It is widely recognized that the development of circulatory support systems for the pediatric population has lagged behind that of the adult systems, partly because of the fewer number of potential patients and the challenges inherent in achieving a small nonthrombogenic pump and acceptable cannulation techniques. The National Heart Lung and Blood Institute of the National Institutes of Health issued a request for proposals in 2002 leading to the funding of five contracts to develop circulatory and respiratory support systems for infants and children. Under this program, The Pennsylvania State University (Penn State) is developing a pulsatile pediatric ventricular assist device (PVAD) for infants and children.
Few pumps have been designed or adapted for pediatric use. Recent reviews by Duncan,1 Fuchs and Netz,2 Reinhartz et al.,3 Throckmorton et al.,4 and Karl and Horton5 describe the devices, indications, and outcomes. The two major categories are continuous-flow pumps and pulsatile (positive displacement) pumps. Indications for pediatric circulatory support may be divided into two groups, as defined by the National Heart Lung and Blood Institute program and others:
- Ventricular dysfunction associated with congenital heart disease, especially hypoplastic left heart syndrome, single ventricle, transposition of the great arteries, tetralogy of Fallot, and anomalous origin of the left coronary artery.
- Ventricular dysfunction due to myocarditis, sepsis, cardiomyopathy, or cardiopulmonary collapse.
These two categories generally define two different populations in terms of age and cardiac output requirements, methods of initiation of support, duration of support, patient mobility, and therapeutic objective. The maximum duration of use has been defined as 6 months for this program.
Pulsatile pneumatic ventricular assist devices have a number of advantages, especially in applications in which a paracorporeal pump is acceptable. Inlet cannulation may be atrial or ventricular; automatic control is easily implemented by fill sensing and adjustment of the pneumatic vacuum during diastole; ejection pressure can be adjusted pneumatically for pulmonary artery or systemic arterial pressures; and, the positive displacement pump is inherently pulsatile. A disadvantage of a pulsatile system is the limiting effect of cannula impedance on filling and ejection, which may be partially overcome by the use of increased pneumatic drive pressures, and the requirement for valves.
Adult Pulsatile Ventricular Assist Devices
The Thoratec VAD System (Thoratec Corp., Pleasanton, CA) is a pneumatically actuated pulsatile assist pump. This pump is the commercially available version of the Pierce-Donachy VAD developed at Penn State.6 More than 3,100 devices have been used in over 2,000 patients worldwide. This device is approved for bridge-to-transplant and postcardiotomy recovery and is the only system approved by the US Food and Drug Administration for right ventricular (RVAD) or biventricular (BVAD) support. Due to the paracorporeal placement and flexibility in cannulation method, it may be used in small patients and has been used to support 176 patients younger than 18 years.3,7 The mean patient age was 13.9 years, with a mean body surface area (BSA) of 1.54 m2. The BSA was less than 1.3 m2 in 35 patients, with the smallest patient being 5 years old with a BSA of 0.73 m2. Survival to transplantation was 60.5%, and survival to weaning was 10.2%.
The Thoratec VAD has a dynamic stroke volume of 65 ml and was designed for the adult cardiac output range. It is generally recommended that the flow rate be maintained at values of more than 2.5 l/min (approximately 40 beats/min) to prevent thrombosis associated with less effective washing of the blood sac. Its use is, therefore, limited in smaller children and infants.
Implantable pulsatile pumps designed for the adult population have been used to a very limited extent in children. Cannulae and/or pump placement is restricted, in some cases preventing sternal closure.5 Use of the adult-sized pumps in children carries an increased risk of thromboembolism due to flow rates below the intended design range.
Pediatric Pulsatile Ventricular Assist Devices
The Berlin Heart EXCOR Pediatric VAD (Berlin Heart AG, Berlin, Germany) is a pneumatically actuated pulsatile assist device available in sizes of 12, 15, 25, 30, 50, 60, and 80 ml. Mechanical tilting-disk valves or polyurethane trileaflet valves are available. The pump chamber is transparent polyurethane, and the diaphragm is a multilayer polyurethane design with graphite interlayer lubrication. Carmeda heparin-coating (Carmeda AB, Upplands Väsby, Sweden) is used on all polyurethane surfaces.8 The Berlin Heart is currently available in the United States under special Food and Drug Administration permission, where over 20 patients have received the device.
