A parametric study on pulse duplicator design and valve hemodynamics

Background: In vitro assessment is mandatory for artificial heart valve development. This study aims to investigate the effects of pulse duplicator features on valve responsiveness, conduct a sensitivity analysis across valve prosthesis types, and contribute on the development of versatile pulse duplicator systems able to perform reliable prosthetic aortic valve assessment under physiologic hemodynamic conditions. Methods: A reference pulse duplicator was established based on literature. Further optimization process led to new designs that underwent a parametric


| INTRODUCTION
3][4] Pulse duplicators (PD) are employed for in vitro valve testing, aiming at both device development and long-term durability assessment, under various physiologic and pathologic hemodynamic conditions.This reduces the need for in vivo animal studies, where apart from the obvious ethical concern, certain hemodynamic settings cannot be studied. 5,6owever, only a few specific PD platforms are recognized by regulatory bodies worldwide and commercialized, such as ViVitro Labs (Victoria, Canada), Dynatek Labs (Galena, MO, USA), and BDC Labs (Wheat Ridge, CO, USA), while other concepts may be used for research purposes, with the basic principle to remain the same.In most cases, only the left heart domain is emulated.The cardiac cycle is induced by piston-driven pumps or with indirect mechanisms, including similar pumps or pneumatic actuators, [7][8][9][10][11][12][13][14][15][16] aiming to generate a pulsatile flow with adjustable cardiac output (CO).
The majority of PDs include an aortic valve (AV) and mitral one (MV) with two chambers that simulate the left ventricle and atrium, additionally featuring anatomical compliance and peripheral resistance.PD valve assessment often employs pressure and flow sensors, as well as optical techniques.While water is commonly employed as the circulating fluid, a water-glycerin mixture (e.g., 40/60% w ∕ w 17 ) is often utilized to approximate better blood's viscosity.Despite basic similarities in design principles of PDs, differences arise in certain characteristics as ISO5840-3 18 provides conceptual guidelines but lacks explicit design specification.It specifies detailed requirements for the measuring equipment accuracy, but equivalent specificity is absent for PD features potentially crucial for physiologic valve performance.This is also depicted in the round-robin study, where commercial PD and systems from top institutions are compared, and results show significant variation. 19stablishing standard parameters such as heartbeat rate or CO in a PD environment can be straightforward, but a physiologic pressure condition and flow environment cannot be trivially achieved. 6Thus, proper calibration of a PD is crucial.Moreover, the interaction between PD design and imposed conditions, especially on pressure history (pressure profile over heartbeat time), has not been thoroughly investigated, despite the fact that proper valve exposure to physiologic pressure environment is crucial.
The objective of this study is to investigate the effects of PD features on valve responsiveness and conduct a sensitivity analysis across valve prosthesis types.This analysis contributes to the development of versatile PD systems able to perform reliable prosthetic aortic valve assessment under physiologic hemodynamic conditions.Starting from a literature-based PD, alternative designs were developed and tested using three different valve prosthesis types (bioprosthetic, mechanical, polymeric).The parametric study identified an optimized PD design and investigated the influence of mechanical features that emulate aspects such as compliance and resistance.The evaluation is based on comparison of pressure and flow history with the physiologic equivalent.For completeness, various hemodynamic conditions (e.g., hypertension, tachycardia, etc.) were also imposed on the optimized PD design, assessing the reliability of replicating a range of physiologic and pathologic hemodynamic conditions.

