Share this article on:

A Novel Pediatric Biventricular Assist Device: In Vitro Test Results

Perrodin, Stéphanie F.; Tozzi, Piergiorgio

doi: 10.1097/MAT.0000000000000060
Pediatric Circulatory Support

Most ventricular assist devices (VADs) currently used in infants are extracorporeal. These VADs require long-term anticoagulation therapy and extensive surgery, and two devices are needed for biventricular support. We designed a biventricular assist device based on shape memory alloy that reproduces the hemodynamic effects of cardiomyoplasty, supporting the heart with a compressing movement, and evaluated its performance in a dedicated mockup system. Nitinol fibers are the device’s key component. Ejection fraction (EF), cardiac output (CO), and generated systolic pressure were measured on a test bench. Our test bench settings were a preload range of 0–15 mm Hg, an afterload range of 0–160 mm Hg, and a heart rate (HR) of 20, 30, 40, and 60 beats/min. A power supply of 15 volts and 3.5 amperes was necessary. The EF range went from 34.4% to 1.2% as the afterload and HR increased, along with a CO from 180 to 6 ml/min. The device generated a maximal systolic pressure of 25 mm Hg. Cardiac compression for biventricular assistance in child-sized heart using shape memory alloy is technically feasible. Further testing remains necessary to assess this VAD’s in vivo performance range and its reliability.

From the Department of Cardiovascular Surgery, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland.

Submitted for consideration September 2013; accepted for publication in revised form January 2014.

Disclosure: The authors have no conflict of interest to report.

This study was supported by public funds from Cardiomet.

Reprint Requests: Piergiorgio Tozzi, MD, PD, MER, Department of Cardiovascular Surgery, Centre Hospitalier Universitaire Vaudois, Rue du Bugnon 21, 1011 Lausanne, Switzerland. Email:

Heart transplantation is the therapy of choice for end-stage heart failure that is refractory to medical treatment. Ventricular assist devices (VADs) were initially developed as a bridge to transplant therapy to compensate the chronic shortage of organs we face and successfully used in adults and children.1,2 Five hundred thirty-two heart transplants in children younger than 18 years were reported to the Registry of the International Society for Heart and Lung Transplantation in 2010. One hundred thirty-five of these recipients were infants (younger than 1 year), 22% of these infants were bridged with mechanical support devices, extracorporeal membrane oxygenation being far more frequent in this age group than VAD.3

The Berlin Heart EXCOR is the most commonly used VAD for bridge to transplant therapy in small children and is one of the rare devices suitable for infants. The main complications are neurologic events and infections,4–6 and anticoagulation therapy management remains one of the major challenges, particularly in children with a body surface area <1.2 m2. For these patients, the Berlin Heart EXCOR can be a successful bridge to transplant therapy, but it is associated with significant morbidity and mortality.7–9

Research currently focuses on minimizing existing adult-sized devices to make them completely implantable in small children and lowering the thrombogenic risk to reduce anticoagulation therapy.10,11 However, most of the complications faced with the Berlin Heart EXCOR or other adult VADs will probably remain with these new devices: thromboembolic complications, the need for anticoagulation therapy and subsequent hemorrhage risk, infections, hemolysis, allosensitization, and secondary right heart failure requiring implantation of a second VAD.

Our group has been working to create a device that is implanted around the heart, allowing biventricular support without direct contact with the blood. This will reduce the thromboembolic risk and therefore suppress the need for anticoagulation therapy. Furthermore, only one device is needed for biventricular support, and it is entirely implantable whatever the size of the child.

We created this device using artificial muscle technology. Smart materials such as nitinol, a shape memory alloy made of nickel and titanium, have inherent properties allowing their use as the contractile element in our device. A nitinol fiber can change its molecular configuration to shorten when heated and lengthen back to its original size when cooled. By subsequently turning the current on and off, we achieve a contraction-like movement.12,13

We used this technology previously to build the Atripump, a dome-shaped device that is placed on the atrium and assists it in its contraction. It significantly improved atrial outflow in a chronic atrial fibrillation animal model.14 We then developed a child-sized biventricular assist device using the same technology and tested it on a test bench.15,16 A solid U-shaped structure was placed around the heart model along the interventricular septum and acted as the skeleton, on which we fastened the nitinol fibers following a spiral line from apex to base. When activated, it reached an ejection fraction (EF) of 12%.

In this study, we will present the in vitro test results of the pediVAD, a new VAD based on the same technology as our previous device and built to fit an infant-sized heart.

