed. Cox proportional hazard modeling was used to evaluate risk factors for rebleeding, including the type of endoscopic treatment, patient characteristics, and endoscopic findings. RESULTSOf 54 total LVAD patients presenting with gastrointestinal bleeding, 23 (42.6%) had AVMs documented on endoscopy. Treatment with argon plasma coagulation (APC) alone was associated with a higher risk of rebleeding compared to no treatment hazard ratio (HR)=4.77, P=0.012, and compared with clip±APC (HR=7.47, P=0.012). The 90-day bleed-free rate was 10.9% with APC, 100% with clipping±APC, and 83.3% with no endoscopic treatment. Additional risk factors for rebleeding included the presence of gastric AVMs (HR=3.64, P=0.024), and presence of hematochezia (HR=5.15, P=0.05). In a multiple Cox regression model, only the presence of gastric AVMs (HR=5.50, P=0.029) and APC use (HR=14.3, P=0.008) remained significant predictors of rebleeding. CONCLUSIONThe use of APC alone for the treatment of AVMs in LVAD patients had a high failure rate. The presence of gastric AVMs was a significant risk factor for rebleeding in LVAD patients. Management decisions should take these factors into account.