Patient-specific in vitro testing for evaluating TAVR clinical performance– a complementary approach to current ISO standard testing

Abstract

Following in vitro tests established for surgical prosthetic heart valves, transcatheter aortic valves (TAV) are similarly tested in idealized geometries- excluding effects that may hamper TAVR performance in situ. Testing in vitro in pulse duplicator systems that incorporated patient specific replicas would enhance the testing veracity by bringing it closer to the clinical scenario. To that end we compare TAV hemodynamic performance tested in idealized geometries according to the ISO standard (baseline performance) to that obtained by testing the TAVs following deployment in patient-specific replicas. Balloon-expandable (n=2) and self-expandable (n=3) TAVs were tested in an idealized geometry in mock-circulation system (following ISO 5840–3 guidelines) and compared to measurements in a dedicated mock-circulation system adapted for the five patient-specific replicas. Patient-specific deployments resulted in a decline in performance as compared to the baseline idealized testing, as well as a variation in performance that depended on the design features of each device that was further correlated with the radial expansion and eccentricity of the deployed TAV stent (obtained with CT-scans of the deployed valves). By excluding deployment effects in irregular geometries, the current idealized ISO testing is limited to characterize baseline device performance. Utilizing patient-specific anatomic contours provides performance indicators under more stringent conditions likely encountered in vivo. It has the potential to enhance testing and development complementary to the ISO standard, for improved TAV safety and effectiveness.

1. Introduction

Aortic stenosis (AS), a fairly common disease whose prevalence increases with age, develops as a result of oxidative, inflammatory and fibrotic processes within the aortic valve, often culminating in progressive buildup and accumulation of calcific hardened deposits in the valve fibrosa [1, 2]. Calcific aortic valve disease (CAVD), in particular, severely restricts valve opening, obstructing flow with resultant clinical sequelae [3, 4]. Calcific aortic valve deposits often occur in distinct, patient-specific patterns distributed amongst the three valve leaflets [5, 6]. In transcatheter aortic valve replacement (TAVR), a minimally invasive transcatheter or percutaneous approach to AS therapy, a stented valve is deployed across the native stenotic valve as an alternative to open prosthetic heart valve (PHV) implantation. Following initial introduction, TAVR quickly emerged as an ideal therapy for the elderly patient cohort with severe AS considered too frail to withstand open surgical valve replacement [7].

Since its introduction TAVR has experienced significant clinical adoption and growth. While initially intended solely for high-surgical risk patients, given its promising outcomes, TAVR was approved in 2016 by the FDA for intermediate-surgical risk patients, and further approved in 2019 for low-surgical risk patients [8, 9]. Nevertheless, severe complications persist even with the latest generation devices. This is particularly alarming given the new TAVR indication for implantation in lower risk and younger patients. Complications of TAVR include: thrombosis and stroke, paravalvular leaks (PVL - leaks around the device between the stent and the calcified native valve), device migration, cardiac conduction abnormalities (CCA) that may necessitate permanent pacemaker implantation (PPI), hypo-attenuated leaflet thickening ((HALT) - thrombus build up on the TAVR leaflets restricting opening), and coronary obstruction [8]. Further, the long-term durability of TAVR remains unknown, with limited data on valve durability and rates of structural valve deterioration (SVD) beyond 7 years of follow-up[10]. This is a particular concern for TAVR procedures in younger patients [11]. Beyond all of the above, recent reports have identified that prosthesis-patient mismatch (PPM), characterized by a hemodynamically underperforming TAVR device, is a significant, potentially modifiable, complication associated with higher mortality; with severe and moderate PPM present in 12% and 25% of 62,125 patients monitored from 2014–2017 [12–14]. We address this issue, in particular, in this study.

Prior to clinical trials TAVR devices are evaluated via a battery of in vitro tests and animal studies[8]. In vitro benchtop testing is guided by ISO 5840–3:2013 which sets a framework, minimum hydrodynamic performance and durability benchmarks for the device, with testing conducted with the TAVR device deployed within a simple, tubular, idealized geometry in a pulse duplicator. This testing standard, adapted from surgical prosthetic valve testing, unfortunately does not provide a realistic deployment and anatomic scenario for TAVR devices, as the landing zone for these devices is typically a knobby, calcified, irregularly contoured, surface with asymmetric dimensions [5, 6]. As such, the standard ISO approach is limited in scope to study the rudimentary performance of devices, and is largely unable to unmask clinically relevant complications [15, 16].

