Share this article on:

In Vivo Experience of the Child-Size Pediatric Jarvik 2000 Heart: Update

Gibber, Marc*; Wu, Zhongjun J.*; Chang, Won-Bae*; Bianchi, Giacomo*; Hu, Jingping*; Garcia, Jose*; Jarvik, Robert†; Griffith, Bartley P.*

doi: 10.1097/MAT.0b013e3181dbe55e
Pediatric Circulatory Support

Data from early in vivo experiments demonstrated that the child-size Jarvik heart was capable of providing partial to nearly complete circulatory support with acceptable adverse effects on blood. However, bearing thrombosis was responsible for device malfunction in most cases. To overcome this problem, original pin bearings were replaced with novel conical bearings. This study evaluated chronic in vivo performance of the modified child-size Jarvik heart in the pediatric setting. Six juvenile sheep were implanted with the modified child-size Jarvik heart. Cardiac and pump output were measured daily. Serial blood samples were drawn to evaluate hematology, biocompatibility, and end-organ function. End-organ damage and device thrombosis were examined at necropsy. No device malfunction occurred during animal experiments up to 70 days. Mean cardiac output of the animals was 3.4 L/min. The child-size Jarvik heart was able to deliver a blood flow ranging from 1.4 to 2.5 L/min at speed from 10,000 rpm to 14,000 rpm. Mean plasma-free hemoglobin was 9.8 ± 5.6 mg/dl, indicating no hemolysis. Acute elevation occurred in some organ function tests after the implant surgery but returned to normal range thereafter. These indices and necropsy showed no end-organ damage. No device thrombosis was observed. The current in vivo experience shows that the modified child Jarvik 2000 heart retained its hemodynamic function and excellent biocompatibility, and the conical bearings permitted it to remain free of thrombus.

From the *Department of Surgery, University of Maryland, Baltimore, Maryland; †Jarvik Heart, Inc., New York, New York.

Submitted for consideration October 2009; accepted for publication in revised form February 2010.

Presented at the Annual Meeting of the American Society for Artificial Internal Organs, May 2009, Dallas, Texas.

Reprint Requests: Zhongjun J. Wu, PhD, Department of Surgery, University of Maryland School of Medicine, 10 S. Pine Street, MSTF Bldg., Rm. 436, Baltimore, MD 21201. Email:

From 1988 to 2009, more than 45,000 heart transplants have been performed in the United States. Although the total number has remained stable, the percentage of heart transplants performed in the pediatric population (age 0–17 years) has increased from 7% to 17%.1 Scarcity of donor organs as the same for the adult population and the difficult donor-recipient size-matching remain the major challenges for the pediatric heart transplantation. Mechanical circulatory supports are widely used to support the failing heart as bridge to recovery or bridge to transplantation.

Historically, extracorporeal membrane oxygenation (ECMO) has been commonly used for pediatric circulatory support.2–4 Several issues associated with the use of ECMO for this type of support are risk of systemic inflammation, complexity of operation, bleeding, and reduced patient mobility. Thus, extended use of ECMO and reduced patient mobility can impair long-term outcome of these patients. Because of the requirement for cannulation, newborns with a body weight <3 kg present difficulties in being placed on ECMO because of their extremely small vessels.

The successful use of ventricular assist devices (VADs) in the adult population5–7 over the last decade as a bridge to transplant or bridge to recovery or destination therapy has lead to the increased interest in this technology for pediatric patients. It has been demonstrated that normalization of cardiac output provided by these devices allows a period of improvement in end-organ function, which, combined with patient ambulation, appropriate nutrition, and cardiac rehabilitation, optimizes patient's condition before transplant. Clinicians have begun to utilize VADs for circulatory support in the pediatric population. The multicenter database of the use of VADs throughout North America in the pediatric population has been maintained by The Pediatric Heart Transplant Study Group. A 10-year review of the data was recently reported by Blume et al.8 Although mechanical assistance devices are increasingly used in the pediatric population, clinicians are still facing a paucity of device options for long-term support in this population.8,9 Furthermore, because of the size and anatomical limitations, the majority of VADs were implanted in patients older than 10 years with a body surface area larger than 1.5 m2.8,9 Currently, the HeartAssist 5™, formerly DeBakey VAD® Child (MicroMed Cardiovascular, Inc., Houston TX), is the only US Food and Drug Administration (FDA) approved device for pediatric use. The Berlin Heart EXCOR VAD (Berlin Heart AG, Berlin, Germany) is available through a special case-based petition from the FDA.

