Fetal Transcatheter Trileaflet Heart Valve Hemodynamics: Implications of Scaling on Valve Mechanics and Turbulence

Abstract

The scarcity of data available on the best approach for pulmonary fetal valve replacement or implantation necessitate an investigation on whether practices using adult transcatheter valves could be translated to fetal applications. The objective of this study is to evaluate the hemodynamic characteristics and the turbulent properties of a fetal sized trileaflet transcatheter pulmonary valve in comparison with an adult balloon-expandable valve in order to assess the possibility of designing valves for fetal applications using dynamic similarity. A 6mm fetal trileaflet valve and a 26mm SAPIEN valve were assessed in a pulse duplicator. Particle image velocimetry was performed. Pressure gradient(ΔP), effective orifice area(EOA), regurgitant fractions(RF), pinwheeling indices(PI) and turbulent stresses were evaluated. ΔP was 8.56±0.139mmHg and 7.76±0.083mmHg with fetal valve and SAPIEN respectively(p<0.0001); EOA was 0.10±0.0007 cm² and 2.1±0.025 cm² with fetal valve and SAPIEN respectively(p<0.0001); RF with the fetal valve was 2.35±1.99% and with SAPIEN 10.92±0.11 %(p<0.0001); PI with fetal valve was 0.404±0.01 and with SAPIEN 0.37±0.07; The flow regime with the fetal valve was turbulent and Reynolds numbers reached about 7,000 while those with the SAPIEN reached about 20,000 at peak velocity. Turbulent stresses were significantly higher with fetal valve compared with SAPIEN. Instantaneous viscous shear stresses with fetal valve were 5.8 times higher than those obtained with SAPIEN and Reynolds shear stresses were 2.5 times higher during peak systole. The fetal valve implantation leads to a turbulent flow (specific to this particular type and design of valve) regime unlike what is expected of a small valve with different flow properties compared to adult valves.

Introduction

The rapid developments in percutaneous surgery (vascular and valvular) over the last three decades made the non-invasive surgical approaches attractive not only for heart valve replacement procedures but also for future fetal applications. Congenital heart defects, in particular single ventricle physiology, present several challenges particularly related to treatment. The available treatment methods comprise interventions that are currently palliative and not curative¹. Unfortunately, most congenital heart defect patients require postnatal valve replacement which is limited by the poor durability of replacement heart valves in the newborns² leading to valve stenosis (pulmonary atresia or stenosis). To overcome the durability challenges, the first step is to develop and optimize novel valve designs in order to mitigate flow abnormalities.

Fetal intervention benefits have been realized for many years³. Recently, prenatal interventions success rates were improved and prenatal cardiac interventions such as fetal aortic valvuloplasty and other non-invasive procedures seem to be promising treatment options to avoid post-birth defects⁴. In 2004, a team from Boston showed feasibility of success in a series of fetal balloon valvuloplasties for severe aortic stenosis⁵. At the Children’s Heart Centre Linz, 57 procedures were performed in 50 fetuses, with a success rate of 83% and a biventricular outcome of 56%⁶, highlighting the emerging trend and necessity of fetal interventions.

Prenatal diagnosis has changed the understanding of congenital heart defects development during the gestational period⁷. Despite the advances in the field, little is known about what the best valve replacement approach would be, if valve replacement were the adopted approach. On the adult side on the other hand, plenty of research whether clinical in-vivo or in-vitro has been performed to assess resulting hemodynamics and consequently plenty of design alterations (size, type, material, effect on the blood cells, blood trauma, patient-specific factors and interaction with the surrounding environment among other factors) of different transcatheter valves^8–15. Given the multitude of studies performed so far on the adult side where the scale is at least a magnitude higher, it is important to evaluate whether the findings could be translated to the fetal scale using similarity approaches.

The principle of dynamic similarity is a widely adopted approach in engineering design. It consists of equating the relevant dimensional parameters of the model and the prototype given that the model and the prototype are geometrically similar or in other words exact replica of each other¹⁶. Dynamic similarity is used in different applications from aerospace design to predicting the movement of mammals^{16, 17}. Testing the applicability of dynamic similarity on heart valve design to translate findings from the adult scale to the fetal scale still needs to be explored.

