Research and development of an implantable, axial-flow left ventricular assist device: the Jarvik 2000 Heart.

Advances in technology and increased clinical need have led to the development of a new type of blood pump. The Jarvik 2000 Heart is an electrically powered, axial-flow left ventricular assist device that has been developed during the past 13 years. Unlike first-generation left ventricular assist devices, which were developed in the 1970s and were designed to totally capture the cardiac output, the Jarvik 2000 is designed to normalize the cardiac output by augmenting the function of the chronically failed heart for extended periods. Design iterations have been tested in 67 animals, and clinical trials have recently begun. Three patients have received the Jarvik 2000 as a bridge to transplantation, and 1 patient is being supported permanently outside the hospital. All 4 patients have improved from New York Heart Association functional class IV to class I, and 2 of them have been discharged from the hospital after heart transplantation. The experimental and clinical results indicate that the Jarvik 2000 can provide physiologic support with minimal complications and is reliable, biocompatible, and easy to implant.

A genetically engineered, nonthrombogenic cellular lining for LVADs: in vitro preconditioning before in vivo implantation.

Because of the clinical success of left ventricular assist devices (LVADs) used for short-term "bridge to transplant" and the limited availability of donor organs, heart assist devices are being considered for long-term implantation as an alternative to heart transplantation. In an effort to improve biocompatibility, our laboratory has developed a nonthrombogenic cellular lining from genetically engineered smooth muscle cells (GE-SMC) for the Thermocardiosystems Heartmate LVAD. Smooth muscle cells have been transduced with the gene for endothelial nitric oxide synthase (NOS III) and produce NO at concentrations that reduce platelet deposition and smooth muscle cell proliferation when tested in vitro. In this investigation, the adhesive capabilities of GE-SMC linings were examined. An in vitro circulatory loop was designed to expose cell lined LVADs to in vivo operating conditions. Cumulative cell loss from cell lined LVADs was less than 10% after 24 hours of flow. Using a protocol for "preconditioning" the cell lining within the mock circulatory loop, the first implantation of an LVAD containing a genetically engineered SMC lining was successfully implemented in a bovine model. Results from this 24 hour study indicate that the flow-conditioned cellular lining remained intact with no evidence of thromboembolization and only minimal changes in coagulation studies.

Minimally invasive access of the normal pericardium: initial clinical experience with a novel device.

The pericardial space is being investigated as a reservoir for local drug delivery to the heart and coronary arteries. Intrapericardial drug delivery is currently limited because the pericardial space is normally small and difficult to access by standard pericardiocentesis without invasive surgery or risk of cardiac injury. Clinical trials are being conducted to evaluate a novel, minimally invasive, pericardial access device (PerDUCER, Comedicus Inc., Columbia Heights, Minn.). As of October 26, 1998, 12 clinical trials have been completed on patients undergoing cardiac surgical procedures. In all patients, a stab incision was made 1" subxiphoid and a 17G angled cannula, with preloaded guidewire, was advanced into the mediastinal space. After cannula removal, a 19F sheath/dilator was inserted over the wire. In eight patients, a median sternotomy was performed and the position of the sheath over the anterior pericardium (PC) was visually verified. Four patients underwent a closed-chest, fluoroscopy-assisted procedure. In all patients, the PerDUCER was inserted into the chest, via the sheath, and positioned over the PC. The PC was captured by suction and a bleb was formed within a side-hole on the PerDUCER tip. A sheathed needle was advanced, puncturing the isolated bleb of PC. A guidewire was advanced through the needle into the pericardial space and the PerDUCER was removed. Guidewire insertion was successful in 10 patients (7 on first attempt, 3 on second) without adverse hemodynamic effects or arrhythmia. Other than the guidewire insertion site, there was no evidence of injury to the PC or the heart. These initial clinical trials suggest that the PerDUCER may provide safe, rapid and effective percutaneous insertion of a guidewire into the normal pericardial space.

Progress in the development of a transcutaneously powered axial flow blood pump ventricular assist system.

