| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 1996;61:437-443
© 1996 The Society of Thoracic Surgeons
Divisions of Cardiothoracic Surgery and Mechanical Engineering, Washington University, St. Louis, Missouri, and Division of Cardiothoracic Surgery, Pennsylvania State University, Hershey, Pennsylvania
Abstract
Background. Despite the successful use of ventricular assist devices in adults over the past 15 years, relatively little has been done to develop similar devices for pediatric patients. Consequently, no such device is currently available. A review of clinical data suggests that the majority of patients in need of a pediatric ventricular assist device, either for postcardiotomy cardiogenic shock or as a bridge to cardiac transplantation, are neonates weighing 3 to 5 kg. Attempts to ``scale down'' an adult blood pump to make an appropriate device for these patients have been difficult due to the lack of sufficiently small, commercially available valves and the tendency for thrombus to develop in these small pumps.
Methods. We report on progress in the development of the Pierce-Donachy pediatric ventricular assist device, which incorporates 10-mm-diameter bileaflet valve prototypes. Particle image velocimetry is used to quantify the velocity field inside the pump.
Results and Conclusions. Particle image velocimetry velocity maps demonstrate the complexity of the flow patterns in these pumps and suggest that improved flow patterns may result from the use of valves with improved hemodynamic performance. Animal tests to determine whether improved flow patterns and better ``washing'' of the pump's blood-contacting surfaces will reduce thrombus formation are underway.
The need for a pediatric ventricular assist device (PVAD) is increasing due to two general trends in pediatric heart surgery. First, corrective operation is now feasible for many previously inoperable forms of complex congenital heart disease, but often requires operating on younger, smaller, and sicker patients than ever before. Postcardiotomy cardiogenic shock (PCCS) remains a significant cause of mortality in this group. Second, the success of pediatric heart transplantation has led to increases both in the number of transplant candidates and in the waiting time for scarce donor organs.
Recently, the use of circulatory support for pediatric patients with PCCS has increased substantially [1]. Surgeons are using an extracorporeal circuit fitted with a roller or centrifugal pump, with or without an oxygenator, to provide support [2, 3].
Although only a limited number of patients have been studied, the overall salvage rate is 25% to 50%, comparable with that in adults. Furthermore, survivors showed significant improvement in left ventricular function during and after left ventricular assist device (VAD) support. These results indicate that circulatory support in children can be at least as effective as in adults. However, extracorporeal VADs have some significant disadvantages and do not meet the needs of every patient.
Current extracorporeal systems require continuous anticoagulation. This exacerbates bleeding, which requires reoperation in approximately 50% of pediatric patients [2–4]. This has been statistically associated with increased mortality in larger adult series. More importantly, extracorporeal systems are intended only for short-term use. These patients are intubated, monitored with multiple transcutaneous catheters, and confined to bed. In this setting, the duration of support is limited by the increasing likelihood of a serious complication developing after a few days. Some potentially salvageable patients may not recover in time, ie, before a complication develops related to this means of support.
At St. Louis Children's Hospital over a 40-month period from January 1989 to May 1992, 570 children less than 6 years of age underwent a corrective operation for congenital heart disease. Severe PCCS developed in 56 patients (10%) in the immediate postoperative period. Twenty-nine patients were supported with extracorporeal membrane oxygenation; 14 of these were successfully weaned and 15 died after decannulation. An additional 27 patients died of PCCS without extracorporeal membrane oxygenation. Extracorporeal membrane oxygenation was not used in these patients because of the concern for bleeding or because in the surgeon's judgment, there was no realistic hope of myocardial recovery in less than 10 days. During the preceding 18 months, there were also 3 cardiac transplant candidates who died of complications of cardiac failure while awaiting a donor heart. We expect this number to grow as pediatric cardiac transplantation grows. Based on a review of these data, we estimate that 10 patients per year would be candidates for PVAD support at St. Louis Children's Hospital: 3 to 4 PCCS patients require a means of support with less anticoagulation than extracorporeal membrane oxygenation mandates, 3 to 4 PCCS patients require a means of support that offers the possibility of intermediate-term recovery or bridge to transplantation, and 2 to 3 cardiac transplant candidates require a bridging device.
Further analysis of these data shows that the majority of these patients were less than 3 months of age and almost all were less than 2 years (mean age, 71 ± 161 days; range, 1 to 758 days). Excluding 1 premature child who weighed 1.2 kg, the average weight of these children was 3.0 ± 1.7 kg, with a range of 1.8 to 10.1 kg. This establishes the need for a PVAD that can deliver a range of flows from 300 to 1,200 mL/min.
