- Email: [email protected]
Use of the Pierce-Donachy ventricular assist device in patients with cardiogenic shock after cardiac operations
Use of the Pierce-Donachv Ventricular Assist Device in Patients With Cardiogenic Shock After Cardiac Operations J
D. Glenn Pennington, MD, Lawrence R. McBride, MD, Marc T. Swartz, BA, Kirk R. Kanter, MD, George C. Kaiser, MD, Hendrick B. garner, MD, Leslie W. Miller, MD, Keith S. Naunheim, MD, Andrew C. Fiore, MD, and Vallee L. Willman, MD Department of Surgery, St. Louis University Hospital, St. Louis, Missouri
In spite of recent improvements in cardiac surgery, a small percentage of patients have severe postcardiotomy ventricular failure refractory to drugs and the intraaortic balloon. In our experience, the Pierce-Donachy external pneumatic ventricular assist device has proved to be one of the most effective devices for these patients. Since 1981,30 patients aged 15 to 71 years (mean age, 52 years) with profound cardiogenic shock refractory to conventional therapy after cardiotomy were supported with the Pierce-Donachy ventricular assist device. Fourteen required left ventricular support, 7 needed right ventricular support with an intraaortic balloon, and 9 had biven-
tricular assistance. Duration of support ranged from three hours to 22 days (mean length, 3.6 days). Seven of the first 11 patients seen died in the operating room of bleeding, biventricular failure, or both. However, 16 patients (53%)had improved cardiac function, 15 (50%) were weaned, and 11 (37%)were discharged. Of the last 19 patients in the series, 47% survived. Factors affecting survival were myocardial infarction (75%) and renal failure (90%). Common complications were bleeding (73%) and biventricular failure (83%).
S
angle-port design containing a flexible, seam-free segmented polyurethrane sac (Fig 1). The VAD utilizes Bjork-Shiley inlet and outlet valves and has a stroke volume of 65 mL with a dynamic ejection fraction of approximately 0.75. An inlet cannula for the atria, a left ventricular cannula, and an outlet cannula that can be used for the aorta or pulmonary artery are available. The atrial cannula is a 51F Sarns venous cannula coated with polyurethane (Biomer) and shaped to a right-angle configuration. The left ventricular apex cannula and the aortic cannula are large-bore, wire-bound, segmented polyurethane (Biomer) tubes. A Dacron graft forms the end of the infusion cannula, thus permitting a standard vascular anastomosis to the aorta or pulmonary artery. The ventricular cannula is fitted with a felt washer to permit suture fixation to the ventricular myocardium. This cannula system allows cannulation of the left ventricle, the left and right atria, and the aorta or the pulmonary artery (Fig 2). The pneumatic power control unit (Thoratec) is capable of functioning in three modes: a manual set-rate mode, a synchronized mode that detects the R wave of the electrocardiogram, or a fill-to-empty mode that operates by activating a Hall effect switch when the sac is full, thus initiating the emptying cycle. When the VAD is operating in the fill-to-empty mode, the rate times the stroke volume (65 mL) equals VAD flow. To avoid injury to left ventricles that we hoped would recover, patients underwent left atrial cannulation by placement of the cannulas through pursestring sutures in
uccessful application of mechanical circulatory support in patients with cardiogenic shock after cardiac operation was done in the mid-1960s by Spencer and associates [ 11 and DeBakey [ 2 ] . Recently, postcardiotomy circulatory support has been accomplished successfully in many patients with a variety of systems, including roller pumps [3], centrifugal pumps [4], and sac-type pneumatic pumps [5]. At St. Louis University, we have employed these devices for a variety of problems, but patients with profound cardiogenic shock after cardiotomy continue to present our greatest challenge. It is now apparent that our best results have been with the PierceDonachy ventricular assist device (VAD) designed by Pierce and Donachy at Pennsylvania State University [6]. This report analyzes our experience with the PierceDonachy VAD in postcardiotomy patients over the last 7 years, and attempts to establish guidelines for appropriate application of the device.
Material and Methods Description of the Device and Technique for lnsertion The Pierce-Donachy VAD (Thoratec Corp, Berkeley, Calif) is a paracorporeal pneumatic sac-type device constructed of a machined polycarbonate housing with an Presented at the Circulatory Support Symposium of The Society of Thoracic Surgeons, St. Louis, MO, Feb 6-7, 1988. Address reprint requests to Dr Pennington, Department of Surgery, St. Louis University, 1325 S Grand Blvd, St. Louis, MO 63104.
