ABSTRACT
Objective To demonstrate a clear link between predicted blood shear forces during valve closure and thrombogenicity that explains the thrombogenic difference between tissue and mechanical valves and provides a practical metric to develop and refine prosthetic valve designs for reduced thrombogenicity.
Methods Pulsatile and quasi-steady flow systems were used for testing. The instantaneous valve flow area was measured using analog opto-electronics with output calibrated to the projected dynamic valve area. Flow velocity during the open and closing periods was determined from the instantaneous volumetric flow rate divided by valve flow area. For the closed valve interval, data obtained from quasi-steady back pressure/flow tests was used. Performance ranked by the derived valvular flow velocity with maximum negative and positive closing flow velocities, indicts experimental evidence for potential clinical thrombogenicity. Clinical, prototype and control valves were tested.
Results Establishment of a link between blood shear force and thrombogenicity led to optimization of a prototype mechanical bi-leaflet valve. The flow velocity metric was used to empirically design a 3-D printed model (BV3D) for softer valve closure dynamics which implicates reduced thrombogenic potential.
Conclusions The relationship between leaflet geometry, flow velocity and predicted shear at valve closure illuminated an important source of prosthetic valve thrombogenicity. With an appreciation for this relationship and based on our experiment generated comparative data, we achieved optimization of valve prototypes with potential for reduced thrombogenicity.
Graphical Abstract Static flow tests illustrate free jets of water passing through a closed mechanical valve (SJM). Investigations with pulsatile flow suggest similar fleeting jets arise near valve closure. Flow velocity determined in-vitro is evidence for shear forces that are predictive of blood cell damage and thrombogenic response. Red star is moment of valve closure.
CENTRAL MESSAGE A derived laboratory metric for valve closing flow velocity offers a way to rank valve models for potential blood damage. These results provide new insight and a mechanistic explanation for prior clinical observations where aortic and mitral valve replacements differ in thrombogenic potential and anticoagulation requirement. The study suggests a path forward to design and evaluate novel mechanical valve models for future development. As multiple modifications to mechanical and bioprosthetic valves have not resolved chronic shortcomings related to thrombogenicity and durability, a new development avenue was required in order to eliminate thrombogenicity in the former and extend durability in the latter.
PERSPECTIVE Prosthetic mechanical valve devices cause blood cell damage. Activation of the coagulation cascade is initiated by dynamic valve function. Design innovation focusing on valve closure behavior may reduce valve thrombogenic potential. Our study demonstrates that valve design can be optimized with emphasis on that phase.
SIGNIFICANCE Attribution of valve leaflet geometry to occluder dynamic behavior and derived regional flow velocity may lead to the development of less thrombogenic valve replacements. Laboratory experiments support the supposition that valve regional flow velocity is associated with valve thrombogenic potential. This study compares three clinical valves and two experimental prototypes.
INTRODUCTION AND BACKGROUND
Our interest in prosthetic valve dynamics was initially stimulated during formation of the Cardiac Research Development Laboratory tasked with supporting the newly opened Cardiac Surgery unit at Royal Jubilee Hospital in Victoria, BC, Canada (1973). Incorporated as Vivitro Systems Inc (VSI), our primary focus was research, design and development of cardiac valve implant devices and on the laboratory test systems required. During this phase we studied valve motion in an early pulse duplicator using high-speed cinematography and photogrammetric analysis [1, 2]. Subsequently, an innovative simpler method was devised to determine dynamic prosthetic valve area similar to planimetric quantification and was published in New Laboratory Technique Measures Projected Dynamic Area of Prosthetic Heart Valves [3]. In 2009, work transitioned into a separate independent research and development enterprise, ViVitro Laboratories Inc. (VLI) also based in Victoria, BC, Canada.
Consistent with a focus on evaluation of various prosthetic valve models, our pulse duplicator was modified to include a unique opto-electronic subsystem which we named Leonardo. The addition of this subsystem redirected our interest in heart valve dynamics to the closure phase and associated supra-physiologic backflow fluid velocities. Driven by ongoing curiosity and armed with new data from Leonardo, we reported our findings through a series of preprints and publications [4–12]. In a recent broader application of our investigative technology, results from in silico computational modeling and in vitro experimental studies provided validation of the characteristics of bioprosthetic valve leaflet flutter along with invited peer commentary [11–15].