The largest experience with this device has been at the Berlin Heart Institute, where 66 patients ranging in age from 2 days to 17 years have been supported with the Berlin Heart (31 left ventricular assist devices [LVADs] and 35 BVADs).3,9 Nine children were successfully weaned from ventricular assist device support, 28 were transplanted, 27 died on support, and 2 were ongoing. Support duration ranged from 1 to 591 days, with a mean of 42 days. All but the first 17 patients used Carmeda-coated pumps.
The MEDOS-VAD System (MEDOS Medizintechnik AG, Stoberg, Germany) is a pneumatically actuated pulsatile assist device developed at the Helmholtz Institute in Aachen, Germany. The LVAD is manufactured with stroke volumes of 10, 25, and 60 ml, and the RVAD in 9, 22, and 55 ml stroke volumes. The pump uses a double-layer polyurethane diaphragm and polyurethane trileaflet valves. The MEDOS-VAD has been used in over 650 patients, including 140 pediatric patients.10 The 10/9-ml size was used in 38 patients and the 25/22.5-ml was used in 51 patients. BVADs were used in 44% of the patients. The longest support time was 216 days. There have also been a number of applications in the United States.
The Penn State PVAD
The Penn State PVAD is a pneumatically actuated, positive displacement pump, similar in design to the adult-sized Thoratec VAD. The important design features that distinguish this design from other pneumatic ventricular assist devices are the use of a seamless blood sac fabricated from segmented poly(ether polyurethane urea) (SPEUU) in combination with a diaphragm, and the use of mechanical heart valves.
The device under development is intended to be placed either paracorporeally, as shown in Figure 1, or to be implanted. Implantability may be advantageous in children in the bridge-to-transplant application where extended durations of support and hospital discharge are likely.
An advantage of the pneumatic system, which has been demonstrated in the application of the Thoratec VAD, is the flexibility of providing right, left, or biventricular support. This flexibility is a result of the cannulation options as well as the ability to adjust the diastolic drive pressures to atrial or ventricular filling pressures, and the systolic drive pressure to either aortic or pulmonary pressures.
An earlier version of the PVAD was under development by our group and 3M Corporation in the late 1980s and early 1990s. The PVAD had a static stroke volume of 15 ml and used 6-mm ball valves (Figure 2). The PVAD was implanted in 20 juvenile goats and 10 calves. The studies were of 4-week maximum duration. Although the pump was intended to be placed paracorporeally in humans, the pump was implanted in animals to prevent pump damage or accidental disconnection. Placement was preperitoneal in goats and intrathoracic in calves. Anticoagulation included combinations of heparin, warfarin, and aspirin. Left atrial cannulation was used in all cases.
With the exception of the last two animals, all pumps were manufactured by 3M. The blood sacs in the first 28 animals were 3M's Hemothane SPEUU. The last three of this group used blood sacs with covalently bonded heparin. The final two animals used Biolon SPEUU sacs manufactured at Penn State.
There were 19 animals supported for more than 14 days. Thrombus was frequently found on the blood sac surface and cannulae. SEM analysis showed regions of platelets and proteinaceous material. This is in contrast to the generally clean surfaces found at explant in the adult-sized pump. A number of design problems were identified, including a 100-μm step at the valve-sac junction in the first six implants, premature fill switch activation and protrusion of the fill switch into the pumping region, cannula disconnection in three cases, driveline disconnection, and a kinked driveline in one case.
A comparison of the pediatric VAD with the adult-sized VAD yields two important differences. First, the pediatric pump did not preserve fluid dynamic similitude relative to the adult-sized pump. Nondimensional numbers that describe the flow were not preserved.11 Second, the ball valves produced an entirely different flow pattern in the pump compared with the tilting disk valves used in the adult pump,12 and the transvalvular pressure gradients were significantly higher because of the design of the ball valves and their small diameter. It has been hypothesized that the high-pressure gradients and associated high shear stresses caused platelet activation, and these activated platelets readily attached to the blood sac surface in regions of low shear rate. Furthermore, the ball acted as a central occluder, whereas the bileaflet valve or tilting disk valve has a central or slightly off-center large orifice.
In summary, animal testing of the original-design pediatric pump showed a tendency to form thrombus in the blood sac, despite anticoagulation. Subsequent testing suggested that the ball valve size, pressure gradients, and pump flow patterns were dissimilar to the adult-sized VAD. In addition, a number of design problems that compromised the desired flow patterns in the pump, or led to inadequate flow rate, may have contributed.
