Arfificiul Orguns 10(3):236-244, Raven Press, New York 0 1986 International Society for Artificial Organs Thromboembolic Complications of the Jarvik-7 Total Artificial Heart: Case Report Mark M. Levinson, Richard G. Smith, Randall C. Cork, Joseph Gallo, Robert W. Emery, Timothy B. Icenogle, Richard A. Ott, *Gregory L. Burns, and Jack G. Copeland Departments of Cardiovascular Surgery, Biomedical Engineering, and Anesthesiology, University Medicul Center, Tucson, Arizona, and *Department of Surgery, Division of Arlificial Organs, The University of Utah, Salt Lake City, Utah, U.S.A. Abstract: The Jarvik-7 total artificial heart (TAH) has now been implanted in seven patients, with four (57%) suffering cerebral embolic events. Functional recovery in two patients was impaired by subsequent intracranial hemorrhage from anticoagulant therapy. The findings of a Jarvik-7 TAH explanted following a cerebral embolus are reported here. Thrombotic deposition was seen in the inner crevice formed between the two components of the aortic quick connector. In addition, deposition was seen around all four valves in the crevice formed between the valve ring and the plastic valve housing. On the atrioventricular valves, the deposition in this crevice was most severe on the outflow surface adjacent to the minor flow orifice. On the pulmonic valve, deposition was seen around the entire circumference of both the ~nflowand the outflow crevices. No thrombi were seen o n the diaphragm, the diaphragm-housing junction, or the housing of the left ventricle. Hypercoagulability of the patient's blood gradually developed during implant, reaching a maximum on the evening before the clinical embolus. Increasing resistance to heparin was also demonstrated by in vitro testing. Key Words: Artificial heart-Jarvik-7 heart-Thromboembolic complications- Hypercoagulation. After prolonged animal testing and documentation of survival, the Jarvik-7 total artificial heart (TAH) has reached the stage of initial human clinical trials. Two separate trial pathways are currently under way, one for permanent implantation, and one for short-term implantation. The most important postoperative problem identified in human trials of the Jarvik-7 TAH is cerebral embolism. Although renal failure, bleeding, infections, respiratory failure, and other serious sequelae have been described following implantation, the neurologic problems have far outweighed all others in importance. Each implanted patient has been treated with a vigorous regimen of anticoagulation, but presently no drug or drug combination has proved effective in eliminating this complication. Indeed, such powerful anticoagulation has been used that severe bleeding complications have been described by all centers participating in these trials. Five of the first seven patients have required reexploration of the mediastinum for perioperative hemorrhage. T h e combined morbidity of the neurologic and bleeding complications has greatly diminished the potential benefit of this promising new therapy for end-stage heart disease. As of this writing, a total of seven patiients have been implanted, with four (57%) suffering embolic neurologic events. Two of the four patients recovered completely, including the one described here. The remaining two patients were left with permanent neurologic deficits; in both, recovery has been impaired by the subsequent development of cerebral hemorrhage from anticoagulant therapy. The presence of surface defects on the bloodcontacting layers of mechanical devices has long been known to favor thrombosis following implantation (1). With improved quality control and elimination of microcracks, microbubbles, and dust particles, thrombosis on the flexing diaphragm or Received February 1986. Address correspondence and reprint requests to M. M. Levinson at University Medical Center, 1501 N. Campbell Avenue, Tucson, A 2 85724, U.S.A. 236 23 7 THROMBOEMBOLIC COMPLICATIONS OF JARVIK-7 HEART semiridged housing has been greatly reduced in the current Jarvik-7 TAH design. When thrombosis or embolization occurs in implanted calves, septicemia is often present (2). Primary nonseptic thrombosis occurs in 25-30% of implanted calves, usually around the atrial cuffs or the mechanical valves (3-5). T h e locations of thrombi associated with t h e Jarvik-7 TAH explanted from a “bridge-to-transplant” patient are reported here. This case constitutes the first successful use of