0 1989 by The Society of Thoracic Surgeons
(Ann Thorac Surg 1989;47:130-5)
0003-4975/89/$3.50
Ann Thorac Surg 1989;47:13&5
PENNINGTON ET AL PIERCE-DONACHY VENTRICULAR ASSIST DEVICE
131
Fig I . Cross-sectional v i m of PierceDonachy ventricular assist device.
the left atrial appendage, the dome of the left atrium, or just anterior to the entrance of the right pulmonary veins. Right-sided VAD inflow cannulas were placed into the right atrium through pursestring sutures. The VAD outflow cannulas were preclotted and sutured to the main pulmonary artery or aorta. The VADs were all inserted during cardiopulmonary bypass (CPB). In 3 patients with left-sided Pierce-Donachy pumps, right heart bypass was accomplished with a Bio-Medicus centrifugal pump (Bio-Medicus, Eden Prairie, Minn) connected to the patient by Tygon tubing (Norton Inc, Akron, Ohio). Standard USCI CPB venous cannulas (William
Harvey, Santa Ana, Calif) were used. The right atrial cannula was placed through a pursestring suture, and the pulmonary artery cannula was placed through a No. 14 woven Dacron graft sutured to the pulmonary artery.
Putien t Selection All patients in the series met a predetermined set of hemodynamic criteria derived from the studies of Norman and associates [7], and adopted by the clinical investigation group of the National Heart, Lung, and Blood Institute [8]. Patients were chosen who, in spite of optimal preload levels, maximum inotropic support, and intraaortic balloon (IAB)support, were unable to maintain the following: cardiac index greater than 1.8 L/min/m2; systemic vascular resistance lower than 2,100 dyne-s/cm5; systolic blood pressure greater than 90 mm Hg; right and left atrial pressures lower than 20 mm Hg; and urine output greater than 20 mL/h. Patients who fail to meet these basic hemodynamic criteria for more than 12 hours have an extremely high mortality [7]. From April 1981 to August 1988, we encountered 30 patients aged 15 to 71 years (mean age, 52 years) who failed to meet these minimum hemodynamic criteria in spite of conventional therapy, including the IAB pump (29 patients). Hernodynamic variables, including right and left atrial pressures, thermodilution cardiac output, mixed venous oxygen saturation, and systemic arterial pressure, were measured before and after implantation of the VADs. Determination of perioperative myocardial infarction was made by analysis of serial electrocardiograms and lactate dehydrogenase and creatine kinase isoenzymes. Ventricular function was assessed by hemodynamic measurements, radionuclide imaging, and echocardiography. Radionuclide imaging of the ventricular function in patients with a VAD required some modifications of established nuclear medicine techniques as previously described [9]. Pathological examinations of the sacs and cannulas were made by gross inspection and light and electron microscopy. Patients who died underwent complete postmortem examination according to a protocol that included myocardial staining for infarction. In a few patients, right ventricular needle biopsies were done at
132
Ann Thorac Surg 1989;4713&5
I'ENNINGTON ET AL PIERCE-DONACHY VENTRICULAR ASSIST DEVICE
the time of insertion and removal of the devices. Approximately 1 month after removal of the VAD, survivors were evaluated by echocardiography, nuclear ventriculography multigated acquisition, or cardiac catheterization. In all patients, anticoagulation with heparin sodium was used during CPB for pump insertion. As soon as patients were weaned from bypass with VADs, full reversal of heparin with protamine sulfate was accomplished. Heparin was not used again until the weaning process was begun. Eight postcardiotomy patients who died in the operating room received no anticoagulants at all; 22 patients received low molecular weight dextran at 25 mL/h. Our previous experience [ 101 suggested that biventricular failure was a common problem in postcardiotomy failure. The diagnosis of predominant left or right ventricular failure was initially made when one ventricle appeared to be contracting less well and its associated atrial pressure exceeded the filling pressure of the other ventricle. In the absence of these findings, balanced ventricular failure was judged to be present. All patients without severe peripheral vascular disease underwent initial placement of an IAB, whether the failure was considered to be predominantly right or left. As these studies evolved, we became more aware of the frequency of biventricular failure and the importance of mechanical support for both ventricles in such instances. In several patients, temporary left or right heart bypass was established using the standard CPB system to determine whether the other ventricle could function without a VAD. When the patient had been in stable condition on the VAD for at least 24 hours, the VAD flow was interrupted for two to three minutes to determine pump-on/pump-off hemodynamic data. During the course of these studies, it became apparent that clamping of the cannulas during periods when the pump was off promoted thrombosis. As a result, the cannulas are no longer clamped during pump-off periods. Pump-on/pump-off data were rarely obtained prior to 24 hours of VAD perfusion because in most patients ventricular recovery was not adequate to support the circulation. During weaning, the pump rate was gradually decreased to a level of 40 beats per minute by one of two methods: reducing the rate in the manual mode or setting the VAD rate at 1 : 2 or 1:3 in the electrocardiogram synchronized mode. Simultaneously, intravenous administration of heparin was begun to maintain the activated clotting time at one and a half times the control value. If satisfactory pump-off data were obtained 12 to 24 hours after the weaning process was begun, the patient was returned to the operating room for repeat sternotomy and removal of the device. We [ll] have previously described our most effective predictors for weaning patients from VADs.
