Despite the advances in the treatment and prevention of cardiac disease, heart valve disease remains an important medical condition, with more than 280,000 patients undergoing a prosthetic heart valve replacement procedure each year.1,2 Although some forms of valve disease can be controlled with medicines, definitive treatment often involves an invasive intervention to repair the dysfunctional valve, and in many cases, valve replacement is required to restore adequate function. Replacement valves are classified as either bioprosthetic or mechanical, and each type of valve has relative advantages and limitations. Bioprosthetic valves, commonly designed from porcine or bovine tissue valves, do not require lifelong usage of anticoagulants.3 However, they are not as durable as mechanical valves, and their functionality is prone to deteriorate over time, increasing the risk of reoperation.3 Mechanical valves are more resistant to structural deterioration and can remain functional for the life of the patient.3 Their disadvantage lies in their propensity for clot formation and requirement for lifelong anticoagulation, with attendant risks.2 Of the mechanical valves, bileaflet valves have become the dominant design because they have excellent hemodynamic performance. Their design results in diminished flow restriction, decreased hemolysis (because of less turbulent flow through the valve), and the use of pyrolytic carbon has resulted in better biocompatibility and durability, compared with materials used historically (e.g., metals and elastomers).
Despite the advantages of bileaflet mechanical heart valves over previous valves, their potential for thrombus formation remains an important issue because of factors including their incorporation of nonbiological materials and nonphysiological flow dynamics.4 This mandates that patients with mechanical valves be maintained on lifelong anticoagulants such as the vitamin K antagonist warfarin.1,3,5 The use of anticoagulants come with inherent hemorrhagic risks, with mechanical valve recipients having significantly higher risk of bleeding (3.4 times) than a patient with a bioprosthetic valve.3
Even with anticoagulation therapy, the risk of thromboembolic events (TEs) in mechanical heart valve patients continues to be a serious problem with a reported incidence rate of 0.6–2.3% TEs per patient-year.1 The TEs are thought to be primarily triggered by interactions between blood and artificial surfaces, regions of low flow, and areas of high shear stresses that can cause platelet activation.2,4,6 Valves in the mitral position have been shown to have high leakage velocity jets between the hinge regions and the valve leaflets.7 Regions of stasis and nonuniform flow fields because of the shape of the hinge recess are also postulated to be contributing factors to TEs.4,6,7 In vitro experiments conducted using rennet milk as a blood analog in a rigid model of the left side of a heart demonstrated that clots formed repeatedly and preferentially in the hinge pivot areas.8,9 These clot formation regions correlate clinically with those observed in explanted bileaflet valves. Because of the recognition of the role that the hinge region plays in TEs,4 hinge designs have been studied to develop more hemodynamically favorable geometries.
Many computational fluid dynamic studies of mechanical heart valves exist, but experimental validation through the visualization of flow is equally important to help improve understanding of valve performance and necessary to substantiate the research for clinical approval. Visualization of the hinge flow fields can be difficult with common methods and is often complex and expensive. Common benchtop fluid flow visualization methods that were taken into consideration in preparation for this study included dye injection, particle image velocimetry (PIV), and microbubble tracking.
Dye techniques are one of the most commonly employed methods for fluid flow visualization because of their availability and low cost.10 A constant dye injection through a delivery port (e.g., needle or tubing) can be used to trace fluid streamlines, which facilitates observation of the fluid dynamics in the areas of interest. This method is particularly useful in revealing areas of recirculation or turbulence. However, an observed limitation is that the dye can diffuse rapidly throughout the fluid being used, making it difficult to localize, especially in very low to zero flows. Furthermore, dye can be difficult to visualize against a dark background color.
Particle image velocimetry is a highly accurate and quantifiable technique and has become a method of choice for many fluid dynamic studies in medical devices such as valves and pumps.11 This technique requires seeding particles into the working fluid, illuminating them with a high-power pulsed laser system and then visualizing them using a high frame rate camera.10 The laser pulse is coordinated with the camera to obtain images at the time the laser fires on the region of interest (ROI) within the flow. Digital cross-correlation of two sequential images is then performed by software to convert time between shots and distance a particle has moved to calculate velocity vectors.10 One drawback of PIV is that it is expensive and requires an extensive setup. In addition, the need for lasers requires additional precautionary safety considerations and instrumentation.