| Reference PD design
Based on a literature review, 12,[20][21][22][23] a reference PD (denoted as 'ref.') was designed and realized, emulating the left heart domain PD, as shown in Figure 1A.In the PD, the fluid is pulled through the MV by a pulsatile piston pump (ViVitro Labs Inc., Victoria, Canada, Cylinder Area: 38.32 cm 2 ), with maximum stroke volume (SV) of ~300 mL, mimicking the left ventricle.The MV experiences an atrial pressure of approximately 5 mmHg 26 imposed by an elevation-induced hydrostatic pressure, by means of a 3 L reservoir filled with ~1.8 L (×20 SV of 70 mL; see section 2.4) that mimics the left atrium.The systole is emulated by the liquid being pushed from the piston pump through the AV via the outlet.The AV is positioned in a compliant elastomeric 3D-printed (Elastic 50A Resin, Formlabs, MA, USA) aortic root model based on design parameters found in Swanson and Clark, 27 with an additional outlet to air bleed the setup, which is used for the MV too.Post-aortic systemic compliance and resistance are crucial.In the PD, flow is directed from the aortic root to a compliance chamber (~0.5 L) filled with water and pressurized air, mimicking systemic compliance contributing to a smooth flow.From the compliance chamber, water flows into a second, column-shaped reservoir (~4.8 L), imposing aortic resistance.This design concept aims to increase aortic resistance while simultaneously providing a continuous hydrostatic pressure on the AV, similar to literature [12,20-23].This setup mimics the anatomical conditions at a standing position and helps maintain diastolic aortic pressure.Via a syphon system, used to maintain the water level in the column, the water flows back into the atrial reservoir closing the circuit.The tubes that allow the flow are rigid, transparent polyvinyl chloride (PVC) ones with an inner diameter of 19 mm and 3 mm thickness, except for an extendable polyethylene (PE) drain hose as the syphon connecting the water column and atrium.The circulating medium is distilled water at room temperature (21 ± 2°C).

| Instrumentation, data acquisition, and analysis
The evaluation of valve behavior involves flow measurements, high-fidelity pressure data, and real-time imaging.These data are acquired and subsequently analyzed using a MATLAB script developed in-house.The output voltage of all devices is recorded by a microcontroller board equipped with a 12-bit analog-to-digital converter, ADC (Arduino Due).The flow is imposed by the pulsatile pump in a sinusoidal form with adjustable amplitude and wavelength/frequency and systole over diastole ratio of 1/2.The piston position (Δ(t)) is used to calculate the imposed volume rate ( V ), heartbeat, and define the start and end of systole.The aortic flow rate (Q AV ) is measured with a flow sensor (SONOTEC GmbH, Halle (Saale), Germany) with an accuracy of ±100 mL/min (in the range of 0-5 L/min) upstream of the AV and is later correlated with Δ(t) .The actual CO (L/min), as well as the regurgitation (V R ), is determined based on averaged (10 cycles) instant flow rate (Q AV ) data (Figure 4).
Two pressure sensors (Sensata Technologies, MA, USA), positioned about 80 mm upstream and downstream of the AV, capture ventricular-aortic (P VA ) and aortic (P AO ) pressure, respectively, as well as atrial (P AT ) and ventricular-mitral (P VM ) pressure for the MV.The pressure sensors being lateral to the flow measure only hydrostatic pressure, within the range of 0-500 mbar (0-375 mmHg) with a fidelity (linearity and hysteresis) of at least 0.025% of the full range, which coincides with the resolution of the ADC system (0.9 mmHg).The reference pressure is taken by the surrounding atmospheric environment.The pressure differential (ΔP(t)) for the AV is evaluated over a full cycle and calculated as The mean pressure differential ΔP, often referred to as mean pressure gradient, 18 is the averaged Δ P(t) per cycle, keeping only positive values. 18o assess valves' responsiveness, two laparoscopic endoscopes (Hopkins II, Karl Storz, Tuttlingen, Germany) are inserted into 45°-Y-fittings, laying coaxially to the AV and MV.The valve's motion is recorded from the downstream side with a high-speed camera (FASTCAM Mini UX100, Photron USA Inc. and MeVis-C 1.6/35 macro lens, LINOS, Göttingen, Germany) at a frame rate of 250 fps and a resolution of 1280 × 1024pix 2 , positioned to the eyepiece of each endoscope.Two continuous light sources arranged around the heart valves are used for illumination.After triggering the camera, a signal is sent, for every captured image, to the microcontroller board (with a fidelity of 125 μs) to allow synchronous data acquisition.Concurrently, pressure, flow, and position data points are recorded with a sampling rate of 3 kHz, enabling the capture of fast events, such as valve opening that occurs in a ms range.Characteristic points such as start of opening and closing, fully opened and closed state, are determined based on the taken images.During post-processing, the final values are averaged over ten consecutive cycles, along with standard deviation.