Concerted efforts and investments were also made by the governments and private sectors to develop novel and innovative pediatric circulatory devices for children and infants.9–11 In 2004, the National Heart, Lung and Blood Institute awarded five contracts for the development of novel circulatory support devices designed for infants and smaller children.9

Jarvik Heart, Inc., in partnership with the University of Maryland under the National Institutes of Health (NIH) Pediatric Circulatory Support Program, jointly developed the pediatric Jarvik 2000 hearts for children and infants. We have previously reported the initial in vivo experience with the child-size Jarvik 2000 heart.12 Despite excellent hemodynamic performance and acceptable biocompatibility of the device, the pin-shaped bearings were the site of thrombi formation and thus the major cause of mechanical failure in most cases.

To overcome this problem, a set of novel conical bearings was developed. The modified child-size Jarvik 2000 heart with the conical bearings was evaluated in six juvenile sheep for hemodynamic performance, biocompatibility, and long-term reliability. The preclinical study for the use of this child-size device in children between 10 and 25 kg is underway. In this article, the updated in vivo experience with the child-size Jarvik 2000 heart with the conical bearings is reported.

Back to Top | Article Outline

Materials and Methods

Device Description

The child-size pediatric Jarvik 2000 heart (Jarvik Heart, Inc., New York, NY) is an electrically powered intraventricular axial flow pump. The pump configuration was adapted from the adult Jarvik 2000 heart.13 The overall size of the child-size Jarvik 2000 heart was scaled down from the adult Jarvik 2000 heart with a redesigned impeller (Figure 1A). The rotor/impeller of the early version of the child-size Jarvik heart was supported by a set of pin bearings (Figure 1B). The stationary component of the front pin bearing was fixed on a three-strut inflow cage and that of the rear pin bearing on the outlet housing wall. The rotating components of the pin bearings were attached at the conical tips at the front and rear ends of the rotor/impeller, respectively. The rotor/impeller of the modified child-size Jarvik heart is supported by a set of conical bearings (Figure 1C). Three stationary supporting struts attached to the housing wall are partially in contact with the conical surface of the front or rear end of the rotor/impeller to form open conical bearings, respectively. The pump weighs 35 g, measures 5.9 cm in length and 18 mm in diameter, and has a priming volume of 4 ml. The pump can be directly inserted into the left ventricle without the need of the inflow cannula, and the outlet of the pump is attached to a 10-mm outflow graft, which can be anastomosed to either the ascending aorta or descending thoracic aorta.

The rotational speed of the child-size Jarvik heart can be operated between 10,000 and 18,000 rotations per minute (rpm) with an increment of 1,000 rpm and is capable of delivering blood flow from 0.5 to 4.0 L/min under the normal physiologic pressure afterload. Sixteen-volt battery packs were used as the power source with continuous electrical power backup during the animal study.

Back to Top | Article Outline

Surgical Implantation

All surgical procedures and postoperative care were carried out according to the approved protocol by the Institutional Animal Care and Use Committee (IACUC) of the University of Maryland School of Medicine and in compliance with the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (National Institutes of Health publication 85-23, revised 1996).

Six Dorsett hybrid sheep weighing between 27 and 37 kg and bred for laboratory use (Thomas Morris, Reisterstown, MD) were used in this study. Surgical implantation of the pediatric Jarvik 2000 heart has previously been described.11 In brief, animals were induced with thiopental sodium (10 mg/kg), intubated, and anesthetized by means of mechanical ventilation with 1.0%–1.5% isoflurane (Drager Anesthesia Monitor, North American Drager, Telford, PA) and continuously monitored during surgery with surface echocardiogram, arterial blood pressure, pulse oximeter, and esophageal temperature. Under sterile surgical conditions, each animal underwent a left anterolateral thoracotomy with excision of the left fifth rib. The outflow graft was anastomosed to the descending thoracic aorta, and a fixation cuff was sewn to the apex of the left ventricle. After administration of systemic heparin (300 IU/kg), the pump was positioned in the left ventricle. Ultrasonic flow probes (Transonic Systems, Ithaca, NY) were placed around the pulmonary artery and the outflow graft for monitoring the cardiac output and the pump flow.