Another technique uses hydrogen bubbles to visualize flow. Hydrogen bubbles are generated using electrolysis, in which direct current is passed through the fluid between two electrodes causing an electrochemical reaction. Hydrogen bubble formation occurs at the cathode, whereas the anode generates oxygen.12 Utilization of fine wires is usually preferred to generate micron-sized hydrogen bubbles, which can be carried along by the fluid flow.10,12 This method can be very versatile because the wire can be placed virtually anywhere in the flow. Although this technique is quite useful for visualizing flow disturbances in certain mock fluid systems, it can be difficult to apply and interpret in practice.
The limitations of these common techniques prompted the need to explore other ways to study and visualize the flow in the hinge recesses of bileaflet mechanical heart valves. The goal of this project was to describe the development of a new flow visualization technique that is applicable to the study of flow in valve hinge regions. This technique involves localized preloading of the area of interest (i.e., hinge recess) with a dairy-based colloidal suspension (DBCS) and tracking its washout by time-lapse photomicroscopy. The images can be then converted into digitized intensity topography maps to enhance visualization of the fluid flow dynamics and relative DBCS concentrations. This technique is intended to supplement the experimental validation required in computational fluid dynamic studies for the design of valves and other mechanical circulatory support systems. The purpose of this study was to develop a fluid visualization technique to enable the study of flow distribution and clearance in hinge regions in current and future mechanical valve designs.
Materials and Methods
The valve housing hinge segments of two commercially available 25 mm aortic bileaflet valves (V1 and V2) were studied using the DBCS technique in individual custom-fabricated fluidic circuits. The valves were selected for their distinct hinge geometry designs and their prevalent use in clinical practice. The channels were constructed from stacked laser-cut acrylic sheets and neoprene gaskets, establishing a 5.0 × 12.5 mm (height × width) fluid channel with tapered inlets and outlets to encourage laminar flow (Figure 1). An adjustable platform was incorporated into the center bottom of the flow channel, which allowed for vertical adjustment of the valve location in the fluid flow. Sections containing the hinge geometries were cut from the actual valve housing and mounted on the valve platform. The total height of the valve protrusion into the main flow was set at 3.1 mm, which was kept constant between valves to create identical protrusion flow profiles. This protrusion height included the thickness of the valve housing (at the hinge region) and its platform, creating a small step similar to the valve–aorta interface in implanted aortic mechanical valves. The channel also contained a retractable 22 gauge needle side port (inserted through the neoprene gasket sandwiched between the acrylic sheets) to locally introduce DBCS into the hinge recesses. The valve leaflets were not included for this study.
Thirty microliters of DBCS (Half and Half, International Delight, WhiteWave Foods, Broomfield, CO) was injected locally through the 22 gauge needle side port into the hinge recess using Syringe Pump 1 (NE-1000X, New Era Pump Systems Inc., Farmingdale, NY). The needle was then withdrawn from the main channel through the neoprene gasket to prevent it from obstructing and disturbing flow. The neoprene gasket not only served to prevent leaks but also added a self-seal feature because of its compliance to prevent fluid volume displacement and disruption upon needle retraction. A constant washout infusion of distilled water was maintained with Syringe Pump 2 (Model 22, Harvard Apparatus, South Natick, MA). Washout phenomena were studied at four calculated average channel fluid velocities at the location where the hinge segments were immersed: 2.5 × 10−3, 5.1 × 10−3, 10 × 10−3, and 17 × 10−3 m/s. These low flow velocities were attained by setting Pump 2 flow rates to 227, 454, 908, and 1550 ml/hr, respectively. Each velocity was tested three times per valve. Figure 1 depicts the complete experimental setup.
