Assessment of Paravalvular Leak Severity and Thrombogenic Potential in Transcatheter Bicuspid Aortic Valve Replacements using Patient-specific Computational Modeling

Abstract

BAV, the most common congenital valvular abnormality, generate asymmetric flow patterns and increased stresses on the leaflets that expedite valvular calcification and structural degeneration. Recently adapted for use in BAV patients, TAVR demonstrates promising performance, but post-TAVR complications tend to get exacerbated due to BAV anatomical complexities. Utilizing patient-specific computational modeling we address some of these complications. The degree and location of post-TAVR PVL was assessed and the risk of flow-induced thrombogenicity was analyzed in 3 BAV patients − using older generation TAVR devices that were implanted in these patients, and compared them to the performance of the newest generation TAVR devices using in-silico patient models. Significant decrease in PVL and thrombogenic potential was observed after implantation of the newest generation device. The current work demonstrates the potential of using simulations in pre-procedural planning to assess post-TAVR complicaitons, and compare the performance of different devices to achieve better clinical outcomes.

Online abstract figure

Patient-specific computational framework to assess post-transcatheter bicuspid aortic valve replacement paravalvular leakage and flow-induced thrombogenic complications and compare device performances

Graphical Abstract

Introduction

Bicuspid aortic valve (BAV) is the most common congenital cardiac abnormality with an incidence of 1–2% in general population [1–3]. Anatomically, BAV is characterized by the fusion of two leaflet cusps and often marked by a stiff raphe at the site of fusion in all the subtypes, except for type 0 BAV. Inherent structural malformation and associated aortopathies result in asymmetric flow patterns and increased mechanical stress on the leaflets, which is hypothesized to lead to early development of valvular calcification and dysfunctions in BAV patients, requiring earlier surgical intervention and the replacement of the bicuspid valves [4].

Transcatheter aortic valve replacement (TAVR), the minimally invasive alternative procedure to traditional surgical aortic valve replacement (SAVR), was initially developed as the only viable therapeutic option for inoperable or high surgical risk patients. TAVR is associated with peri-procedural complications, like paravalvular leakage (PVL), cardiac conduction abnormalities, stroke and vascular complications [5]. However, improvement of the quality of pre-interventional analysis, physicians experience and development of newer generation devices have mitigated some of these clinical complications and enhanced its performance in patients with various pre-operative risks. The high prevalence of BAV in younger aortic stenosis patients, coupled with the worldwide expansion of TAVR use among younger and lower surgical-risk patients, including BAV patients [6–10], establish a compelling need for strategies to avoid procedural complications, which appears to be frequent in this subgroup of patients. After being used as an off-label treatment option for a long time, TAVR was recently being approved for BAV patient population by United States Food and Drug Administration and European Conformity [9]. Nevertheless, the expansion of TAVR in BAV population still remains a hurdle to surgical community worldwide, given the distinct pathological characteristics of BAV patients and the lack of randomized clinial trial comparing TAVR vs. SAVR performance in this patient population [7].

BAV patients tend to have larger annulus and severe and asymmetric leaflet calcification which often prevent proper interaction between the TAVR device and the native BAV tissue [10]. This may result in malpositioning of the valve and incomplete and eccentric device deployment, which are major risk factors in the occurrence and severity of paravalvular leakage [11]. According to a recent multicenter retrospective clinical study, BAV patients tend to have lower device success rate, higher intra-procedural second device implantation and higher post-TAVR moderate-severe paravalvular aortic insufficiency compared to tricuspid patients [12]. PVL is associated with serious complications including increased chances of requiring a second valve replacement, heart failure, hemolytic anemia and increased mortality [11]. Various hemodynamic and structural parameters including aortic regurgitation index and device landing zone calcification are often used for assessing the risk of PVL and predict PVL severity [11,13]. However, due to anatomical complexity and uniqueness, the range of these clinical parameters may vary for BAV patients, limiting their applicability to assess the risk of post-procedural PVL accurately in BAV patients, and making it crucial to analyze the risk of PVL in each BAV patient, individually. Due to the high predisposition of BAV patients to PVL and the associated complications, we have considered PVL as a major focus of this study.

