Abstract
We evaluated our novel method of controlling a continuous flow artificial heart. This method employs the detection of the total-assist and sucking points by means of pump pulsatility. Acute animal experiments were carried out using beagle dogs and our mixed flow pump. The pump was installed as the left ventricle-descending aorta bypass (LVAD: n=8) or the right atrium-pulmonary artery bypass (RVAD: n=5) through left thoracotomy. Hemodynamic parameters, pump flow rate, and motor current were monitored. To estimate the pump's pulsatility without any specific sensor, we calculated the index of current amplitude (ICA), which was obtained by dividing the amplitude of the motor current waveform by the simultaneous mean value. In the LVAD, the ICA plotted against the pump speed had a peak point (t-point) which corresponded highly with the turning point from partial to total assistance. The ICA also had a trough (s-point) which in most cases corresponded with the beginning point of severe sucking. Among preload (LVEDP), afterload (mAoP), and contractility (max LV dP/dt), only preload significantly influenced the pump flow rate at the t-point. In the RVAD, the ICA had the s-point but showed no t-point. However the pump assist flow at the s-point was larger than left heart output. These data in the LVAD indicated that the control of the pump speed towards the t-point was preload dependent, allowing us to simulate Starling's law of the natural heart. Control of RVAD by detecting the sucking phenomenon only was considered inappropriate in light of excess assist flow.