| Parametric study on PD designs
The literature demonstrates various PD design approaches; however, there is limited understanding regarding the distinct effects of these variations in order to determine a superior design.Therefore, more PD designs were developed herein by not only changing the resistance, enhancing compliance, but also addressing secondary features.
The column-shaped reservoir was removed, the aortic side was elevated, and a restrictor was added after the compliance chamber for variations ((i)-(vi)) depicted in Figure 2. The restrictor, models peripheral resistance, replacing the hydrostatic column, and can be adjusted for desired peak systolic pressure.These variations still comply with ISO5840-3 18 as there are very few specific design requirements, 19 while compliance chambers are optional.
The evaluation of the PD designs is based on the following criteria:  3) polymeric (polyurethane, PU-CARBOTHANE™ PC-3585A) (PHV) with a 22 mm ID following the design by Leat and Fisher [28,29].The latter was prepared in-house, with PU leaflets dip-coated by means of an aluminum mold, over a 3D-printed stent (Therma 289, DWS).For the mitral position, the aforementioned BHV model was used.The three AVs are assessed for each alteration under normotensive hemodynamic setting 1 (see section 2.4).Due to system complexity, the focus was on isolation of dominant parameters to understand individual effects.Given the scale of potential combinations, a full exploration of all parameters was beyond the scope of this study and can be the content of future studies.PDs are meant to assess pressure and flow history of various valve types under physiologic and pathologic conditions.To evaluate hemodynamic versatility of the optimized PD, a BHV was attested in aortic position based on the aforementioned criteria (see section 2.3), for five hemodynamic settings listed in Table 1 following ISO 5840-3. 18The restrictor setting, as well as pump frequency and amplitude, was adjusted to achieve these hemodynamic settings.The nominal CO was referenced by the value identified by the flow sensor for (ref. ) with the BHV, to serve as benchmark to identify Δ (t), which was maintained for all variations (per hemodynamic setting).For this benchmark, a nominal of 5 L/min correlates to an imposed V of ~6.3 L/min.As explained in section 3.1.4,the actual CO is always lower (10%-90%) than V, as fluid is also flowing through the mitral outlet.This discrepancy depends on valve type and PD design.

| RESULTS AND DISCUSSION
In summary, the parametric study demonstrated strong influence of various features on pressure and flow history.The subsequent analysis identified alteration (v) (Figure 3), as the most optimal configuration, reliably replicating natural physiologic conditions, having a compliance chamber and flow restrictor.The results of this configuration are used as a baseline for comparison.

| PD response as a function of design and valve
The pressure and flow history with hemodynamic setting 1 on (v) can be seen in Figure 4A-C.The aortic pressure curves closely resemble the physiologic equivalent (Figure 1B), achieving a hemodynamic condition akin to 120/80 mmHg with a PP of ~50 mmHg.In systole, P VA exhibits a slight elevation compared to P AO , both following a similar waveform.Furthermore, P AO decreases toward the end of systole, creating a distinct peak as the AV closes.In diastole, P AO steadily declines, whereas P VA experiences a sudden drop (see Figures 1B and 4A-C) that may even reach negative values, out of the sensing range, attributed to the piston actuation (see section 3.2).The flow curve exhibits a physiologic pattern, characterized by a sharp rise and a gradual decline.Images of the bioprosthetic AV and MV in (v) at characteristic points are displayed in Figure 4D, confirming AV's physiologic behavior, showing complete opening and closing synchronized with the cardiac cycle.Figure 5 shows that the MHV exhibits higher V R due to late closing.This is evident in Figure 4A-C, where P VA drops later (0.52 s for MHV vs. 0.49 s for BHV and PHV).Furthermore, the PHV has the highest ΔP (on average 1 mmHg more than the BHV) as illustrated in Figure 5.
3.1.1| Influence of column-shaped reservoir With all valve prostheses using the column-shaped reservoir in (ref. ), the actual CO remains consistent, ΔP experiences either no change or a slight increase, and V R decreases.On one hand, the pressures in Figure 6A-C illustrate accelerated ascension of aortic pressure due to a delayed AV opening.The delay is due to constant exposure of the AV to water column pressure, subsequently inducing a hydrostatic pressure (meant to be P AO ), which apparently leads to a delayed AV closing subsequent to a prolonged pressure decline.The consequence is an increased PP of about 100 mmHg (125/20 mmHg), accompanied by a gradual Q AV reduction during the latter half of systole.
Furthermore, P VM increases by 15-48 mmHg compared to (v) (or more as sensing limit is often reached), due to delayed ultimate opening of AV (in the range of 12-24 ms), causing a prolonged pressure build-up on the mitral side (see Figure 6A-C).Despite minimal flow disturbance, the reservoir-induced resistance fails to achieve a physiologic pressure history.