The Penn State PVAD Prototype
In 1986, a collaborative effort between the Division of Artificial Organs at Penn State and Sarns/3M, Inc (Ann Arbor, MI) was established to develop a PVAD. This device was a scaled-down version of the Pierce-Donachy adult VAD (Fig 1
). It consisted of a pneumatically powered sac-type blood pump, a pneumatic power unit, and a control unit. The seamless U-shaped blood sac was made of segmented polyurethane (Hemothane). The sac was encased in a rigid housing of polycarbonate that contained a diaphragm made of polyurethane. A port on one side of the diaphragm admitted pulses of air, a port on the other side permitted deairing, and a Hall-effect switch detected when the pump was full. Specially designed inlet and outlet ball valves were precisely seated in the case and attached to custom-made cannulas. The valves and cannulas were mated; both had an inner diameter of 6 mm. The pump was activated by cyclically delivering and withdrawing air to and from the space between the diaphragm and case. This generated positive and negative pressures on one side of the blood sac, causing alternate emptying and filling of the blood sac. Unidirectional flow was ensured by the valves. The blood sac held a total volume of 17 mL; under standard operating conditions the pump had a stroke volume of 11 mL.
|
Animal studies were undertaken to evaluate the in vivo performance of the PVAD. Over a period of years, 28 devices were implanted in baby goats and calves treated with various combinations of anticoagulants including heparin, warfarin, and aspirin. These experiments demonstrated the repeated propensity of thrombus to form on the surface of the blood sac (Fig 2A). This material was often seen in the flexing regions of the sac, but could be located anywhere on the sac's surface. Scanning electron microscopy demonstrated that this material was rich in fibrin and platelets (Fig 2B). Postmortem histologic analysis of the kidneys from these animals frequently demonstrated embolization, presumably from the pump. In contrast, thrombus formation and thromboembolism were infrequently seen in the adult pump.
|
The fluid dynamic factors that contribute to thrombus formation likely come into play either as they cause damage to blood elements potentiating coagulation or as they govern the development of viscous wall boundary layers and therein the convective transport of blood elements and plasma solutes to and from the surface. For example, high shear rates in the vicinity of the valves may activate platelets as they enter the blood sac. These activated platelets are then transported to the sac's surface, and if the shear forces here are too low to detach microscopic platelet aggregates, ie, to ``wash'' the sac's surface, visible thrombus may form. Subsequent embolization of this material may cause organ damage.
Comparison of the adult and pediatric VADs revealed two obvious differences that were likely to cause significant differences in the fluid dynamics in the two pumps. First, of clinical necessity, the PVAD was smaller and operated at lower average flow rates than the adult VAD. The fluid dynamic consequences of scaling down the pump were analyzed by comparing values of common dimensionless parameters, such as the Reynolds number, that help to characterize various aspects of the flow in the two systems. We focused on the effects of the inlet jet during pump filling because flow visualization studies suggested that this jet initiates the diastolic vortex, which is thought to be important in washing the surfaces of the blood sac. The conclusions of this analysis were that the flow regimens in the adult and pediatric devices were not, and could not be, dynamically similar. The Reynolds number of the flow in the PVAD will always be lower; hence, viscous effects will have greater influence in the smaller device. Furthermore, the boundary layers in the PVAD generally will be thicker, relative to the flow passage width, than in the adult VAD. The consequences of greater viscous effects and thicker boundary layers may be that platelets and soluble coagulation factors that find their way to the blood sac's surface tend to remain there longer and experience less shear force that washes the surface.
The second difference between the adult and pediatric VADs, likely to cause significant differences in fluid dynamics, was that the adult VAD used relatively large-diameter tilting-disc valves (24 mm inlet orifice, 21 mm outlet orifice), whereas the PVAD used relatively small-diameter (6 mm) inlet and outlet ball valves (Fig 3). The hydrodynamic performance of the ball valves was assessed in detailed mock circulation studies. Pressure and flow measurements over a range of physiologic conditions demonstrated excessive pressure gradients and energy losses across these valves (Fig 4). Excessive energy losses may potentiate blood element damage in the vicinity of the valves and may also reduce the momentum of the inlet jet, which is needed to drive the diastolic vortex that helps to wash the surfaces of the pump.
|
|
). The conduit was designed to minimize junctions between components where thrombus may form. In our experience, even small discontinuities in the flow stream, on the order of 100 µm, can cause thrombus formation. A bileaflet design with an 85-degree opening angle was chosen because for a valve with this small of a diameter central flow via an open orifice was expected to give optimal hydrodynamic performance. In fact, subsequent in vitro hydrodynamic studies using the PVAD demonstrated minimal pressure gradients and energy losses across the bileaflet compared with the ball valves. The time required to fill or empty the PVAD using the same operating conditions (eg, drive pressure, systole duration) was about 1.6-fold less with the bileaflet valves. These data supported the hypothesis that the fluid dynamics in the PVAD system could be significantly altered by changing the valves. Whether, and how, these changes might result in reduced thrombogenicity was the next question.
