Gastrointestinal bleeding and subsequent risk of thromboembolic events during support with a left ventricular assist device

Abstract

Modern left ventricular assist devices (LVAD) require anti-coagulation (AC) with warfarin and anti-platelet therapy to prevent thromboembolic complications in patients. Gastrointestinal bleeding (GI) is a significant adverse event in these patients and treatment typically requires reduction or elimination of AC or anti-platelet therapy. It is not known whether alterations in AC to treat GI bleeding influence subsequent risk of thromboembolic (TE) events during LVAD support. Between July 2003 and September 2011, 389 patients (308 male) underwent implantation of a continuous-flow LVAD at the University of Michigan Health System and the Mayo Clinic. Median age at implant was 60 years (range 18 to 79 years). Outcomes were analyzed for the association of GI bleeding events and subsequent TE events, defined as stroke, transient ischemic attack, hemolysis or suspected or confirmed pump thrombosis. Median survival was 10 months (maximum 7.2 years, total 439 patient-years). TE events occurring within the first 30 days were not counted. Overall survival and freedom from an outcome event were assessed using the Kaplan-Meier method. Associations between GI bleeding and subsequent TE events and survival impact were analyzed as time-dependent covariates. One hundred ninety-nine GI bleeding episodes occurred in 116 of 389 patients (30%) for an event rate of 0.45 GI bleed/patient-year of support. One hundred thirty-eight TE events occurred in 97 of 389 patients (25%) for an event rate of 0.31 TE event/patient-year of support. Median time from LVAD implant to first GI bleed was 5 months (range 1 to 116 months) and to first TE event was 6 months (range 1 to 29 months). For patients who had a TE event after GI bleed, the median interval was 5 months (range 0.5 to 25 months). TE events were 7.4-fold more likely in patients who had a prior GI bleed (range 4.9- to 11.1-fold) (p < 0.001); however, neither the presence of GI bleeding (0.7 to 1.2) nor a TE event (0.8 to 2.0) portended a lower overall survival. Patients who had GI bleeding were at significantly higher risk for a subsequent TE event. Although the exact cause of this relationship is unknown, it suggests that a reduction in anti-coagulation and anti-platelet management to treat GI bleeds may contribute to this risk.