Recently, our group published an alternative treatment for atrial fibrillation by using the mechanical support around both atriums using artificial muscles (AM) and avoiding the use of anticoagulation with its associated risks.1,2 By using the same technology, we addressed the problem of heart failure. To achieve success, the biomimetic material had to contract under the electrical impulsion and development of unique tridimensional designed disposure of the AM around the heart.3,4 To address these issues, we designed a specific ventricular assist device (VAD). Because our VAD was implantable around both ventricles, it created the possibility of external differential compression, and for this reason, we called it a biVAD with one device producing two different compressions. Until now, one device was used for each ventricle in children with total heart failure. These VADs have been proven as useful during years of their clinical use, but delayed secondary effects were responsible of long-term limited efficacy.5–8
The phase of processing and engineering of prototypes started in October 2008. We obtained prototypes able to reproduce the contraction activity of the heart, but we needed to test them and to verify whether they were capable of generating enough mechanical work to fill the void, in the case of heart failure. The creation of an in vitro test, the test bench (TB), was proven useful to verify the efficacy of the latest progress of the prototypes, to create a favorable environment for the design, and to provide information for the improvement of our prototypes.
Because of this effective new technology, the biVAD can usher in a new era of implantable VADs. These devices needed to be tested as mechanical circulating support before their use in vivo in animals. That was why this approach would need the creation of an experimental protocol to validate the TB and to make the results reproducible. The data obtained would constitute a reference for their comparison between in vitro and in vivo results after the phase of animal experimentation.
Our aim was to determine the real-time measurement of the modulation of mechanical movements, which recreated heart compression, and the impact of the biVAD mechanical movements on the static system that meant without the heart intrinsic beating. We chose to measure parameters of fluid displacement. To complete the TB, model heart was constructed based on the characteristics of the human heart and is described and tested herein.9–13
Materials and Methods
As far as we know, there was no existing TB to assess the hemodynamic performances of VADs for pediatric application. This was the reason why we decided to aim children's heart. We took children from newborns to teens. Our technology, derivate of our previous experience, was already efficient for small-size device. We looked for dimensions of a physiologic child's heart. The dimensions varied proportionally to child's age. At the beginning of our research, one example, at random, was enough as model close in size to reality. The dimensions, tissue characteristics, series of children heart measured, way of how we obtained it, and what kind of technique was used to acquire images were not significant if we supposed that our TB was used as the means to record the capacity to mechanically displace the fluid with the biVAD's prototypes. The computed tomography (CT) scan of a 10-year-old girl's failing heart was obtained (Figure 1). For example, she had end diastolic volume of 86 cm3, end systolic volume (42%) of 38 cm3, stroke volume of 48 cm3, and ejection fraction of 55% of the left ventricle, no representative of the left heart failure. The right ventricle showed end diastolic volume of 84 cm3, end systolic volume (42%) of 42 cm3, stroke volume of 42 cm3, and ejection fraction of 50%. The diastolic dimensions of child's heart were 83.9 mm on vertical line and 85.5 mm on horizontal line (Figure 1). The CT scan measures were the most accessible tool for us, comparing with magnetic resonance imaging or 3D ultrasound more precise.
These were the dimensional data used to construct our model. A synthetic foam rubber heart model with approximately the same sizes was scanned in 3D (NextEngine; Europe). The software SolidWorks processes these data. The heart model was formed with the 3D printer (Fab@Home) in silicone (Figure 2). The silicone used was MED 10-6607 NuSil Technology Europe, possessing mechanical properties similar to human muscle of the same thickness.14,15 We respected the fact that the myocardial wall thickness difference varies, where the left ventricle wall is thicker than the right ventricle wall. This produced a greater resistance to the compression of the left ventricle. Two chambers representing ventricles were created, without atriums. The largest size of the silicone model was 8 cm (±0.2 cm) in diameter when empty and 9 cm (±0.2 cm) when filled. The smallest size was 5.5 cm (±0.2 cm) diameter when empty and when filled. The height of the model was 7 cm (±0.2 cm). The volumes of both ventricles were 55 ml (±5 ml) each. With this model, the biVAD prototypes were attached, and the platform with the measuring systems was set up (see the following section).