Another focus of our study is to analyze the development of PVL flow-induced thromboembolic events and its potential relation to stroke risk, which is an insufficiently researched post-TAVR complication. In addition to the procedural events, including vascular access and valvular debris embolization, non-procedural events can often lead to thrombotic events, which implies a prothrombotic environment in these patients [14]. Currently, the antithrombotic therapy used to treat post-TAVR thrombotic events is not standardized across clinics, which makes it crucial to assess and attempt to mitigate the risk and long-term potential of thromboembolic events in each patient case individually. This is of particular interest to applying TAVR in BAV patient population since they tend to require aortic valve replacement at an earlier age. Due to high pressure gradient across narrow leaks during diastole, PVL tends to induce abnormal flow and can expose platelets to high shear stress. This may lead to platelet activation, microthromboemboli formation, and eventually increase the risk of stroke in TAVR patients [14,15]. Post-procedural stroke is one of the major clinical complications with devastating consequences and it did not show any notable improvement over the past several years despite improved device design and operators experience [16]. In addition, a recent study showed that BAV patients tend to have increased 30-day risk for post-TAVR stroke compared to tricuspid patients [17]. Clinically it is challenging to assess the risk of PVL induced thrombogenicity and minimal amount of computational study has focused on this issue, which marks the significance of thrombogenic risk assessment as another major focus of our study.

Patient-specific computational modeling of TAVR procedure is gaining traction for assessing various post-TAVR complications, often utilizing finite element (FE) analysis and computational fluid dynamics (CFD) simulation [15,18–23]. However, studies focused on patient-specific TAVR modeling in BAV are scant [24,25]. In the USA, BAV patients have been excluded from all the major clinical trials that led to TAVR commercialization; which is why the outcomes and performance of those TAVR devices cannot be generalized towards BAV populations. With the help of a well established, credible and rigorus workflow, computational modeing techniques have the potential to play a pivotal role in the patient-specific pre-procedural planning for these challenging patient population and depict the feasibility of TAVR in them.

In the current study, we have demonstrated the potential of computational techniques to analyze post-TAVR PVL complications in patient-specific BAV models, assess the risk of PVL induced thrombogenicity and compare self-expandable device performances in the same patient anatomies. The computational models were further validated by utilizing the patient’s post-TAVR clinical data- establishing a proof of concept for a computational framework for patient-specific modeling of TAVR in BAV that could potentially aid in TAVR pre-procedural planning for improving clinical outcomes.

Methodology

The patient-specific simulations followed several sequential steps – patient-specific model creation, structural simulation of TAVR procedure in these models with devices that they originally received (Corevalve or Evolut R – referred to as the original patient cases) as well as with a newest generation device (Evolut Pro+), CFD studies to assess the PVL degree location and the risk of thrombogenicity for each patient, and finally comparing the computational data to clinical results in order to validate the models.

Patient-specific model creation and structural simulation

Three type-1 BAV patients, each receiving an older generation self-expandable TAVR device (CoreValve or Evolut R) from CoreValve series (Medtronic plc, Minneapolis, MN, USA), were selected for this study based on varying degrees of post-TAVR PVL (Table 1). De-identified Cardiac CT scans of these patients were obtained from Rabin Medical Center (Tel Aviv, Israel) under a Stony Brook IRB approved protocol (522113). Patient-specific surface meshes were obtained from segmenting the cardiac CT scans, and prepared for the FE simulation using methods that we described previously [26]. BAV leaflets were reconstructed from the anatomical landmarks of the patient-specific lumen and the location of the calcification. Fused leaflet material property was defined based on averaging the native leaflet material testing data available on the literature [27,28]. Native tissue and calcium deposits were modeled using hyperelastic and linear elastic material models, respectively [26]. The annulus dimensions of the in-silico models were in agreement with the CT dataset, which ensures consistency between the CT dataset and the 3D model annular dimensions (Table 1).

