Abstract 19810: A Significant Decrease in Early Confirmed Left Ventricular Assist Device Pump Thrombosis: Recent Observations from a Large Multi-Institutional Experience

Abstract

Introduction: Recent reports have documented an abrupt increase in 3-month confirmed HeartMate II left ventricular assist device (LVAD) thrombosis beginning in 2011 without a plateau observed to date. Because very little data exist from multi-institutional efforts, we report our experience. Methods: From January 2004 to December 2013, 516 pt underwent HeartMate II LVAD implant at three institutions (University of Michigan (n=254), Mayo Clinic Rochester (n=205), and Vanderbilt Heart and Vascular Institute (n=57)). Median age at implant was 60 yr (range, 18-82) and 428 (83%) were male. Confirmed pump thrombosis at 3 months was defined as thrombus found in the blood-contacting surfaces of the pump at the time of urgent transplant, device exchange or autopsy. Pt without at least 3 months of follow-up were excluded. Results: Confirmed pump thrombosis occurred in 9 pt within 3 months post implant. The incidence was greatest in the year 2012 (5.3%). Following this peak, the incidence of 3-month confirmed pump thrombosis decreased significantly to 2.9% (Figure). Significant differences in preoperative characteristics between patients with pump thrombosis and without thrombosis included female sex (36% vs. 16%, p=0.01), bridge to transplant (77% vs. 54%, p=0.03), atrial fibrillation (55% vs. 32%, p=0.04), and preop platelet count (231,000 vs. 172,000, p<0.001). Follow-up was available in all for a total of 857 patient years of support. At a median follow-up of 1.2 years (max., 9.8 years), a similar proportion of patients were alive (73% vs. 74%, p=0.9). Conclusions: We report a recent decrease in early, confirmed HeartMate II LVAD thrombosis to historically observed levels, which is disparate to recent reports. While we identified several differences in clinical characteristics between patient groups, the issue of pump thrombosis is complex and multifactorial. Most importantly, our findings demonstrate the need for more granular analyses from collaborative efforts.