Measuring Platform and Systems
The biVAD was designed to supply heart compression by exerting its mechanical action directly on both ventricles. This property made our biVAD a pulsatile VAD, and to measure the efficacy and the work capacity of this “pump,” we chose the height level of the liquid displaced in two vertical pipes connected in series, one for each of the ventricles. Other measured values were the ejected volume, the flow, the afterload, the temperature, and the contact pressure between the wall of the heart model and the biVAD for each of the ventricles. The platform consisted of a child's heart model affixed to a stiff oval pedestal made of Plexiglas (Figure 2). This structure was attached vertically to the upper part of the platform. This part was made up of two standard laboratory pipettes fixed on the stiff oval Plexiglas base, where each of these corresponded to one ventricle. The size of each pipette was 243-mm (±3 mm) long and 8-mm (±0.1 mm) of internal diameter. The working liquid used for the experience was NaCl 0.9% saline. Blood was not used, because in our institution, we need approval of ethical committee to use it, and this was not obtained.
In this vertically connected system, the level of replenishment of both pipettes with the liquid gave us the direct value of preload (Figure 3). The liquid level height in each of the pipettes could be measured by the instruments giving the precise values in real time. We chose the ultrasonic liquid level sensor with beam columnator, the UNAM 12U9914/S14D (Baumer), to measure this parameter. The main characteristic of this sensor was that it used ultrasonic waves to measure the distance between the emitter and the surface of the liquid in the pipettes. The range of its detection was 2–82 mm with a resolution of ±0.3 mm. This sensor generated an output signal ranging from 0 to 10 V according to the difference in height, where an analog signal was produced. One of these sensors was placed at the top of each pipette. We measured the temperature at the contact points between each ventricle and our biVAD during the activation of the prototype, because its intrinsic proprieties generated the heat in the system. A temperature probe sensor distributed by Distrelec was used for these measures. The initial temperature was 20°C. The measuring systems were held in place by a metallic skeleton in inversed Y-shape configuration and fixed with simple screws (Figure 4). Once the platform was constructed, we need to display the data and to supervise in real time, the functioning of our prototypes.
The signals generated by the ultrasonic sensors and the temperature probes were analog. To have a clear vision of what these signals meant, it was necessary to digitalize the signal, then to display it through a visual interface. We chose to use the LabVIEW software (National Instrument). This software permitted the creation of each of the parameters in a graphic representation screen containing their real-time values. For each measured value, we created the visual representation in two graphic screens, and the numerical representation for each ventricle with the instantaneous values and the mean values.
There were two sensors generating analogical signals. These ultrasonic sensors measured the liquid height level in pipettes. We programmed the software and created a “Virtual Interface Front Panel” containing the graphic screens. The screens showed the curves of liquid displacement inside the pipettes in real time. The y axis scale was adapted to the real time–generated liquid displacement in millimeters (Figure 4).
To obtain other values, the LabVIEW software was used to extrapolate these values from the measured height level values. That was the case for the ejected volume. We knew the internal diameter of the pipette (8 mm ± 0.1 mm). From that moment, the surface area of the cross section of the pipette was calculated, and these values were included into the software as the constant. The following mathematical formula was applied to calculate the volume:
where Π was the Pi number constant 3.14, r was the radius and constant 4 mm, and h was height variable in millimeter, which was measured by the sensors.
We obtained a volume of liquid displaced at each contraction of the biVAD, which meant the ejected volume. The y axis scale in graphics was adapted to the real-time ejected volume value in cubic millimeter (Figure 5).
The next parameter was the flow generated by each of ventricles. The values were calculated with the previously calculated ejected volume. The following mathematical formula was applied to calculate the flow:
where Q was the flow in milliliter per second, V was the ejected volume in cubic millimeter, and t was the time in seconds.
We obtained the flow of liquid displaced at each contraction of the biVAD. This parameter was calculated by derivation over time of the ejected volume, meaning the curves were derived from the ascending and descending phase. This provided the positive and negative values of this parameter. Only the positive phase was considered, because the described system did not have the reflux valve.