Two new PVAD pump sizes are being developed: an infant-sized device with a dynamic stroke volume of approximately 12 ml, and a 25-ml stroke volume child-sized device. The weight and age range for theses two sizes are shown in Figure 3. For a given age, the cardiac output required is shown for the mean body mass (heavy line) as well as the 5th and 95th percentiles (lighter lines). The 12-ml PVAD supports infants from birth up to a small 1-year-old (5th body mass percentile) and the 25-ml child PVAD supports a large 3-month-old (95th body mass percentile) up to a small 9.5-year-old (5th body mass percentile). Outputs exceeding 3 l/min may be provided by the adult VADs. The maximum duration of use has been estimated at 6 months.
The prototype 12-ml PVAD is shown in Figure 4. The following key design variables are being evaluated though analysis and in vitro testing: valves (type, size, and orientation), pneumatic drive pressure parameters (pressure profile, dP/dt, systolic duration, systolic/diastolic timing), cannulae design (length, angle, taper, tip design, materials), and blood sac polymer.
In Vitro Testing
Performance testing is intended to measure pressure-flow relationships and the sensitivity of the PVAD to preload and afterload. Performance testing of an optically clear test pump with a 12-ml (nominal) stroke volume was performed on a Penn State mock circulatory loop.13 Room-temperature 30% glycerin-water solution with a viscosity of a 2.6 centipoise was used. The cannulae had a 6.35 mm (0.25 in) internal diameter and were 15 cm long, which simulated the intended clinical application. The mean inlet pressure was varied from 0 to 26 mm Hg and the arterial pressure was maintained at 90/60 mm Hg, with mean 75 ± 3 mm Hg.
The systolic pneumatic drive pressure was 225 mm Hg with a fixed duration of 300 milliseconds. Diastolic pneumatic drive vacuum was varied in 15-mm Hg increments from 0 to 60 mm Hg. Two valve types were evaluated: the Björk-Shiley monoleaflet valve and the Carbomedics bileaflet valve.
Pump output and stroke volume data is shown in Figure 5. The pump under test contained a diaphragm but no blood sac or fill switch. Filling was estimated by visual inspection of the pump and flow/pressure waveforms. The maximum flow rate (at 102 beats per minute) was 1.43 l/min for the monoleaflet valves and 1.56 l/min for the bileaflet valves. The monoleaflet valve pump stroke volume ranged from approximately 13.2 to 14.2 ml, whereas the bileaflet stroke volume ranged from approximately 12.0 to15.3 ml, over a flow range of approximately 0.6 to 1.5 l/min. Backflow in the bileaflet valve was greater than in the monoleaflet valve, which contributed to the more pronounced effect of rate on stroke volume.
In summary, initial in vitro evaluation of the PVAD demonstrated a maximum flow rate of approximately 1.5 l/min with clinically sized cannulae and modest drive pressures. In addition, stroke volume with the bileaflet valves exhibited a greater dependency on pump rate compared with that of the monoleaflet valves.
A high-speed video and data acquisition system (Redlake MotionPro 2000 Color with MiDAS DA132 M 100kHz A/D board; Redlake, San Diego, CA) was used to simultaneously record video images and flow waveforms. The videos are used to mark valve positions (opened or closed) relative to pressure and flow waveforms acquired simultaneously (Figure 6). These results are being combined with the cavitation and hemolysis data to study the effect of operating conditions on valve dynamics.
Hemolysis in pulsatile pumps is mainly associated with valve closure, primarily of the inlet valve.14 Previous studies have also found a correlation between hemolysis and cavitation intensity.15 Studies are underway to measure hemolysis using a low-volume mock circulatory loop and calf blood.16 A hydrophone is used to detect cavitation in the 50–500 kHz frequency range. Experimental variables include valve type and drive pressure dP/dt (systolic and diastolic). Valve closure is also affected by the delay between end-diastolic filling and systole, and the delay between end-systolic ejection and diastole.
Experimental fluid dynamic studies are being employed to evaluate the effect of pump geometry, valve type and orientation, and operating conditions, on the velocity fields and fluid stresses.17 Particle image velocimetry (PIV) is a non-invasive optical measurement technique in which small particles (<10 μm diameter) suspended in a test fluid are illuminated with a thin laser sheet. Sequential images are cross-correlated to estimate two-dimensional (and three-dimensional) velocity components. The PIV technique allows near-instantaneous whole field measurements (rather than individual points), and is therefore useful in identifying flow structures such as regions of recirculation and stasis.