Back to Top | Article Outline

Postoperative Care

The sheep were monitored daily until the study end point. The sheep received continuous heparin infusion via the indwelling jugular venous catheter. Anticoagulation was monitored by activated clotting time (ACT) with a targeted range of 160–200 seconds. Activated clotting times were initially collected daily and then biweekly after a steady state was achieved. Each pump, when not being manipulated for obtaining hemodynamic data, was kept at a constant pump speed of 11,000 or 12,000 rpm. At the end of the study period, each animal underwent device explantation with examination for thrombus formation, graft patency, and pump position. In addition, the internal organs were examined grossly and microscopically for evidence of thromboembolism.

Back to Top | Article Outline

Hemodynamic Measurements

Blood flows from the graft and main pulmonary artery were recorded in synchrony with heart rate for the duration of the study under physiological conditions and at five varying pump speeds (from 10,000 to 14,000 rpm with the increment of 1,000 rpm). The ratio of graft to total cardiac output was calculated for each pump speed.

Back to Top | Article Outline

Blood Chemistry, Hematologic, and Biocompatibility Measurements

Blood samples from an indwelling jugular venous catheter were collected at baseline, after implantation, and twice a week thereafter for determination of complete metabolic panel (CMP), complete blood count (CBC), plasma-free hemoglobin (PFH), lactic acid dehydrogenase (LDH), and plasma-soluble P-selectin. Collected blood samples were sent to an outside laboratory (Antech Diagnostics, Lake Success, NY) for CBC, CMP, and LDH determination.

Measurement of PFH was carried out with the use of a modified cyanomethemoglobin method for colorimetric determination of hemoglobin. Plasma-soluble P-selectin levels were measured by enzyme-linked immunosorbent assay (ELISA) to indicate the platelet activation. The ELISA was developed in our lab specifically for ovine. A monoclonal antibody (Psel.K.O.2.7 and AbD Serotec, Raleigh, NC) was used as the capture antibody and coated on MaxiSorp plate (Nunc, Rochester, NY) overnight at 4°C. Human recombinant P-selectin protein (R&D Systems, Inc., Minneapolis, MN) was used to generate a standard curve. The ovine plasma was diluted and incubated in the coated plate for 3 hours at room temperature. After washing, biotinylated P-sel.K.O.1.12 monoclonal antibody (AbD Serotec, Raleigh, NC) was added as the detection antibody and followed by incubating with avidin-peroxidase (Pierce, Rockford, IL). The concentration of soluble P-selectin was determined by incubating with tetramethylbenzidine (TMB) (Pierce, Rockford, IL) as the substrate, and the absorbance at 450 nm was measured using Microplate Spectrophotometer (SPECTRAmax Plus 384, Molecular Device, Sunyvale, CA). The minimum detectable concentration of soluble P-selectin was 0.1 ng/ml.

Back to Top | Article Outline

Statistical Analysis

Data are given as mean ± sd. Comparisons between baseline and measurements at subsequent times were carried out by using paired t tests. A p value < 0.05 was considered to be statistically significant.

Back to Top | Article Outline


Study Summary

The results of the device implantation and support in the six sheep are summarized in Table 1. During surgical implantation, the child-size Jarvik pumps were easily placed entirely within the confines of the thoracic cavity. One animal unexpectedly died on the third postoperative day due to sudden cardiac arrhythmia. The necropsy examination indicated that the pump was not appropriately positioned in the left ventricle. The pump was pointed toward the ventricular septum and caused suction on the septum, indicated by the bruise on the septum wall. One animal died due to bleeding of gastrointestinal tract on the 54th postoperative day. One animal was terminated on the 41st postoperative day because of infection, and the outflow graft was completely blocked due to thrombosis in the graft while the impeller was still rotating. The other three animals survived to the 60-day study end point and were electively terminated at 60, 63, and 70 days, respectively. In all the animal experiments, the pump was fully functioning either at the death or the termination time, including the one which was infected during the study. No thrombosis or clot was observed inside the explanted devices. Tissue overgrowth was noticed on the orifices of the device explanted from the infected animal. Thrombosis was also noticed in the lumen of the graft of the same animal. Pathology reports determined that this thrombus was bacterial in nature and not due to pump malfunction.