Given the arc of success with prosthetic valves over decades including initiation and progressive expansion of transcatheter devices, long-term durability and thrombosis issues persist. Thrombosis is a physics-based phenomenon that nature evolved to stem bleeding after an injury. For both transcatheter and surgically implanted bioprosthetic valves, limited durability related to multiple factors has stimulated introduction of a variety of “rescue devices”. Intended to provide transcatheter based mitigation of complications in primary and valve-in-valve bioprosthetic valves implants, these devices are associated with their own unique complications.
The longer term consequences of a transcatheter based multiple valve approach for patient morbidity, mortality and overall cost are yet to be determined. In the current paper, our focus returns to identification and assessment of sources of thrombogenicity in contemporary clinical and also experimental mechanical and bioprosthetic heart valves with particular attention to the central role of conspicuous transient fluid velocities during valve closure. Flow velocity with intrinsic velocity gradients sufficient to induce blood damage can prompt clot formation from shear forces in vascular disease processes and prosthetic cardiovascular devices [16, 17]. Evidence comparing dynamic behavior across a range of both clinical and experimental devices then stimulated provocative conclusions regarding development of less thrombogenic yet durable prosthetic valves.
METHODS
Pulse Duplicator Experiments and Computational Modeling
Over time, progressive adaptations to the pulse duplicator apparatus have been reported [1–12]. This included the optical measurement of valve kinematics and in silico evaluations [3–15]. Aforementioned conditions and settings used in this study are applicable to Figure 1 and are related to fluid-structure interaction (FSI) studies and associated boundary conditions [11–15]. Also, in comparing results from a computational model with results from the pulse duplicator, experimental outcomes showed close agreement [11, 12, 14, 15]. Test valves are listed in Table 1. Flow and pressure signals are filtered by analogue circuitry (bandwidth BW~ 0-100 Hz). The valve flow area signal was unfiltered and had a measured rise time ~ 1.04 μs (BW ~ 0-337 KHz). Saline was used as an accepted test fluid per ISO 5840-3 with density ρ = 1.0 g/cm3 and dynamic viscosity μ = 1.0 cP [18].
Pressure measurement
Left atrial, left ventricle aortic outflow tract, and aortic sites pressures were measured via short catheters (~ 7.5 cm length) with internal diameter of ~1.8 mm connected to disposable pressure transducers (Deltran™, model PT43-604)**. Catheter ports were located at sites where measured pressure was referenced to the mid-plane of the tested valve. In Figure 1, aortic transvalve pressure is measured between the left ventricular outflow tract (LVP) and the aorta (AP). Mitral transvalve pressure is measured between the left atrium (LAP) and the left ventricular outflow tract (LVP).
Significant waveform characteristics
Several traits are clear in Figures 3–5. All signals were sampled synchronously. Regions relevant to the valve closure moment are near the dashed red line. Of importance are: initial minimum valve flow area values (VFA) and initial peak negative values attained in transvalve pressure, volume flow rate, and closure flow velocities. For bioprosthetic valves, we observed an upward-downward movement of both the valve frame and leaflets demonstrative of compliance reactivity. Hydrodynamic oscillations are also present post valve closure as seen in the unadjusted volume flow rate in some of the waveforms. Comparing the phasing of volume flow rate and valve flow area near valve closure, regional components having compliance influenced hydrodynamic patterns.
Signal synchronicity
As indicated in Figures 3 and 4, in the aortic and mitral valve sites, instantaneous minimum regurgitant volume flow rate and minimum valve flow area are synchronous with valve closure. Results for aortic and mitral valves show that maximum negative instantaneous volume flow rates (regurgitation) are synchronous with valve closure. The flow and pressure signals are processed through 100 Hz low-pass analogue filters whereas the valve area signal is unfiltered which preserves maximum frequency response (~ 0-337 KHz).
Synchronized hydrodynamic oscillations emerge post valve closure due to the pooled reactive compliances of the test valve and holder. These damped oscillations are observed in the unadjusted volume flow rate signals (dashed grey) ascribed to the combined movement of test valve components (i.e., leaflets plus stent) within the elastic silicone rubber holder. Oscillations evident for all valves and sites have similar periodicity ~ 20ms (50 Hz).
