Cerebrovascular Events in Patients With Centrifugal-Flow Left Ventricular Assist Devices Propensity Score-Matched Analysis From the Intermacs Registry

Abstract

BACKGROUND: Ischemic and hemorrhagic cerebrovascular accidents remain common among patients with centrifugal-flow left ventricular assist devices, despite improvements in survival and device longevity. We compared the incidence of neurologic adverse events (NAEs) associated with 2 contemporary centrifugal-flow left ventricular assist devices: the Abbott HeartMate3 (HM3) and the Medtronic HeartWare HVAD (HVAD). METHODS: Using the Society of Thoracic Surgeons Interagency Registry for Mechanically Assisted Circulatory Support (Intermacs), we collected data on adult patients who received a centrifugal-flow left ventricular assist device as a primary isolated implant between January 1, 2017, and September 30, 2019. Major NAEs were defined as transient ischemic attack, ischemic cerebrovascular accident, or hemorrhagic cerebrovascular accident. The association of HVAD with risk of NAE in the first year after implant was evaluated using propensity score matching to balance for preimplant risk factors. After matching, freedom from first major NAE in the HM3 and HVAD cohorts was compared with Kaplan-Meier curves. A secondary analysis using multivariable multiphase hazard models was used to identify predictors of NAE, which uses a data-driven parametric fit of the early declining and constant phase hazards and the associations of risk factor with either phase. RESULTS: Of 6205 included patients, 3129 (50.4%) received the HM3 and 3076 (49.6%) received the HVAD. Median followup was 9 and 12 months (HM3 and HVAD, respectively). Patients receiving HVAD had more major NAEs (16.4% versus 6.4%, P<0.001) as well as each subtype (transient ischemic attack: 3.3% versus 1.0%, P<0.001; ischemic cerebrovascular accident: 7.7% versus 3.4%, P<0.001; hemorrhagic cerebrovascular accident: 7.2% versus 2.0%, P<0.001) than did patients receiving HM3. A propensity-matched cohort balanced for preimplant risk factors showed that HVAD was associated with higher probabilities of major NAEs (% freedom from NAE 82% versus 92%, P<0.001). Device type was not significantly associated with NAEs in the early hazard phase, but HVAD was associated with higher incidence of major NAEs during the constant hazard phase (hazard ratio, 5.71 [CI, 3.90-8.36]). CONCLUSIONS: HM3 is associated with lower hazard of major NAEs than is HVAD beyond the early postimplantation period and during the constant hazard phase. Defining the explanation for this observation will inform device selection for individual patients.