Multicomponent laser-Doppler velocimetry18 is then used to measure two-dimensional mean (ensemble averaged) velocities and principal turbulent stresses at regions of interest identified by PIV with one millisecond time resolution. The viscous wall shear stress may also be estimated from the near-wall mean velocity. Previous studies with the ball valve PVAD11 found the wall shear rates and Reynolds stresses to be significantly lower than in the adult pump.
The interpretation of PIV and laser-Doppler velocimetry results depends on the blood analog solution. Blood is a non-Newtonian fluid that displays Newtonian behavior at high shear rates. For this reason, Newtonian analogs such as glycerin/water mixtures are commonly used. However, at shear rates of less than 100 s−1, blood displays a marked shear thinning viscosity and a nonnegligible elastic component, because it is a suspension of elastic particles. Use of a non-Newtonian analog will display more accurate results for in vitro flow experiments and will be essential in determining areas of the flow field prone to clotting.
Previous studies11 demonstrated differences between a Newtonian analog (20% glycerine, 79% saturated sodium iodide [aqueous], 1% water solution) and non-Newtonian analog (Newtonian analog plus 0.433 g Xanthan gum). The non-Newtonian fluid resulted in lower values for wall shear stress and Reynolds stresses, and thereby more accurately reflects the tendency for surface thrombus formation. The Xanthan gum solution is currently being modified to provide better refractive index matching to the acrylic pump model, and to evaluate the viscoelastic stability over time.
Viscoelasticity measurements of pediatric blood using an oscillatory rheometer (Vilastic 3; Vilastic Scientific, Inc., Austin, TX) found that pediatric and adult blood viscoelasticity are similar and both strongly dependent on hematocrit.19
To efficiently evaluate blood contact polymers, various in vitro test methods are being developed. The two major approaches are to study platelet adhesion and activation of the intrinsic coagulation cascade, using model test systems—platelet-rich plasma and platelet-poor plasma—as opposed to the more complex whole blood system. Additionally, because fabrication, cleaning, and sterilization methods may affect the biologic response, we use polymer fabrication techniques that replicate the procedures used to fabricate blood sacs, thereby ensuring consistent surface chemistry.
Platelet adhesion is measured using a rotating disk system.20–22 A sample disk of polymer is suspended in platelet-rich plasma solution. At constant rotational speed, the shear stress can be shown to vary linearly with radial distance, thereby providing a known range of shear stress exposure. Furthermore, the flux of platelets to the surface is uniform across the entire disk surface. Adhered platelet density is mapped and the ratio of platelets adhered to the number exposed to the surface can then be calculated. Platelet activation in the bulk solution is also measured by flow cytometry to detect activated but nonadherent platelets.
In the second approach, we have used a blood plasma coagulation protocol that assesses the time to formation of a clot after then addition of a recalcified 50% plasma solution to a test vial. We have previously used this assay to look at the response to varying levels of test procoagulant (either soluble protein activators or well-characterized model materials).23 For this assay, the test vial itself is used as the activator, similar to the Lee-White whole blood clotting assay. Initial experiments have focused on verifying experimental technique, including adequate mixing and cleaning. The polymer test vials are fabricated under similar conditions as the PVAD blood sacs.
In Vivo Testing
An in vivo study is underway to measure baseline hematologic parameters in the juvenile goat and lamb, and determine the dose-dependent response of various anticoagulants and platelet inhibitors, including heparin, warfarin, aspirin, and clopidogrel. After the completion of this study, specific anticoagulation and/or platelet inhibition objectives will be developed for the pump implant studies, which are scheduled to begin later in 2005. A final choice of animal model will be made at that time.
The pulsatile pneumatic ventricular assist device continues to play an important role in the field of mechanical circulatory support. For pediatric applications, in which the applications are primarily bridge-to-transplant and myocardial recovery, a pneumatic ventricular assist device meets the requirements of a relatively short-term device that can be applied in left, right, or biventricular applications, a robust control mode, and physiologic pulsatility. The 12-ml and 25-ml Penn State PVAD will provide support for infants and children. Future development will focus on the initiation of in vivo studies and design of a portable pneumatic driver.
This research is funded by the National Heart, Lung, and Blood Institute contract N01-HV-48191.
1.Duncan BW, Hraska V, Jonas RA, et al: Mechanical circulatory support in children with cardiac disease. J Thorac Cardiovasc Surg
117: 529–542, 1999.