Table 1:

Pre and post-TAVR clinical information and difference between clinical and in-silico annulus diameter

	Patient-1	Patient-2	Patient-3
Gender, Age	87, Male	89, Female	83,Female
BAV subtype	1C	1A	1A
TAVR device type	CoreValve 29mm	CoreValve 29mm	Evolut R 29mm
New left bundle branch block	No	No	Yes
Area derived annulus diameter (clinical), mm	27	24.8	24.4
Area derived annulus diameter (in-silico model), mm	27.03	25.47	24.21
% difference between clinical and computational area derived annulus diameter	0.11	2.63	0.79
Past atrial fibrillation, flutter, cerebrovascular accident or transient ischemic attack	No	No	No
Prior pacemaker	Yes	No	No
Post TAVR outcomes (PVL degree, mean transvalvular pressure gradient)	Mild, 5 mmHg	Mild to moderate, 5 mmHg	Very mild 16 mmHg
30 day outcomes (Death, stroke, hospitalization, myocardial infraction)	No	No	No
30 day outcomes (PVL degree, mean transvalvular pressure gradient)	Mild to moderate, 4 mmHg	Trivial, 18 mmHg	Trivial, 17 mmHg

BAV - Bicuspid aortic valve

TAVR – Transcatheter aortic valve replacement

The self-expandable TAVR devices were modeled and finite element (FE) simulation of TAVR procedure was performed following our previously described method [15]. The implantation depth was obtained from invasive angiography. In the simulation, the implantation depth was defined by the distance between the distal end of the stent and the annular plane. For the simulation with Evolut Pro+, same implantation depth as the original device was maintained in each patient. A detailed image of the patient-specific FE models with the deployed stents in them is illustrated in Figure 1. These deformed models were then prepared for CFD to model transient diastolic flow through the PVL channels.

Fig. 1.

(Top row, a - c) Patient-specific finite element BAV models, with leaflet calcium distribution; (Middle row, d - f, and bottom row, g - i) Post deployment configuration of the self-expandable stents (originally implanted) in each patient model. The annular calcification in patient-2 is circled in red. (BAV – Bicuspid aortic valve)

CFD study for PVL and thrombogenicity analysis and clinical validation

Patient-specific models were prepared for the CFD simuliaton based on the methods that we described previously [15]. In order to assess the PVL jets and their locations, a transient simulation was performed using Ansys Fluent (Ansys, Inc. Canonsburg, PA). Patient-specific transvalvular pressure gradients were obtained from an in-vitro flow study using the following physiological conditions- cardiac output of 5 L/min, heart rate of 60 bpm, and a mean aortic pressure of 100 mmHg.

For the thrombogencity study, particles representing platelets (Figure 2) were seeded ~20mm distally from the annulus plane and CFD simulations were performed following a method that we described previously [15]. The number of platelets varied from 23,000 to 68,600 with the variation of the cross-sectional area. The PVL channels were traced and the regurgitation volume was calculated using the PVL flowrate data from the CFD (Figure 3 and 4). In-silico PVL data was then compared to the post-TAVR two-dimensional echo-doppler data in order to assess the validity of these computational models. The echocardiography data was limited to the PVL degree based on the 5 class grading system. As such, the leak volume’s exact values could not be fully established from two-dimentional echocardiography.

Fig. 2.

A freeze frame of injection of platelets (particles in blue) flowing through the PVL channels in the aortic root of one patient case. The injection plane, which was placed ~20 mm above the annulus plane, is highlighted in the zoomed in section. (PVL – Paravalvular leak)

Fig. 3.

(Top row, a - c) Lateral view of the velocity streamlines illustrating the flow paths of the PVL jets during peak diastole. (2nd row, d - e) Polar projection of the velocity streamlines illustrating the entire circumferential image of aortic root and all the PVL that were detected in each patient case. The dashed line highlights the attachment region. First two rows represent the CFD data obtained from the original patient cases. (3rd row, g - i) Freeze frame from Echo Doppler color flow, demonstrating the PVL jets in the patients and the location of the native leaflets. (Bottom row, j - l) Polar projection of the velocity streamlines through the aortic root in the newest generation device (Evolut Pro+) cases. PVL locations in each case are circled in red in the streamline images. (RCL – Right coronary leaflet, LCL – Left coronary leaflet, NCL – Non-coronary leaflet, PVL – Paravalvular leak)

Fig. 4.