Results Of 4,696 patients undergoing cardiac operations at St. Louis University Hospital from February 1982 to August 1988, 30 patients aged 15 to 71 years (mean age, 52 years) were treated with Pierce-Donachy VADs. The cardiac
Table I. Diagnosis in 30 Patients Diagnosis Coronary artery disease Mitral valve disease and/or coronary artery disease Left ventricular aneurysm and/or coronary artery disease Congenital heart defect Ascending aortic aneurysm Prosthetic valve endocarditis
No. of Patients 17 6
3 2 1 1
operations were performed during standard CPB. In 1 patient, the aorta was not clamped. The aortic clamp times in 29 patients ranged from 30 to 156 minutes (mean time, 83 minutes). One patient was maintained on extracorporeal membrane oxygenation in the operating room for 12 hours before insertion of a right VAD. Four patients had IABs inserted preoperatively, 25 patients had lABs inserted during operation, and 1 patient had severe peripheral vascular disease, which prohibited IAB placement. As expected, coronary artery disease, with or without ventricular aneurysm, was the most frequent diagnosis (Table 1). Mitral valve replacement was necessary in 6 patients (with or without coronary artery disease), most of whom required biventricular support. One 28-year-old man with Down's syndrome had a septum primum atrial septal defect, cleft mitral valve, and unrecognized hypoplastic left ventricle. A 27-year-old woman with a septum secundum atrial septal defect and anomalous right pulmonary veins may have sustained coronary air embolization in spite of measures to prevent it. Another patient underwent aortic valve replacement for prosthetic valve endocarditis and aortic insufficiency, and 1 patient had replacement of an ascending aortic aneurysm. All patients met our criteria for severe ventricular failure modeled after the studies of Norman and associates [7] for patients with cardiogenic shock who are on IAB support and have optimal preload levels. The VAD was inserted in 20 patients because they could not be weaned from CPB. Eight patients were initially weaned from CPB but sustained cardiac deterioration in the operating room and required reinstitution of CPB and VAD replacement. One patient was maintained on partial CPB for 12 hours until a right VAD was inserted. The condition of another patient deteriorated six hours postoperatively in the intensive care unit, and the patient was returned to the operating room for insertion of a left VAD. All patients were eventually weaned from CPB with maximum VAD flows ranging from 1.6 to 2.9 L/min/m2. The length of VAD perfusion was less than 12 hours in 9 patients, between one and seven days in 15 patients, and greater than seven days in 6 patients. Bleeding, biventricular failure, and multiple-organ failure accounted for the deaths of 14 patients who had no improvement in cardiac function. Sixteen patients (53%) demonstrated improvement in cardiac function, 15 (50%)
Ann Thorac Surg
Table 2. Complications in 30 Patients Complication
133
PENNINGTON ET AL PIERCE-DONACHY VENTRICULAR ASSIST DEVICE
1989;47130-5
No. of Patients
Biventricular failure
24
B1eedin g
22
Respiratory failure Renal failure Dialysis Infection Thrombus Embolus Neurological deficit Cerebrovascular accident Hemolysis (plasma hemoglobin > 40 mgldL) Mechanical failure Intraaortic balloon pump (limb ischemia)
10
10
9 8 3 0 2 2 2 1 1
were weaned from the VAD, and 11 (37%) were discharged from the hospital. Although 7 of the first 11 patients in the series died in the operating room, there was only 1 intraoperative death among the last 19 patients. Sixty-three percent of the last 19 patients were weaned, and 47% (9 patients) survived. Of the 11 survivors followed 6 to 60 months after hospital discharge, 1 died of severe cardiomyopathy 6 months after discharge, 1 died 26 months after discharge of unknown causes, and 1 died of metastatic cancer 53 months after discharge. Four patients returned to work or school, 2 retired, 1 is recovering, and 1 has recurrent episodes of congestive heart failure but refuses further treatment. Biventricular failure occurred more commonly than isolated ventricular failure (Table 2). Of the 30 patients, 14 received only a left VAD, 7 received only a right VAD (with an IAB to support the left ventricle), and 9 received biventricular assist devices. Eight of the 16 patients who received biventricular support were weaned, and 6 were long-term survivors. In contrast, of 8 patients with biventricular failure who received only a left VAD just 2 were weaned (25%) and 1 survived. Of 6 patienkwith isolated left ventricular failure, 5 received only a left VAD and survived. The sixth patient died in the operating room of a tear in the left atrium. In 2 of the 4 surviving patients with a left VAD, minimal use of inotropic drugs was required during the period of full left VAD support. In the other 2, inotropic drugs were not required until the weaning process was begun. Bleeding led to at least one reoperation in 18 of the 30 patients (see Table 2), two reoperations in 2 patients, and four reoperations in 2 patients. Early in our experience, we left the sternum open and closed only the skin, a technique that facilitated reoperation in the intensive care unit. In the latter part of the study, we found that sternal closure decreased the amount of bleeding and allowed for endotracheal tube removal. Disseminated intravascular coagulation was documented in only 1 patient; coagulation disorders in the other patients were of diverse etiol-