The afterload was the next measured value. We took the height level value as calculated previously. The value of 0 was the baseline of liquid height level in the inert system, without contraction of the biVAD. This value corresponded to the preload. This parameter was modifiable and controllable. It directly influenced the strength necessary to displace the fluid in the system, because the bigger the preload, the more the column of fluid weighted on the ventricular walls. For this reason, the preload influenced indirectly the height of liquid in the pipette that meant the afterload. The software was programmed to calculate the height level value at the end of each contraction, meaning at each maxima. As we knew the diameter of each pipette, we used the mathematical formula to calculate the end-contraction pressure:
where ΔP was afterload value in millibar, ρ was the relative density of NaCl 0.9% saline and constant, h was the height level value in meter, and g was the acceleration because of the gravity.
We obtained the afterload generated at each contraction of the biVAD. To calculate this parameter, only the positive values of the flow were used, and so the after load values at the end of systole (telesystolic) were obtained. The y axis graphic scale was adapted to real time after the load values in millibar. The biVAD frequency of contraction, or beating, was calculated by counting the number of maxima in the curves during 1 minute. Then this number was reported in the box reserved for the level of the biVAD's frequency (Figure 6).
Temperature was a value also measured in this system. The prototypes were based on the use of AM. These AM were composed of Nickel-Titanium (Nitinol) alloy. During its functioning, the Nitinol generates heat. To quantify this heat and because the Nitinol in our concept could potentially be in contact with the myocardial tissue, we put the temperature probe in direct contact with the heart model. The probe was also connected to the software. The temperature values generated around the wires during the work of the biVAD were obtained. These values are represented in the box showing in real time the mean temperature (degree Celsius) of each ventricles (Figure 7). The coefficient of thermal conductivity of the blood is approximately the same as that of the water.16 Therefore, we believe that temperature measurements are reliable even if we use saline water instead of blood.
When the biVAD contracted, it compressed the heart model externally. To measure this contact pressure, the pressure sensors were placed at several points of contact between the heart model and the biVAD. These pressure values were required because they could determine the threshold of pressure, which would produce the tissue necrosis. These measurements were made using the Tactilus (Sensor Products Inc.) sensors interfaced with their own computer display. The sensors were introduced between every direct point of contact of the biVAD and the heart model. The computer program showed pressure (y axis) versus time (x axis) in millimeters mercury and the contact area (y axis) versus time (x axis) in square inches. We saved and stocked all data for each of the quantified parameters in Excel files (Microsoft) under tabloid forms for the future statistical applications.
The height level displacement of the liquid in the pipettes with the ultrasonic sensors in the range of 0–80 mm was measured in the vertical dimension. The NI LabVIEW software extrapolated the precise ejected volume in the range of 0–4019.20 mm3 at each contraction of the biVAD, the flow was generated in the range of 0–4.0192 ml/s, and the afterload in the range of 0–80.8344 mbar produced by each of the ventricles supported with our biVAD prototypes. We also used two other types of sensors. The first was the temperature probe, which gave the precise values of heat generated during the functioning of the biVAD with starting temperature at 20°C. The importance and the impact of the heat on the myocardial tissue could be estimated in terms of its impact. The contact pressure sensors were also used to quantify whether there was a risk of tissue necrosis during the biVAD functioning. This necrosis could be generated by the direct contact between the prototypes and the myocardial surface. All these values depended on the characteristics of biVAD prototype. The NI LabVIEW software controlled the functioning of the biVAD and processed in real time, the data coming from the three sensors. A virtual interface front panel (VI Front panel) was created. Through extrapolation, we were able to display measured values of the ejected volume (cubic millimeter), flow (milliliter per second), and afterload (millibar) for each ventricle directly on the VI front panel. The values directly obtained were the height level of liquid (millimeter), the temperature (degree Celsius), and the contact pressure (millimeters mercury) (and square inches). The experimental protocols were defined to obtain predictable results based on the smart biVAD's contraction. The cost for the construction of the complete TB was kept to a strict minimum. The purchase price for necessary materials was less than $2000 (Table 1).