| Influence of compliance chamber
The created compliance chamber emulates aortic and systemic compliance, with a compressible air-filled cavity, adding dampening ability on the flow and inertia to the system.All these promote the transition of pulsatile to continuous flow, as well as the shortening and delaying of volume passing through the AV, 30 contributing to a delay and slower decrease in Q AV , in its absence.Moreover, the reduced aortic resistance leads to a lower ΔP and an elevated actual CO for a given V .The overlapping of P AO and P VA (Figure 6D-F) suggests an elevated PP alongside a shortened duration of increased pressure.Throughout systole, P VM decreases, displaying a faster initial increase followed by a gradual decline (Figure 6D-F).Reduced compliance, according to O'Rourke's and Hashimoto's 31 human model, leads to elevated systolic pressure due to reduced vessel distensibility and wave reflection.Notably, the plots lack a rise in systolic pressure since the aortic restrictor is adjusted to approximate a peak of ~120 mmHg.Rigid tubes allow for minimal wave dampening, causing earlier wave reflection, elevating pressure in late systole instead of diastole, ultimately enhancing PP. 31 Early wave reflection potentially explains the challenge in sustaining P AO over diastole given the absence of continuous pressure from water storage and pressurized air.Decrease in compressed air within the chamber, to reduce compliance, yields comparable, yet less pronounced patterns than the total absence of the compliance chamber, as depicted in Figure 7A-C.The CO remains relatively consistent or slightly elevated, while the average ΔP experiences a marginal reduction.Both PP and P VM decrease, too.However, reduced compliance, as opposed to complete absence, leads to higher V R .This rise can be attributed to the compliance chamber serving as a small water reservoir, ensuring greater water availability downstream to the AV during reverse flow.Compared to high compliance scenarios, a reduction or absence of compressible air cavity yields reduced water storage and more disrupted flow in the drainage tube.

| Influence of atrial reservoir
Without the atrial reservoir, both MV and AV (Figure 7D-F) are directly affected due to their series connection.Instead of water storage along with a dampening mechanism, the flow is immediately directed from the compliance chamber through the MV.Atrial systole effect is induced through the atrial reservoir utilizing imposed hydrostatic pressure.Despite regarded optional according to the ISO 5840, 18 this reservoir proves to be of high importance.Moreover, the AV remains open over an extended period and closes only during diastole leading to a significantly slower decrease of Q AV .Piston's retraction is forcing the AV to close from the ventricular side, while simultaneously pushing it open from the aortic side through the open MV.This results in elevated CO and PP.Despite very similar behavior of the three different AVs, there is no discernible trend in V R .Namely, the BHV shows a minor increase, the MHV a decrease, and the PHV no change.ΔP is reduced for both MHV and PHV.The examination was limited to the two extremes: employing a large reservoir (×20 SV) and its complete absence.Nevertheless, the study showed that the provided inertia to the system is essential for physiologic behavior.