a TAH as an interim step prior to cardiac transplantation. In addition, this is the first report of the areas of thrombotic deposition in the Jarvik-7 TAH following human implantation, embolization, and explantation. It is believed that this study identifies areas of increased thrombogenicity where attention t o redesign and modification can be centered in future models of this device. This patient also illustrated the dynamic and changing nature of the blood-material interface during the period of implantation. Just prior to the neurologic deficit, a relative hypercoagulable state existed, and in vitro coagulation studies returned abruptly to normal. Elevated serum coagulation factors, rising platelet counts, and increasing resistance to continuous heparin infusion were an integral part of these changes. The combination of all these factors predisposed the patient to an acute thrombotic episode on the 7th postimplant day. Improvements in the device and the methods of anticoagulation show considerable promise in eliminating embolic complications of this TAH device. CASE REPORT M.D. was a 25-year-old man with end-stage idiopathic cardiomyopathy. Following marked clinical deterioration, the patient developed refractory cardiogenic shock despite high-dose pressor and inotropic therapy. No suitable donor could be located; and both the patient and the family accepted the alternative of TAH implantation as an interim step until human cardiac transplantation could be performed. On August 29, 1985, the Jarvik-7 TAH was implanted. Hemodynamic stabilization after bypass occurred without incident. Initial cardiac outputs were 6.0 Llmin. Protamine (550 mg) was given to reverse intraoperative heparin. No further anticoagulation therapy was given until verification of the absence of mediastinal bleeding. Fifteen hours following closure of the chest, heparin was begun at 100 Uih by continuous infusion. Partial thrombo- HEPARIN DOSE - UNITS PER HOUR 2500, IMPLAt4l TRANSPUNT STROKE ‘ 7 ‘ 1500 0 S/~S/SS 1 2 3 4 5 DAYS A F E R 6 7 8 9 10 IMPLANT FIG. 1. Heparin dose (U/h) as a reflection of time during the implant period. plastin times (PTT) were monitored every 4 h but failed to respond to increasing doses of heparin. Figure 1 displays the rising dose of heparin during the first 2 days. After reaching a steady dose of 1,400 U/h with no alteration in the PTT (Fig. 21, thrombin times (TT) were monitored, since this test is more sensitive t o the agent. Thrombin times were maintained between two and three times control values from the 3rd to the late 6th postimplant day (Fig. 3). Continuous heparin, between 1,400 and 1,800 U/h, was needed to maintain this level of effect. No antiplatelet agents were given prior to the neurologic event. No bleeding complications from mediastinal, gastrointestinal, other mucosal, o r needle puncture sites were identified. Following prolonged serous drainage, the mediastinal tubes were removed on the 5th postoperative day. Severe pulmonary edema was treated with aggressive diuretic management. Contraction of the interstitial fluid compartment induced gradual resolution of the postoperative pulmonary edema. A total of 17 L of negative fluid balance was incurred in the first 6 days following operation. Successful extubation on the afternoon of the 4th postoperative day demonstrated the effectiveness of this therapy. In addition, there was dramatic clearing of t h e chest radiograph a n d f u r t h e r pulmonary problems were absent following extubation. The remainder of the postoperative course was uneventful until the morning of the 7th day. At this time, the patient abruptly developed a dense aphasia affecting the expressive and receptive areas of the cerebral cortex (Table 1). In addition, the subcortical white matter was involved, as was evidenced by a conductive aphasia clearly apparent M . M . LEVINSON ET AL. 238 PARTIAL THROMBOPLASTIN TIMES TABLE 1. Clinicalfindings T I M E S CONTROL I 0 . ~ ~ . ~ ~ 1 0 . . . 2 * . . . ~ 4 3 . . .. ~ I . ._ .. . I .I . .. . I I . . * . . .I . 