The objective of this study is to evaluate the hemodynamic characteristics and the turbulent properties of a fetal trileaflet transcatheter pulmonary valve in comparison with an adult balloon-expandable valve. The overarching aim is to investigate if the flow regimes change significantly or not from turbulent to laminar regime between the trileaflet adult and fetal heart valves.

Methods

Valve Selection and Hemodynamic Assessment

A zinc-aluminum alloy made of 96% Zinc and 4% Aluminum (Zn-4Al) was used to manufacture the bioabsorbable fetal valve stent. A repetitive thermal process was performed to achieve the required alloy microstructure starting from blocks of Zn-4AL. Tubes were extruded from the blocks and then laser cut to achieve the current stent design as shown in Figure 1. The prototype tissue-engineered heart valve scaffold was developed from electrospun polycaprolactone (Figure 1). Tubular scaffold of 20μm wall thickness constructed with polycaprolactone fiber of 400 nm diameter was used as leaflet compartments of the valve. The valve scaffold was shaped into a 6mm trileaflet valved-conduit and attached to the biodegradable zinc-alloy stent designed for low profile delivery. The PCL membrane was not shaped but just sutured circumferentially inside the stent at the inlet and outlet. The inlet has more number of suture points while the outlet has only 3 sutures to form the commissures. A 26mm SAPIEN 3 transcatheter aortic valve was selected to be tested as well. Both valves were placed in a pulse duplicator under physiological flow and pressure conditions. The fetal valve was subjected to a heart rate of 90 beats per minute, a cardiac output of 0.5 L/min^{18, 19} and an aortic pressure ranging from 4 to 28 mmHg. The SAPIEN valve was subjected to a heart rate of 60 beats per minute, a cardiac output of 5 L/min as dictated by ISO 5840–3 standard of heart valves and aortic pressure conditions that range from 80 to 120 mmHg. More details on the experimental setup can be found in previous publications^20–22.

Figure 1:

Photographs of the fetal transcatheter valve (a) in profile, (b) aortic view and (c) ventricular view.

Hemodynamic Parameter Calculations

A hundred consecutive cardiac cycles of aortic pressure, ventricular pressure and flow rate data were recorded at a sampling rate of 100Hz. The mean transvalvular pressure gradient of both valves (ΔP) is defined as the average of positive pressure difference between the ventricular and aortic pressure curves during forward flow.

The effective orifice area (EOA) is an important parameter to evaluate valve orifice opening in addition to the efficiency of the valve²³. EOA was computed using the Gorlin’s equation:

$$EOA=\frac{Q}{51.6\sqrt{PG}}$$ (1)

Where Q represents the root mean square aortic valve flow over the same averaging interval of the ΔP.

Regurgitant fractions (RF) were calculated as the ratio of the closing (CV) and leakage volumes (LV) to the forward flow volume (FV) as follows:

$$RF=\frac{LV+CV}{FV}$$ (2)

The pinwheeling index (PI) is an indication with implications on leaflet durability and resilience^{24, 25}. It is computed from still frames obtained from high-speed imaging as per the following equation and in accordance with previous publications^{26, 27}:

$$PI=\frac{L_{actual}-L_{ideal}}{L_{ideal}}$$ (3)

Where L_actual represents the deflected free edge of the leaflet and L_ideal represents the unconstrained ideal configuration of the leaflet free edge.

Particle Image Velocimetry (PIV)

For particle image velocimetry experiments, the flow was seeded with fluorescent PMMA-Rhodamine B particles with average diameter of 10 μm. The velocity field downstream of the valves was measured using high spatial and temporal resolution PIV. PIV involves illuminating the flow region using a laser sheet created by pulsed Nd:YLF single cavity diode pumped solid state laser coupled with external spherical and cylindrical lenses; while acquiring high-speed images of the fluorescent particles within the region. Time-resolved PIV images of the SAPIEN were acquired with resulting spatial and temporal resolutions of 0.0723mm/pixel and 1000Hz respectively. Time-resolved PIV images of the fetal valve were acquired with resulting spatial and temporal resolutions of 0.0296mm/pixel and 500Hz respectively. Phase locked measurements were recorded for 4 phases of the cardiac cycle (acceleration, peak, deceleration and diastole) repetitively 250 times with a spatial resolution of 0.0723mm/pixel and 0.0296mm/pixel for the SAPIEN and the fetal valve respectively. Refraction was corrected using a calibration in DaVis particle image velocimetry software (DaVis 7.2, LaVision Germany). Velocity vectors were calculated using adaptive cross-correlation algorithms. Further details of PIV measurements can be found in Hatoum et al^{10, 11, 20, 26}.