Patient-specific in vitro models, combined with advances in 3D printing, are increasingly utilized to study various TAVR complications [15, 17–19]. Testing a TAVR device performance under patient-specific conditions would yield clinically relevant information that current ISO testing guidelines cannot provide. We evaluated the performance of a novel polymeric TAVR device, advocated as the future of TAVR technology by addressing many TAVR’s clinical complications[20–22], in order to optimize its design for better clinical performance prior to clinical trials. The deployment and hemodynamic performance of TAVR valves was studied in custom fabricated patient-specific calcified aortic valve replicas, reconstructed from CT scans of pre-surgical TAVR patients. The deployed TAVR device eccentricity and deformation and resultant hydrodynamics was evaluated and compared to their performance in the idealized ISO based controls. This strategy provides detailed information of device performance in an environment which is a closer clinical facsimile, thereby providing a better estimation of expected TAVR clinical performance.

2. Materials and methodology

The goal of this study was to show efficacy of a complementary patient-specific hydrodynamic testing protocol that follows similar data collection practices of the current standard ISO 5840 hydrodynamic evaluation, in order to evaluate TAVR devices with more realistic clinical-like outcomes. In this study, two TAVR devices were first evaluated in a standard ISO compliant left heart simulator. Next the devices were deployed in 5 patient-specific CAVD replicas and evaluated with the same instrumentation and target cardiac conditions. The results were evaluated to show performance differences in the TAVR devices that would not be seen in standard ISO complaint testing as well as for clinical emulation. Additional advanced imaging techniques are used to evaluate the device deformation (Section 3.1) that could not be measured with current methods.

2.1. Hydrodynamic testing

Two TAVR devices were evaluated (Figure 1): PolyV-1 (PolyNova Cardiovascular, Stony Brook, NY USA), a novel polymeric self-expandable TAV (n=3) [16], and the Inovare (Braile Biomèdica, Brazil), a bioprosthetic balloon-expandable (n=2) [23], both size 20 mm. Both devices are designed to treat AS, however the two have distinct design features, with the Inovare containing bioprosthetic leafelts on a balloon-exandable stent and the PolyV-1 with polymeric leafelts on a self-expandable stent. ISO 5840–3 standard -compliant measurements were performed in a Left Heart Simulator (ViVitro Labs Inc., Victoria, BC, Canada) with an idealized rigid aortic root and a compliant silicone annulus, using blood analog fluid (50.3% glycerin, 0.9% NaCl) [15, 16] at a constant 37°C with a temperature controller (μ_dynamic= 3.5 mPa s, ρ= 1121 kg/m³). ISO standard mandates valves to meet target hydrodynamic performance parameters for forward and regurgitant flows. Systolic forward flow performance must achieve an average systolic pressure gradient (ΔP, defined as a time average of the entire positive transvalvular pressure gradient [ventricular pressure – aortic] in mmHg) and an effective orifice area (EOA, Equation 1) target. EOA is calculated in units of cm², Qrms,systole is the root mean square of the systolic forward flow (ml/s), and ρ is the density of the working fluid (g/cm³).

Figure 1: TAVR devices and patient-specific simulator.

The two TAVR devices evaluated in the study are shown (top left). The layout of the patient-specific simulator used to study the devices (Top Right). The basic experimental/anatomical setup and a sample pressure/flow waveform of PolyV-1 deployed in anatomy A (CO: 5 LPM) (Bottom).

$$EOA=\frac{Q_{RMS,systole}}{51.6\sqrt{\frac{\Delta P}{\rho }}}$$ (1)

Regurgitant flow performance must meet targets of per beat diastolic regurgitant flows with paravalvular leak (PVL) flows and closing flows (combined flows termed total regurgitant flows % stroke volume). According to the ISO 5840–3 standard, the devices of the size used in this study must meet EAO ≥0.85 cm², PVL flows ≤ 10% SV, and total regurgitant fraction ≤ 15% SV. Aortic and ventricular pressure and flow waveforms were recorded at normotensive conditions (120mmHg/80mmHg, 70 BPM), at 4 cardiac outputs (3, 4, 5, 6.5 LPM).