ogy. Hemolysis (plasma hemoglobin level > 40 mg/dL) occurred in 2 patients, but cleared during the course of VAD perfusion, suggesting that hemolysis was unrelated to the VAD. Maximum plasma hemoglobin levels per patient ranged from 5 to 139 mg/dL (mean value, 15 mg/ dL). Thrombus was noted in three VAD sacs. In 1 patient with biventricular assist devices, we had intermittently clamped the right VAD cannulas to perform thermodilution cardiac output measurements. At the time of right VAD removal, there was a large thrombus (3 x 4 cm) in the right VAD sac and small thrombi on the right VAD valves. The other 2 patients had small thrombi (less than 3 mm in diameter) firmly adherent to the VAD sac. We subsequently discontinued the practice of briefly clamping the cannulas during pump-on/pump-off periods, and have had no visible thrombi present in the last 18 postcardiotomy patients who were perfused from eight to 528 hours. After 48 to 72 hours, a thin, fibrous layer composed largely of acellular elements forms on the sac and can be gently lifted away at the time of VAD removal. In spite of the finding of thrombus in some sacs, there were no documented instances of embolism from the sacs. Two patients had strokes during the perioperative period. In 1 of them, the postoperative course was complicated by renal failure, sepsis, and a large right-sided stroke. The etiology of the stroke was never determined because both the left and right VAD sacs were free from thrombus at removal. The second patient sustained a stroke of the left parietal lobe during the weaning period. Both VAD sacs were free from thrombus at the time the device was removed. However, as the patient sustained the stroke during the period of lowest flow, it is possible that the stroke was related to the device. The patient regained full function of his extremities within 6 months and was working full-time within a year of discharge. Postmortem examinations were performed on 15 of the 19 patients who died during the immediate postoperative period. Eleven patients had evidence of extensive myocardial infarction, which occurred during the perioperative period (Table 3). Five of these patients were known to have sustained acute myocardial infarctions preoperatively, which led to acute cardiogenic shock necessitating emergency cardiac operations. One patient had an extensive myocardial infarction during bypass grafting. Two
Table 3 . Survival Anion,q Patients With Myocardial Infarction Myocardial Infarction
Proven (autopsy or ECG and enzymes) Likely (positive ECG o r enzymes) Possible (inadequatedata) None
A
Survivors
Nonsurvivors
1
11
2
1
1
3
7
4
Total
11
19
Incidence of infarction (76)"
27
63
This is based on proven or likely diagnosis of myocardial infarction
ECG
=
electrocardiogram.
134
PENNINGTONET AL PIERCE-DONACHY VENTRICULAR ASSIST DEVICE
patients without acute myocardial infarctions had evidence at autopsy of either diffuse ischemic changes or focal interstitial myocarditis. Four patients did not have a postmortem examination. In 2 of them, right ventricular biopsy specimens demonstrated myocardial necrosis; 1 died in the operating room of a tear in the left atrium and the other of biventricular failure. Based on these pathological data, we estimated that at least 11 patients had myocardial injuries that in themselves were severe enough to preclude survival. The deaths of the other 8 patients were attributed primarily to bleeding, biventricular failure, and renal failure. Of the 11 survivors, 3 had some evidence of perioperative myocardial injury. One patient had elevated cardiac enzyme levels but no changes from the preoperative electrocardiogram, and 1 patient had electrocardiographic evidence of an infarction but normal postoperative enzyme levels. The latter patient lived for 6 months after discharge and then died of extensive myocardial infarctions in spite of patent coronary arteries. Only 1 discharged patient had positive electrocardiographic and enzymatic changes clearly indicating a perioperative acute myocardial infarction. Ten of the 11 survivors underwent echocardiographic or nuclear angiographic multigated acquisition evaluation approximately 1 month after operation. The left ventricular ejection fractions ranged from 32% to 72% (mean value, 53.5%), and 8 patients had normal ejection fractions.
Comment This series demonstrates the increasing success of mechanical support of the failing ventricle after cardiac operations. Although our survival of 37% is not as high as desired, it has gradually improved. Seven of our first 11 patients died in the operating room, but of the subsequent 19 patients, 63% were weaned and 47% survived. We attribute our increasing success to improved patient selection, earlier application of devices, better control of postoperative bleeding, and more aggressive treatment of biventricular failure. Fifty-three percent of our patients required biventricular support: 9 received biventricular assist devices and 7 were supported with right VADs and IABs. The rate of weaning and the survival in the 16 patients who received biventricular support were substantially better than the results in patients with biventricular failure who received mechanical support of the left ventricle only. However, of the 6 patients with isolated left ventricular failure who received only a left VAD, 5 were weaned and survived (83%),thus indicating that some VAD candidates do not require biventricular devices. Isolated left ventricular failure does occur, and it can be treated effectively with a left VAD alone [lo]. We continue to prefer left atrial cannulation for patients with postcardiotomy heart failure to avoid injury to the left ventricle. Although Bernhard and co-workers [12] were successful with ventricular recovery after apex cannulation, we believe left atrial cannulation allows the ventricle more potential for recovery. Although apex cannulation allows for better ventricular decompression,