Back to Top | Article Outline

Materials and Methods

Device Description

We designed the pediVAD using BMX 100 (BioMetal, Toki Corporation, Tokyo, Japan), a coiled nitinol fiber made of 50% nickel and 50% titanium. With a diameter of 100 μm and a coil diameter of 400 μm, it has a kinetic displacement of up to 200% and an electric resistance of 900 Ohm/m. The maximum force developed by this fiber is 10–20 g.

We placed one wooden disc at the apex and one at the base of our heart model, on which we fastened 16 nitinol fibers. Each fiber was 2.2 ± 0.1 cm long and stretched to 4.4 ± 0.1 cm before it was mounted vertically and regularly spaced around the heart, as seen in Figure 1. When activated, the nitinol fibers shorten back to their original length (2.2 ± 0.1 cm) and bring the two discs closer together, compressing the heart. This device was created to support the entire heart, as the contraction is vertical from apex to base. We cannot activate selectively one ventricle or the other as allowed by configuration described in our previous work.15

Figure 1

Figure 1

As the fibers were in contact with our heart model at its widest diameter, we surrounded the heart with an isolating stripe of Kevlar tissue.

Back to Top | Article Outline

Mock Circulatory System

The pediVAD was built around a heart model made of synthetic rubber. It had an end-diastolic volume (EDV) of 26 ml. As mentioned previously, as our device is to provide vertical compression of the heart from apex to base and does not allow for selective left or right ventricular compression, we elected to test it on a heart model with only one ventricle. This heart model is about the size of the left ventricle of a healthy 3 year old or that of an infant in severe heart failure.17

The heart was connected to a plastic pipe with an internal diameter of 1 cm, representing the aorta. Our test bench has previously been described.16 The height of the water column in the pipe was measured in centimeters of water (cmH2O) and converted to millimeters of mercury (mm Hg). It represents the preload. To change the afterload, we connected the end of the pipe to a manometer via a rubber tube (Figure 2A). We also used an ultrasonic probe to measure water height variation precisely during device activation. The results were recorded in a dedicated software (LabVIEW, National Instruments Co, Austin, TX) (Figure 2B). Finally, we measured the pressure generated by the pediVAD using the same technology as an arterial line, with a catheter connected to the pipe and a monitor.

Figure 2

Figure 2

A laboratory power supply connected to an arbitrary waveform generator was used to activate our device, as seen in Figure 2C. We selected a square wave, its frequency, and duty cycle that were shown on an oscilloscope. The frequency at which the current is activated and turned off by the arbitrary waveform generator determines the heart rate (HR). The duration of current flow in one cycle is the stroke length.

Back to Top | Article Outline

In Vitro Test Protocol

We set the arbitrary waveform generator to deliver a HR of 20, 30, 40, and 60 beats/min with the systole lasting 30% of the cardiac cycle length. The afterload ranged between 0 and 160 mm Hg. The preload range was 0–15 mm Hg.

For each of the previously mentioned settings, we measured the generated pressure and water height variation in the pipe. We calculated the stroke volume (SV), the EF, and the cardiac output (CO) with the following equations: SV (ml) = Δh · π · r 2; EF (%) = SV/EDV; CO (ml/min) = SV · HR. Each set of measurements has been carried out for more than 4 hours.

Back to Top | Article Outline


To reach maximal contraction, the pediVAD required a power supply of 15 volts (V) and 3.5 amperes (A). The ejection fraction achieved by this device is 16.3 ± 0.3% with a HR of 30 beats/min, 9.7 ± 0.3% at 40 beats/min, and 7.9 ± 0.3% at 60 beats/min, as seen in Figure 3. It reaches 34.4 ± 0.3% when the HR is 20 beats/min. There is a slight elevation in the EDV as the pediVAD is activated at HRs of 40 and 60 beats/min. Our device shows signs of diastolic dysfunction when activated at a faster rate than 30 beats/min. This was minimized by shortening the stroke length to 30% of the cardiac cycle length instead of 50%, allowing more time for the fiber to cool down, without loss of performance.

Figure 3

Figure 3

The pediVAD is able to deliver a maximal CO of 0.18 ± 0.02 L/min (Figure 4). The blood flow ranges between 100 ± 20 and 180 ± 20 ml/min depending on the HR.

Figure 4

Figure 4

The EF diminishes drastically when the afterload is increased, as illustrated in Figure 5. Although efficient, our device is not powerful enough to contract against high afterload pressures.