Vorticity Dynamics

Using the velocity measurements from PIV, vorticity dynamics were also evaluated downstream of the valve. Vorticity is the curl of the velocity field and therefore captures rotational components of the blood flow shearing as well as visualizing turbulent eddies²⁸. Regions of high vorticity along the axis perpendicular to the plane indicate both shear and rotation of the fluid particles. Vorticity was computed using the following equation:

$${\omega }_z=-(\frac{d{V}_x}{dy}-\frac{d{V}_y}{dx})$$ (4)

Where ω_z is the vorticity component with units of s⁻¹; V_x and V_y are the x and y components of the velocity vector with units of m/s. The x and y directions are axial and lateral respectively with the z direction being out of measurement plane.

Viscous Shear Stress (VSS)

Viscous shear stresses (VSS) reflect the physical environment of the cells and they were calculated as follows:

$$\tau =\mu (\frac{d{V}_x}{dy}+\frac{d{V}_y}{dx})$$ (5)

Where τ is in Pa and μ is the dynamic viscosity in N.s/m²

Reynolds shear stress (RSS)

Reynolds shear stress has been widely correlated to turbulence and platelet activation^29,30 It is a statistical quantity that measures the shear stress between fluid layers when fluid particles decelerate or accelerate while changing direction³¹. These stresses quantify the transport of momentum by fluctuating velocity components.

$$RSS=\rho \sqrt{{\left(\frac{\overline{u\prime u\prime }-\overline{v\prime v\prime }}{2}\right)}^2+{\left(\overline{u\prime v\prime }\right)}^2}$$ (6)

Where ρ is the blood density and u′ and v′ are the instantaneous velocity fluctuations in the x and y directions respectively.

Statistics

Statistical analysis in this study was performed using JMP® Pro, 13.0.0, (SAS Institute Inc., Cary, NC). All data are presented as mean ± standard deviation. For data that were found to follow the normal distribution (based on Kolmogorov-Smirnov test with p-value > 0.05), ANOVA followed by the t-test between each pair were performed. For the data that do not follow the normal distribution, the non-parametric Kruskal–Wallis test was performed followed by Man-Whitney test for the direct comparisons between each pair. P-value < 0.05 was considered statistically significant. Analyses were performed over 100 replicates.

Results

Hemodynamic Parameters

Figure 2 shows the hemodynamic parameters of the fetal valve and the balloon-expandable valve. The pressure gradient obtained with the fetal valve was 8.56±0.139mmHg and that with the SAPIEN 7.76 ± 0.083mmHg (p<0.0001). The EOA was 0.10±0.0007 cm² and 2.1 ± 0.025 cm² with the fetal valve and the SAPIEN respectively (p<0.0001). Normalizing the EOA obtained to the area of each stent (of diameters 6mm for fetal and 26mm for SAPIEN), we get for the fetal valve 35.4% and for the SAPIEN 39.6%. The RF with the fetal valve was 2.35±1.99% and with the SAPIEN 10.92 ± 0.11% (p<0.0001).

Figure 2:

Hemodynamic parameters and pinwheeling index results. ΔP denotes the transvalvular pressure gradient (mmHg), EOA denotes the effective orifice area (cm²), RF denotes the regurgitation fraction (%) and PI denotes the pinwheeling index. The results are presented ± standard deviation.

Leaflet Kinematics

Figure 3 shows the en-face imaging figures of the fetal valve and the SAPIEN at different phases during the cardiac cycle: acceleration, peak systole, deceleration and diastole. The acceleration and deceleration phases show an asymmetric opening and closing of the leaflets with the fetal valve compared to a more symmetric opening and closing obtained with the SAPIEN. In addition, both valves show twisting of the leaflets or pinwheeling. Figure 2 shows the pinwheeling index results for the two valves. The PI obtained with the fetal valve was 0.404±0.01 and that with the SAPIEN was 0.370±0.07.