5 7 6 8 9 10 DAYS A F E R I M P L A N T 8/29/85 FIG. 2. Partial thromboplastin times (PTT) graphed over the period of implantation. Patient values expressed as a ratio against laboratory control values. Dotted line represents equality to control. Sharp peaks in PTT represent errors in obtaining blood samples. Aphasia Expressive Receptive Conductive Right hemiparesis Arm drift Decreased hand grip Diminished dorsiflexion of foot Bilateral Babinski's Mild parietal lobe signs to 2,000 U/h and maintained a t this level until transplantation was performed (Fig. 1). Dextran 40 (25 mlfh) was also added. Neurologic resolution began -4 h following onset. At this time, the receptive aphasia cleared, followed shortly thereafter by a considerable improvement in the expressive aphasia. For several days, the conductive aphasia was apparent, but it then gradually resolved The right hemiparesis cleared >90% by the first 4 h following onset. Subsequent to transplantation and recovery from posttransplant complications, there are no focal neurologic deficits, and speech is normal. Nine days following implantation, the patient was returned to the operating room, where cardiac transplantation and explantation of the device were performed. Figure 4 shows a photograph of the left ventricular prosthesis. Immediately apparent is a thrombus that lines the circumference of the inner crevice created where the two components of the aortic quick connector snap together. Part of this crevice, however, is devoid of macrothrombus. In the closeup view, Fig. 5, a nonadherent freely mobile remnant of the clot is visible. It is suspected that a clot previously occupied the remainder of this crevice but broke free and embolized to the cerebral circulation. Patchy thrombotic deposition can also be seen on the titanium valve ring, particularly in che minor flow orifice position. Careful inspection shows a small amount of deposition on the under surface of the aortic valve along the crevice where the titanium and the hard plastic valve housingjoin. In addition, there are tiny thrombi at the junction between the Dacron aortic graft and the polyurethane ring that constitutes the female portion of the aortic quick connector. Figure 6 shows the histologic appearance of the thrombus along the aortic valve ring. It is a very fresh, unorganized, unmodified acute thrombus with intact red blood cells, white blood cells, and immature fibrin. Thrombi associated with the other ~ later that day. A patchy right hemiparesis involving the forearm flexors and the distal leg muscles with partial involvement of the proximal right upper extremity muscle groups also occurred. Bilateral Babinski's and mild parietal lobe signs were also detectable. Cerebral computer axial tomographic (CAT) scanning 2 h, 5 days, and 8 days post-event failed to show evidence of cerebral necrosis. In view of the absence of intracranial hemorrhage on CAT, increasing levels of heparin were administered. Following a bolus of 7,500 U, the infusion was raised THROMBIN TIMES T I M E S CONTROL ........................................y........ 1- 0 I >7 IYFIAHT .,.. 0 8/29/85 I I 1 . I . I 2 . . . I . 3 . . I . . . 4 l . 5 DAYS A F T E R . . ' . . 6 . I . . 7 . L A 8 L 9 L 10 IMPLANT FIG. 3. Thrombin times (TT), with patient values expressed as a ratio against laboratory control values. Dotted line represents equality to control. Note return of TT to control values at t h e time of this patient's stroke. Artif Orgcms. Vol. 10, N o . 3 , I986 THROMBOEMBOLIC COMPLICATIONS OF JARVIK-7 HEART FIG. 4. Explanted left ventricle. Note thrombus located in the inner crevice where the two components of the aortic quick connector snap together (arrow). valve rings were of similar composition. The mitral valve as seen from the atrial side is pictured i n Fig. 7. The inflow surface is free of thrombi, but the outflowiventricular s u r f a c e s h o w s a rim of 239 FIG. 5. Closeup of Medtronic Hall aortic valve. Note loosened remnant of thrombus displaced out of the aortic quick connector crevice (arrow). thrombus adherent to the crevice formed between the valve ring and the valve housing. This thrombus is present only in the minor flow orifice and extends from the rod-guide of the Medtronic Hall FIG. 6. Photomicrograph, with hematoxylin and eosin stain, of thrombus from around aortic valve No evidence of organization I S present, i m p l y i n g recent thrombus formation x 250 240 M . M . LEVINSON ET A L . FIG. 7. Mitral valve. Note thrombus adherent to the outflow valve ring crevice adjacent to the minor flow orifice of this tilting disc valve. FIG. 8. Ventricular surface of mitral valve (after formalin preservation). Arrow highlights the fine rim of thrombus, which starts at the central strut and extends along the valve ring crevice. valve to the retaining pin. After