2.1. Patient selection, model creation and patient-specific testing

Cardiac CT scans of five (de-identified) patients (Stony Brook University Hospital, IRB approval 2013–2357-R5) were selected based on the clinical complications type (Figure 2:). The selected patients had a large range of valvular calcific masses, pre-procedural hemodynamic performance, and variety of aortic root and left ventricular outflow tract (LVOT) morphologies and dimensions. These CT scans were segmented, leaflet shape and masses followed similar protocols to previous studies[24], (Mimics v21, Materialise NV) from the left ventricle up to the brachiocephalic artery and reconstructed for fabricating patient-specific replicas (details below) by Vascular Simulations (Stony Brook, NY, USA).

Figure 2: Summary of patient-specific models and simulations.

Left to right: Images for each patient-specific anatomy (A-E). Top to bottom: Relevant clinical data, rendering of the patient vasculature, and renderings of top and side view of the calcified valves. Photos of the systolic opening of the calcified valves and the TAVR valves deployed in each model.

To record ISO-compliant measurements in the patient-specific replicas, the Vascular Simulations Replicator Pro simulator[15, 16] that contains a functional left heart with pneumatically actuated ventricle and atrium was used (Figure 1:). A mechanical valve 29mm (St Jude Medical) was used in the mitral position. The Replicator Pro was modified to control systemic compliance with a pressurized air chamber surrounding the aortic root replicas, allowing the addition/removal of air pressure and control of the aortic distention (maximum compliance is limited to the silicone stiffness) to reach the desired pressure targets. The aorta and valve was mounted with a compression clamp system, allowing the free expansion and movement of the aortic root up to 3 radial cm with the silicone distention and air chamber providing the resistive force. The setup included a generic aorta, provided by Vascular Simulations, with average lengths, diameters, and shape of a healthy male population. Compliance, systemic resistance and ventricle actuation is automatically controlled at a specified cardiac output with the feedback control system of the Replicator.

The patient-specific silicone and valve anatomies were manufactured and provided by Vascular Simulations. The silicone models were cast with uniform wall thickness offset from the segmented diastolic lumen geometry and with a soft silicone (15A silicone elastomer) (Supplementary Video 1). The functional valve models (25A PU elastomer, PU for increased durability during deployment) were molded in the diastolic configuration, with calcific masses (stiffer 40A PU elastomer) affixed to aortic side of the leaflets[16] with an additional over molding process. Valve anatomies were scaled from the original annulus average diameters (between 22–27 mm) to 18.5 mm to provide equivalent oversizing to each device. The same matching patient-specific replicas were used with both calcified stenotic valve models (Ca) and non-calcified stenotic valve models (NCa, Supplementary Figure 1).

2.3. Testing protocol

The Replicator system was placed in a supine position and the native stenotic valves performance were recorded to establish a baseline. Each test condition (same as in the ISO idealized testing conditions) was recorded and reported as an average of 10 consecutive cardiac cycles. The TAVR devices were deployed in a randomized order into the stenotic valve models (Figure 1). Balloon expandable valves were deployed with the manufacturer’s delivery catheter and the self-expandable polymeric valves were self- deployed. The neo-commissures of the TAV devices were aligned to the native commissures between the two coronary cusps for consistency. Deployment, axial positioning and consistency (±1 mm) were confirmed with two orthogonal X-ray images (Artis Zeego, Siemens Healthcare GmbH, Germany).

2.5. Statistical analysis

The data was grouped by testing anatomy (including the idealized), presence of calcification, and by the deployed TAVR device (including native stenotic valve). Data points were considered for each cardiac output and deployed valve and were an average of the 10 consecutive cycles as defined by the ISO 5840 standard. The data normality was confirmed with Shapiro-Wilk test and noted in Table 1. Initially a MANOVA was performed using IBM SPSS Statistics 25 (IBM Corporation, Armonk NY) with all three groups with post-hoc Tukey tests. Between subjects results show significance with EOA, leak flows and total regurgitation, and no significance (p>0.05) with ΔP, closing flows, cardiac outputs and stroke volumes. A three-way MANOVA was done to confirm the level of significance. The more informative two-way MANOVA tests (with Tukey post-hoc tests) within each device between the idealized and patient specific geometries, and within each anatomy (and calcific configuration) between the replica and the two tested TAVR devices, were performed to determine if each device performed uniquely in each configuration.

Table 1: Correlations between valve hydrodynamics and deployment.