Figure 3:

En-face imaging of the valves at different phases in the cardiac cycle.

Flow Velocity Field

Figure 4 shows the phase averaged velocity vectors and vorticity contours at different phases in the cardiac cycle for both valves. Higher vorticity in the shear layers is found with the fetal valve compared to the SAPIEN. While in the acceleration phase the starting vortex in the SAPIEN case shows clearly, the acceleration with the fetal valve is already characterized by fully formed shear layers with more fluctuations and higher velocities. Figure 5 shows the vorticity variation with time for both valves at the exit of the leaflets tip and at 18 mm downstream the valves in both the upper and lower shear layers. Figures 5a and 5b show clear differences in vorticity magnitudes between the two valves in both shear layers reaching up to ±2500 s⁻¹ with the fetal valve while reaching only up to less than ±1000 s⁻¹ with the SAPIEN valve. In addition, the fluctuations showing differences in not only magnitudes but also direction (clockwise for negative and counterclockwise for positive) are more significant with the fetal valve compared to the SAPIEN. Now with respect to the velocity, Figure 6 shows the probability distribution of the instantaneous velocity with the fetal valve. The figure shows higher velocity magnitudes with the fetal valve compared with the SAPIEN (reaching up to 3.1m/s and 2.8m/s respectively) and higher probabilities for higher velocities downstream with the SAPIEN (ranging from 1.5 m/s and higher).

Figure 4:

Phase averaged velocity vectors and vorticity contours at different phases in the cardiac cycle.

Figure 5:

Time-varying vorticity fluctuation with the fetal valve and the SAPIEN valve at the tip of the leaflets and at 18mm downstream in (a) the lower shear layer and (b) the upper shear layer.

Figure 6:

Probability distribution of the instantaneous velocity downstream the fetal valve and the SAPIEN valve.

Reynolds Shear Stresses (RSS) and instantaneous viscous shear stresses (VSS)

Figure 7 shows the contours of the principal RSS with both valves. The figure shows quite well that there exist tremendous differences in RSS magnitudes and distribution between the two valves with higher distribution downstream the valves accompanied by higher magnitudes with the fetal valve. To assess the differences quantitatively, Figure 8 shows the probability distribution of the RSS magnitudes in the measurement region for both valves. The RSS with the fetal valve is 6 times higher than that obtained with SAPIEN during acceleration (550 Pa versus 90 Pa respectively), 2.5 times larger during peak systole (450 Pa versus 180 Pa respectively) and around 1.5 times higher during deceleration (150 Pa versus 100 Pa respectively).

Figure 7:

Principal Reynolds shear stress (RSS) at different phases in the cardiac cycle.

Figure 8:

Probability distribution of the Reynolds shear stresses for both valves during acceleration, peak systole and deceleration.

Figure 9 shows the probability distribution of the VSS downstream the valves. The VSS magnitudes were found to be 5.8 times higher with the fetal valve compared with the SAPIEN (35 Pa versus 6 Pa respectively).

Figure 9:

Probability distribution of the instantaneous viscous shear stress downstream the fetal valve and the SAPIEN valve.

Discussion

In this study, a comparison of flow dynamics between a large transcatheter aortic valve (26mm SAPIEN) and a fetal valve of diameter 6mm. Hemodynamic parameters, flow turbulence and flow regime details were compared between the two valves.