Visualization and Analysis
A high-resolution digital camera (Nikon D3100, Nikon USA, NJ) mounted on a stereomicroscope (Leica MZFLIII, Leica Microsystems Inc., Buffalo Grove, IL) with a 100 mm focal depth was used to obtain videos of the valve hinge region. The video frame extraction software (Free Video to JPG Converter, DVDVideoSoft Ltd., London, UK) was used to extract frames from the video for analysis as individual time-lapse images. By using an image-processing software package (ImageJ, NIH, Bethesda, MD), the valve hinge recess was designated as the ROI and used to perform histogram analysis and correlate pixel intensities with DBCS concentrations. Intensity topography maps were generated from ROI pixel intensities to visualize the characteristic washout of each valve’s hinge recess. Normalization was achieved by subtracting the lowest intensity value (0 concentration) in each experiment and then dividing it by the maximum value (DBCS concentration before washout); this yielded percent concentration values irrespective of background brightness differences between experiments. Statistical comparison was done by first fitting the normalized average washout data series to a nonlinear exponential decay model using MATLAB (MathWorks, Natick, MA) to determine the decay and intercept constants and corresponding standard errors. This was followed by an unpaired two sample t-test analysis of the decay constants of both valves at each tested velocity. The time constants, τ (i.e., corresponding to the time required for DBCS concentrations to reach 1/e = 36.8%), were obtained by taking the reciprocal of the decay constants.
The DBCS provided insightful visualization of washout dynamics in the hinge region of the valves. The suspension’s white color was easy to differentiate from the dark pyrolytic carbon valve because of its high contrast. Figure 2 shows selected frames (0, 24, 48, and 72 seconds) from the video recordings (see Video, Supplemental Digital Content 1, http://links.lww.com/ASAIO/A78) of the hinge region of both valves at a local fluid washout velocity of 5.1 × 10−3 m/s and the corresponding top view of the pixel intensity–generated color topography maps (V1 i-Map and V2 i-Map) of the DBCS concentrations. By 72 seconds, all of the DBCS was washed out of the hinge recess in V2, whereas a significant amount remained in V1. This was also demonstrated in the topography maps of V1 i-Map by the large yellow high-intensity region. Figure 3 shows normalized washout data and exponential decay fits for V1 and V2 at the four tested velocities. The maximum SEM in the normalized data was ±3.9. For the nonlinear decay model fits, the R2 values were 0.979, 0.988, 0.981, and 0.983 for V1 at 2.5 × 10−3, 5.1 × 10−3, 10 × 10−3, and 17 × 10−3 m/s, respectively. For V2, R2 values were 0.967, 0.975, 0.976, and 0.974 at the aforementioned fluid velocities.
There was an increased washout rate of both valves as the fluid velocity increased. Of note, there was a difference in washout between valves, with V2 having a significantly faster washout than V1, at each tested flow velocity (p < 0.01, comparison by t-test of the nonlinear exponential decay fit constants), demonstrating that V1 had a longer fluid residence time than V2, with the τ values shown in Table 1. Based on the τ values, DBCS concentrations reached 36.8% in V2 41.1 seconds faster than V1 at 2.5 × 10−3 m/s washout fluid velocity, 23.8 seconds faster at 5.1 × 10−3 m/s, 8.7 seconds at 10 × 10−3 m/s, and 6.6 seconds faster at the highest fluid velocity tested (17 × 10−3 m/s). On average, V2 washed out 40.6% faster than V1 (V2 took 36.7, 47.5, 38.0, and 40.0% less time to wash out than V1 at 2.5 × 10−3, 5.1 × 10−3, 10 × 10−3, and 17 × 10−3 m/s, respectively; see Video, Supplemental Digital Content 1, http://links.lww.com/ASAIO/A78).
The intensity topography maps provided a visual representation of how the DBCS concentration decreased over time as washout occurred. It also helped visually identify regions of potential stagnation. The top view of the intensity topography maps (Figure 2, i-Maps) demonstrated a unique washout pattern for each hinge. The DBCS in V1 was washed out starting at the periphery of the upstream channel wall. V2 demonstrated a multimodal pattern having two distinct washout sites (“bow-tie” corners) per hinge recess (see Video, Supplemental Digital Content 2, http://links.lww.com/ASAIO/A79).
Despite advances in their materials and design, bileaflet mechanical heart valves are still susceptible to TEs and require the lifelong use of anticoagulants. Hemorrhagic complications associated with anticoagulation therapy are not infrequent and can cause considerable disability and mortality. Clinically, these valves are prone to develop thrombi primarily localized in the hinge area. Therefore, there is considerable interest in understanding the flow dynamics within this region. Although computational techniques in recent years have been developed to model the fluid dynamics in the hinge region, direct visualization can help validate and improve the models.