Development of the Jarvik 2000 intraventricular assist system for long-term support is ongoing. The system integrates the Jarvik 2000 axial flow blood pump with a microprocessor based automatic motor controller to provide response to physiologic demands. Nine devices have been evaluated in vivo (six completed, three ongoing) with durations in excess of 26 weeks. Instrumented experiments include implanted transit-time ultrasonic flow probes and dual micromanometer LV/AoP catheters. Treadmill exercise and heart pacing studies are performed to evaluate control system response to increased heart rates. Pharmacologically induced cardiac dysfunction studies are performed in awake and anesthetized calves to demonstrate control response to simulated heart failure conditions. No deleterious effects or events were encountered during any physiologic studies. No hematologic, renal, hepatic, or pulmonary complications have been encountered in any study. Plasma free hemoglobin levels of 7.0 +/- 5.1 mg/dl demonstrate no device related hemolysis throughout the duration of all studies. Pathologic analysis at explant showed no evidence of thromboembolic events. All pump surfaces were free of thrombus except for a minimal ring of fibrin, (approximately 1 mm) on the inflow bearing. Future developments for permanent implantation will include implanted physiologic control systems, implanted batteries, and transcutaneous energy and data transmission systems.

Traumatic aortic transections: eight-year experience with the "clamp-sew" technique.

BACKGROUND: Because traumatic aortic transection is associated with high mortality rates, great debate exists about the appropriate operative technique for treatment of patients who have acute traumatic aortic transection. METHODS: To determine the safety and efficacy of the "clamp-sew" method, we retrospectively reviewed our 8-year experience treating 75 patients who had aortic injuries secondary to blunt trauma. Seventy-one of these patients were treated surgically. The clamp-sew method was used in all of these operations. RESULTS: Aortic cross-clamp time averaged 24 minutes (range, 14 to 36 minutes), with 4/71 having times in excess of 30 minutes. One patient (clamp time, 28 minutes) became paraplegic. Significant associated injuries were seen in 51/75 patients (48/71 patients with operation), including intrathoracic (35 patients), orthopedic (28 patients), intraabdominal (24 patients), and central nervous system (17 patients) damage. No patient died within 24 hours of operation. Overall 30-day mortality was 12% (9/75), with 7/9 having two or more aforementioned associated injuries. Of these 7, 5 had central nervous system injuries. Two of 9 died within 30 days without two or more associated injuries: 1 Jehovah's Witness of low hemoglobin, and 1 patient of sepsis. CONCLUSIONS: Although any of several maneuvers may be appropriate in managing traumatic aortic injuries, the simple "clamp-sew" technique is a safe and effective method for the treatment of traumatic aortic transections.

Development of an implantable ventricular assist system.

BACKGROUND: This study describes the present state of progress in the development of the Jarvik 2000 ventricular assist system. METHODS: Designed for implantation in the human thorax, the system consists of a small (25 cm3, 90 g) intraventricular axial-flow blood pump that transmits power and data via internal electronics and a transcutaneous energy transfer system. The pump is powered by portable internal and external polymer lithium ion batteries. The only moving part, the pump rotor, contains a permanent magnet of a brushless direct-current motor that mounts an axial-flow impeller and partial magnetic thrust support, with blood-immersed radial and thrust bearings. The motor uses a redundant coil and electric lead design, which permits continued operation in case of wire breakage. RESULTS: Seven calves have been supported for an average of 107 days (range, 40 to 162 days) with prototypes of the Jarvik 2000 ventricular assist system. No physiologic complications have occurred. When its user is at rest, the pump produces flows of 5 to 6 L/min with a decreased arterial pulse contour. Renal and hepatic functions have remained normal throughout the duration of all studies. Mean plasma free hemoglobin levels ranged from 4.3 to 11.4 mg/dL (mean, 6.3 mg/dL) for each study. Pathologic analyses of the heart and kidneys revealed no damage related to the device. CONCLUSIONS: These studies indicate that the Jarvik 2000 ventricular assist system is feasible in animals and holds promise for long-term support of patients.

Improved left ventricular function after chronic left ventricular unloading.