Matrix of Pearson correlation (r) of the key variables/results in the patient-specific simulations for both TAVR device (PolyV-1 top, Inovare bottom). The top right of the matrix represents correlation between the other variables within each test (for the TAVR device). Bottom left is the correlation between CAVD replica and the deployed TAVR device. Cyan signifies a strong positive linear correlation and red, a strong negative. Sample Pearson correlation coefficients are reported were calculated in SPSS. * denotes data violates Shapiro-Wilk normality.

3. Results

Testing demonstrated unique TAVR device performance and characteristics (Idealized-refers to testing in the ViVitro Left Heart Simulator, patient-specific calcified and non-calcified replicas refer to testing in the Replicator Pro simulator). Each TAVR device passed ISO 8450–3 requirements in each testing protocol employed (EOA≥ 0.85cm², total regurgitant fraction ≤15% of SV) [25] PolyV-1 had an average EOA of 1.41±0.15, 1.21±0.18, 1.31±0.14 cm² in the respective idealized, calcified and non-calcified protocols; while Inovare had 1.06±0.14, 1.04±0.08, 1.14±0.09 cm² in the same respective protocols. The average systolic pressure gradient of PolyV-1 was 12.2±3.3, 36.7±8.4, 31.5±7.1 mmHg in the respective idealized, calcified and non-calcified protocols; while Inovare had 19.1±5.1, 43.5±6.3, 37.4±5.7 mmHg in the same respective protocols (Supplementary Figure 2). The results (Figure 3, top symbols) show that out of the 11 tested configurations, the two TAVR devices had statistically significant differences in performance to each other in 8 cases, with the PolyV-1 having larger EOA values. The PolyV-1 EOA performance in the patient-specific cases were statistically significant to the idealized case in 7/10 cases and the Inovare in 3/10 cases.

Figure 3: Hydrodynamic results in idealized and patient-specific simulators.

Histograms of the average value over all the tested cardiac outputs and the repeated TAVR devices. Idealized refers to testing in the ViVitro LHS. Patient anatomies A-E are shown with Ca representing the calcified valves and NCa representing the non-calcified valves.

PolyV-1 had an average leak flow of 7.5±3.0, 5.8±1.3, 4.4±1.75 cc in the respective idealized, and the patient-specific calcified and non-calcified replicas, while Inovare had 2±1.3, 3.3±2.9, 1.8±1.6 cc in the same respective protocol. PolyV-1 had larger leak flows in 8/11 cases. However, those were generally smaller in the patient-specific cases; whereas for the Inovare valve the leaks were larger in three cases. For the closing flows, PolyV-1 had an average flow of 1.3±1.6, 2.9±0.43, 2.6±0.26 cc in the respective idealized, calcified and non-calcified protocols; with Inovare having 1.9±0.7, 3.7±0.2, 4.2±0.7 cc for the same respective protocols. In general, all the patient-specific cases had larger closing flows compared to the idealized, with the Inovare valve having significantly larger flows in all the cases.

3.1. Flat panel cone-beam CT scans

During patient-specific testing, a flat panel cone-beam CT scan (DynaCT) was taken on each of the TAVR device (randomized). The Replicator was set to mimic a rapid pacing protocol with a heart rate of 180 BPM and a 5 second spin to reduce valve motion and radiation dose. Axial slices were reconstructed with isotropic voxels of 110μm and the stent frame was segmented. The stent was extracted with a custom MATLAB program to threshold the higher density voxels of the stent frame (manual selection of this threshold depended on the scan quality) and smoothing of the resulting surface mesh. The stent center, extracted from the stent deformation maps- Figure 4) that shows the average radial distance (center of stent struts) of the segmented frame voxels from the lumen centerline, 1° azimuthal and 250μm axial spacing, were used to calculate stent expansion by averaging the radii over the bottom 15mm of each stent (avoiding the crown region of the PolyV-1 and capturing the annulus) reported in Section 3.2. Stent eccentricity was calculated as the standard deviation of these radii. A secondary consecutive 5 second spin subtraction spin was also collected, during the rapid pacing, while injecting 50 ml of a 50/50 mixture of Omnipaque 270 (lohexol, GE Healthcare Inc, NJ) and 0.9% NaCl saline at 10 ml/s with a 5F pigtail catheter placed in the right coronary cusp. Visualization of the major PVL channels were reconstructed and presented in similar projection density maps (Figure 5).