The DBCS technique proved to be inexpensive, easy to use, and informative. Under flow conditions, it formed identifiable streamlines, and its white color provided excellent contrast against the dark pyrolytic carbon valve. To enhance the interpretation of regional washout characteristics, the change in the concentration of the DBCS was correlated to pixel intensity and converted to DBCS intensity topography maps. Dairy-based colloidal suspension served as a predictor to reveal potential clot deposition zones in a particular hinge design rooted on its residence time, which is a measure related to platelet interaction time and subsequent activation. The colloidal properties of DBCS are particularly effective at localizing in areas of interest and the visualization of streamline traces in areas of recirculation.
An important difference between the two valves is their apparent hinge recess geometry and volumes, despite the fact that they are the same valve size, 25 mm. The hinge volume of V2 is approximately 3 μl, whereas V1 was 5 μl. This may be an important contributing factor to the hinge washout time constants observed. In addition, the shape of the hinge and the washout flow patterns it generates differentiates one valve from the other, which can be seen in the intensity maps in Figure 2. This indicates that the technique has high sensitivity for identifying washout differences between valves and could be very useful for comparing hinges of valves of different sizes or manufacturers.
In an open pivot hinge setup, it would appear that higher washout is beneficial to prevent stagnation regions. This study does not intend to address the superiority of one valve over the other and cannot conclude categorically that the higher washout rate is better. Clinically, with the presence of leaflets, higher washout implies higher flow, which could result in jetting or turbulence because of high velocities, which are also contributing factors to valve performance and TEs. This study emphasizes hinge design and associated fluid dynamic characteristics that would allow for optimization of leaflet ear design to promote clearance and washout.
This study evaluates only one factor out of numerous factors that influences valve performance. Although this technique could be indicative of regions with high thrombogenic propensity, the results cannot be translated directly into the clinical setting as additional factors other than stasis and residence time are involved in TEs (i.e., high shear, blood element damage).
In situ, the hinge recess washout is aided when the leaflets sweep open/close and during retrograde high velocity jetting through the gap between the hinge and the leaflet ear. Despite the action of the leaflets and jets, significant areas of very low flow still exist. Although the experiments conducted were open pivot studies (the leaflets were not included), the information herein is valuable, because it can help identify areas of concern within the hinges that need to be evaluated and targeted when designing leaflet ears and its sweep action to facilitate the removal of hinge residual volume.
Mitral valve experiments using laser Doppler velocimetry have revealed hinge recess velocities to be around 0.5 m/s (with variations stemming from cardiac cycle, measurement plane, and location within the hinge recess), with a maximum reported velocity of 0.75 m/s and a minimum of 0.18 m/s.7 There is significant experimental and clinical evidence that thrombus formation in the hinge occurs partly because of areas in which blood moves at the lower spectrum of these measured velocities. The flow regime used in this study was far below these reported values. Interestingly, in clinical situations, anticoagulation can be discontinued for several days without high thromboembolic risk, implying that the process time constant for a TE is longer than one cardiac cycle. Areas of sustained low flow can contribute over time to TEs. However as such, the washout times presented cannot be correlated directly to the clinical setting. The aim was not to duplicate the in vivo flow parameters during the cardiac cycle. The study was specifically aimed at systematically visualizing the effect of small increments of hinge flow in two geometrically distinct valves using the DBCS method.
The flow field out of an implanted mechanical valve is more unsteady than what has been studied here, and it is due in part to the presence of the leaflets and the pulsatile nature of the ejected flow volume. Ventricular contraction also gives rise to a natural helical flow pattern contributing further to the unsteady flow and is not within the scope of this study. The effects of pulsatile flow on DBCS performance were also outside the scope of this study and have therefore yet to be assessed. The Reynolds number for each experimental run was far lower than what is seen clinically because of the low flow velocities used. Hence, an important parameter to consider in the future will be unsteady flow and its dependence on the underlying Reynolds number. The visualization of turbulence is entirely possible using this imaging system.