BACKGROUND: This study assessed the effect of prolonged left ventricular unloading on native ventricular function. METHODS: We reviewed data from 31 patients (30 men, 1 woman) supported more than 30 days (mean, 137 days; range, 31 to 505 days) with the HeartMate left ventricular assist system. The patients' mean age was 46 years (range, 22 to 64 years); 17 had idiopathic and 14 had ischemic cardiomyopathy. Data (anatomic, physiologic, hemodynamic, histologic, and biochemical) were collected at the time of HeartMate implantation, during support with the device temporarily off, and at the time of device explantation. RESULTS: Routine chest roentgenogram showed improvement in cardiothoracic ratio (0.62 +/- 0.04 to 0.55 +/- 0.03; p < 0.0001). Echocardiography performed with the pump off showed a significant decrease in left ventricular end-diastolic dimension (6.81 +/- 0.87 cm to 5.39 +/- 1.08 cm; p < 0.0005) and a significant improvement in ejection fraction (0.11 +/- 0.05 to 0.22 +/- 0.17; p < 0.02). Cardiac index increased (1.96 +/- 0.52 L.min-1.m-2 to 2.93 +/- 0.73 L.min-1.m-2; p < 0.0001), mean aortic pressure increased (71.40 +/- 10.63 mm Hg to 76.33 +/- 16.84 mm Hg; p = 0.48), pulmonary capillary wedge pressure decreased (24.18 +/- 6.27 mm Hg to 14.48 +/- 3.01 mm Hg; p < 0.0001), and pulmonary vascular resistance decreased (3.34 +/- 2.00 Wood units to 2.51 +/- 0.88 Wood units; p < 0.05). Comparisons of tissue samples taken at the time of implantation and at the time of transplantation showed a marked reduction in myocytolysis. Calcium uptake, calcium-binding rates, and lipid levels normalized in patients studied. Plasma norepinephrine levels decreased to near normal levels. CONCLUSION: Prospective studies are now indicated to determine whether device removal without transplantation may be beneficial in selected patients.

Retrospective analysis of infection in patients undergoing support with left ventricular assist systems.

Infection is a problem in patients undergoing support with left ventricular assist systems. To better understand the nature of this problem, we retrospectively analyzed data on 56 patients supported by the HeartMate (Thermo Cardiosystems, Inc, Woburn, MA) left ventricular assist system. Infection was defined as fever > 38 degrees C, white blood count > 12,000 cells/ml, and a need for antimicrobial therapy. Of the 56 patients, 25 (41%) had an infection. Device related infections (as determined by positive culture from driveline, housing, or inflow or outflow tract) occurred in eight patients (14.3%). The most common sites of infection were the respiratory system (42.4%), the central venous catheter (27.8%), and blood (18.3%). Of the positive cultures, 84% were bacterial and 16% fungal. There were no positive viral cultures. Positive cultures from left ventricular assist system related sites made up only 8.7% of the total. All but one of the patients with device related infections survived to transplantation. The long-term survival rate for patients in this group after transplantation was 77.8%. Two patients required surgical revision of the driveline because of infection. Both were free of infection postoperatively. Patients who stayed in the intensive care unit for longer periods had a greater risk of infection (uninfected, 35 days; infected, 78 days). In conclusion, although infection is a problem in patients undergoing support with left ventricular assist systems, it does not preclude survival to transplantation or alter the survival rate after transplantation.

Localized administration of sodium nitroprusside enhances its protection against platelet aggregation in stenosed and injured coronary arteries.

Sodium nitroprusside, a potent vasodilator, was evaluated for its effect on platelet aggregation in stenosed and endothelium-injured coronary arteries in a canine model. Twenty-five anesthetized dogs were studied; coronary blood flow velocity was continuously monitored. Recurrent intracoronary platelet aggregation and dislodgment (indicated by cyclic variations in coronary blood flow) were induced by mechanically injuring and stenosing the left anterior descending coronary artery. Sodium nitroprusside was administered either intrapericardially or intravenously 30 min after cyclic flow variations were established. Intrapericardial administration of saline (control) did not affect cyclic flow variations in any of 6 tested dogs. Sodium nitroprusside abolished cyclic flow variations in all 7 dogs (100%) when given intrapericardially and in 5 to 7 dogs (71%) when given intravenously (compared to intrapericardial salines, p < 0.01). A smaller dose of sodium nitroprusside was required to abolish cyclic flow variations when given intrapericardially than when given intravenously (1.6 +/- 0.5 micrograms.kg-1.min-1 vs 4.8 +/- 0.8 micrograms.kg-1.min-1, p < 0.01). The mean aortic pressure was reduced by 10 to 20 mmHg after intrapericardial sodium nitroprusside administration and by 30 to 40 mmHg after intravenous sodium nitroprusside administration. To investigate the mechanism of protection by sodium nitroprusside, NG-monomethyl-L-arginine, an inhibitor of nitric oxide synthetase, was used to induce cyclic flow variations in mildly injured and stenosed left anterior descending coronary arteries in 5 dogs. Intrapericardial sodium nitroprusside abolished the cyclic flow variations in all 5 dogs. Then oxyhemoglobin, an inhibitor of nitric oxide, was administered into the left anterior descending coronary arteries of these dogs, and it restored the sodium nitroprusside-abolished cyclic flow variations in all 5 dogs. Thus, sodium nitroprusside protects against platelet aggregation and cyclic flow variations in stenosed and endothelium-injured canine coronary arteries, probably by the action of nitric oxide, and it is more effective and hemodynamically safer when administered intrapericardially than when administered intravenously.

