Fluid-Structure Interaction Simulations for Transcatheter Aortic Valve Replacement Surgeries
Transcatheter aortic valve replacement (TAVR) has become a standard treatment option for intermediate and high-risk patients suffering from aortic stenosis. As a result, clinicians have a wide array of transcatheter aortic valves to choose from, some of which are shown below in Figure 1.
Figure 1: Selected examples of transcatheter aortic valve devices that are routinely used in TAVR.
To select the most suitable device for a given patient, among many options, clinicians benefit from multimodality imaging methods such as echocardiography, computed tomography, and magnetic resonance imaging. These tools are also indispensable to develop an optimum surgical plan for TAVR. However, such approaches alone are often not enough to fully understand the interaction between the device and the patient’s body (device-host interaction), leading to post-TAVR complications, including flow regurgitation, device migration, cardiac conduction abnormalities, thrombus formation and coronary obstruction. The PARTNER trial showed that device-related post-surgical complications after TAVR are a traumatic experience for patients and represent a considerable burden on healthcare agencies regarding resources and costs associated with re-intervention procedures. It is estimated that one-third of patients suffer from device-related post-TAVR complications within the month after the surgery (Fassa et al., 2013), resulting in additional hospital costs of approximately $12,500 per patient (Arnold et al., 2014). With the elderly population set to double by 2028 (d’Arcy et al., 2011), this burden is only estimated to increase over time.
One of the most effective ways to anticipate or even eliminate device-related post-TAVR complications is to virtually perform the surgery during the planning phase to test different TAVR devices at various implantation locations and predict their post-operative efficacy on a patient-specific (personalised) basis. Fluid-structure interaction (FSI) simulations lie at the very heart of such virtual surgical platforms. However, performing such FSI simulations is an intricate and challenging task because the extreme and sudden deformations of the valve leaflets can lead to numerical instabilities. Traditional computational methods typically used to perform heart valve FSI simulations, e.g., immersed boundary (IB) method and arbitrary Lagrangian-Eulerian (ALE) method, are computationally costly in terms of memory and time, which make it ineffective for real-life clinical use.
To address the challenge of making FSI simulations suitable for real-life practical applications, FlowVision has developed a revolutionary boundary conforming FSI approach called the sub-grid geometry resolution (SGGR) method, which can effectively overcome the abovementioned numerical limitations. In this method, a Cartesian mesh is created initially for the entire fluid domain, which is then intersected by the wall boundaries, leading to a layer of perfectly conforming polyhedral cells around the wall. As a result, the precise topology of the underlying geometry on which FSI simulations are to be performed is maintained throughout, as highlighted in Figure 2. Furthermore, as this is a boundary conforming approach, the velocity gradient at the moving walls can be resolved effectively, which facilitates the quantification of wall shear stress at the boundaries. Hence, SGGR based FSI simulations are as stable as the IB method and as accurate as the ALE method.
Figure 2: The moving valve leaflets intersect the initial Cartesian mesh for the fluid domain to give a layer of perfectly conforming polyhedral cells around the moving boundaries, thus maintain their geometric topology.
Other features that enable FlowVision to perform clinically realistic TAVR FSI simulations are:
⦁ The ability to easily assign lumped parameter networks, such as three-element Windkessel models as outflow boundary conditions.
⦁ A proprietary gap model which accurately resolves the flow field in the narrow clearances (down to 1 micron) within the computational domain without the need for an ultra-refined grid.
⦁ Ability to simulate full closure of the valve leaflets with and contact without additional ‘tricks.
FSI simulations are often perceived as complex as it may be challenging to couple the flow and structural solvers and control the data exchange during the simulation whilst ensuring numerical stability. FlowVision offers a straightforward and intuitive workflow to enable coupling with many industry-leading FE solvers such as Abaqus and Ansys Mechanical. For example, FlowVision can be coupled directly and efficiently with Abaqus using its co-simulation engine. Due to above mentioned lucrative features of FlowVision, several research labs and industry leaders from around the world are using the FlowVision platform to tackle the most advance and complicated TAVR FSI simulations, e.g., Kandail et al. (2018), Sodhani et al. (2018), Ghosh et al. (2020), among others. Key results from Kandail et al. (2018) are highlighted below.
In their numerical investigation, Kandail and colleagues deployed a 29 mm CoreValve (Medtronic Inc., USA) at two different locations within a patient-specific aortic root (Figure 3). They quantified the post-operative blood flow patterns and wall shear stress using FlowVision and Abaqus/Explicit.
Figure 3 : Virtual deployment of a 29 mm CoreValve inside a patient-specific aortic root at an annular (left) and supra-annular (right) location.
Through their FSI simulations, they delineated the precise effect that each deployment location had on the resulting flow patterns. For example, they reported that when the maximum flow deceleration occurs in the cardiac cycle, valve leaflets of the annularly deployed CoreValve closes relatively asymmetrically in contrast to the supra-annularly deployed CoreValve (Figure 4), thus leading to very different blood flow patterns in the ascending aorta (Figure 5).
Figure 4: Valve leaflets closed relatively asymmetrically for the annular location compared to leaflets of the supra-annularly deployed CoreValve.
Figure 5 : Blood flow patterns in the ascending aorta during maximum flow deceleration. The blood flow patterns in the annular location are relatively chaotic as compared to the supra-annularly deployed CoreValve.
Exploiting FlowVision’s in-built gap model, they could also simulate the paravalvular flow effectively, or the blood flow squeezed between the aortic wall and the transcatheter valve. They found that paravalvular flow for this specific patient occurs only if the 29 mm CoreValve is deployed in the supra-annular location, as shown in figure 6.
Figure 6 : Paravalvular blood flow can be seen clearly in the supra-annularly deployed CoreValve.
Finally, as the SGGR is a boundary conforming method, they were also able to quantify the post-operative wall shear stress on the walls of the ascending aorta (figure 7) and the valve leaflets. Wall shear stress is a critical flow index for biomedical engineers and doctors alike as it can be used to predict the risk of thrombus formation after surgery. Platelets, which play a crucial role in thrombus formation, get activated in high wall shear stress regions. When these activated platelets are trapped in the region of low wall shear stress, they can initiate the thrombus coagulation cascade (Menichini et al., 2016).
Figure 7 : Sample wall shear stress patterns from the FSI simulations of Kandail et al. (2018) during maximum flow deceleration. Supra-annularly deployed CoreValve led to a region of broader high wall shear stress than its annularly deployed CoreValve. Aortic sinuses in both deployment locations are characterised by low wall shear stress.
These findings from Kandail et al. (2018) would have been impossible to predict by looking at echocardiography and imaging data alone. Such models lay the foundation for the SGGR-FSI approach to be used in the TAVR planning phase to understand the device-host interaction in detail. By translating this methodology into a clinically compliant simulation platform, clinicians can optimise and personalise each TAVR procedure on a patient-specific basis.