Ann Thorac Surg 1989;4713&5
it is apparent that ventricular recovery can occur without complete decompression, as evidenced by our results utilizing atrial cannulation. Two important factors are strong negative determinants of survival: perioperative myocardial infarction and renal failure [13, 141. Bleeding was a frequent problem in these patients, who had all undergone cardiac operations and periods of CPB. In spite of complete reversal of heparin anticoagulation with protamine, postoperative bleeding was frequent. Although disseminated intravascular coagulation was rarely documented, many patients have platelet disorders and other coagulation defects [15]. In the latter part of the study, we wired the sternum when possible, and the bleeding seemed to decrease. Meticulous surgical technique, the withholding of heparin, and liberal administration of platelets and fresh frozen plasma seem to have also helped decrease the incidence of bleeding. In 1987, we [13] reported the important effect of a perioperative myocardial infarction on survival of VAD patients, and our present data support that thesis. Of the 11 survivors in this series, only 1 had clear evidence of a perioperative infarction, whereas 11 of the 19 nonsurvivors had acute myocardial injuries severe enough to preclude survival. Previously, we [ 101 demonstrated that preoperative left ventricular function was not as important in determining survival as the effects of biventricular failure or perioperative infarction. Considering our present series, it continues to appear that acute perioperative myocardial infarction is an important negative determinant of survival. Renal failure requiring dialysis is also an important factor influencing survival. In this series of 30 patients, renal failure developed in 10 patients, 9 of whom had dialysis; there was 1 survivor. In an earlier evaluation of 27 VAD patients at our institution [14], 13 of 15 nonsurvivors had renal failure. It is clear that renal failure requiring dialysis in our VAD patients was highly predictive of death. Infection in patients with VADs is also an important event that can influence survival [16]. In this series, infections developed in 8 patients, pneumonia in 7 patients and empyema in l patient. Mediastinitis developed in 2 patients who had undergone multiple sternotomies for bleeding complications. Only 1 of the 8 patients with a major infection survived. All organisms isolated were typical nosocomial pathogens and not opportunistic agents. Surprisingly, the cannula exit sites were frequently uninvolved with infection and when infection did occur at the cannula sites, mediastinitis was not inevitable. Duration of support appears to be a factor in the development of infection; hence, weaning should be initiated after the first 24 hours of stabilization. Previous studies [ 111 suggested that hemodynamic measurements of atrial pressures and cardiac index with the VAD off were predictive of successful weaning. Therefore we perform pump-odpump-off measurements as soon as the hemodynamic data suggest that weaning might be successfully accomplished. Of the 11 patients discharged, 3 died. Only 1 of these deaths was cardiac related. The survivors are leading
Ann Thorac Surg
PENNINGTON ET AL PIERCE-DONACHY VENTRICULAR ASSIST DEVICE
1989;47130-5
active lives, and many have returned to work or school. Follow-up studies demonstrate normal or nearly normal ventricular function in more than half of the recipients. It is now apparent that many patients with postcardiotomy failure have a stunned myocardium that can undergo full recovery [S, 17). The PierceDonachy VAD has proved to be an effective device for postcardiotomy patients, and salvage rates are increasing as we improve patient selection and device applications. Supported by grant No. N01-HV12909 from the National Heart, Lung, and Blood Institute.
References 1. Spencer FC, Eiseman 8, Trinkle JK, Rossi NP. Assisted circulation for cardiac failure following intracardiac surgery with cardiopulmonary bypass. J Thorac Cardiovasc Surg 1965;49:56. 2. DeBakey ME. Left ventricular bypass pump for cardiac assistance: clinical experience. Am J Cardiol 1971;27:3. 3. Rose DM, Laschinger J, Grossi E, Krieger KH, Cunningham JN Jr, Spencer FC. Experimental and clinical results with a simplified left heart assist device for treatment of profound left ventricular dysfunction. World J Surg 1985;9:11-7. 4. Park SB, Liebler GA, Burkholder JA, et al. Mechanical support of the failing heart. Ann Thorac Surg 1986;42:627-31. 5. Gaines WE, Pierce WS, Donachy JH, et al. The Pennsylvania State University paracorporeal ventricular assist pump: optimal methods of use. World J Surg 1985;9:47-57. 6. Donachy JH, Landis DL, Rosenberg G, et al. Design and evaluation of a left ventricular assist device: the angle port pump. In: Unger F, ed. Assisted circulation. Berlin, Heidelberg, New York: Springer-Verlag, 1979:13E46. 7. Norman JC, Cooley DA, Igo SR, et al. Prognostic indices for
8.
9.
10.
11.
12.
13.
14.
15. 16. 17.
