Our recent STAR Global Conference 2017 showed the growth of modeling and simulation in the Life Sciences sector. Over the three-day conference, our customers gave wide-ranging presentations, showing the huge benefit that STAR-CCM+® software is bringing to this sector. One use of was presented by Professor Klaus Affeld, from Charité – Universitätsmedizin Berlin. Professor Affeld, with co-workers Bente Thamsen and Michael Lommel, is using it to predict the formation of blood clots in heart pumps, aiming to improve pump design and reduce the formation of potentially life-threatening clots. In this blog post, I have asked him to describe his research and share his latest results with us.
Heart failure affects millions of people: 40 percent of deaths each year are related to heart failure. The lack of donor hearts limits the number of heart transplantations which can be carried out – only a small fraction of people can profit from this therapy. There is, however, an alternative therapy: the temporary or permanent implantation of a rotational blood pump which assists the failing heart. These blood pumps usually are fast rotating axial or centrifugal pumps, miniature in size and implanted in the vicinity of the heart. Figure 1 shows a cross section of such a blood pump.
Since these pumps have entered clinical practice, about 35,000 patients have so far received one. The most prominent recipient is Dick Cheney, former Vice President of the United States. He was fitted with a device in 2010 to compensate for his worsening congestive heart failure, and used the device for 22 months until a donor heart was found and successfully implanted. Using a pump in this way is known as “Bridge to Transplant.”
As successful as these blood pumps are, they are not without problems for many patients: after one year, 80 percent of recipients suffer a major adverse event, which can either be a stroke, a bleeding or an infection. The basic cause of all these events is believed to be the high speed of the rotor blades in the pump, which affects the composition of the blood. This high rotor speed is required to produce the required blood pressure, but it causes damage to the blood cells. If the red blood cells are affected, their membrane is torn and the interior hemoglobin is released – the blood is “hemolysed.” Platelets can also be affected. They become activated by the shear stress in the blood flow, deposit and form a thrombus. This thrombus impedes the function of the pump and – worse – can dislodge, cause a blockage and create a stroke, severely crippling the patient.
A thrombus is formed when both the following conditions occur:
- The platelets are activated. Normally they float passively through the circulatory system. Activation occurs when the platelets meet a lesion on the vessel wall, or, are subjected to a high shear stress (as in the blood pump).
- There is an area of low flow, where platelets have the chance to attach to the wall. This condition usually is met in an area of flow stagnation. Experiments have been performed to show this in the laboratory and to quantify the flow conditions.
The objective of our current research is to understand the flow inside the blood pump and identify critical areas where damage to the blood is likely to occur. Flow simulation is one of the main tools used to achieve this objective.
STAR-CCM+ software from Siemens PLM Software is used to compute the blood flow within the rotational blood pump. The exact geometry of the blood pump is considered intellectual property by the company, so to create the correct geometry a clinically used and later explanted pump is measured in all relevant dimensions and the pump geometry is generated by reverse engineering. Figure 2 shows the design obtained in this way. The rotational blood pump creates the blood flow by a rapidly spinning rotor. This rotor is pivot mounted between a flow straightener on the inflow side and a set of stationary blades on the outflow side. The latter are intended to convert the rotational momentum of the flow into pressure.
Once the geometry is created, a mesh is generated for the application of computational methods. A mesh of 11 million cells is used. The continuously changing geometry of rotor and housing is taken into account with “the sliding mesh” method. With this method, the mesh of the rotor is rotated within the mesh of the housing. Every two degrees a newly composed mesh is generated for the simulation. It takes about 40 iterations before convergence is achieved. About two revolutions of the rotor are required until a stable flow simulation is achieved.
Two modes are used to visualize the results: plots of the path-lines of virtual particles (figure 3) and plots of shear stress at the surface (figure 4). The former is of interest for the general flow through the pumps, to recognize larger areas of flow separations for instance. The latter – the plot of shear stress at the surface – is of greater interest: it can serve to identify the spots where a thrombus is likely to develop.
Figure 3 shows path lines of virtual particles flowing through the pump. The path lines in the middle are referenced to the spinning rotor, while the others are referenced to the stationary housing. Figure 4 shows the shear stress on the surface of the pump. The red areas at the leading edge and back of the blades show areas of high shear stress where platelets are likely to become activated. These platelets can then deposit in areas of low shear stress downstream, shaded blue, such as the end of the cylindrical part of the rotor. In this area, a boundary layer flow from right to left along the cone meets the boundary layer flow from left to right coming from the rotors. This creates a stagnation point area in a ring shape and a large flow separation is generated. A thrombus is highly likely to start to grow in this area – an example of this is shown in figure 5. Since a thrombus changes the geometry, it also changes the flow. It can itself generate a stagnation flow both in front and behind, leading to additional platelet deposition and growth of the thrombus both upstream and downstream from its initial location.
Thrombi can also form in other parts of the blood pump. For example, high shear stress occurs on the back of the rotating blades. The blades glide with 5 to 7 m/s over the cylindrical housing. The gap is only 100 microns wide. Platelets are activated here, but because of the high flow velocity cannot deposit on either the back of the blades or the housing. At the leading edge of the blade, however, both conditions for a thrombus are met: high shear stress and a stagnation point flow. The two conditions are also met on the stationary blades downstream of the rotor. Figure 6 shows a thrombus generated in these two locations.
The ability to identify critical regions in the rotational blood pump makes CFD an important tool for the development of better pump designs with reduced likelihood of thrombus formation. The main conditions in a rotational blood pump are difficult to change, and a rotational pump which will not generate any thrombi at all is probably impossible to design. To create the required pressure of about 15,000 Pa in the pump, the blood has to be accelerated to velocities much higher than found in the human body. Thus, the blood is subjected to high shear stress and is hit by the blades at high speed. CFD enables identification of the regions where thrombus generation is likely to occur, and with a change in the pump design, this flow can be modified.
You can view Prof. Affeld’s full presentation at SGC2017 here. Research like this highlights the benefits of simulation in the Life Sciences sector, helping develop a greater understanding of flow physics and delivering improved outcomes for patients.