Statistical Analysis
For each experiment, we acquired 10 consecutive cycles and report the mean and confidence limits (CL) of the 10 negative peak flow velocities recorded. Summary of the valve flow velocity test datasets utilize EXCEL*** data analysis tool labeled descriptive statistics with options -Analysis ToolPak and -Solver Add-in. In Figure 3–5, mean negative peak flow velocities and CL=95% indicated by verti-bars (whiskers) adjacent to the flow velocity waveforms show the predicted uncertainty.
Experimental valve design
The BV3D rapid prototype valve design shown in Figures 2A and 2B show a working model that resulted from several designs we trialed [7]. Consideration of competing factors produced 56 laboratory constructed prototype valves with empiric data from experiments producing optimization through interaction of leaflet profile, pivot location and surface characteristics. These led to the rapid printed prototype model BV3D with advantageous forward and especially closing flow dynamics. In this work, Leonardo provided immediate feedback on the impact of even subtle geometric changes on valve dynamic behavior.
RESULTS
Aortic vs. Mitral Dynamics
When fluid in motion is abruptly halted or changes its direction (momentum change) water hammer is observed. This unwanted transient motion closely associates with valve closure timing, property of valve mountings and has been previously reported in vitro [5]. Figure 3–5 illustrate shock water-hammer dynamics with mean flow velocities reaching −29.0 m/s for the On-X mitral and −65.2 m/s for the On-X aortic valve and is absent for the Lapeyre-Triflo -F6, prototype BV3D, mock-TAVI, and Edwards-Perimount valves. The mock-TAVI valve was untested in the mitral site.
For the On-X, SJM, and LT valves in the mitral site, a brief fractional reopening is evident in valve flow area post initial closure. In Figure 4, negative peak transvalve pressure spikes range over comparable levels in the mitral site (−75 to −90 mmHg). However, in the aortic position (Figure 3) pressure spikes encompass a wider range (−40 to −95 mmHg).
Signature shock water hammer dynamics for aortic and mitral valves are evident for valve flow area, transvalve pressures, volume flow rates, and derived valve flow velocity waveforms. We found that shock water hammer dynamics can be mitigated by valve designs optimized for soft valve closure which reduces retrograde valve flow velocity in the early closing phase as observed for the BV3D and EP valves.
Edited vs. unedited instantaneous flow rates
In Figures 3 and 4, measured and derived aortic and mitral valve test results include 10 consecutive cycles each experiment. Data acquisition sampling interval is ~3.36 ms (white data points). Evident are maximum negative valve closure flow velocities ranging up to −65.2 m/s (On-X aortic). The oscillatory character seen in the primary unadjusted (grey) flow rate waveforms were edited to produce the black volume flow rate waveform data prior to deriving the flow velocity data (red waveforms). The spatially averaged metric of mitral flow velocities (red curves) are obtained by dividing time-periodic volume flow rate by time-periodic valve flow area. Average flow velocity and confidence limits are depicted by whiskers and show where 95% of the derived negative peak valve flow velocities fall. An inherent advantage of the Leonardo sub-system is that flow area is determined by a high resolution optical approach projecting geometric valve boundaries. In Figure 4, the mock-TAVI valve produced aortic but not mitral site data. Leakage rates are indicated either intra-valvular (IVL) or paravalvular (PVL). Post valve closure, valve fractional re-openings, highlighted in yellow, are attributable to shock water hammer phenomena.
Readers will note that in prior publications, closing flow velocities in excess of those observed in this study were reported [5–10]. Previous records with excessive closing flow velocities have been rectified in this report by re-analysis of the historical experiment test datasets imported into revised EXCEL templates.
Other noteworthy points in Figures 3–5 are:
For the SJM valve, the differential between aortic and mitral maximum volumetric backflow rates consistently exceeded that observed in other tested valves ~ (−180 ml/s-max vs. −150 ml/s-max).
For current clinical mechanical mitral valves, abrupt closure and ensuing shock water hammer dynamics highlighted in yellow, triggers a brief partial valve re-opening ~ 6 ms after valve closure and rebound duration of ~ 4-10 ms.
Mitral valve closure generates flow rate and transvalve pressure oscillations of greater amplitude but less dampened than valves in the aortic site.
In the mitral site, valve occluder rebound (highlighted yellow) associates with high negative regurgitant volume flow rate and transvalve pressure peaks and flow velocities impacting on valve re-closure.