formalin preservation, the ventricles were opened and this thronbus could then b e seen from the ventricular side (Fig. 8). T h e body of the left ventricle was free of thrombus. The polyurethane monolayer that covers the housing, the diaphragm-housing junction, and the blood-contacting surfaces of the diaphragm did not show any signs of thrombi followirig this 9-day implantation (Fig. 9). The right venf ricle, when opened, did display some tiny white patches on the diaphragm-housing junction, and a small thrombotic stain at the surface defect created where the de-airing needle punctured the polyurethane monolayer (Fig. 10). The right ventricular valves demorstrated many FIG. 9. The body and pumping diaphragm of the left ventricle were free of visible thrombi. A r r i f O r g m s , Val. 10. hi#.3. 1986 FIG. 10. Right ventricle, showing thrombotic stain on de-airing port and flecks of white material (arrow) along diaphragm ,?ousing junction. THROMBOEMBOLIC COMPLICATIONS OF JARVIK-7 HEART of the same findings as did the left. Figure 1 I shows the tricuspid valve from the atrial view. Again, there is thrombus adherent to the crevice formed between the valve ring and the valve housing. The thrombus is adherent to the ventricular surface of this crevice and extends from the rod-guide to the first retaining pin in a fashion similar to that seen on the mitral valve. In addition, there are a few small thrombi adherent to the crevice on the atrial side of this valve. The pulmonic valve is pictured in Fig. 12. The crevice created between the titanium valve ring and the hard plastic housing can easily be seen. Small thrombi are scattered throughout the entire circumference of both the inflow and outflow crevices of this valve. The acute thrombotic event occurred as a result of a complex of multiple factors, many of them unrelated to the device. It was discovered that progressive hypercoagulability developed during the implant, reaching a maximum on the 6th day. Not only was resistance to heparin manifest at that time, but analysis of coagulation factor levels revealed simultaneous supranormal concentrations of all important coagulation mediators on the same day. The platelet count underwent a two-fold increase the day preceding the stroke. Fibrinogen peaked at almost twice normal. Factor X, albumin, and antithrombin 111 all underwent similar, though less dramatic, increases. Coupled with these trends was a progressive fall in cardiac output during the previous several days, reaching a minimum of 4.0 Limin on the sixth postoperative day. Lower pump flows were seen at this time because of severe volume contraction following aggressive diuresis. Although this therapy was effective in removing pulmonary edema fluid, the resulting hemoconcentration and hypovolemia were deleterious to pump output and the effectiveness of anticoagulation therapy. It is believed that the unfortunate combination of increasing serum cofactor concentrations with lowered cardiac outputs predisposed this patient to acute thrombosis at the sites described above. DISCUSSION Cerebral thromboembolic complications have proved to be the most serious drawback uncovered in the initial human trials of the Jarvik-7 TAH. Despite aggressive anticoagulation therapy of different types, no definitive solution has been found. Of the first seven implanted patients, four have had embolic strokes, and two of these patients have also deve I op e d i n t rac rani al hemorrhage secondary t o 241 anticoagulant therapy. Although two of the initial four patients have resolved their strokes, the risk of permanent neurologic deficit is a major clinical concern when TAH replacement is recommended to a patient with dwindling cardiac function. A solution to this problem must be obtained before widespread clinical use of this device can be recommended. The first step in this solution certainly must be an analysis of the source of emboli from affected patients. Of these seven patients receiving implants, the patient described here is the first individual reported from whom the device was explanted shortly following a cerebral embolic event. Localization of the sites of thrombosis was possible. It was also possible to intelligently speculate that the aortic quick connector was the site of origin of the cerebral embolism in this particular case. Although similar sites of thrombosis are seen in animal implants ( 3 - 3 , bacteremia or septicemia is postulated