| Mitral valve type and secondary effects
A similar effect to the one of MHV on the aortic side (see Figure 5), was observed when MHV is in mitral position, showing delayed closing 17,32 resulting in higher V R .Despite assumptions about larger diameter to contribute to this, no direct connection exists. 12The delayed closing results in a notable reduction in actual CO on the aortic side, accompanied by a significant decrease in P VM with less pronounced local peaks, as illustrated in Figure 8A-C.This phenomenon arises from the circumstance that, during the initial systolic phase, a greater volume can pass through the MV at a reduced resistance, bypassing the naturally closed AV.For instance, the PP of the BHV decreased from 61 mmHg (124/63 mmHg) to 29 mmHg (127/98 mmHg).The apparent lower ΔP and PP strongly relate to the highly reduced actual CO, 33,34 highlighting the significant mismatch between the pumpdisplaced volume, that is, imposed volume rate V, and the achieved hemodynamic condition.
While endoscopes are essential for imaging purposes, ensuring minimal influence on pressure and flow history is also imperative.Their presence does not affect the actual CO, yet their absence increases V R , by 2.37 mL (35%) for BHV and 3.08 mL (35%) for PHV (Figure 5).Generally, the highest flow velocities are anticipated at the center of the orifice section, 35 and thus the absence of endoscope allows for a fully developed central jet, consequently elevating V R during closing.Nonetheless, this pattern does not hold for the MHV, showing an 11.06 mL (45%) decrease.ΔP increases across all valves, ranging from 0.19 to 0.78 mmHg (~2% to 19%).No obvious change on pressure history occurs, except for an elevated P VA positive pressure peak during diastole for the BHV and PHV.In contrast, both minima and maxima for P VM are less pronounced throughout the entire cycle across all valve types, suggesting reduced resistance (Figure 8D-F).

| Influence of hemodynamic settings
A complete valve assessment requires valve's functionality evaluation under various hemodynamic settings, and this was done with (v) and a BHV.A comparison between the P AO history of hemodynamic setting ① in Figure 9 and their physiologic equivalent demonstrates close alignment.Data for Δ P, V R , and CO for individual hemodynamic settings are available in Figure 9F.As actual CO increases, there is a corresponding rise in Δ P. A further increase can be seen for the ratio between actual CO and V as heart rate and pumping rate increase.Figure 9A piston), and V R of (v) for hemodynamic settings 1 − 5 .Similar to the MHV's effect on the aortic side (see Figure 5), cycles, and thus PD's strong stability.Negative pressure peaks for P VA during diastole are absent in healthy human individuals. 36However, this phenomenon arises in certain piston-driven PDs, a consequence of pump's inherent high suction capability.Additionally, the elevated P VM at systole's onset aligns with observations by Lanzarone et al. 6 and Durand et al. 37 They reported negative pressure at diastole onset and substantial positive pressure peaks at the beginning of systole within pump's chamber.These phenomena persist across various hemodynamic settings and most PD modifications, spanning all valve types, which is an inherent effect of piston pumps (e.g., Figures 4 and 6).For settings ③-⑤, the actual P VM peak cannot be entirely displayed as the pressure exceeds sensor's acquisition range.
In fact, due to anatomical accessibility, sub-mitral ventricular pressure has not been measured to date.A CFD simulation investigating left-ventricular pressure distribution revealed that pressure directly beneath the MV surpasses that beneath the AV by ~30 mmHg, 38 while this case shows a peak-to-peak difference of 100 mmHg.The use of a rigid pulsatile pump, lacking pressure relaxation as in the human body and therefore the passive filling of the ventricle, 39 may induce negative pressure.Nevertheless, the optimized PD (v) replicates well both physiologic and pathologic AV conditions.

CONCLUSIONS
This study showed that engineered features that serve flow and pressure characteristics may equivalently lead to a reliable PD model exhibiting physiologic conditions, reducing the effort needed for high-fidelity biomimetic features.This is somewhat described in ISO 5840, 18 yet this study showed the direction on how to achieve physiologic conditions in a robust manner.For instance, rigid tubes can be combined with a compliance chamber and a flow restrictor to act as circulatory pressure resistance.Furthermore, the presence of a buffer reservoir in the atrial line proves indispensable for replicating the cardiac cycle.
Noteworthy disparities in pressure and flow history emerged across various valve prostheses.While evaluating the AV, a BHV positioned in the mitral position is favored over a mechanical counterpart, avoiding the adverse repercussions of late valve closing.Moreover, piston-based PDs may have inherent dependencies on pressure (e.g., elevated peaks or even negative ones) and flow characteristics, with respect to imposed volume and beat rate.Note also that the waveform generation capacity of the used one was confined to a systole/diastole ratio of 1/2.It is also worth mentioning that the chosen flowing medium, namely distilled water, is less viscous than blood, and this may have some minor effect on the increase of V R (compared to blood 40 ).
The most optimal of the attested PD systems is able to mimic the global framework of a native hydrodynamic environment, overlooking detailed exploration of localized fluid dynamics, biomaterial, and cellular interactions.Future investigations should delve into these aspects to provide a comprehensive understanding of valve's behavior, potential failure modes, and longevity within the physiologic context.Future studies can also explore the interaction of individual mechanical features providing further system optimization.
In summary, the thorough analysis led to a robust, versatile PD setup able to efficiently emulate both physiologic and pathologic AV hemodynamic conditions for various prosthesis types, while exposing intrinsic limitations complementing the directions proposed by the standard.