Figure 4: Stent deformation projections of each testing scenario showing the impact of calcific masses.

“Maps” or projections of the native lumen (gray) colored by distance to the centerline, calcifications (red) colored by mass of calcifications, and average distance of the stent frame (green to blue) (Right). Two sample dyna-CT scan slices with the PolyV-1 in patient C (Left) with overlaid 4mm and 15mm radial arrows corresponding to the stent deformation coloration.

Figure 5: Compiled comparison of clinical data and visualization of PVL gaps.

Top- Clinical data compared to the collected performance of the CAVD replicas. * denoted an attempted match of the clinical CO based on HR and LV SV. Bottom- shows the projection maps from the subtracted CT imaging showing the relative density of the contrast dye washing into the PVL gaps and into the LVOT. The dashed white lines help visualize the right (RCC), left (LCC) and non-coronary cusps and the blue arrow shows the approximate direction of the PVL flow.

3.2. Stent deformation

The average radius results from the stent deformation maps showed that both devices expanded the native lumen (Figure 4), and the conformational behavior of the stents deforming onto the calcifications and the native commissures– with decreased eccentric deployment in the non-calcific models. PolyV-1 had an average radius (with deviation designating the eccentricity, Supplementary Figure 3) of 7.94±1.41 mm (calcified: 7.71±1.63, non-calcified: 7.91±1.46), with Inovare having 8.26±1.46 mm (calcified: 8.16±1.69, non-calcified: 8.16±1.52). The native lumens had an average radius of 7.60±1.75 mm. The Inovare valve tended to have a larger deployment radius, indicating that the balloon-expandable frame had larger radial forces compared to the self-expandable frame.

3.3. Correlated results

Pearson coefficients results of the ten patient-specific cases (NCa models included), are summarized in Table 1 (including relationships between the stenotic valve model and the post-deployment performance, and the post-deployment performance variables within each device-with response differences of each TAVR device). Correlations were considered for each cardiac output tested with each valve type. Highlighting the prominent and key revealed relationships, first, the TAVR performance is directly related to the native valve performance, with a less stenotic valve replica producing a better performing deployed TAVR device (regardless of annulus/LVOT geometry). PolyV-1 was more dependent on the diseased valve anatomy, with a stronger correlation and a steeper response (r=0.75 vs r=0.63 for the Inovare, Supplementary Figure 4). Similar performance trends were seen in the CAVD replica systolic pressure gradient to TAVR systolic pressure gradient. The deployed stent dimensions correlated to many of the flow characteristics, corresponding to the relation between the average radius and the performance (EOA) of the TAVR valves (PolyV-1 r=0.72 vs Inovare r=0.58). Stent eccentricity had a greater impact on the PolyV-1 performance (EOA, PolyV-1 r=−0.63 vs Inovare r=−0.35, Supplementary Figure 4).

3.4. Clinical Emulation

We define emulation of the clinical behavior of the CAVD replica and the deployed device as an approximation of the valve performance measured with echo/Doppler in the clinic with our in vitro models, as well as reasonable/expected performance given the device size. The echo/Doppler derived EOA (at various CO and HR) of the native stenotic valve was compared to the performance of the CAVD replicas (Figure 5) indicates no statistical difference (p=0.1, two-tailed t-test). The deployed TAVR valves performance was similar in range to the echo/Doppler derived pressure drops. In the three patients studied with moderate PVL flows, the figure shows that the location of the PVL flows visualized in the projection maps of injected contrast directly corresponding to the location specified in the clinical reports.

With all the devices passing the current ISO-8540 requirements, estimating the actual valve clinical performance parameters becomes critical. Classifying PPM requires an assumed body surface area (BSA) for a patient with the same annulus size based on a regression [26]. In this in vitro study the PPM classification is assumed as the deployed valve performance for the given annulus size. Based on these results and classifications, the PolyV-1 had moderate PPM[12] (male BSA) in 3/11 cases, where the Inovare had moderate PPM in 7/11 cases (Supplementary Table 1). The PolyV-1 had more severe cases of PVL with 6/11 cases estimated to have mild/moderate flows [27], compared to a single case for the Inovare. PVL severity appeared to be linked to the location and severity calcific masses in CAVD replicas as well as to the original clinical complications.