Progress in the development of the ABIOMED total artificial heart.

The ABIOMED implantable total artificial hearts in the final phase of engineering development. The system has a compact electrohydraulically driven energy converter sandwiched between two blood pumps, an internal electronics pack, an internal battery, a transcutaneous energy transmission coil for power transmission, and external wearable electronics pack and battery. The current effort is to complete development of the system during 1996 in preparation for formal pre clinical testing of the device. In vivo studies with the current thoracic unit (ABH II) have achieved 108 days of survival verifying the thermal, physiologic, and hematologic compatibility of the system. The abdominal implantable electronics pack showed no thermal dissipation problem. System improvements include scaling down the size of the thoracic unit, and efficiency enhancement in the power and hybrid electronics. The new system (ABH III) retains the flow capacity of greater than 10 L/min. Size reduction results in an atrial to sternal dimension that would fit 98% and 75% of men and women, respectively.

Five month survival in a calf supported with an intraventricular axial flow blood pump.

We are studying in vivo an intraventricular axial flow blood pump (Jarvik 2000) designed for long-term left ventricular support. The small (25 cc, 85 g) valveless pump has been placed intraventricularly in seven calves; pumps have functioned for as long as 5 months. In the four most recent long-term studies completed, calves have survived for 70, 120, 155, and 162 days (in that order); weight gain has averaged 0.56 kg/day. One study is ongoing at more than 30 days. Under resting physiologic conditions in the normal calf, the continuous flow pump produces flows of 5-6 L/min with a decreased arterial pulse contour. The device has caused no physiologic complications. Calves in the completed studies had mean free plasma hemoglobin levels of 11.4, 7.1, 6.5, and 4.3 mg/dl, respectively. We have modified the inflow structures of the device, and these results suggest that a thrombus free design with no pannus at or around the inlet of the pump can be achieved. Histopathologic analyses of the heart and kidneys in studies of as long as 5 months show no deleterious effects of this device. These studies demonstrate the feasibility of a small implanted intraventricular blood pump for long-term use. Future developments for permanent implantation will include implanted physiologic control systems, transcutaneous energy transmission systems, and implanted batteries.

Hospital charges for conventional therapy versus left ventricular assist system therapy in heart transplant patients.

An important goal of a left ventricular assist system (LVAS) is to provide long-term, safe, cost-effective mechanical circulatory support. The Heartmate LVAS (Thermo Cardiosystems, Inc., Woburn, MA), used in recent clinical trials as a bridge to transplantation, has proven extremely reliable for long periods, and its use has improved the survival rates in patients who subsequently received a donor heart. Patients who have been implanted with an LVAS often have significantly improved physical status and can leave the intensive care unit (ICU) and be treated at less expense elsewhere in the hospital. In this preliminary report, the authors analyzed hospital charges for three groups of patients: heart transplant patients who received conventional medical therapy before transplantation (Group 1), transplant patients who received in-hospital LVAS therapy (Group 2), and one patient who awaited transplantation at home while undergoing LVAS support. Group 1 patients (n = 6) received intraaortic balloon pump (IABP) support and inotropic therapy in an ICU before transplantation. Group 2 (n = 6) patients were receiving IABP support and inotropic therapy when the LVAS was implanted and eventually underwent heart transplantation. The length of hospital stay in Group 1 (51 days) was significantly shorter than in Group 2 (185 days). Mean hospital charge in Group 1 was $5,150/day; in Group 2, $3,178/day. The patient living at home incurred an average medical expense of $27/day. Although LVAS supported patients remain in the hospital much longer than do medically treated patients, their average daily in-hospital charges are much less. If the LVAS could be used as an alternative to transplantation, it might offer not only a longer, better quality life, but also a potential cost savings to patients with end-stage heart failure.

Permanent mechanical replacement of the left ventricle.

Patients with congestive heart failure have a poor prognosis and their care is expensive. Treatment with the HeartMate left ventricular assist device (Thermo Cardiosystems Inc, Woburn, MA) has indicated that life for such patients can be prolonged, quality of life increased, and cost of care decreased. Although currently approved only as a bridge to transplantation, the HeartMate and other such devices may one day offer permanent support of the left ventricle as an alternative to cardiac transplantation.

