Predictors of Recurrent Gastrointestinal Bleeding in Patients With Left Ventricular Assist Device: Device Settings

Background: Use of left ventricular assist device (LVAD) as a bridge to transplant (BTT) and especially as a destination therapy (DT) is increasing. Gastrointestinal bleeding (GIB) is a common complication observed in LVAD recipients. Underlying hematologic abnormalities in LVAD patients are well described. However, the risk factors associated with GIB remain unclear. Methods: Electronic medical records of 59 HeartMate II (HM-II) recipients at a university hospital and 52 HM-II recipients at a community hospital were reviewed for demographic data, social history, medication use, and co-morbidities. Data were analyzed using analysis of variance, chi-square analysis and logistic regression to identify predictors of GIB and rebleeding. Results: Of the 111 patients analyzed 31 had at least one GIB (27.9%). Caucasians represented the majority of the patients (n=55; 49.5%) followed by African-Americans (n=44; 39.6%). Age, gender, BMI, presence of hypertension, peripheral vascular disease, diabetes, aortic stenosis, cerebrovascular disease, chronic kidney disease (CKD), and history of GIB, were not associated with GIB after LVAD implant. A logistic regression model including history of smoking (OR = 3.13, 95% CI=1.13- 8.68, p= 0.028) and NSAID use (OR=4.28, 95% CI= 1.03-17.81, p=0.045) was as predictive of GIB post-LVAD. The most predictive model for rebleeding included indication for LVAD (OR=0.068, 95% CI=0.008-0.601, p=0.016), history of smoking (OR = 5.06, 95% CI=0.97-26.49, p=0.05) and NSAID use (OR=9.46, CI=1.21-73.87, p=0.03). 66.7% (10/15) of patients with recurrent bleeding had CKD versus 36.5% (36/95) of patients without recurrent bleeding (p=0.045). Patients with LVAD as a BTT were found to be 93.2% less likely to rebleed. Rebleeders had significantly higher LVAD power (5.50 ± 0.732; CI=5.09 - 5.91 vs. 4.85±1.16; CI=4.60-5.09; p=0.038), pulsatility index (6.04 ± 0.94; CI=5.52-6.56 vs. 5.27 ± 1.27; CI=5.01-5.53; p=0.027) and cardiac output (4.81 ± 1.27; CI= 4.10 - 5.51 vs. 3.89±1.66; CI 3.53-4.26; p=0.046). Conclusion: Patients with BTT as indication for LVAD are much less likely to rebleed. Moreover, lower power, pulsatility index, and cardiac output might prevent rebleed. Larger studies are required to confirm such associations.