Transient volumetric flowrate through the PVL channels during one cardiac cycle in each case. No PVL flow was present in patient-3 Evolut Pro+ case. (PVL – Paravalvular leak)

In order to analyze the stress history experienced by the platelets that serves as a proxy to their activation state (thrombogenic potential), the stress accumulation (SA) value on each platelet was calculated along their individual flow trajectories through the PVL channels. The statistical distribution of this large SA ensemble is collapsed into a probability density function (PDFs) curve (Figure 5) − representing the device ‘thrombogenic footprint’ [29–31] which can compare the thrombogenic potential generated in the varying device designs. By interpolating between the smaller and larger populations’ statistical distributions using bootstrapping statistics [29], we have ensured that PDFs from the different population sizes are compatible and comparable. This method was previously used in order to assess the risk of flow induced thrombogenicity in both mechanical and bioprosthetic heart valves and TAVR devices [15,29].

Fig. 5.

PDFs of the accumulated stress values in all cases (a – c). This function is independent of the number of platelets seeded. (PDF – Probability density function; SA – Stress accumulation)

Results

Analyzing PVL degree and locations

CFD results were analyzed in order to assess the flow dynamics and the severity of PVL in all cases, and were compared to post-TAVR echo-doppler results in order to validate the computational models.

Figure 3 depicts the velocity streamlines in all cases at peak diastole (at approximately 35% of the cardiac cycle). In patient-1 CoreValve case, two major leak channels were spotted near the fused cusp. The largest leak was located near the belly region where the raphe was located (Figure 3, circled). The second largest leak was located near the attachment region of the left cusp, a little below the coaptation region of the left and right cusp. No leak was detected near the right coronary cusp. This agrees with the clinical echo-doppler image of this patient, where similar PVL jets were also spotted near the fused cusp (Figure 3). In patient-1 Evolut Pro+ Case, only one major leak was detected, located near the raphe region of the fused cusp and follows the same trajectory as seen in the CoreValve case.

In patient-2 CoreValve case, major leaks were spotted near the non-coronary and left coronary cusps. The first major leak was located near the attachment region between left and non-coronary cusp and it flows down to the location between the left and right cusp attachment region. Another major leak starts near the attachment region of the non-coronary leaflet where calcium deposit extends distally, into the left ventricular outflow tract (LVOT) (Figure 3, circled). The third leak path was located near the attachment region between left and right cusp. The echo-doppler of this patient indicated similar regurgitant jets near the non-coronary and left side of the fused cusp (Figure 3). In patient-2 Evolut Pro+ case, a small leak path was detected below the attachment region of the left cusp.

In patient-3, minor leak channels were spotted near the left and the non-coronary cusps. Two of them were located along the attachment region of the left coronary cusp and slightly below that region, and another was located near the attachment region of the non-coronary cusp (Figure 3, circled). The echo-doppler image also shows the location of minor PVL jets near the non-coronary and the left coronary cusps (Figure 3). In patient-3 Evolut Pro+ case, no PVL channel was detected.

Transient volumetric flowrates were measured at the ventricular boundary of the CFD domains for three cardiac cycles. Total regurgitant volume was then calculated by integrating over the area under these volumetric flow waveforms correspondingly for the last cycle, yielding the regurgitant flow volume, which quantifies the PVL severity in each case (Table 2). Patient-2 who originally received CoreValve, had the highest flowrate through the PVL channels (Figure 4), with the highest degree of regurgitant volume, 26.4 ml/beat (Table 2), which falls within the range of mild-moderate PVL severity [32]. Patient-3, who originally received Evolut R, had the lowest PVL flowrate as well as the lowest amount of leak volume 1.1 ml/beat, which indicates a very mild severity (Figure 4, Table 2). Patient-1, who originally received CoreValve, had a regurgitant volume of 17.9 ml/beat, which falls towards the lower end of the mild-moderate PVL severity (Table 2). Overall, these categorizations were in agreement with the clinically assigned severity. In- silico implantation of the newest generation self-expandable device (Evolut Pro+) clearly indicated a significant decrease in the regurgitant volumes in patient-1, 2 and 3 (by 37.4%, 93.9% and 100%, respectively).