135
survival during postcardiotomy intra-aortic balloon pumping. J Thorac Cardiovasc Surg 1977;74:709. Pennington DG, Bernhard WF, Golding LR, Berger RL, Khuri SF, Watson JT. Long-term follow-up of postcardiotomy patients with profound cardiogenic shock treated with ventricular assist devices. Circulation 1985;72(Suppl 2):21626. Pennington DG, Samuels LD, Williams G, et al. Experience with the Pierce-Donachy ventricular assist device in postcardiotomy patients with cardiogenic shock. World J Surg 1985;9:37-46. Pennington DG, Merjavy JP, Swartz MT, et al. The importance of biventricular failure in patients with postoperative cardiogenic shock. Ann Thorac Surg 1985;39:16-26. Termuhlen DG, Swartz MT, Pennington DG, et al. Predictors for weaning patients from ventricular assist devices (VADs). ASAIO Trans 1987;33:683. Bernhard WF, Clay W, Gernes D, et al. Temporary and permanent left ventricular bypass: laboratory and clinical observations. World J Surg 1985;9:54. Pennington DG, McBride LR, Kanter KR, Swartz MT, Miller LW. The effect of acute perioperative myocardial infarction on survival of postcardiotomy patients supported with ventricular assist devices. Circulation (in press). Kanter KR, Swartz MT, Pennington DG, et al. Renal failure in patients with ventricular assist devices. ASAIO Trans 1987; 33:426-8. Joist JH, Pennington DG. Platelet reactions with artificial surfaces. ASAIO Trans 1987;33:341-4. McBride LR, Ruzevich SA, Pennington DG, et al. Infectious complications associated with ventricular assist device support. ASAIO Trans 1987;33:201. Ruzevich SA, Kanter KR, Pennington DG, Swartz MT, McBride LR, Termuhlen DG. Long-term follow-up of survivors of postcardiotomy circulatory support. ASAIO Trans 1987;33(Suppl 3):177.
D. Glenn Pennington, MD, Lawrence R. McBride, MD, Marc T. Swartz, BA, Kirk R. Kanter, MD, George C. Kaiser, MD, Hendrick B. garner, MD, Leslie W. Miller, MD, Keith S. Naunheim, MD, Andrew C. Fiore, MD, and Vallee L. Willman, MD Department of Surgery, St. Louis University Hospital, St. Louis, Missouri
In spite of recent improvements in cardiac surgery, a small percentage of patients have severe postcardiotomy ventricular failure refractory to drugs and the intraaortic balloon. In our experience, the Pierce-Donachy external pneumatic ventricular assist device has proved to be one of the most effective devices for these patients. Since 1981,30 patients aged 15 to 71 years (mean age, 52 years) with profound cardiogenic shock refractory to conventional therapy after cardiotomy were supported with the Pierce-Donachy ventricular assist device. Fourteen required left ventricular support, 7 needed right ventricular support with an intraaortic balloon, and 9 had biven-
tricular assistance. Duration of support ranged from three hours to 22 days (mean length, 3.6 days). Seven of the first 11 patients seen died in the operating room of bleeding, biventricular failure, or both. However, 16 patients (53%)had improved cardiac function, 15 (50%) were weaned, and 11 (37%)were discharged. Of the last 19 patients in the series, 47% survived. Factors affecting survival were myocardial infarction (75%) and renal failure (90%). Common complications were bleeding (73%) and biventricular failure (83%).
S
angle-port design containing a flexible, seam-free segmented polyurethrane sac (Fig 1). The VAD utilizes Bjork-Shiley inlet and outlet valves and has a stroke volume of 65 mL with a dynamic ejection fraction of approximately 0.75. An inlet cannula for the atria, a left ventricular cannula, and an outlet cannula that can be used for the aorta or pulmonary artery are available. The atrial cannula is a 51F Sarns venous cannula coated with polyurethane (Biomer) and shaped to a right-angle configuration. The left ventricular apex cannula and the aortic cannula are large-bore, wire-bound, segmented polyurethane (Biomer) tubes. A Dacron graft forms the end of the infusion cannula, thus permitting a standard vascular anastomosis to the aorta or pulmonary artery. The ventricular cannula is fitted with a felt washer to permit suture fixation to the ventricular myocardium. This cannula system allows cannulation of the left ventricle, the left and right atria, and the aorta or the pulmonary artery (Fig 2). The pneumatic power control unit (Thoratec) is capable of functioning in three modes: a manual set-rate mode, a synchronized mode that detects the R wave of the electrocardiogram, or a fill-to-empty mode that operates by activating a Hall effect switch when the sac is full, thus initiating the emptying cycle. When the VAD is operating in the fill-to-empty mode, the rate times the stroke volume (65 mL) equals VAD flow. To avoid injury to left ventricles that we hoped would recover, patients underwent left atrial cannulation by placement of the cannulas through pursestring sutures in
uccessful application of mechanical circulatory support in patients with cardiogenic shock after cardiac operation was done in the mid-1960s by Spencer and associates [ 11 and DeBakey [ 2 ] . Recently, postcardiotomy circulatory support has been accomplished successfully in many patients with a variety of systems, including roller pumps [3], centrifugal pumps [4], and sac-type pneumatic pumps [5]. At St. Louis University, we have employed these devices for a variety of problems, but patients with profound cardiogenic shock after cardiotomy continue to present our greatest challenge. It is now apparent that our best results have been with the PierceDonachy ventricular assist device (VAD) designed by Pierce and Donachy at Pennsylvania State University [6]. This report analyzes our experience with the PierceDonachy VAD in postcardiotomy patients over the last 7 years, and attempts to establish guidelines for appropriate application of the device.