2.Fuchs A, Netz H: Ventricular assist devices in pediatrics. Imag Paediatr Cardiol
9: 24–54, 2002.
3.Reinhartz O, Stiller B, Eilers R, Farrar DJ: Current clinical status of pulsatile pediatric circulatory support. ASAIO J
48: 455–459, 2002.
4.Throckmorton AL, Allaire PE, Gutgesell JH, et al: Pediatric circulatory support systems. ASAIO J
48: 216–221, 2002.
5.Karl TR, Horton SB: Options for mechanical support in pediatric patients, in Goldstein, DJ, Oz MC (eds), Cardiac Assist Devices
, Armonk, NY, Futurea Publishing, 2000, pp. 37–62.
6.Pierce WS, Rosenberg G, Donachy JH, et al: Postoperative cardiac support with a pulsatile assist pump: techniques and results. Artif Organs
11: 247–251, 1987.
7.Reinhartz O, Farrar DJ: Thoratec ventricular assist devices in pediatric patients. ASAIO J
51: xxx–xxx, 2005.
8.Von Segesser LK, Weiss BM, Hanseler E, et al: Improved biocompatibility of heparin surface-coated ventricular assist devices. Int J Artif Organs
15: 301–306, 1992.
9.Hetzer R, Stiller B, Weng Y, et al: Pediatric mechanical circulatory support (MCS): Current practices and devices in Europe and the Berlin experience [abstract], in First International Pediatric Mechanical Circulatory Support Systems and Pediatric Cardiopulmonary Perfusion Conference
, Hershey, PA, 2005, vol. 1, p. 14.
10.Gulbins H, Mair H, Kozlik-Feldman R, et al: Mechanical circulatory support in pediatric patients with the MEDOS assist device. ASAIO J
51: xxx–xxx, 2005.
11.Bachmann C, Hugo G, Rosenberg G, et al: Fluid dynamics of a pediatric ventricular assist device. [Published erratum appears in Artif Organs 24: 989, 2000.] Artif Organs
24: 362–372, 2000.
12.Daily BB, Pettitt TW, Sutera SP, Pierce WS: Pierce-Donachy pediatric VAD: Progress in development. Ann Thorac Surg
61: 437–443, 1996.
13.Rosenberg G, Phillips WM, Landis DL, Pierce WS: Design and Evaluation of the Pennsylvania State University Mock Circulatory System. ASAIO J
4: 41–49, 1981.
14.Zapanta CM, Stinebring DR, Sneckenberger DS, et al: In vivo observation of cavitation on prosthetic heart valves. ASAIO J
42: M550–555, 1996.
15.Lukic B, Weiss WJ: The effect of pump rate and filling on hemolysis in a pulsatile left ventricular assist device, in IEEE 28th Annual Northeast Bioengineering Conference
, Philadelphia, PA, IEEE Press, 2002, pp. 181–182.
16.Lukic B, Zapanta C, Griffin K, Weiss W: A study of the effect of diastolic and systolic duration on valve cavitation in a pulsatile pediatric ventricular assist device. ASAIO J
51: xxx–xxx, 2005.
17Manning K, Wivholm B, Long J, et al: Whole flowfield PIV measurements in a 10 cc pulsatile pediatric assist pump, in ASAIO 51st Annual Conference
, Washington, DC, 2005, vol. 51, p. 9A.
18.Baldwin JT, Deutsch S, Geselowitz DB, Tarbell JM: LDA measurements of mean velocity and Reynolds stress fields within an artificial heart ventricle. J Biomech Eng
116: 190–200, 1994.
19.Long J, Undar A, Manning K, Deutsch S: The viscoelasticity of pediatric blood and its implications for the testing of a pulsatile pediatric blood pump. ASAIO J
51: xxx–xxx, 2005.
20.Benton ER: On the flow due to a rotating disk. J Fluid Mech
24: 781–800, 1966.
21.Shive MS, Brodbeck WG, Colton E, Anderson JM: Shear stress and material surface effects on adherent human monocyte apoptosis. J Biomed Mater Res
60: 148–158, 2002.
22.Siedlecki CA, Lestini BJ, Marchant K, et al: Shear-dependent changes in the three-dimensional structure of human von Willebrand factor. Blood
88: 2939–2950, 1996.
23.Zhuo R, Miller R, Bussard KM, et al: Procoagulant stimulus processing by the intrinsic pathway of blood plasma coagulation. Biomaterials
26: 2965–2973, 2005.