Table 2:

PVL severity comparison

Patient	Patient-1	Patient-2	Patient-3
PVL Degree (From clinical data)	Mild	Mild to moderate	Very mild
PVL degree (Obtained in-silico using TAVR devices originally received by the patients)	Lower end of mild to moderate (17.9 ml/beat)	Mild to moderate (26.4 ml/beat)	Very mild (1.1 ml/beat)
PVL degree (Obtained in-silico using newest generation TAVR device)	Mild (11.2 ml/beat)	Very mild (1.6 ml/beat)	None (0 ml/beat)

PVL – Paravalvular leak

TAVR – Transcatheter aortic valve replacement

Thrombogenic potential assessment

The statistical distribution of the platelet stress accumulation was represented by the PDF curves (as described in Methodology) which are generally bimodal in nature. While each peak is located within an SA range where platelets were most frequently observed, we are interested in the second peak and the tail end to the right in particular, which indicates the amount of platelets experiencing the highest stress accumulation values (the more thrombogenic). Based on the Hellums criteria that was used in previous studies to assess the risk of platelet activation, SA values > 3.5 $Pa\cdot$ s are considered high stress accumulation range that may drive platelets beyond their activation threshold [29]. In patient-1 CoreValve and Evolut Pro+ cases, the rightmost peak was located between 0.1–1 $Pa\cdot s$. Percent of platelets populating the high stress accumulation (SA) range (>3.5 $Pa\cdot s$) were 2.48% and 1.93%, respectively. Similarly, for patient-2 CoreValve and Evolut Pro+ cases, the rightmost peak was located within the 0.1–1 $Pa\cdot s$ range. Percent of platelets populating the high SA range were 2.61% and 0%, respectively- indicating that the latter would have significantly reduce the thrombogenicity. In patient-3 original case (Evolut R), the rightmost peak was less prominent, and was located around the 0.001–0.01 $Pa\cdot s$ range (Figure 5) with the percent of platelets experiencing high level of SA ~0.15%.

Discussion

In this study, the degree and location of post-TAVR PVL, and the risk of PVL flow-induced thrombogenicity were investigated in BAV patients who underwent TAVR, and these results were compared between older and newer generation devices using patient-specific computational modeling.

Various anatomical features, including the degree and location of calcium deposits and the asymmetry of the aortic roots, are risk factors of post-TAVR PVL and ensuing complications. Previous clinical and computational studies investigated the relationship between device landing zone calcification and the resultant PVL degree. While some work indicated that device landing zone calcifications are correlated to the PVL degree, especially for the self-expandable devices, others casted doubt on this correlation [13,18], clearly indicating that this matter should be investigated in a patient-specific manner. In our current study, the calcifications distribution patterns varied significantly between the patients. In patient-1, calcium deposits were distributed throughout the leaflets, with significant amount being located near the free edge (Figure 1). Whereas in patient-2, the main calcium deposit expanded from near the free edge of the noncoronary leaflet and extended towards the entire leaflet and onto the LVOT region (Figure 1). The presence of calcification near the free edge of a leaflet, combined with annular calcium deposits extending toward the LVOT disrupt the native tissue-stent interaction, which may worsen the PVL. This is supported by comparing these three patient cases, where patient-2 had annular calcification and the worst PVL degree in the original cases where patients received older generation devices (CoreValve and Evolut R). In patient-3, who had the least severe PVL, no calcification was present in the LVOT. However, patient-3 originally received an Evolut R, which is a newer generation self-expandable valve compared to CoreValve, and it is associated with decreased PVL degree due to design improvements.

