Impact of heart failure drug therapy on rates of gastrointestinal bleeding in LVAD recipients: An INTERMACS analysis

Gastrointestinal bleeding (GIB) remains a common and vexing complication of left ventricular assist device (LVAD) support. Recent single-center analyses suggest that ACE inhibitors (ACEi)/angiotensin receptor blockers (ARB) and digoxin may prevent GIB in LVAD patients. Here we evaluate the effect of guideline-directed medical therapies (GDMT) for heart failure (HF) on rates of GIB through analysis of the INTERMACS registry database.Thirteen thousand seven hundred thirty-two patients who received a continuous-flow LVAD and were on antiplatelet therapy and anticoagulation with warfarin after 3 months of pump support were included in the analysis. GIB events following implant were assessed based on receipt of ACEi/ARB, beta-blockers (BB), mineralocorticoid receptor antagonist (MRA), amiodarone, digoxin, loop diuretics, and phosphiesterase-5 inhibitors (PDE5). Backwards stepwise cox regression was used to control for confounding of each drug class on each other, as well as for clinical variables like age, gender, renal function, HF etiology, and device strategy.After 3 months of pump support medications used in LVAD patients were BB (65.0%), ACEi/ARB (51.7%), Amio (43.7%), MRA (37.9%), and loop diuretics (70.1%). In patients with available data, PDE and digoxin use were 18.2% and 16.9%, respectively. The overall incidence of GIB was 19.5% at 2 years of support. After adjustment for other clinical variables, loop diuretics (HR 1.274, p < 0.001) and PDE5 (HR 1.241, p < 0.001) use were associated with increased risk of GIB, while use of BB (HR 0.871, p = 0.006) was associated with lower risk of GIB. ACEi/ARB (HR 1.002, p = 0.971), Amio (HR 1.083, p = 0.106), AA (HR 0.967, p = 0.522) or digoxin (HR 1.087, p = 0.169) did not affect GIB rates on LVAD support (Figure).Despite recent reports, ACEi/ARB, MRA, Amio, and digoxin use does not appear to be associated with GIB during LVAD support. The heightened risk seen in those on loop diuretics may reflect venous congestion in these patients, while antiplatelet effects of PDE5 could drive the higher risk of GIB.