Progress in cardiac transplantation.

This article summarizes general practices in cardiac transplantation, including the selection of recipients and pretransplant management, selection of donors, surgical technique, management of perioperative complications, and induction and maintenance protocols for immunosuppression. Advancements in the management of rejection and infection also are included. Coronary artery disease is the most frequent cause of late death in spite of improved immunosuppressive regimens, antiplatelet agents, and lipid-controlling drugs. Still, long-term survival has improved markedly. A brief description of the Texas Heart Institute experience with 532 cardiac transplants is provided.

Use of a left ventricular assist device in an outpatient setting.

The vented electric Heartmate LVAD (VE-LVAD) (Thermo Cardiosystems, Inc., Woburn, MA) is a reliable, fully portable system that allows selected patients with end-stage cardiomyopathy to undergo outpatient treatment while waiting for heart transplantation. This implantable, pusher-plate LVAD is actuated by an electric motor located within the pump housing. The patient wears external batteries and a system controller, which power and control the LVAD motor through a percutaneous lead. Since May 1991, four men have been supported with the VE-LVAD. They ranged in age from 33 to 50 years (mean, 44.3 years); two had idiopathic cardiomyopathy, and two had ischemic cardiomyopathy. Of the four patients, three underwent support of 196, 219, and 504 days; support in the fourth patient is ongoing at more than 90 days. All four patients were fully rehabilitated to New York Heart Association Class I status. Because they were well and fully mobile, the protocol was amended to allow these patients to leave the hospital in a four phase program that begins with 16 hr day passes and leads to hospital discharge. When patients leave the hospital, they are accompanied by trained family members or friends. The patients who have participated in the program have performed routine activities, attended social events, and spent the night at home. The VE-LVAD system seems safe and appropriate for the outpatient setting in selected patients. Patients have been able to manage the system without assistance from medical or engineering personnel. This initial positive experience with outpatient LVAD treatment demonstrates the potential for providing long-term cardiac support with this type of implantable technology.

Chronic in vivo evaluation of an electrohydraulic total artificial heart.

Development of the Abiomed total artificial heart (TAH) designed for human use is progressing. Implant durations of longer than 60 days have been achieved in calves. The device consists of blood pumps, valves, and a hydraulic atrial flow balancing chamber fabricated from polyetherurethane. The energy converter, a centrifugal hydraulic pump with a rotary fluid switching valve, is positioned between the blood pumps. In two consecutive chronic in vivo studies (47 days and longer than 60 days), cardiac output was maintained in excess of 8 l/min. The atrial flow balancing chamber maintained a mean right-to-left pressure gradient of 7.5 and -1.4 mmHg in each respective study. There were no pulmonary complications. Platelet counts, fibrinogen concentrations, and hematocrit values returned to baseline levels within 20 days, whereas bilirubin, serum glutamic-oxaloacetic transaminase, blood urea nitrogen, and creatinine levels returned to normal within 1 week of implant. After the first post-operative day, plasma free hemoglobin levels of less than 10 mg/dl indicated no device-related hemolysis throughout the duration of the studies. At explant (47 day study), pathologic analysis showed no renal infarcts, no tissue necrosis, and no thermal damage. The device was fully encapsulated by 2-4 mm thick fibrous connective tissue. A newly designed textured-to-smooth surface inflow showed no signs of pannus ingrowth or thrombotic complications. These studies demonstrate that this TAH is suitable for long-term implantation.

In vivo evaluation of a trileaflet mechanical heart valve.