to be a major causative factor in this investigative model ( 2 ) . The exact sites of thrombosis of the Jarvik-7 TAH when i t is in contact with aseptic human blood have not previously been described. It was observed in this one case that the critical areas with a high propensity for thrombus formation are the following (Fig. 13): (a) the inner crevice of the aortic quick connector, (b) the junction of the quick connector and the aortic Dacron graft, and (c) the inflow and outflow crevices of each valve where it is seated into the hard plastic valve housing. In addition, there is a possibility that thrombus can deposit at a surface defect created where the de-airing needle punctures through the polyurethane membrane of the housing. In order to reduce the risk of thromboembolic disease with the Jarvik-7 design, several possibilities must be considered. First, a reduction in surface defects where the inflow and outflow valves are mounted is certainly important. A new technique for mounting the valves must be designed to minimize the formation of anatomic crevices. The quick connector system should also be changed to provide a crevice-free, smooth-surfaced transition that is less thrombogenic. It appears that t h e housing and the diaphragm of the current Jarvik-7 design pose a minimal risk for thrombosis in shortterm human implants. Second, anticoagulation systems that are more effective in preventing thrombosis without significant hemorrhage are desperately needed. Warfarin, aspirin, dipyridamole, and heparin have all been used in various combinations without complete success in preventing embolization. In addition, se- 242 M . M . LEVINSON ET AL. FIG. 11. Tricuspid valve, showing sites of thrombotic deposition similar to those seen in the mitral valve. Deposits also occur o n the i n f l o w surface of the valve r i n g crevice (arrows). FIG. 13. Conceptual cutaway diagram of the aortic outflow and quick connector system, indicating sites of thrombosis in this patient and the corresponding surface defects. Components of the Jarvik-7 outlet port: 1: Dacron aortic graft; 2: Skirt of aortic quick connect; 35: Valve mountirg; apparatus; 4: Titanium ring of Medtronic Hall outflow valve: 6: Polyurethane housing of blood pump. Sites of thrombosis: A: Dacron-quick connect junction: B: Aortic quick connect crevice (possible source of cerebral embolus); C: Outflow valve ring crevice: D: Inflow valve ring crevice. vere bleeding problems in TAH recipients have been reported in most patients treated with these agents (R. J. Jarvik, et a]., unpublished observations). It is difficult to determine exactly which in vitro assay best follows the degree of the in vivo anticoagulation effects obtained with these medications. Thus, it is also difficult to decide which dose range is the ideal target for their antithrombotic effects. In addition, multicomponent therapy has increased the complication rate without protecting against thromboembolism. Research into new anticoagulants such as protease inhibitors, low-molec- ular-weight heparin, fibrinolytic agents (tissue-type plasminogen activator), and new antiplatelet drugs such as prostaglandin analogs and calcium channel blockers provides a new frontier of research and potential therapy for this difficult problem. A third component of the solution to this important complication is the recognition of airerations in the patient that may induce coagulation (Table 2). Following implantation, the liver may respond dramatically to the relief of heart failure and shock. Subsequent outpouring of coagulation factors can occur, which induces a hypercoagulable state. In FIG. 12. Pulmonic valve, showing thrombotic deposits adherent to both valve ring crevices around the entire circumference (arrows). Artif Organs, Vol. 10. N o . 