Back to Top | Article Outline

Chronic Hemodynamics

It was noticed that the implanted flow probe on the Dacron graft caused a kink, inducing deposits in the first four animals at the placement location. Thus, the flow probe was not used in the last two animals. The daily hemodynamic data sets from four chronic animals are summarized in Figure 2. The measured flow through the outflow graft at increasing speeds from 10,000 to 14,000 rpm with the increment of 1,000 rpm were 1.31 ± 0.52, 1.61 ± 0.62, 1.93 ± 0.62, 2.16 ± 0.52*, and 2.32 ± 0.38* L/min (*p ≤ 0.05 with reference to values at 10,000 rpm). The contribution of pump flow to total cardiac output was 40.9% ± 0.12%, 51.8% ± 0.17%, 61.9% ± 0.17%, 68.6% ± 0.17%*, and 73.5% ± 0.18%* (*p ≤ 0.05 with reference to values at 10,000 rpm).

Back to Top | Article Outline


Hematological data [hematocrit (Hct), PFH, and LDH] are shown in Figure 3. Only data obtained during the normal operation of the implanted devices are presented. The decrease in Hct from 28.8% ± 4.1% at baseline to 21.3% ± 4.7% (p < 0.05) postoperatively can be attributed to blood loss and hemodilution during surgery (Figure 3A). Two weeks after the implant surgery, the Hct gradually returned to baseline and increased for five long-term animals. As mentioned above, one animal (animal #3) had internal bleeding during the 39th to 51st postoperative day. The Hct of this animal was low during this time period, and blood transfusion was initiated during this time. Thus, the data for this animal during this time period was excluded for analysis.

The measured PFH values over the postoperative days for all five long-term animals are shown in Figure 3B. The average value was 9.7 ± 5.1 mg/dl over the postoperative period for five long-term animals. There was no statistical difference compared with the preoperative baseline PFH (6.8 ± 4.3 mg/dl, p < 0.05). The LDH level (in U/L) over the postoperative days is shown in Figure 3C. The preoperative baseline level was measured to be 533 ± 159. There was a spike in the measured LDH level (1109 ± 367, p < 0.05 from baseline) in the first week after the implant surgery. However, the LDH level returned to the baseline level after 2 weeks and remained in the normal range throughout the remainder of the postoperative study period.

Back to Top | Article Outline

Platelet Response and Activation

Serial platelet counts and concentration of soluble P-selectin for platelet activation measured with ELISA are shown in Figure 4. The total number of platelets had an initial drop from the baseline value on the first postoperative day, which was attributable to a dilution during the implant surgery. Subsequently, the platelets increased in the first 3 weeks and returned to the normal range thereafter. The platelet counts of the animal during the time when the transfusion took place were excluded for analysis. The ELISA measurement of the soluble P-selectin revealed a significant rise in platelet activation state immediately after the surgery. The concentration of soluble P-selectin increased from 1.89 ± 0.84 ng/ml to 9.26 ± 0.93 mg/ml (p < 0.05) immediately postimplantation. However, the concentration of soluble P-selectin decreased to the level close the baseline 1 day after the surgery and remained slightly elevated during the remaining time of the study period compared with the baseline level.

Back to Top | Article Outline


Heparin was infused via an indwelling catheter to maintain a target ACT of 160–200 seconds. Escalating doses of heparin infusion was required initially. However, the required dosage decreased and stabilized after the third week. The need to escalate the heparin dose to maintain the targeted ACT, interestingly, corresponds with the observed increase of the platelet counts. This may be a unique property of the ovine model in response to the left thoracotomy or to the pediatric Jarvik 2000 hearts.

Back to Top | Article Outline

End-Organ Function

Selected laboratory values for kidney function [creatinine (Cr)] and liver function [aspartate aminotransferase (AST), alanine transaminase (ALT)] are given in Figure 5. The creatinine value for four long-term sheep remained in the normal range throughout the study period. The creatinine value of one sheep elevated during the first week. However, the creatinine of this sheep returned to the normal range thereafter and remained in the normal range throughout the device support duration. The abnormal data points from this particular animal were excluded because the abnormal data point from this animal was not device related. This unusual elevation from this particular sheep may be attributed to the acute kidney injury induced by the implantation surgery and not the device itself. The liver transaminases show a transient elevation with peak values at the postoperative days 1–7 and returned to the normal range at day 14.