Material and Methods Description of the Device and Technique for lnsertion The Pierce-Donachy VAD (Thoratec Corp, Berkeley, Calif) is a paracorporeal pneumatic sac-type device constructed of a machined polycarbonate housing with an Presented at the Circulatory Support Symposium of The Society of Thoracic Surgeons, St. Louis, MO, Feb 6-7, 1988. Address reprint requests to Dr Pennington, Department of Surgery, St. Louis University, 1325 S Grand Blvd, St. Louis, MO 63104.
0 1989 by The Society of Thoracic Surgeons
(Ann Thorac Surg 1989;47:130-5)
0003-4975/89/$3.50
Ann Thorac Surg 1989;47:13&5
PENNINGTON ET AL PIERCE-DONACHY VENTRICULAR ASSIST DEVICE
131
Fig I . Cross-sectional v i m of PierceDonachy ventricular assist device.
the left atrial appendage, the dome of the left atrium, or just anterior to the entrance of the right pulmonary veins. Right-sided VAD inflow cannulas were placed into the right atrium through pursestring sutures. The VAD outflow cannulas were preclotted and sutured to the main pulmonary artery or aorta. The VADs were all inserted during cardiopulmonary bypass (CPB). In 3 patients with left-sided Pierce-Donachy pumps, right heart bypass was accomplished with a Bio-Medicus centrifugal pump (Bio-Medicus, Eden Prairie, Minn) connected to the patient by Tygon tubing (Norton Inc, Akron, Ohio). Standard USCI CPB venous cannulas (William
Harvey, Santa Ana, Calif) were used. The right atrial cannula was placed through a pursestring suture, and the pulmonary artery cannula was placed through a No. 14 woven Dacron graft sutured to the pulmonary artery.
Putien t Selection All patients in the series met a predetermined set of hemodynamic criteria derived from the studies of Norman and associates [7], and adopted by the clinical investigation group of the National Heart, Lung, and Blood Institute [8]. Patients were chosen who, in spite of optimal preload levels, maximum inotropic support, and intraaortic balloon (IAB)support, were unable to maintain the following: cardiac index greater than 1.8 L/min/m2; systemic vascular resistance lower than 2,100 dyne-s/cm5; systolic blood pressure greater than 90 mm Hg; right and left atrial pressures lower than 20 mm Hg; and urine output greater than 20 mL/h. Patients who fail to meet these basic hemodynamic criteria for more than 12 hours have an extremely high mortality [7]. From April 1981 to August 1988, we encountered 30 patients aged 15 to 71 years (mean age, 52 years) who failed to meet these minimum hemodynamic criteria in spite of conventional therapy, including the IAB pump (29 patients). Hernodynamic variables, including right and left atrial pressures, thermodilution cardiac output, mixed venous oxygen saturation, and systemic arterial pressure, were measured before and after implantation of the VADs. Determination of perioperative myocardial infarction was made by analysis of serial electrocardiograms and lactate dehydrogenase and creatine kinase isoenzymes. Ventricular function was assessed by hemodynamic measurements, radionuclide imaging, and echocardiography. Radionuclide imaging of the ventricular function in patients with a VAD required some modifications of established nuclear medicine techniques as previously described [9]. Pathological examinations of the sacs and cannulas were made by gross inspection and light and electron microscopy. Patients who died underwent complete postmortem examination according to a protocol that included myocardial staining for infarction. In a few patients, right ventricular needle biopsies were done at
132
Ann Thorac Surg 1989;4713&5
I'ENNINGTON ET AL PIERCE-DONACHY VENTRICULAR ASSIST DEVICE
the time of insertion and removal of the devices. Approximately 1 month after removal of the VAD, survivors were evaluated by echocardiography, nuclear ventriculography multigated acquisition, or cardiac catheterization. In all patients, anticoagulation with heparin sodium was used during CPB for pump insertion. As soon as patients were weaned from bypass with VADs, full reversal of heparin with protamine sulfate was accomplished. Heparin was not used again until the weaning process was begun. Eight postcardiotomy patients who died in the operating room received no anticoagulants at all; 22 patients received low molecular weight dextran at 25 mL/h. Our previous experience [ 101 suggested that biventricular failure was a common problem in postcardiotomy failure. The diagnosis of predominant left or right ventricular failure was initially made when one ventricle appeared to be contracting less well and its associated atrial pressure exceeded the filling pressure of the other ventricle. In the absence of these findings, balanced ventricular failure was judged to be present. All patients without severe peripheral vascular disease underwent initial placement of an IAB, whether the failure was considered to be predominantly right or left. As these studies evolved, we became more aware of the frequency of biventricular failure and the importance of mechanical support for both ventricles in such instances. In several patients, temporary left or right heart bypass was established using the standard CPB system to determine whether the other ventricle could function without a VAD. When the patient had been in stable condition on the VAD for at least 24 hours, the VAD flow was interrupted for two to three minutes to determine pump-on/pump-off hemodynamic data. During the course of these studies, it became apparent that clamping of the cannulas during periods when the pump was off promoted thrombosis. As a result, the cannulas are no longer clamped during pump-off periods. Pump-on/pump-off data were rarely obtained prior to 24 hours of VAD perfusion because in most patients ventricular recovery was not adequate to support the circulation. During weaning, the pump rate was gradually decreased to a level of 40 beats per minute by one of two methods: reducing the rate in the manual mode or setting the VAD rate at 1 : 2 or 1:3 in the electrocardiogram synchronized mode. Simultaneously, intravenous administration of heparin was begun to maintain the activated clotting time at one and a half times the control value. If satisfactory pump-off data were obtained 12 to 24 hours after the weaning process was begun, the patient was returned to the operating room for repeat sternotomy and removal of the device. We [ll] have previously described our most effective predictors for weaning patients from VADs.