Further reduction in the PVL degree that was observed after simulating the implantation of newer generation self-expandable device in these patient models is also likely attributed to the design improvements. The stent frame design of the Evolut Pro+ is based on the stent frame of its predecessor, Evolut R. However, a new outer pericardial skirt was added to its design in order to enhance the sealing between the TAVR device and native tissue and reduce PVL. The number of PVL channels and overall leak degree in each patient case significantly decreased after in-silico implantation of the Evolut Pro+ device. This observation is consistent with a recent clinical study that demonstrated a significant reduction of moderate to severe PVL in BAV patient groups who received Evolut Pro+, as compared to BAV patients who received Evolut R [33]. By comparing the outcome of older and newer generation TAVR devices in the same patient anatomy, this study demonstrates how profoundly the improvement in device design impacts the outcome of TAVR for each BAV patient. Furthermore, it demonstrates the ability of our methodology to capture these differences in outcomes between two devices and potentially help the clinicians in selecting the best device with further simulations of current commercial devices.

For the original cases, the in-silico results of the PVL severity in patient-2 and 3 are consistent with the echocardiography data. However, a slight overestimation was noticed in patient-1 data. Based on clinical observations, this patient had a mild PVL which is <15 ml/beat according to the 5 class grading system [32]. The in-silico volume was slightly higher (17.9 ml/beat), which is on the lower end of the mild-moderate severity. This discrepancy may have stemmed from the two-dimensionality of the imaging modality which limits our ability to visualize the origin and course of the leak channels and estimate their severity accurately. It is also associated with the lack of clinical information on the post-deployment orientation of the TAVR device. The design of the CoreValve allows it to seal some areas that are located above the base height of the skirt, which makes certain orientations more favorable than the others. Differences in device orientation and position may vary PVL values observed clinically, and these differences would be impact the outcomes of computational modeling as well. Performing a computational study like this could help interventional cardiologists not only in identifying the optimal device but also, the optimal device position and orientation. While none of the current generation TAVR devices are re-orientable, devices that are fully repositionable could still be benefited from this workflow.

Among these three patients, patient-2 CoreValve case had the highest thrombogenic potential since the majority of the platelets populated the higher stress accumulation range. Patient-1 CoreValve case had the second highest thrombogenic potential. The thrombogenic potential in both of these patients decreased with the newest generation device. In patient-2 Evolut Pro+ case, the PDF peak shifted significantly to the left, and the amount of platelets residing in the higher stress range also decreased significantly (Figure 5). Similarly, in patient-1 Evolut Pro+ case, the amount of platelets located at the high stress range (>3.5 $Pa\cdot s$) decreased. While the location of the second peak did not shift significantly, the amplitude of this peak decreased with the new device. Patient-2 CoreValve case appeared to be the most thrombogenic among all these cases. Cyclic exposure to high shear stress may eventually lead to platelet activation, formation of thromboemboli and increase the risk of stroke.

While the thrombogenic risk in these patients decreased with the newest generation device and were correlated with their PVL severity, this relationship is not straightforward. Previous work has shown increased thrombogenic risk with the reduction of PVL, thus implying that the stress accumulation on the platelets are multifactorial and may depend on the spatial features of the leak channels and their interaction with the platelets. While clinical studies demonstrated higher incidence of stroke in BAV patients[17], it is unclear if those increased incidences of stroke were induced by the nature of PVL jets, the type of the TAVR device (e.g. older vs. newer), or certain patient specific anatomical features of those BAV patients. Hence, a comprehensive computational study like the one presented herein could demonstrate the comparative risk of PVL flow induced thrombogenity between two devices, regardless of the degree of PVL and find the optimum set of parameters in order to mitigate the risk of PVL flow-induced thrombogenicity in each patient as well as improve the TAV performance.