Back to Top | Article Outline

Necropsy Findings

The hearts from four animals were normal in size and shape. There was no evidence of endocardial bruising to the interventricular septum and left ventricle. A thin layer of the neointima covered the textured surface outside of the pump but did not grow into the smooth surface (Figure 6A). There was massive fibrotic tissue accumulation around the pump textured surface and tissue overgrowth into the pump inflow orifices in one infected animal. The tissue overgrowth almost completely blocked the pump inflow orifices. There was apparent endocardial injury to the interventricular septum and left ventricle in the animal that died on the third postoperative day. The pump was not completely inserted into left ventricle and directed toward the interventricular septum. Suction and compression on this structure may have triggered the fatal arrhythmia.

All the pumps explanted from the six sheep were free of thrombosis. Both the front and rear bearing were clean and free of deposited materials or clots (Figure 6B). The outflow grafts in five animals were covered with smooth neointima and no gross thrombosis was noticed (Figure 6C). It was noticed that the implanted flow probe on the Dacron graft caused a kink, inducing unfavorable patterns of neointima formation at the kinked location. The kink was not observed in the grafts in the animals in which no graft flow probe was implanted. The graft from one infected animal had a mixture of thrombus and bacteria inside the graft at the location of the flow probe placement. Cultures from the thrombus revealed Streptococcus mitis.

The kidneys from two animals (one infected and one with elevated creatinine in the first week after surgery) had gross infarcts. The kidneys from the other four animals were almost free of infarcts. Only one or two tiny infarcted spots (<1 mm) were noticed in two animals. These infarcts might have been caused during the surgical procedure because the descending aorta was partially clamped while the anastomosis was performed during the implantation surgery. Overall, the kidneys from these animals had fewer and smaller infarcts compared with those from the animals with the old child-size pumps with pin bearings. There was no visible evidence of thromboembolic phenomenon to any of the animals' hearts, lungs, liver, spleens, brains, or bowels.

Back to Top | Article Outline


Our ongoing experience with the pediatric Jarvick mechanical assistance device confirmed the implantation feasibility, its biocompatibility profile, and its reliability in long-term settings. Throughout our studies, we have consistently seen an early elevation of platelet count, platelet activation, LDH, and liver transaminases. The surgery itself with partial clamping of the descending thoracic aorta to anastomosize the device's outflow graft, the inflammatory response after surgery, and the impact of continuous flow circulation are the determinants of the observed early end-organ response. These data are consistent with our initial observations with the old version of the child-size pump. As already observed, the trend of the serum markers are positive, with an achievement of a new and stable steady state.

In the previous in vivo evaluation of the child-size pump, we used oral aspirin and clopidogrel to antiaggregate the animals. However, in our recent studies, all animals were anticoagulated with continuous heparin infusion with a targeted ACT of 160–200 seconds. Heparin followed the trend of needing escalating doses to maintain the required ACT for the first 2 weeks and was then decreased dramatically until a steady state was achieved approximately 3 weeks postoperatively.

One concern we have had was the line infection. Although we take the necessary precautions such as subcutaneous tunneling of the intravenous lines, using antibacterial-coated lines, and cleaning the sampling ports before and after use, when those ports are maintained for extensive periods of time, infection rates are increased. Since we started using the heparin, we check ACT levels daily. However, after 2 weeks, the heparin doses exhibit a plateau and multiple blood draws became unnecessary; so we decreased our bloods draws from daily to twice weekly in our later experiments. This improved our rates of line infection in all our animal studies.

Although the intended use of this pediatric circulatory assist device is to provide ventricular assistance to a failing heart, the aim of this study was to evaluate the hemodynamic performance, biocompatibility, and long-term durability of the device. The focus, in particular, was to assess the impact of the new structural modification of the pump in its performance and biocompatibility. Once the device is proved to be biocompatible and reliable during this phase, implantation and performance evaluation of the device in a model of cardiomyopathy should be a logical step toward human clinical trial. However, such a study is beyond the scope of this study. Thus, healthy sheep were used in this study. We understand that the healthy left ventricle could be able to eject through the pump and the situation is different from that of a failing ventricle. Although it was not the major aim of this study, we performed the intraoperative echocardiographic imaging of the left ventricular unloading of this device. As the rotational speed was increased, the intraoperative echocardiography showed that the left ventricle could be completely unloaded at the higher operating speed. We also assessed the complete unloading by visualizing no aortic valve opening after the speed of the device was increased. This parameter is considered the most reliable for complete ventricle unloading in the literature. Because the aim of this article was more about the hemodynamics, biocompatibility, and long-term reliability, the echocardiographic assessment was not presented.