AUTHOR CONTRIBUTIONS
Caroline C. Smid was involved in conceptualization, methodology, investigation, formal analysis, software, data curation, and writing -original draft.George A. Pappas was involved in conceptualization, methodology, project administration, software, and writing -review and editing.Volkmar Falk contributed to funding acquisition and critical revision.Paolo Ermanni contributed to funding acquisition, supervision, and critical revision.Nikola Cesarovic contributed to conceptualization, methodology, supervision, and critical revision.

( 2 )
PP = P Systolic − P Diastolic F I G U R E 2 Top: Schematic (not to scale) of investigated PD configurations (i)-(vi), to be compared with (ref.)(Figure 1), note the absence of the column reservoir, elevation of aortic side, and addition of a restrictor, while other modified areas are circled in red.Bottom: The three different valve prostheses attested imaged during testing at an open state.

1 F I G U R E 3
Hemodynamic settings for valve assessment.Note that actual CO corresponds to flow sensor-based measurement.Pictures of the optimized PD (v) and schematic.F I G U R E 4 (A)-(C) Pressure and flow history (mean of 10 cycles) with (v) for a (A) BHV, (B) MHV, and (C) PHV in aortic position with hemodynamic setting 1 .A: aortic; M: mitral; OB: opening begins; FO: fully opened; CB: closing begins; FC: fully closed.(D) Images with (v) for BHV in aortic and mitral position with hemodynamic setting 1 .15251594,0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/aor.14757by ETH Zurich, Wiley Online Library on [29/04/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

F I G U R E 5
ΔP, CO (actual), V (piston), and V R (mean of ten cycles) of (ref.) and all alterations (i)-(vi) for (A) BHV, (B) MHV, and (C) PHV in aortic position for hemodynamic setting 1 .F I G U R E 6 Pressure and flow history (mean of ten cycles) with (ref.)(A-C) and alteration (i) (D-F) for BHV, MHV, and PHV in aortic position with hemodynamic setting 1 .A: aortic; M: mitral; OB: opening begins; FO: fully opened; CB: closing begins; FC: fully closed.Note that the dashed line indicates pressure beyond detection range.Consequently, during systole, pump's volume change retains the MV closed from the ventricular side, while simultaneously forcing it to open from the atrial side given that the AV has already opened.Additionally, the elevated hydrostatic pressure caused by the height difference forces the MV to open, all resulting in a shortened closed period with early mid-systolic opening.

F I G U R E 7
Pressure and flow history (mean of ten cycles) with alterations (ii) (A-C) and (iii) (D-F) for BHV, MHV, and PHV in aortic position with hemodynamic setting 1 .A, aortic; CB, closing begins; FC, fully closed; FO, fully opened; M, mitral; OB, opening begins.15251594, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/aor.14757by ETH Zurich, Wiley Online Library on [29/04/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License presents remarkably low standard deviation, indicated by the shaded area along the curves, encompassing a mean of 10 F I G U R E 8 Pressure and flow history (mean of ten cycles) with alterations (iv) (A-C) and (vi) (D-F) for BHV, MHV, and PHV in aortic position with hemodynamic setting 1 .A, aortic; CB, closing begins; FC, fully closed; FO, fully opened; M, mitral; OB, opening begins.F I G U R E 9 (a-e): Physiologic pressure and flow history with hemodynamic setting (A) 1 , (B) 2 , (C) 3 , (D) 4 , and (E) 5 and standard deviation represented as shaded area.Note that the dashed line indicates pressure beyond detection range.(F) ΔP, CO (actual), V