3 , 1986 THROMBOEMBOLIC COMPLICATIONS OF JARVIK-7 HEART this respect, the sharp rise in “acute phase reactants” normally seen after any surgical stress assumes a great deal more importance in recipients of TAH devices. The acute increase in platelet count just prior to embolism in this patient was also seen in at least two of the permanent Jarvik-7 TAH recipients (W. C. DeVries, personal communication) and is a potentially important observation. This phenomenon may be the earliest sign of impending thrombotic events. Further experience will determine if changes in these “acute phase reactants” following TAH implantation actually provide an early clue to thrombotic complications. If so, anticoagulant therapy could be modified at these times if such trends are recognized prior to neurologic sequelae. iatrogenic factors worsening these effects include dehydration and transfusion of plasma cofactors to treat bleeding complications. Volume contraction also increases serum viscosity and osmolarity, which may predispose acute thrombosis. Hemoconcentration may account for a great deal of the increased levels of serum cofactors seen in this case. In addition, dehydration lowers venous filling pressure, pump filling volumes, and mean cardiac output. This, in turn, causes decreased velocity of blood flow across the critical thrombogenic areas of this device. After the 1st or 2nd postoperative day, cardiac outputs should probably be maintained at a 3 L/min/m2 to counter this effect. Unfortunately, the patient described here had a hypercoagulable state at the same time low cardiac outputs were obtained. This unfortunate combination may well have overcome the effects of the anticoagulant system used in this patient (Table 2). 243 TABLE 2. Cause of stroke: identifiable factors Thrombogenicity of the device Aortic quick connector crevice Valve ring crevices Minor flow orifice characteristics Step-off at aortic quick connector Dehydration effects Decreased pump flows Decreased surface washings Increased eddy currents in minor flow orifices Hemoconcentration Increased levels of coagulation cofactors H yperosmolarity-hyperviscosity Recovery of hepatic function Increased synthetic function, contributing to elevated cofactor levels Thrombocytosis Acute two-fold increase in platelet concentration noted during the 48 h preceding the neurologic event It is believed that heparin has not been given an adequate trial as a single agent for the prevention of thromboembolism in the Jarvik-7 TAH. Heparin retains many ideal properties for an anticoagulant in this setting (Table 3). It may be that with more experience this agent could be successfully used for short-term implants. Prospective monitoring of important laboratory values, including the serum fibrinogen, Factor x,albumin, and antithrombin IiI levels; platelet count; and serum osmolarity, may detect the simultaneous concentration of these factors before resistance to anticoagulation therapy develops. The role of adjunctive antiplatelet agents such as Dextran 40 may also be important but is still unproven. TABLE 3. Selection of anticoagulants f o r short-term total artificial heart recipients: advantages vs. disadvantages Mechanism Duration Route Acutely “reversible” Ease of titration Monitoring assay PTT WBaPTT TT PT BTG TX b,, aggregation BTG TX b2, aggregation ? None Risk of bleeding at TXN ~ Heparin Antithrombotic 30 rnin i.v. Yes Yes Warfarin Aspirin Antithrombotic Antiplatelet 3-5 days 7- 10 days p.0. p.0. No No None None Dipyridamole Antiplatelet 6-8 h p.0. No None Sulfinpyrazone Dextran 40 Antiplatelet Anti platelet 18h 6-12 h p.0. i.v. No No None Minimal Minimal Moderate High Moderate ? Minimal TXN, heart transplantation; PTT, partial thromboplastin time; WBaPTT, whole blood activated PTT; TXb,, thromboxane b2; TT, thrombin time; BTG, beta thromboglobulin. M . M . LEVINSON ET A L . 244 REFERENCES 1. Olsen DB, Unger F, Oster H, Lawson J, Kessler T, Kolff J , Kolff WJ. Thrombus generation within the artificial heart. J Thorac Cardiovasc Surg 1975;70:248-55. 2. Hastings WL, Aaron JL, Deneris J, Kessler TR, Pons AB, Razzeca KJ, Olsen DB, Kolff WJ. A retrospective study of nine calves surviving five months on the pneumatic total artificial heart. Trans A m Soc Art$ Intern Organs 1981;27: 71-6. 3. Vasku MJ. Artificial heart: pathophysiology of the total artificial heart and of cardiac assist devices. Acta Facultatis Art$ Orgiiris. Vol. 10, No. 3 , 1986 Medicae Univ Brunensis, JE Purkyne University Brno, Medical Faculty, Bmo, 1982:180-25. 4. Kessler TR, Pons AB, Jarvik RK, Lawson JH, Razzeca KJ, Kolff W E Elimination of predilection sites for thrombus formation in the total artificial heart-bef~3re and after. Trans Am SOCArtif Intern Organs 1978;24:5?r2-5. 5 . Olsen DB, Kessler TR, Pons AB, Razzeca I(,Lawson JH, Kolff WJ. Fabrication, implantation, and pathophysiology of the total artificial heart in calves for 6 months. In: Pierce WS, ed. US-USSR Joint Symposium on Circidatory Assistance and the Artificial Heart, Thilisi, USSR. Washington, D.C., NIH publication No. 80-2032, 1981:3155.