Although the number of animals fulfilling the whole protocol is small, it is sufficient to outline that the new pump design is safe and biocompatible, as our main goal. The problems we encountered also have clinical relevance. The improper position of the pump and its mechanical suction of the septum triggered fatal arrhythmia, a situation that can also occur in clinical practice. The infection and bleeding event that we experienced are also the major causes of ventricular assistance failure as seen in the human clinical series. So, our model of even healthy sheep replicated the most common situations of the clinical practice and thus providing the reliability of our study design.

Overall, our recent experiments with the child-size pediatric Jarvik heart are encouraging, showing anatomic fit, partial to complete circulatory support, and good biocompatibility. The utilization of the conic bearing has mitigated the thromboembolic phenomenon and bearing seizure. Because these changes have been made, we have not experienced device clotting events or pump malfunction. Our next goal is to complete the preclinical study and to initiate the clinical trial for the use of this child-size device in children.

Back to Top | Article Outline


Supported by NHLBI Pediatric Circulatory Support Contract: HHSN28600448190C and a grant from the Joseph and Corrine Schwartz Foundation (to M.G.).

Back to Top | Article Outline


1. The United Network for Organs Sharing. Available at: Accessed February 22, 2006.
2. Del Nido PJ, Armitage JM, Ficker FJ, et al: Extracorporeal membrane oxygenation support as a bridge to pediatric heart transplantation. Circulation 90: II66–II69, 1994.
3. Gajarski RJ, Mosca RS, Ohye RG, et al: Use of extracorporeal life support as a bridge to pediatric cardiac transplantation. J Heart Lung Transplant 22: 28–34, 2003.
4. Fiser WP, Yetman AT, Gunselman RJ, et al: Pediatric arteriovenous extracorporeal membrane oxygenation (ECMO) as a bridge to cardiac transplantation. J Heart Lung Transplant 22: 770–777, 2003.
5. Rose EA, Gelijns AC, Moskowitz AJ, et al; Randomized Evaluation of Mechanical Assistance for the Treatment of Congestive Heart Failure (REMATCH) Study Group: Long-term mechanical left ventricular assistance for end-stage heart failure. N Engl J Med 345: 1435–1443, 2001.
6. Long JW, Kfoury AG, Slaughter MS, et al: Long-term destination therapy with the HeartMate XVE left ventricular assist device: improved outcomes since the REMATCH study. Congest Heart Fail 11: 133–138, 2005.
7. Deng MC, Edwards LB, Hertz MI, et al: Mechanical Circulatory Support Device Database of the International Society for Heart and Lung Transplantation: second annual report—2004. J Heart Lung Transplant 23: 1027–1034, 2004.
8. Blume ED, Naftel DC, Bastardi HJ, et al; for the PHTS Investigators: Outcomes of children bridged to heart transplant with ventricular assist devices: a multi-institutional study. Circulation 113: 2313–2319, 2006.
9. Baldwin TJ, Borovetz HS, Duncan BW, et al: The National Heart, Lung and Blood Institute Pediatric Circulatory Support Program. Circulation 113: 147–155, 2006.
10. Dasse KA, Gellman B, Kameneva MV, et al: Assessment of hydraulic performance and biocompatibility of a MagLev centrifugal pump system designed for pediatric cardiac or cardiopulmonary support. ASAIO J 53: 771–777, 2007.
11. Fynn-Thompson F, Almond C: Pediatric ventricular assist devices. Pediatr Cardiol 28: 149–155, 2007.
12. Kilic A, Nolan T, Li T, et al: Early in vivo experience with the Pediatric Jarvik 2000 heart. ASAIO J 53: 374–378, 2007.
13. Frazier OH, Myers TJ, Gregoric ID, et al: Initial clinical experience with the Jarvik 2000 implantable axial-flow left ventricular assist system. Circulation 105: 2855–2860, 2002.
Copyright © 2010 by the American Society for Artificial Internal Organs