Measurements of transvalvular pressure gradients and energy losses reflect the integrated effects of all of the infinitesimal fluid elements moving through the pump. The velocities of these elements vary continuously with location in the pump and time during the pumping cycle. A record of the velocities of these elements constitutes the flowfield. The finding that the transvalvular pressure gradients and energy losses (per milliliter of blood pumped) were greater in the PVAD than in the adult VAD indicated differences in the flowfield in the two devices. To understand how differences in the flowfield might cause differences in thrombogenesis, the impact of the forces that generate the flowfield on blood elements that generate thrombus needs to be investigated. Ideally, the flowfield would be resolved on a scale that would explain the motion of individual blood elements. Thus, a characteristic spatial scale that enters this problem naturally is given by the diameter of the platelet (
3 µm) and a time scale by a fraction of the period of pump filling or emptying (200 ms).
Several experimental methods in fluid dynamics are available to study the flowfield inside VADs. Conventional flow visualization methods involve direct observation of the movement of reflective particles as they travel with the flow. These techniques are easy to use and have been useful qualitatively for identifying areas of stagnation or recirculation in VADs [5–7], but extracting quantitative information such as velocities or Reynolds stresses has been difficult. Recently, quantitative techniques capable of providing information about two of the three spatial components of the flowfield's velocity vectors have been used to study the flow in VADs [8–11]. Two of these techniques, laser Doppler anemometry and particle image velocimetry, have been used to study the flowfield inside the PVAD. These techniques are complimentary in that laser Doppler anemometry provides a high degree of temporal resolution whereas particle image velocimetry provides a high degree of spatial resolution. Results from these initial studies are emerging.
Koehler and associates [12] at the University of Utah have used laser Doppler anemometry to estimate the velocity and Reynolds shear stress profiles in an infant VAD, similar to the PVAD, fitted with 6-mm ball valves. During ventricular filling they found axial velocities downstream of the inlet valve up to 1.1 ± 0.3 m/s and high turbulent shear stresses at the inlet jet boundaries with a maximum of 65 ± 190 Nm-2, lasting about 0.1 second. They concluded that these Reynolds shear stresses are present for a sufficient duration to potentially cause blood damage.
We have used particle image velocimetry to provide two-dimensional velocity vector maps of the flowfield inside the PVAD fitted with ball valves and bileaflet valves. In general, for both types of valves, these studies demonstrated that the flow in the PVAD was highly complex and three-dimensional. During pump filling, the inlet jet initiated a diastolic vortex, which rotated around the pump's short axis in the plane of largest cross-sectional area (Fig 5). Recirculation consistently developed downstream of the inlet valve, along the curved pump wall. Regions of slow flow were identified over variable portions of the pumping cycle adjacent to the surface of the pump in areas remote from the valves and their associated jets. The onset of pump emptying is reflected by an abrupt change in the flowfield as the velocity vectors align toward the outlet in response to pressure from the pump's diaphragm (Fig 6). This convergence of fluid upstream of the outlet valve is generally more uniform than the divergence of fluid downstream of the inlet valve during pump filling, although small areas of recirculation can be identified at the boundaries of the outlet jet. The flowfield changed continuously with time, but at a given time during the pumping cycle the flowfields from different cycles were remarkably similar, except near the valves in the case of the ball valves. Here, bouncing of the ball in the valve housing caused nonperiodic flow variations.
|
|
Preliminary animal studies in weanling lambs are underway using the PVAD with bileaflet valves. Hemothane, a Biomer derivative with acceptably low thrombogenicity in the adult Pierce-Donachy device, is no longer available so a similar but nonidentical polyurethane (Biospan) has been used for blood sac fabrication. In addition, because the bileaflet valve conduits have a larger diameter than the ball valves, the cannulas were changed from a uniform 6 mm inner diameter to a custom-designed 10 mm to 7 mm inner diameter taper. Whether these changes alter the thrombogenicity PVAD system, independent of the valves, is unknown.
Two PVADs have been implanted sequentially in a single lamb using left atrial to aortic cannulation and a paracorporeal pump location. The devices were run in the full-to-empty mode at a rate of approximately 90 beats/min. The animal was initially anticoagulated with aspirin (15 mg/kg) and heparin, maintaining the activated clotting time at approximately twice its baseline value. Warfarin administration was started the night after operation, and when the prothrombin time reached twice baseline the heparin administration was discontinued. The first pump ran for approximately 2
days when we noted extravasation of blood between the sac and case. Because the pump was outside the body it was a relatively simple task to clamp the cannulas briefly and insert a new pump. This device ran 11 days before the same extravasation was identified. The animal was sacrificed and a complete autopsy performed.