Transvalve negative pressure spikes are nearly synchronous with valve closure.
Important features underscored in the valve closure phase are: near synchronous timing of valve closure (i.e., valve minimum flow area), negative transvalve closure pressure spikes, maximum retrograde closure volume flow rate, and maximum retrograde flow velocities.
The velocity of unsteady pressures in water is ~1,500 m/s. Therefore, pressures in immediate proximity to the aortic test valve for example would have trivial phase (~0.07 ms) relative to the aortic wall pressure site which is ~10 cm distant from the mid-plane of the valve.
Study results are summarized in Figure 5 for aortic and mitral valves tested during the valve closing phase based on predicted flow velocities. Notable are positive flow velocities for SJM, On-X and LT. An interesting pattern in the mitral site is that shock water hammer phenomena are associated with three of the four mechanical mitral valves tested (On-X, SJM, LT), a response of momentary post closure fractional reopening and absent in the two bioprostheses (mTAVI and EP). The EP aortic and mitral control valves have low closure flow velocity (< 10 m/s).
Important elements in Figure 3–5:
The EP aortic and mitral control valve demonstrated the lowest predicted valve closing flow velocity which is consistent with their reported clinical experience of low thrombogenic potential.
Driven by shock water hammer dynamics, backflow velocity is higher in the mitral than aortic site.
Slow-motion and real time visualization of aortic and mitral valves revealed qualitatively both symmetrical and non-symmetrical closing behavior of occluders. Quantitatively, flow velocity whisker length may be useful as a measure of occluder motion evenness over 10 consecutives cycles. Results show that the On-X and SJM have greatest whisker lengths.
The mTAVI valve sample with trivial paravalvular leak (PVL) had a peak average negative valve closing flow velocity of −10.7 m/s (±5.25). However, it has been noted that valve samples with PVL = zero and counter intuitively, with > PVLs, have reduced valve closing flow velocities [5, 6].
Based on the control valve results, mean closure flow velocities < −10 m/s suggests low/non-thrombogenic shear force potential.
Valve closing dynamics are associated with shock water hammer are also associated with maximums in flow velocity, backflow rate and transvalve pressure.
DISCUSSION
Over the past 60 years, heart valve design and performance evolved to provide improved durability and hemodynamic function. While preferences for mechanical vs. bioprosthetic valves fluctuated widely, the advent of transcatheter delivered devices and their less invasive insertion methodology, the balance shifted progressively in favor of bioprostheses. Although initially restricted to use in the elderly or patients designated too fragile for conventional surgical implantation, use is now extended to younger and lower risk candidates. This despite prior clinical data that demonstrated reduced durability of bioprosthetic valves in younger recipients. To respond to an incidence of technical failure in contemporary transcatheter valves and perhaps in anticipation of increasing frequency of degenerated valves in younger patients, a variety of transcatheter “rescue” devices are offered. Inserted within a degenerated or malfunctioning primary valve or a previously implanted rescue device, the predicted durability of rescue devices is speculative. In a prophetic article Russian Dolls; when valve-in-valve implantation is not enough [19], Tseng called attention to the possibility of rescue valve recipients who will represent therapeutic challenges that may not be resolvable by transcatheter devices. As a response to a possible increase in this initially small subset of patients, for those valve replacement candidates who favor the alternative of one valve durable for life but are hesitant because of concerns over anticoagulant related issues and the as-yet unfulfilled need by rheumatic patients in developing countries [20–23] it is essential that research driven by purposeful curiosity continues toward achievement of a long awaited anticoagulant independent mechanical valve by research endeavors such as recently reported by Wium et al. [22, 23].
At or near valve closure, flow dynamics can be considered analogous to transitory valve stenosis, whereby regurgitation is increasingly constrained until complete, motionless, closed-valve conditions arise. Thus, during brief crucial moments preceding valve closure, localized prothrombotic microenvironments may be relevant to generation of high velocity leakage jets that may induce blood element damage [24–26]. These influences may impact multiple valve types and potential consequences of these phenomena may include but not limited to:
reduced valve leaflet mobility, sub-clinical thrombosis [27],
potential for pannus formation,
transient ischemic attack (TIA), embolic acute/sub-acute stroke, other silent micro-infarction and adverse cerebrovascular events [28],
cavitation and high intensity transient signals (HITS) [30].
