Experimental investigation of blood damage in ventricular assist devices

In recent years, ventricular assist devices (VADs) have been advanced to such an extent that the survival rate of patients has increased signifcantly. This has been achieved primarily through the use and further development of rotary blood pumps as VADs, which are implanted completely in the body with the exception of the energy supply and control units. However, VADs are currently only implanted in patients with end-stage heart failure: patients with terminal heart failure, and their use in patients with a milder form of heart failure is limited by the high complication rate during therapy. Within one year of implantation, 70% of patients experience complications, including coagulation disorders such as intestinal bleeding and thromboembolic events such as vascular occlusions and strokes. In addition to the complications caused by blood contact with foreign materials, the unphysiologically high stress on the blood in the fow feld of the VADs has been identifed as a critical factor. To provide the required pumping capacity, the VAD rotors generate fow velocities of up to 6 m/s with shear stresses of up to 1200 Pa, which is more than 100 times higher than in the natural heart. Various blood damage models have been developed to understand the mechanisms of blood damage and the relationship between the level and duration of non-physiological stress and the complications that occur. The frequently used stress-based models are based on the power law, in which the individual parameters are ftted to the results of experiments in simple shear fows. They calculate the blood damage based on the shear stress and the corresponding exposure time. The models are used in numerical fow simulations, which have become an essential tool in the development of VADs in recent years as a result of increased computing power and improved software. Despite the progress made in numerical simulations of complex fow in VADs, the hemolysis models still provide unsatisfactory results. They can currently only be used for a rough qualitative estimation and not for a quantitative evaluation of VADs. Therefore, the need for further experiments investigating blood damage, especially in more complex fows, has been expressed in the literature to provide a basis for further improvements of the models. This dissertation presents diferent approaches to fll this research gap. It consists of three publications that provide experimental studies of blood damage resulting from stresses typical of VADs and the predictive power of hemolysis models. The frst paper describes a novel shear device that, for the frst time, allows the study of blood damage at the stress peaks characteristic of VADs under the infuence of multiple stresses. This shear device was used to study the bleeding tendency of patients after implantation. For this purpose, the infuence of shear stress on the impairment of primary hemostasis, the ability of the blood to close vascular lesions, was investigated. The extent of impairment of primary hemostasis was assessed using the Platelet Function Analyzer (PFA) device, in which a vascular injury is simulated in vitro. The results were used to create a model for increasing PFA values, which correlates with reduced coagulability, as a function of shear stress. The application of the model to the fow feld of a VAD shows that the stresses occurring lead to an impairment of primary hemostasis and thus to a coagulation disorder. In the second publication, a revised version of the shear chamber was presented and used to validate the most common hemolysis models. The models were applied to the recorded shear rates and the resulting prediction was compared to the experimental results. It could be shown that two sets of parameter of the power-law hemolysis model show a high agreement with the measured values, although they are based on a diferent design of a Couette chamber in which only single loads with a fxed shear stress threshold were investigated. However, these models signifcantly overestimate the blood damage when used in simulations of VADs. In the discussion, it was hypothesized that this overestimation is mainly due to the complex fow in VADs, with characteristic short-term load peaks in the millisecond range and diferent stress types. This is not refected by the models, as they are only based on the simple shear fow with longer loading times. This hypothesis was tested in the third publication in the form of a new test stand, that can measure the hemolysis caused by short-term stress loads under controlled conditions. Three diferent geometries were used to investigate the infuence of normal and shear stresses in the range of milliseconds on hemolysis. Remarkably, no signifcant diferences were found between the channels and the reference channel after 1200 stress repetitions, although the hemolysis models predicted increased hemolysis with signifcant diferences. The results of this publication indicate that the damage is overestimated by the models at the loads investigated, as they occur in the small gaps between the rotor and the housing of VADs, because crucial mechanisms are not taken into account. In contrast, a more complex hemolysis model, which calculates the time-dependent deformation and efective stress in the cell membrane from the stresses in the fow, shows a high level of agreement with our measured values. This type of model is referred to as “strain-based”. These three studies have expanded the knowledge of hemolysis in VADs and highlighted the weaknesses of current models. In our next studies, we will focus on measuring hemolysis in defned, more complex fow situations, varying individual parameters such as stress duration and normal and shear stress components to extend and improve the blood damage models.