This study introduces DBCS as a means to better visualize fluid flow dynamics and a simple method to allow comparison with computational models in the field of fluid mechanics, specifically in the study of prosthetic heart valves. Dairy-based colloidal suspension has shown promise with providing quantifiable high contrast flow visualization. By using this technique, two different valve housing segments containing their respective hinge geometries were compared, demonstrating significantly different washout dynamics. Although similar in overall hinge-pivot function, the specific hinge geometries (with the valve leaflets absent) created unique stagnation zones and clearance rates. In our model system, DBCS washout of valve 2 is markedly faster than valve 1, implying less inherent stagnation. This factor may be of importance to the thrombogenicity of valve designs. Although the relationship between experimental hinge washout dynamics and thromboembolic complications of the valves need to be compared with in vivo studies, the methodology shows promise to assist in investigating the contributing factors for TEs. In addition to its utility in investigating heart valves, this approach may be generally useful in evaluating stasis, turbulence, and disturbed blood flow in other cardiovascular devices, such as pumps and vascular prosthesis.
The authors thank Drs. Hongseok (Moses) Noh, Antonios Zavaliangos, and Patricia Shewokis for their advice and guidance.
Supplemental Video 1: Washout of Valve 1 and Valve 2 at 5.1x10-3 m/s. The DBCS in Valve 2 washed out 47.5% faster than Valve 1.
Supplemental Video 2: Top view of the intensity topography maps (i-Maps) depicting washout patterns of Valve 1 and Valve 2. The brighter color denotes higher DBCS concentrations, while the darker blue denotes lower or absence of DBCS. Note: The bright “sun-spot” that remains in the intensity plots after washout is a reflection artifact resulting from the microscope illumination.
1. Pibarot P, Dumesnil JG. Prosthetic heart valves: Selection of the optimal prosthesis and long-term management. Circulation. 2009;119:1034–1048
2. Forsberg P, DeSancho MT. Role of novel anticoagulants for patients with mechanical heart valves. Curr Atheroscler Rep. 2014;16:448
3. Tillquist MN, Maddox TM. Cardiac crossroads: Deciding between mechanical or bioprosthetic heart valve replacement. Patient Prefer Adherence. 2011;5:91–99
4. Yun BM, Wu J, Simon HA, et al. A numerical investigation of blood damage in the hinge area of aortic bileaflet mechanical heart valves during the leakage phase. Ann Biomed Eng. 2012;40:1468–1485
5. Grzymala-Lubanski B, Labaf A, Englund E, Svensson PJ, Själander A. Mechanical heart valve prosthesis and warfarin—Treatment quality and prognosis. Thromb Res. 2014;133:795–798
6. Jun BH, Saikrishnan N, Arjunon S, Yun BM, Yoganathan AP. Effect of hinge gap width of a St. Jude medical bileaflet mechanical heart valve on blood damage potential—An in vitro
micro particle image velocimetry study. J Biomech Eng. 2014;136:091008
7. Ellis JT, Yoganathan AP. A comparison of the hinge and near-hinge flow fields of the St Jude medical hemodynamic plus and regent bileaflet mechanical heart valves. J Thorac Cardiovasc Surg. 2000;119:83–93
8. Martin AJ, Christy JR. An in-vitro technique for assessment of thrombogenicity in mechanical prosthetic cardiac valves: Evaluation with a range of valve types. J Heart Valve Dis. 2004;13:509–520
9. Scharfschwerdt M, Thomschke M, Sievers HH. In-vitro localization of initial flow-induced thrombus formation in bileaflet mechanical heart valves. ASAIO J. 2009;55:19–23
10. Smits AJ Flow Visualization: Techniques and Examples. 20122nd ed. London Imperial College Press
11. Lee H, Tatsumi E, Taenaka Y. Flow visualization of a monoleaflet and bileaflet mechanical heart valve in a pneumatic ventricular assist device using A PIV system. ASAIO. 2010;56:186–193
12. Ezzat AW, Mansoor TM. Water flow visualization and velocity measurement using hydrogen bubble generation technique in low speed open channel. J Eng. 2012;18:844–858
mechanical heart valve; flow visualization; colloidal suspension; thromboembolism; flow velocity; flow stagnation
Supplemental Digital Content
Copyright © 2016 by the American Society for Artificial Internal Organs