Careful inspection of the blood sacs from the two pumps demonstrated a small tear in each sac. The tears occurred at the same location, in the flexing region near the periphery of the sac adjacent to the case. At the site of the tears, wear marks were identified on the external surface of the sac, suggesting friction between the case and sac. After reviewing the pump assembly procedure, we concluded that we had not injected an adequate volume of lubricant between the sac and case. Previous experience during development of the adult-sized device demonstrated the necessity of this lubricant to prevent such failures from blood sac wear. In addition to identifying the cause of blood sac failure, inspection of both blood sacs also demonstrated no microscopic thrombus on the surface of either sac. Thrombus was identified in the valve conduits of the first device along a junction between polysulfone components that developed during valve fabrication. No thrombus was identified in a second set of valve conduits that were used in the second device. The cannulas were also free of thrombus, and examination of the animal's kidneys and other organs yielded no evidence of embolization.
Future Directions
The development of a VAD for use in pediatric patients has lagged behind the development of devices for adults. This is principally due to the smaller number of children that require a VAD and the historically better results of extracorporeal membrane oxygenation in children than adults. In addition, the lack of availability of a small-diameter valve and the apparently increased thrombogenicity of small devices operated at low flow rates have slowed the development of a PVAD. Recently, as developers and users of implantable adult VADs have focused on lengthening and expanding the use of these devices, it has become clear that all such currently available devices are too large to fit easily into patients with body surface areas less than 1.5 m2. Poor anatomic fit may increase the risks of cannula obstruction, thromboembolism, or device infection. Thus, knowledge relating to the successful design of scaled-down devices has implications not only for the development of a PVAD but also for implantable adult pumps.
Thrombus formation and thromboembolism have prevented the use of the original Penn State PVAD prototype in patients. The factors that determine the thrombogenicity of VADs are understood only in the most general terms, which provide little more insight than the ideas Virchow first expressed in the 1850s. The inherent complexity of these systems (cannulas, valves, blood sacs, pneumatic drivers, whole animal models) makes it difficult to identify and investigate the mechanisms that enhance or reduce thrombogenicity. Nevertheless, an understanding of these mechanisms incorporated in a description of the fluid dynamics in these devices, with temporal and spatial resolution that correspond to the scales (beat frequency, blood element dimensions) that are relevant to thrombus formation, is an important part of the scientific base of information that is needed to make steady, long-term progress in VAD design possible. The advantage of studying thrombus formation in the PVAD is that this device is a geometrically scaled-down version of the adult Pierce-Donachy VAD, which is made of the same materials and has clinically proven low thrombogenicity. This suggests that differences in fluid dynamics may be key to the differences in thrombogenicity.
Knowledge of the flowfield inside of VADs is currently limited because all available information comes from two-dimensional techniques. Although three-dimensional techniques are under development, it may be some time before they are applied to these complex, unsteady flows. Because there is minimal cycle-to-cycle variability in the two-dimensional flowfield it may be possible to reconstruct the three-dimensional flowfield by assembling multiple two-dimensional images from two orthogonal planes obtained at corresponding times of the pumping cycle. Another limitation of current knowledge relating to VAD thrombogenicity is the lack of information concerning the initiation and progression of thrombus formation. Using an extracorporeal device and platelet or fibrinogen scintigraphy it may be possible to determine where on the surface of the blood sac these elements are first deposited and how these depositions change with time. In combination with three-dimensional flowfield data, this information on the ``kinetics'' of thrombus formation may provide new insights into the mechanisms of flow-related thrombus formation in VADs.
Footnotes
Presented at The Third International Conference on Circulatory Support Devices for Severe Cardiac Failure, Pittsburgh, PA, Oct 28-30, 1994.
Address reprint requests to Dr Daily, Division of Cardiothoracic Surgery, Jewish Hospital, 216 S Kingshighway, St. Louis, MO 63110.
References
This article has been cited by other articles:
|
|
 |
|
  J. T. Baldwin, H. S. Borovetz, B. W. Duncan, M. J. Gartner, R. K. Jarvik, W. J. Weiss, and T. R. Hoke The National Heart, Lung, and Blood Institute Pediatric Circulatory Support Program Circulation, January 3, 2006; 113(1): 147 - 155. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. W. Duncan Mechanical circulatory support for infants and children with cardiac disease Ann. Thorac. Surg., May 1, 2002; 73(5): 1670 - 1677. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A K Mahmood, J M Courtney, S Westaby, M Akdis, and H Reul Critical review of current left ventricular assist devices Perfusion, September 1, 2000; 15(5): 399 - 420. [PDF]
|
|
|
|
 |
|
 B Stiller, I Dahnert, Y G Weng, E Hennig, R Hetzer, and P E Lange Children may survive severe myocarditis with prolonged use of biventricular assist devices Heart, August 1, 1999; 82(2): 237 - 240. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
of the total pump cycle time the mean diastolic flow is approximately 3/2; of the mean pump output.)