4. Discussion

TAVR significantly impacted and largely transformed the heart valve replacement landscape. However, persistent clinical adverse events remain, hampering its further expansion into lower risk/younger patients. Complications and ensuing adverse events are directly related to the calcified aortic valve and landing zone irregular and asymmetric anatomy into which devices are deployed. Pre-clinical determination of performance to date has not been predicative of in vivo performance, due in part to standard testing protocols in idealized in vitro geometries. We present a new benchtop testing platform for TAVR devices in complex patient-specific anatomies, which better emulates and evaluates the clinical performance of TAVR devices. Utilizing the patient-specific replicas demonstrated that testing TAVR devices in the ISO compliant idealized geometries overestimated the performance of both valves across the board- as compared to the testing in the ten patient-specific scenarios (Figure 3). The performance was highly dependent on the underlying geometry of the diseased anatomy into which the device is deployed (Figure 3). These variations are critical for understanding the source of prosthesis-patient mismatch (PPM) and its effects on the devices’ hemodynamic performance. While the polymeric TAV tested had a larger variation in performance in the patient-specific models, it exceeded and outperformed the bioprosthetic valve’s EOA performance. Such nuanced performance differences are masked by measurements in standard ISO pulse duplicator systems. We offer our approach as a complimentary, potential alternative, to the current ISO testing methods, for better evaluation of TAVR device actual performance in more realistic geometries.

Tested according to ISO, the polymeric valve had higher PVL (Figure 3/ Supplementary Table 1), possibly attributed to the lack of PVL reducing features in the current design (such as a skirt or frame wrap). However, in the patient-specific replicas, the valve showed lower or equivalent PVL flows– attributed to the stent conforming to the native diseased valve leaflets, resulting in a larger contact area. The Inovare valve on average had lower PVL compared to the polymeric valve, however, in the specific anatomy C (annular calcifications) and in both calcified and non-calcified anatomy D, there was an increase in PVL flow. The discrepancy between the PolyV-1 and Inovare PVL in irregular anatomies demonstrates the ability of the nitinol frame of the PolyV-1 to better conform to annular calcifications than that of the rigid steel frame of the Inovare stent, which resulted in larger PVL gaps. In both valves there was an increase in the closing flows volumes as compared to the idealized case. However, the polymeric valve, with its optimized leaflet configuration and reduced stress design[28], had lower closing flows (Figure 3).

The correlated trends and results offer useful indicators of the design behavior and performance (Table 1). As expected, larger calcifications of more stenotic valves impede deployment had greater impact on the TAVR performance. Larger Pearson coefficients indicate the valve sensitivity, with the PolyV-1 being more susceptible to variations in performance. The PolyV-1 had a better correlation with both the average radial deployment (direct relationship) and the eccentricity (inverse relationship) impacting the forward flow performance of the valve. A similar trend was noted for the Inovare; however, to a smaller effect. Another key finding is the better conformation of the self-expanding PolyV-1 to the native lumen eccentricity as compared to the Inovare valve possibly attributed to the stent technology differences. The Inovare stent was able to expand the LVOT dimensions and produce a more circular and larger deployed state, but in some anatomies/patients this increases the risk of annular rupture and the PolyV-1 stent frame was able to match the LVOT radius and eccentricity.

The ability to demonstrate the sensitivity to stent deformation in patient-specific anatomies suggests that this may be a valuable development tool for TAVR device manufacturers for optimizing TAV stent design (e.g., radial force needed during valve expansion, and improving conformation to irregular geometries). Additionally, leaflet design may be modified to be less impacted by stent deformation– making PPM less likely. A major advantage of our approach is that such design effects can be assessed in vitro, during the R&D phase of design, prior to design freeze and expensive regulatory testing. An added benefit is of deploying different devices in the same patient specific anatomy replica, facilitating more accurate comparative performance testing and farther iterating anti-PVL design/testing for their efficacy prior to clinical trials.

4.1. Clinical significance

Our testing approach offers a platform to study, reduce, and possibly eliminate, some of the common clinical complications of TAVR devices. The simulator can be used to optimize deployment strategies in a clinical setting, under fluoroscopic imaging, and to serve as an agnostic decision tool for evaluating advantages and disadvantages of each device where the device can be deployed in a clinical setting without preference to the device deployed. The PolyV-1 device with its self-expanding design showed better adaptability to patient-specific anatomical contours that resulted in better hydrodynamic performance, though with a larger variation in performance. This highlights the utility of testing in a dedicated patient-specific system that can inform the devices designers how to further improve the stent design of PolyV-1 for achieving a reduction in its performance variance. Such more realistic performance advantages or disadvantages are otherwise masked by the testing in ISO idealized geometries systems.