Figure 5

Figure 5

The EF also diminishes as the preload rises (Figure 6). As it is made of synthetic rubber, its resiliency is low and so is its compliance.

Figure 6

Figure 6

The pediVAd generates a maximal pressure of 25 ± 1 mm Hg, lowering as the HR elevates. The pressure–volume loops at various HRs are illustrated in Figure 7.

Figure 7

Figure 7

Back to Top | Article Outline


The pediVAD has a maximal EF of 34.4%, which corresponds to a SV of 9 ml. It is a little more than half the normal values for a healthy left ventricle of this size (EDV of 26 ml).17,18 This represents a 50% improvement of heart function when assisted with our device.

These results far exceed those we managed with our previous device.15 The vertical orientation of the nitinol and the use of coiled fibers generate a better hemodynamic effect than our earlier design.

The CO is 0.18 L/min, which is approximately 10% of the value expected for this size ventricle.18 Although the SV is high, the HR is very low, resulting in a low CO. The nitinol fibers are capable of generating enough force to shrink to their shortest configuration when activated, thus producing high SVs, but with our current method of activation, they need a longer time to cool down and stretch back to their longer configuration. This provides us from shortening the diastolic duration and explains the poorer results obtained with higher HRs. We can bypass that problem by using an electronic control unit (ECU) with pulse-width modulation stimulation, delivering a precise quantity of power to each fiber individually.15 This prevents the nitinol from being overheated and optimizes relaxation time, which in turn allows for higher HRs. Moreover, for in vivo setting, the blood itself will act as a heat exchanger, further reducing the cooling time of the fibers and the diastolic dysfunction. Another scenario could be to assist one ventricular contraction over two when the HR is higher than 100 beats/min.

Other pediatric VADs being currently in development, such as the Penn State device or the PediaFlow for example, have better hemodynamic results. They can maintain a flow rate from 0.5 to 1.4 L/min.19–21 However, all these devices are intended to replace the heart by deviating the blood flow through the machine, whereas our device is designed to enhance existing heart function to improve hemodynamic performance.

The pediVAD generated a pressure of up to 25 mm Hg. When we compare the stroke work (PV loop area, Figure 7) at various HRs, we can see that the higher the HR, the lower the stroke work. This illustrates the diastolic dysfunction our device suffers from. There are two probable causes to this issue, the first being the time the nitinol fibers need to regain their original configuration after having been activated. They contract quickly and easily, but need a long time to lengthen. When the HR rises, the diastolic time becomes too short to allow for complete elongation. We minimized this problem by reducing the stroke time to 30% of the duration of the cardiac cycle, but some dysfunction remains. By further reducing the stroke time, we lose efficiency with the energy supplying method we used in this study. However, by using the previously described ECU,15 we will probably be able to optimize systole and diastole duration and test our pediVAD with higher HRs more suitable to infant physiology. The second probable cause is the material our heart model is made of. This synthetic rubber is not very extensible and has low compliance. When we elevate the preload for example, the performance drops instead of following the Frank–Starling law. By using a silicone heart model, a more compliant material, we could avoid this problem.

For in vivo setting, the discs placed at the base and the apex of the heart model will be replaced, respectively, by Kevlar ring, at the base, and disk, at the apex, sutured to the epicardium. A silicon rubber coating will prevent heart damage and diastolic dysfunction.

Finally, our device does not perform very well when pushing against high afterload pressures. Our goal is to get the same EF with an afterload pressure of 0–100 mm Hg, to accommodate infant physiology. The nitinol fibers we used and the position they were in achieved great results but were not powerful enough. Increasing the number of nitinol fibers around the heart could easily raise the power of contraction generated by the pediVAD. We elected to initially build our device with as few fibers as possible because we wanted to determine the efficiency of the design with minimal costs.

Back to Top | Article Outline


Hemodynamic support of an infant with this device applied around the heart as an artificial cardiomyoplasty is technically feasible. Even if there are still technical challenges to overcome, artificial cardiomyoplasty should be able to temporarily improve the hemodynamic of diseased heart facilitating the weaning of extracorporeal circulation after complex reconstructive surgery. The pediVAD could bring a significant amelioration to the patient’s New York Heart Association class, allowing for greater mobility of the child and a better quality of life. Our next steps include increasing the number of nitinol fibers on the device and improving energy delivery to the device to minimize diastolic dysfunction and elevate HR. Further testing also remains necessary to assess this VAD’s in vivo performance range and its reliability.

