Echocardiographic Predictors of Hemodynamics in Patients Supported with Left Ventricular Assist Devices

INTRODUCTION: The assessment of hemodynamics in patients supported with left ventricular assist devices LVAD is often challenging. Physical exam maneuvers poorly correlate with true hemodynamics. We assessed the value of novel, echocardiographic (TTE) derived variables to reliably predict hemodynamics in patients supported with LVAD.
METHODS: 102 Doppler-TTE images of the LVAD outflow cannula were obtained during simultaneous invasive right heart catheterization in 30 patients (22 HMII, 8 HVAD) supported with CF-LVAD either during routine RHC or during invasive ramp testing. Properties of the Doppler signal though the outflow cannula were measured at each ramp stage (RS) including the systolic slope (SS), diastolic slope (DS) and velocity time integral (VTI). Hemodynamic variables were concurrently recorded including Doppler opening pressure (MAP), heart rate (HR), right atrial pressure, pulmonary artery pressure, pulmonary capillary wedge pressure (PCWP), Fick cardiac output (CO) and systemic vascular resistance (SVR). Univariate and multivariate regression analyses were used to explore the dependence of PCWP, CO and SVR on DS, SS, VTI, MAP, HR and RS.
RESULTS: Multivariate linear regression analysis revealed significant contributions of DS on PCWP (PCWPpred = 0.164DS + 4.959, R = 0.68). Receiver operator characteristic curve (ROC) analysis revealed that PCWPpred could predict an elevated PCWP ≥ 18 mmHg with a sensitivity (Sn) of 94% and specificity (Sp) of 85% (AUC 0.88). Cardiac output could be predicted by RS, VTI and HR (COpred = 0.017VTI + 0.016HR + 0.12RS + 2.042, R = 0.61). COpred could predict a CO ≤ 4.5 L/min with a Sn of 73% and Sp of 79% (AUC 0.81). SVR could be predicted by MAP, VTI and HR (SVRpred = 15.44MAP – 5.453VTI – 6.349HR + 856.15, R = 0.84) with a Sn of 84% and Sp of 79% (AUC 0.91) to predict SVR ≥ 1200 Dynes-sec/cm5.
CONCLUSIONS: Doppler-TTE variables derived from the LVAD outflow cannula can reliably predict PCWP, CO and SVR in patients supported with LVAD and may mitigate the need for invasive testing.