This testing platform can be further utilized to develop and compare new diagnostic techniques, with repeatable and zero-risk imaging; and further illustrates the validity of advanced techniques such as flat-panel cone beam CT used in this study. The utility of the post-deployment CT was exemplified by the ability to distinguish the effectiveness of the device design under patient-specific scenarios. Advancement of in vitro patient-specific models and incorporation to valve development performance evaluations offers ability to estimate the clinical performance of TAV devices prior to trials, and further opportunity for device manufacturers to iterate and mitigate clinical complications. It also offers clinicians an opportunity to train and study challenging patient scenarios in an agnostic platform and to compare devices and procedural techniques combined with imaging.

4.2. Limitations and future directions

Given the challenge of reconstructing and fabricating the patient-specific anatomy replicas, their number was limited. Nevertheless, we specifically selected scenarios covering a range of CAVD patients who typically undergo TAVR, with these replicas exhibiting hyperelastic and semi-rigid structures mechanical response and correct anatomical features similar to what is usually found in CAVD patients. However, those are not necessarily patient-specific. Importantly, it was not our aim to identically simulate a given, specific patient, but to recreate equivalent functional anatomical models, in which transcatheter devices may be tested in a comparative manner. Additional anatomies would further support confirming the correlations established in this study, as well as provide an opportunity to study unique and challenging clinical cases. Future selection of anatomies and ongoing and future work will focus on mimicking such challenging anatomies to test the limits of a device performance. Limited availability of TAVR devices restricted this study to the investigational polymeric valve and a clinically utilized bioprosthetic valve, while excluding other commonly employed clinical devices. The selection of TAVR devices dictated the scaling of the anatomies to a smaller and less common clinical size [29] and therefore a direct comparison of the patient’s pre-procedure stenosis and performance to the valve model could not be realized. However, the validity of our models is demonstrated by the values of the pressure gradients and EOA for the stenotic valves that match physiological values, as well as their performance that follows similar trends of clinical data.

5. Conclusions

A new platform was developed for benchtop testing of TAVR devices in complex patient-specific anatomies which better emulates and evaluates anticipated clinical performance of TAVR devices in more realistic geometries; here presented as a complementary approach to the idealized geometry pulse duplicator system that is currently used to test the performance of TAVR devices following the ISO 5840–3 guidelines. In our system, TAV valves are deployed in patient-specific replicas which better emulate the irregular contours of calcified stenotic aortic valves, thus providing a more robust and accurate testing platform for evaluating device performance and hemodynamics. This unique patient-specific testing platform allows the collection of ISO comparable hydrodynamics, clinical imaging and measurements, as well as advanced imaging and collection of stent deformation and eccentricity measurements. We have demonstrated that testing under patient-specific conditions enhanced the veracity of the performance results as well as the information obtained. We showed that testing TAVR devices in patient-specific systems yielded a decline in their hydrodynamic performance, likely predicting their clinical performance more accurately as compared to the idealized geometries currently used in the present ISO standard. Of the valves tested, the polymeric valve on average had better performance as compared to a bioprosthetic TAV, while demonstrating a larger variance in performance - indicating that it is highly dependent on patient-specific features. Such discrepancies in performance are masked when testing TAV devices using the idealized geometries of the ISO standard. Similarly, stent deformations were shown to have different degrees of sensitivity to the performance of each device. Incorporating patient-specific models into current benchtop testing paves the way for more accurate, early, “upstream” study of clinical TAVR deployment scenarios and ensuing potential complications. Utilization of the enhanced information obtained from this approach in the device development cycle, in the long run, offers potential for more rapid development of devices, at lower cost, ultimately with enhanced patient safety and effectiveness.

Abbreviations and Acronyms

AS

Aortic stenosis

CAVD

Calcific aortic valve disease

CO

Cardiac output

EOA

Effective orifice area

HR

Heart Rate

SV

Stroke volume

TAVR

Transcatheter aortic valve replacement

PPM

Patient-prosthesis mismatch

PVL

Paravalvular leak

RF

Regurgitant fraction

LHS

Left heart simulator