The design, construction, and validation of the TB (Figure 8) were an essential step in the field for designing a cardiac VAD based on the use of AM. After the phase of conception of a prototype, it was ethically and legally indispensable to start animal experiments that meant in vivo tests, to confirm and to prove the efficacy of the new prototype. The TB used highly precise measuring instruments giving the opportunity to develop the standardized protocols permitting in vitro evaluation of a biVAD based on AM, which meant reproducibility for each result and the capability to compare different prototypes. This is the main reason why one device that is able to do two different compressions was created. This reproducibility was to verify or to certify the efficacy of the biVAD. Another reason that we determined a particular TB was the cost. Its construction was cost effective, and during the phase of in vitro tests, and no animal was killed. The use of animals for experiments is very expensive and labor intensive. So this application gave the possibility to save money and to invest it in the development of the prototype. If the biVAD prototypes prove their efficacy in vitro, it would be first necessary to create an in vitro heart failure model to improve their efficacy before they could be in the future tested under physiologic and pathologic conditions during in vivo with animals.
We believe that the TB made the development of cardiac devices based on AM safer, more cost effective, more time effective, more precise, and spared the sacrifice of numerous animals.
Supported by The Swiss National Science Foundation (SNSF).
1. Abdelnour-Berchtold E, Tozzi P, Siniscalchi G, et al
: Atrial assist device, a new alternative to lifelong anticoagulation? Swiss Med Wkly
139: 58–59, 2009.
2. Tozzi P, Hayoz D, Siniscalchi G, et al
: Artificial muscle to wash blood out of fibrillating atrium: An alternative to lifelong anticoagulation. ASAIO J
55: 24–27, 2009.
3. Sawyer PN, Page M, Baseliust L: Further studies of nitinol wires as contractile artificial muscle for an artificial heart. Cardiovasc Dis
3: 65–78, 1976.
4. Levi DS, Kusnezov N, Carman GP: Smart materials applications for pediatrics cardiovascular devices. Pediatr Res
63: 552–558, 2008.
5. Kamdar F, Boyle A, Liao K, et al
: Effects of centrifugal, axial and pulsatile left ventricular assist device support on end-organ function in heart failure patients. J Heart Lung Transplant
28: 352–359, 2009.
6. Radovancevic B, Vrtovec B, de Kort E, et al
: End-organ function in patients on long-term circulatory support with continuous or pulsatile-flow assist devices. J Heart Lung Transplant
26: 815–818, 2007.
7. Klotz S, Deng MC, Stypmann J: Left ventricular pressure and volume unloading during pulsatile versus nonpulsatile left ventricular assist device support. Ann Thorac Surg
77: 143–149, 2004; discussion 149–150.
8. Garcia S, Kandar F, Boyle A, et al
: Effects of pulsatile- and continuous-flow left ventricular assist device on left ventricular unloading. J Heart Lung Transplant
27: 261–267, 2008.
9. Timms D, Hayne M, McNeil K, Galbraith A: A complete mock circulation loop for the evaluation of left, right, and biventricular assist devices. Artif Organs
29: 564–572, 2005.
10. Patel S, Allaire PE, Wood HG, et al
: Design and construction of a mock human circulatory system, in Summer Bioengineering Conference
. Florida, Sonesota Beach Resort in Key Biscayne, 2009, pp. 965–966.
11. Pantalos GM, Koenig SC, Gillars KJ, et al
: Characterization of an adult mock circulation for testing cardiac support devices. ASAIO J
50: 37–46, 2004.
12. Liu Y, Allaire P, Wood H, Olsen D: Design and initial testing of a mock human circulatory loop for left ventricular assist device performance testing. Artif Organs
29: 341–345, 2005.
13. Swier P, Bos WJ, Mohammad SF, et al
: An in vitro test model to study the performance and thrombogenecity of cardiovascular devices. ASAIO Trans
35: 683–687, 1989.
14. Gregory S, Timms D, Pearcy MJ, Tansley G: A naturally shaped silicone ventricle evaluated in a mock circulation loop: A preliminary study. J Med Eng Technol
33: 185–191, 2009.
15. Kolipaka A, McGee KP, Araoz PA, et al
: MR elastography as a method for the assessment of myocardial stiffness: Comparison with an established pressure-volume model in a left ventricular model of the heart. Magn Reson Med
62: 135–140, 2009.
Copyright © 2011 by the American Society for Artificial Internal Organs
16. Ponder E: The coefficient of thermal conductivity of blood and of various tissues. J Gen Physiol
45: 545–551, 1962.