Despite the significant reduction of moderate-severe PVL, mild PVL remained frequent with the newer generation devices [34]. While moderate-severe PVLs are associated with impaired survival, the long-term detrimental effect of trivial and mild PVL, especially in BAV patients, are unclear because of multiple limitations. With a large percentage of TAVR patients being discharged with mild PVL and due to progressive expansion of TAVR towards younger patients, understanding the long-term effect of the mild PVL is getting increasingly important. While this appears to be benign, these mild leaks can worsen over time in some patient cases. Clinical studies, including a 5 year follow-up study of PARTNER 1 trial, indicated that even the presence of mild post-TAVR PVL was associated with impaired survival [35,36], which highlights the importance of analyzing the risk associated with mild PVL. In patient 1 from the current study, who demonstrated mild PVL after TAVR, 30 day outcome showed worsened PVL (mild to moderate). Hence it is important to closely monitor each patient cases and capture the the leak channels and their flowpath, regardless of how harmless they appear. However, further research is necessary before generalizing the application of this framework towards all BAV patients. By analyzing pre-TAVR hemodynamics, validating this framework for severe degrees of PVL and extending this study by including balloon-expandable device cases, this framework has the potential to be expanded to the clinical setting and incorporated into pre-procedural planning.

Clinical Significance

Clinical studies demonstrating the performance of TAVR in BAV are based on BAV patients who were considered favorable after careful evaluation and discussion by the heart team. However, this patient selection criteria was not clear and may not be consistent across centers, which makes it difficult to apply those observations to the entire and morphologically heterogenous BAV population who may become candidates for TAVR [9,37]. Hence, while performing the pre-procedural planning for TAVR, it is tremendously helpful to be able to visualize the 3D model of the patient anatomies, the device-tissue interaction during the procedure and quantify how different devices perform in each patient anatomy. This is of particular concern for implanting self-expandable devices in BAV, which are associated with a higher risk of PVL complications than balloon-expandable devices [34]. This model has the ability to capture the spatial details of the leak origins and path in complex BAV anatomies. While echocardiograpy is the gold standard for PVL assessment, visualization and quantification may often get challenging due to the presence of heavy calcification, presence of raphe and thicker leaflets in BAV patients. Our computational framework could mitigate the shortcomings of current imaging modalities and help the clinicians visualize the post-TAVR deformation of different devices, find the severity of the PVL and their locations, assess the patient-specific hemodynamic performances. In addition, no current clinical modality is capable of quantifying the comparative risk of flow induced thrombogenicity due to post-TAVR PVL from different devices in patient-specific anatomies before the procedure, which is another gap that could be filled by the current framework. The capability of this framework to address all these uncertainties in each patient case underscores the clinical significance of computational modeling.

Conclusion

In conclusion, this study marks the development of a computational framework to analyze the risk of post-TAVR PVL and flow-induced thrombogenicity in BAV patients and compare the performance of older and newest generation self-expandable devices in these patients using patient-specific computational modeling. We have demonstrated how our rigorous computational modeling techniques enhance the granularity to detect the exact location and volume of the leakage- also when involves what may appear as harmless minor leakages. Significant decrease in PVL was noticed after implantation of the newest generation device in the same cases analyzed. An overall good agreement was found between the clinical and the in-silico data, serving as clinical validation of the models and complementing the clinical data with information that is beyond the reach of clinical modalities. Since TAVR was approved for BAV patients population without a randomized clinical trial, the careful selection of BAV patients for TAVR has now became even more critical, and ultimately, it depends on the judgement of the individual clinician. This study has the potential to compare the performance of different self-expandable devices, and associated PVL and thrombogenic complications in the same patient anatomy, help physicians with the device selection and provide them with assurance for the feasibility of TAVR procedure in each BAV patients, which is the ultimate goal of this framework. In order to enhance the strength of this study, and increase its applicability and reliability as a pre-operative tool, future work will involve investigation of additional patient cases and expand the scope of this study towards balloon-expandable devices which will enhance the robustness of the presented framework.

Abbreviations and Acronyms

BAV

Bicuspid aortic valve

CFD

Computational fluid dynamics

FE

Finite element

LCL

Left coronary leaflet

LVOT

Left ventricular outflow tract

NCL

Non-coronary leaflet

PPI

Permanent pacemaker Implantation

PVL

Paravalvular leak

RCL

Right coronary leaflet

SAVR

Surgical aortic valve replacement

TAVR

Transcatheter aortic valve replacement