Shock water hammer phenomena and occluder rebound
Mitral valve closure dynamics shown in Figure 4 include a partial post closure transient opening (rebound) attributable to shock water hammer, were found less prominent in the aortic site (Figure 3). Occluder rebound driven by water-hammer power (product of transvalve pressure and volumetric flow rate) for the SJM mitral valve is observed as a momentary post closure partial re-opening. This has been previously reported with high resolution magnified examples of valve flow area rebound data [3, 5, 7]. Relative to the bioprosthetic control valve, mechanical mitral valves (On-X, SJM, LT) are driven by higher-level transvalve pressures and retrograde flows and reveal rebound behavior prior to final closure. Additional pro-thrombotic aspects may be related to sub-optimal valve forward flow energy losses (e.g. arterial sclerosis) and biomechanical and biochemical responses sufficient to exacerbate risk of pathologic thrombus formation and propagation.
Study outcomes (previous and present reports) [3–10]
High amplitude valve closing flow velocities may result in supra-physiologic shear forces originating in small intra- and para-valvular leak gaps and which become mechanistic initiators of the clotting cascade.
Platelet activation is induced by high shear even with short exposure time (transient) [26].
Closing dynamics disparity between mechanical (MHV) and bioprosthetic valves is chronically overlooked as a primary indicator of valve thrombogenicity.
Current MHVs have overt valve closing flow velocity transients relating to occluder non-response to flow deceleration and residual valve flow area.
For a given volumetric backflow rate near valve closure, the smaller the total residual leakage area, the greater the valve closing flow velocity.
The highest valve closing flow velocity was for the SJM Aortic valve compared to the tissue control valve (Edwards pericardial).
The prototype BV3D valve had lowest predicted valve closing flow velocities in both aortic and mitral sites. Assessment of this dynamic behavior represents a practical means to qualitatively screen valves and controls for thrombogenic potential.
Valve BV3D test data suggest that specific MHV leaflet geometries generate a closing force during forward flow deceleration and prior to flow reversal, a potentially beneficial “soft closure” response.
Our results infer shear damage of formed blood elements and potential for continuing thrombogenic response. This constitutes a mechanistic explanation for observed thrombogenic disparity in prosthetic valve types and designs, now broadened to include observations of TAVR related thromboembolic events.
For MHVs, cyclic valve closing flow velocities appear to be related to a prothrombotic state requiring chronic anti-coagulation.
Bioprosthetic valves appear not to generate a pro-thrombotic closing phase.
In both experimental and computational pulse duplicator studies, bioprosthetic valves in smaller diameters and/or with thicker leaflets generate higher flutter frequencies [11, 12].
Limitations
We give consideration here to a derived valve closing flow velocity metric. Also, we noted a brief positive flow transient for some mechanical mitral valves. Such transients manifest along with a shock water hammer which are indicative of high shear force, blood cell damage, biomechanical and biochemical responses that promote pathologic thrombus formation and propagation.
Strengths
Mechanical valve tested in the mitral position consistently manifested leaflet rebound observed as a transitory post closure partial re-opening driven by water-hammer power (product of transvalve pressure and volumetric flow rate). This has been previously reported with magnified examples of high resolution valve flow area rebound data [5].
High fidelity computational simulations are required to resolve small scale valve geometries and flows etc. This prerequisite may benefit from recently developed computational approaches and needed extensions to help fill lingering computational gaps [33, 34].
CONCLUSIONS
This work exposes the central relationship between thrombogenicity and predicted high velocity flows and shear forces at valve closure. As practical application of our findings, specific valve geometric features were identified that led to prototype designs with potential for further development. The application of unique technology, rapid prototyping and ranking of flow velocity patterns near valve closure optimized experimental valve geometry for reduced thrombus potential compared with control valves.
Our results identify and quantify crucial valve closure factors such as closing flow velocity but also raise a challenging question: does optimum mechanical valve performance mandate similar, identical or lower valve closing flow velocities to those of contemporary clinical bioprostheses?
While development and introduction of any new and innovative prosthetic heart valve is a lengthy, complex, and costly process, our work here opens an experimentally assisted pathway for development of durable but less thrombogenic devices. Prototype model BV3D is tangible evidence that focused laboratory efforts can yield rewarding breakthrough results. The holy grail of a mechanical valve independent of chronic anticoagulation still beckons.
