Three-dimensional flow characteristics in ventricular assist devices: impact of valve design and operating conditions.

OBJECTIVE

The use of paracorporeal ventricular assist devices has become a well-established procedure for patients with cardiogenic shock. However, implantation of ventricular assist devices is often associated with severe complications, such as thrombosis inside the ventricular assist device and subsequent embolic events. It was the purpose of this study to use flow-sensitive 4-dimensional magnetic resonance imaging for a detailed analysis of the 3-dimensional (3D) flow dynamics inside a clinical routine ventricular assist device and to study the effect of different system adjustments and a new valve design on flow patterns.

METHODS

A routinely used clinical paracorporeal ventricular assist device was integrated into a magnetic resonance-compatible mock loop. Flow-sensitive 3D magnetic resonance imaging was performed to measure time-resolved 3-directional flow velocities (spatial resolution ∼ 1.2 mm, temporal resolution = 42.4 ms) in the entire device under ideal conditions (full fill, full empty, ejection fraction = 88%), insufficient filling (ejection fraction = 81%), and insufficient emptying (ejection fraction = 67%) of the pump chamber. In addition, a new valve design was evaluated. Flexible control and monitoring of pressures at inlet and outlet were used to generate realistic boundary conditions.

RESULTS

Flow pattern changes for different operating conditions were clearly identified and included reduced velocities during systolic outflow for impaired filling (78% reduction in pump flow compared with optimal operating conditions) and impaired clearing of the pump chamber for insufficient emptying (52% reduction). For all operating conditions, 3D visualization revealed vortex flow inside the ventricular assist device at typical locations of thrombus formation near the valve systems. Most noticeably, the new valve design provided similar global ventricular assist device function (pump flow 3.6 L/min), but vortex formation was eliminated.

CONCLUSIONS

The results of this study provide insight into the mechanisms underlying possible thrombus formation inside a ventricular assist device and the effect of different system adjustments. The presented methods may permit the optimization of future ventricular assist device systems with respect to optimal flow conditions.