Design goals for a mechanical heart valve include duplicating the hemodynamic performance of the natural valve, eliminating the need for anticoagulants, and maintaining safety. The Lapeyre-Dassault (Dassault-Aviation, Paris, France) prosthetic valve, undergoing development, addresses these goals. The unique trileaflet design consists of a solid titanium ring and three leaflets. Prototypes of the valve fabricated with Delrin leaflets were implanted in the mitral position in six calves (70-90 kg). Four calves (Studies 1, 2, 3, 5) had long-term survival of 165, 158, 219, and 281 days, respectively. Two calves were killed, one on Day 37 and one on Day 39, after complications unrelated to the valve developed. In all calves heparin was given intravenously to maintain partial thromboplastin time at 1.5 to 2.0 x baseline for approximately 1 week. In Studies 1 and 2 full anticoagulation and antiplatelet therapy was given (orally administered sodium warfarin to maintain prothrombin time at 1.5 to 2.0 x baseline, along with aspirin (1 g/day) and dipyridamole 400 mg/day). In Study 3, all anticoagulation and antiplatelet therapy was discontinued at 1 month after implant. In Study 5, no anticoagulation therapy was given after the initial week of intravenous heparin; however, antiplatelet therapy was started on the fifth postoperative month and maintained until the study's end at 9 months. At 1, 2, 3, and 5 months, the mean plasma free hemoglobin level in the four long-term animals was 5.0 +/- 2.16, 6.0 +/- 3.83, 8.5 +/- 4.93, and 11.3 +/- 6.74 mg/dl, respectively. Hemolysis was not a problem. Valve performance during normal activity was excellent in all the calves, as evidenced by echocardiography and the overall appearance of good health. In the four completed long-term studies, left heart catheterization showed a mean valve pressure gradient of 11.57 +/- 1.26 mmHg and no apparent valvular regurgitation. Histopathologic examination of major organs showed no evidence of thromboembolic events. This study shows that the innovative design of this trileaflet valve performed well in initial in vivo testing, justifying further development.

In vivo evaluation of an intraventricular electric axial flow pump for left ventricular assistance.

In vivo studies have begun to evaluate a new intraventricular electric axial flow left ventricular assist device (LVAD), the Jarvik 2000, which is a small, valveless pump that is placed inside the left ventricle through the left ventricular apex. The operation, which is performed through a left thoracotomy, may be done without cardiopulmonary bypass and aortic cross-clamping. Outflow is provided through a 16 mm softly woven, Dacron graft anastomosed to the descending thoracic or abdominal aorta. Pump flow, which varies from 2 to 16 l/min in vitro, is changed by adjusting the speed of pump rotation. Preliminary studies were done to evaluate the ease of implantation, hematologic and anatomic compatibility, and pump performance. The device has been implanted in seven healthy, preconditioned calves (83-138 kg), one of which is currently undergoing support. The implantation procedure averaged 3 hours. There were no operative deaths, and blood transfusions were not required. Postoperatively, anticoagulation was achieved with heparin followed by warfarin sodium to maintain prothrombin time or partial thromboplastin time at 1.5-2.0 times baseline. In the six completed studies, support time ranged from 2 to 120 days (mean, 36 days). The seventh calf has been supported for 30 days. In the four long-term studies (20, 70, 120, > 30 days), the mean plasma free hemoglobin values during support were 11.0, 7.7, 6.6, and 3.4 mg/dl, respectively. Under normal conditions, the average daily flow rate ranged from 5 to 6 l/min. During treadmill exercise (10% grade, 1.5 km/h) lasting 20 minutes, peak flow rates exceeded 8 l/min. These pilot studies suggest that this intraventricular axial flow pump is relatively easy to implant, operate, and control. In addition, it is hemocompatible, provides physiologic flow rates, and may be able to provide long-term circulatory support.

Improved survival after extended bridge to cardiac transplantation.

In the past, left ventricular assist device (LVAD) support was frequently plagued by complications; thus, bridge to transplantation times were kept short. Increasing evidence suggests that extended bridging provides greater benefit due to improved end-organ perfusion and, thus, generally improved physical condition. To assess whether extended bridging translates into improved long-term survival after transplantation, we reviewed our experience with the HeartMate 1000 IP LVAD (Thermo Cardiosystems, Inc, Woburn, MA). Since January 1988, 19 patients (mean age, 45 +/- 9 years) have undergone extended bridging (mean time, 106 +/- 57 days). Their mean weight was 82 +/- 16 kg, and their mean body surface area was 2.0 +/- 0.2 m2. We define "extended" as the length of support necessary for systemic organ recovery after prolonged heart failure. During support, average pump flow indices ranged from 2.3 to 3.3 L.min-1.m-2, and all patients underwent physical rehabilitation. Between the time of LVAD implantation and explantation, the mean serum creatinine value decreased from 1.63 +/- 0.6 to 1.25 +/- 0.6 mg/dL (p = not significant), and the mean serum total bilirubin value decreased from 2.8 +/- 2.0 to 0.63 +/- 0.11 mg/dL (p < 0.05). All but 1 patient improved from New York Heart Association class IV to class I. Device-related complications were minimal. Twelve control patients ("de facto randomized") who did not receive the LVAD also were evaluated: actuarial survival at 1 year was 0% (p < 0.05); 3 (25%) underwent transplantation and died within 2 months; 9 (75%) died before transplantation.(ABSTRACT TRUNCATED AT 250 WORDS)