Conflict of Interest Statement
All authors report no conflicts of interest.
Contributions
(I) Methodology, conception and design: LN Scotten; (II) Project administrative support: LN Scotten; (III) Provision of study materials: LN Scotten; (IV) Collection and assembly of data: LN Scotten; (V) Data analysis, interpretation, validation: LN Scotten, DJ Blundon…; (VI) All authors have contributed.
Acknowledgements
Competing Interests: None declared.
Financial Disclosure: This research has been done on a pro bono basis by all authors with no financial support of others.
Data availability statement: Core data that support the findings of this study are available from the corresponding author, [LNS], upon reasonable request.
Open access: Anyone can share, reuse, remix, or adapt this material, providing this is not done for commercial purposes and the original authors are credited and cited.
APPENDIX 1. SUPPLEMENTAL VIDEO
See MPEG 4, animated rotating visualization of prototype BV3D valve structures and full range motion potential. https://youtu.be/5QmDPbfUvTM U-tube link
APPENDIX 2. SUPPLEMENTAL INFORMATION
A, Viscoelastic Impedance Adapter (VIA)
Figure 1 includes a schematic of a Windkessel unit comprised of one resistive and two compliance elements. The VIA was characterized a reduced-order model which defines driving and loading conditions for a fluid-structure interaction (FSI) computational modeling method [11–13]. In a saline based test fluid, the reduced-order model elements characterizing the VIA unit are: CVIA1 = 0.0010 mL/mm Hg; CVIA2 = 0.1456 mL/mm Hg; RVIA = 0.15 mm Hg/mL/s.
B, Compliance modeling
The bulk pump source compliances simulated in VIA contain two air volumes CVIA1 and CVIA2. These volumes are adjustable and cover physiological range. Air volume settings we used are: CVIA1 = 120 ml (0.0010 mL/mm Hg); CVIA2 = 50 ml (0.1456 mL/mm Hg); aortic root (Croot) and systemic arterial compliance Cper = 615 ml. Compliance is defined as the ratio of volume change to pressure difference as follows:
Where:
ΔV = change in contained air volume in ml
ΔP = pressure difference (mm Hg) caused by volume change ΔV
ΔP = P2 - P1
P1 = initial static pressure in mmHg
P2 = final static pressure in mmHg
Conversion factor: 1 mmHg = 1333.2 Dynes/cm2
Air volumemax values found experimentally that simulate left ventricle, aortic root, systemic arterial compliance and also calibrated parameters for the reduced-order models with a saline test fluid [11–13] facilitated realistic pressure and flow wave forms under pulsatile flow conditions were:
RVIA = 0.15 mm Hg/mL/s.
Aortic root CVIA1 = 120 ml = 0.1456 mL/mm Hg
Output compliance CVIA2 = 50 ml
Aortic root Croot = 640 ml= 0.0010 mL/mm Hg
Left ventricle source compliance air volume CVIA2 = 640 ml
Output compliance air volume = 50 ml
Peripheral systemic Cper = 615 ml
C, Resistance modeling
Resistance to flow causes frictional loss of energy and flow container chamber radius is the dominant determinant of resistance. In Figure 1, RVIA and Rper offer flow resistance. RVIA consists of a micro-porous water filter section which offers a low fixed resistance to flow (200 c.g.s. units). The peripheral resistance Rper offers alterable resistance allowing for operator adjustment of end diastole aortic pressure.
D, Light source
Phlox® SLLUB Backlight* 50×50mm × 8.5 mm, LED back light source (diffuse red), luminance 5,780 cd/m2, uniformity 99.24%, wavelength 625±15 nm.
E, Inter-connection of testing equipment
Footnotes
In an early coal mining era, small songbirds accompanied workers in subterranean extraction sites. As these avian species are highly sensitive to changes in ambient air quality, they provided early warnings of impending hazardous breathing conditions.
Competing Interests: None declared.
Financial Disclosure: This research has been done on a pro bono basis by all authors.
No significant differences in text or graphics.
↵** Utah Medical Products Inc., Midvale, Utah 84047-1048, USA
↵*** Microsoft, Redmond, WA, USA
↵* PHLOX Corp., ZAC de l’Enfant, Aix-en-Provence, France