Back to Top | Article Outline


1. Terracciano CM, Miller LW, Yacoub MH. Contemporary use of ventricular assist devices. Annu Rev Med. 2010;61:255–270
2. Esper SA, Subramaniam K. Heart failure and mechanical circulatory support. Best Pract Res Clin Anaesthesiol. 2012;26:91–104
3. Kirk R, Dipchand AI, Edwards LB, et al.International Society for Heart and Lung Transplantation. The Registry of the International Society for Heart and Lung Transplantation: Fifteenth pediatric heart transplantation report–2012. J Heart Lung Transplant. 2012;31:1065–1072
4. Clark JB, Pauliks LB, Myers JL, Undar A. Mechanical circulatory support for end-stage heart failure in repaired and palliated congenital heart disease. Curr Cardiol Rev. 2011;7:102–109
5. Morales DL, Almond CS, Jaquiss RD, et al. Bridging children of all sizes to cardiac transplantation: The initial multicenter North American experience with the Berlin Heart EXCOR ventricular assist device. J Heart Lung Transplant. 2011;30:1–8
6. Yuki K, Sharma R, DiNardo J. Ventricular-assist device therapy in children. Best Pract Res Clin Anaesthesiol. 2012;26:247–264
7. Brancaccio G, Amodeo A, Ricci Z, et al. Mechanical assist device as a bridge to heart transplantation in children less than 10 kilograms. Ann Thorac Surg. 2010;90:58–62
8. Gandhi SK, Huddleston CB, Balzer DT, Epstein DJ, Boschert TA, Canter CE. Biventricular assist devices as a bridge to heart transplantation in small children. Circulation. 2008;118(suppl):S89–S93
9. Fan Y, Weng YG, Xiao YB, et al. Outcomes of ventricular assist device support in young patients with small body surface area. Eur J Cardiothorac Surg. 2011;39:699–704
10. Baldwin JT, Borovetz HS, Duncan BW, Gartner MJ, Jarvik RK, Weiss WJ. The national heart, lung, and blood institute pediatric circulatory support program: A summary of the 5-year experience. Circulation. 2011;123:1233–1240
11. Jefferies JL, Morales DL. Mechanical circulatory support in children: Bridge to transplant versus recovery. Curr Heart Fail Rep. 2012;9:236–243
12. Levi DS, Kusnezov N, Carman GP. Smart materials applications for pediatric cardiovascular devices. Pediatr Res. 2008;63:552–558
13. Hassoulas IA, Ladopoulos VS, Kalogerakos PD. Study of shape memory alloy fibers for the development of artificial myocardium. Hellenic J Cardiol. 2010;51:301–309
14. Tozzi P, Hayoz D, Taub S, Muradbegovic M, Rizzo E, von Segesser LK. Biometal muscle to restore atrial transport function in a permanent atrial fibrillation animal model: A potential tool in the treatment of end-stage heart failure. Eur J Cardiothorac Surg. 2010;37:870–874
15. Tozzi P, Michalis A, Hayoz D, Locca D, von Segesser LK. Artificial muscle for end-stage heart failure. ASAIO J. 2012;58:103–108
16. Muradbegovic M, Taub S, Rizzo E, von Segesser LK, Tozzi P. Ultimate test bench for pediatric biventricular assist device based on artificial muscles. ASAIO J. 2011;57:62–67
17. Poutanen T, Jokinen E, Sairanen H, Tikanoja T. Left atrial and left ventricular function in healthy children and young adults assessed by three dimensional echocardiography. Heart. 2003;89:544– 549
18. Cattermole GN, Leung PY, Mak PS, Chan SS, Graham CA, Rainer TH. The normal ranges of cardiovascular parameters in children measured using the Ultrasonic Cardiac Output Monitor. Crit Care Med. 2010;38:1875–1881
19. Roszelle BN, Deutsch S, Manning KB. A parametric study of valve orientation on the flow patterns of the Penn State pulsatile pediatric ventricular assist device. ASAIO J. 2010;56:356–363
20. Schönberger M, Deutsch S, Manning KB. The influence of device position on the flow within the Penn State 12 cc pediatric ventricular assist device. ASAIO J. 2012;58:481–493
21. Maul TM, Kocyildirim E, Johnson CA Jr, et al.The Pediaflow Consortium. In vitro and in vivo performance evaluation of the second developmental version of the pediaflow pediatric ventricular assist device. Cardiovasc Eng Technol. 2011;2:253–262

artificial organ; circulatory assist device; mechanical circulatory support; ventricular assist device; heart failure surgery

Copyright © 2014 by the American Society for Artificial Internal Organs