The geometric orifice areas of the two valves that are different by an order of magnitude, which dictated the flow rate, led to an EOA with the fetal valve that is one magnitude lower than that obtained with the SAPIEN. After normalizing the EOA with the stent area of each valve, the values obtained turn up to be close despite the geometric differences showing that the fetal valve utilizes its area in a closer manner to SAPIEN. The pressure gradient on the other hand was almost the same with the two valves. The velocities obtained downstream of the valves reached about 2.0 - 2.5 m/s. Interestingly, contrary to the general expectation of laminar flow for small valves, it turns out that the flow regime with both valves is turbulent. Figure 10 shows plots of the Reynolds numbers over the cardiac cycle based on averaged measurement of flow to pipe area ratio (a and b) and experimental velocity measurement (c and d). Figures 10a and 10b clearly show that the flow regime is laminar with the fetal valve and turbulent with the SAPIEN valve respectively. Whereas, Figures 10c and 10d show that the flow regime is turbulent for both valves, which raises the question about the validity of the simple velocity obtained from dividing the flow with pipe area ratio to predict the flow regime. Due to the generated instabilities affecting the formation of shear layers (as shown in Figure 4), a more accurate estimation of the diameter would be that obtained from the effective orifice area, or in other words, the distance between the shear layers. Therefore, a more accurate estimate of the flow regime would be derived from the ratio of flow to the effective orifice area. Despite the small scale of the fetal valve along with the lower flow rate, the jet flow does not fall within the limits of laminar regime.

Figure 10:

Reynolds number variation with time for the different valves based on (a, b) ratio of flow to the tube area with the fetal valve and the SAPIEN valve respectively and (c, d) particle image velocimetry calculated jet velocity with the fetal valve and the SAPIEN valve respectively.

The nature of the flow greatly affects blood damage^{32, 33}. Turbulence after device implantation for instance, due to viscous shearing, was shown to induce non-physiological physical forces on cells particularly platelets, which leads to their activation and consequently thrombus formation – a form of blood damage^{34, 35}. In turbulent blood flow, the assessment of momentum fluxes combined with instantaneous viscous shear stresses is important to capture the overall impact that turbulent flow has on platelets³⁶. This is captured in the Reynolds shear stresses results³⁷. Viscous shear stresses on the other hand exert physical forces on the cells. Several blood trauma studies sought to identify the mechanisms of blood damage along with shear stress thresholds and exposure time³⁵. Platelet activation is thought to happen at Reynolds shear stress thresholds ranging from 10-100 Pa^{20, 38–40} and instantaneous viscous shear stresses of less than 10 Pa^{33, 36, 41}.

The results of this study show that there is a great possibility of blood damage to occur with after fetal valve implantation as RSS magnitudes can reach 5 times the upper limit of the blood damage RSS range. With respect to VSS results, while VSS post SAPIEN implantation were found to be less than 10 Pa, those obtained after fetal valve implantation reached about 35 Pa. Several studies have shown that thrombosis and hemolysis are observed after implantation of SAPIEN valves^42–44. Relating these findings to the results of this study, the SAPIEN shows lower magnitudes of RSS and VSS in comparison to the fetal valve, therefore highlighting a higher blood damage probability.

Thrombus formation is one of the factors that affect the durability of the valve especially the bioprosthetic type. Pinwheeling is another factor correlated to leaflet durability and captures the twisting of the leaflet free-edges resulting from leaflet redundancy²⁶. Pinwheeling can lead to accelerated wear and indicates a high degree of strain on the leaflet⁴⁵. The SAPIEN and the fetal valve in this study show similar degree of pinwheeling (37% and 40.1% respectively) which show significantly higher pinwheeling values compared to other transcatheter aortic valves such as Medtronic Evolut²⁶, and textile transcatheter valves¹¹.

Limitations

This study assessed the hemodynamics of a trileaflet transcatheter fetal valve only. Differences in hemodynamics are prone to occur should the design of the valve or the type of the valve be different. In addition, the diameter of the artery in the chamber was selected to be 8mm, a value that may be slightly larger (around 1-1.5 mm) than that documented in Cartier et al⁴⁶.

Conclusion

In this study, a comparison between a 26mm balloon-expandable transcatheter valve (SAPIEN) and a 6mm balloon-expandable fetal sized valve was made. It was shown that despite the small scale of the testing chamber and the valve size, the properties of the induced flow were similar in terms of turbulence and impact on blood cells and platelets. Contrary to expectation, Reynolds shear stresses and instantaneous viscous stresses were higher with the fetal valve compared with the SAPIEN. However this may be attributed to valve design itself. The pressure gradient across both valves was similar and there was an order of magnitude of difference between the effective orifice area obtained with the fetal valve and the SAPIEN valve.