Results Of 4,696 patients undergoing cardiac operations at St. Louis University Hospital from February 1982 to August 1988, 30 patients aged 15 to 71 years (mean age, 52 years) were treated with Pierce-Donachy VADs. The cardiac
Table I. Diagnosis in 30 Patients Diagnosis Coronary artery disease Mitral valve disease and/or coronary artery disease Left ventricular aneurysm and/or coronary artery disease Congenital heart defect Ascending aortic aneurysm Prosthetic valve endocarditis
No. of Patients 17 6
3 2 1 1
operations were performed during standard CPB. In 1 patient, the aorta was not clamped. The aortic clamp times in 29 patients ranged from 30 to 156 minutes (mean time, 83 minutes). One patient was maintained on extracorporeal membrane oxygenation in the operating room for 12 hours before insertion of a right VAD. Four patients had IABs inserted preoperatively, 25 patients had lABs inserted during operation, and 1 patient had severe peripheral vascular disease, which prohibited IAB placement. As expected, coronary artery disease, with or without ventricular aneurysm, was the most frequent diagnosis (Table 1). Mitral valve replacement was necessary in 6 patients (with or without coronary artery disease), most of whom required biventricular support. One 28-year-old man with Down's syndrome had a septum primum atrial septal defect, cleft mitral valve, and unrecognized hypoplastic left ventricle. A 27-year-old woman with a septum secundum atrial septal defect and anomalous right pulmonary veins may have sustained coronary air embolization in spite of measures to prevent it. Another patient underwent aortic valve replacement for prosthetic valve endocarditis and aortic insufficiency, and 1 patient had replacement of an ascending aortic aneurysm. All patients met our criteria for severe ventricular failure modeled after the studies of Norman and associates [7] for patients with cardiogenic shock who are on IAB support and have optimal preload levels. The VAD was inserted in 20 patients because they could not be weaned from CPB. Eight patients were initially weaned from CPB but sustained cardiac deterioration in the operating room and required reinstitution of CPB and VAD replacement. One patient was maintained on partial CPB for 12 hours until a right VAD was inserted. The condition of another patient deteriorated six hours postoperatively in the intensive care unit, and the patient was returned to the operating room for insertion of a left VAD. All patients were eventually weaned from CPB with maximum VAD flows ranging from 1.6 to 2.9 L/min/m2. The length of VAD perfusion was less than 12 hours in 9 patients, between one and seven days in 15 patients, and greater than seven days in 6 patients. Bleeding, biventricular failure, and multiple-organ failure accounted for the deaths of 14 patients who had no improvement in cardiac function. Sixteen patients (53%) demonstrated improvement in cardiac function, 15 (50%)
Ann Thorac Surg
Table 2. Complications in 30 Patients Complication
133
PENNINGTON ET AL PIERCE-DONACHY VENTRICULAR ASSIST DEVICE
1989;47130-5
No. of Patients
Biventricular failure
24
B1eedin g
22
Respiratory failure Renal failure Dialysis Infection Thrombus Embolus Neurological deficit Cerebrovascular accident Hemolysis (plasma hemoglobin > 40 mgldL) Mechanical failure Intraaortic balloon pump (limb ischemia)
10
10
9 8 3 0 2 2 2 1 1
were weaned from the VAD, and 11 (37%) were discharged from the hospital. Although 7 of the first 11 patients in the series died in the operating room, there was only 1 intraoperative death among the last 19 patients. Sixty-three percent of the last 19 patients were weaned, and 47% (9 patients) survived. Of the 11 survivors followed 6 to 60 months after hospital discharge, 1 died of severe cardiomyopathy 6 months after discharge, 1 died 26 months after discharge of unknown causes, and 1 died of metastatic cancer 53 months after discharge. Four patients returned to work or school, 2 retired, 1 is recovering, and 1 has recurrent episodes of congestive heart failure but refuses further treatment. Biventricular failure occurred more commonly than isolated ventricular failure (Table 2). Of the 30 patients, 14 received only a left VAD, 7 received only a right VAD (with an IAB to support the left ventricle), and 9 received biventricular assist devices. Eight of the 16 patients who received biventricular support were weaned, and 6 were long-term survivors. In contrast, of 8 patients with biventricular failure who received only a left VAD just 2 were weaned (25%) and 1 survived. Of 6 patienkwith isolated left ventricular failure, 5 received only a left VAD and survived. The sixth patient died in the operating room of a tear in the left atrium. In 2 of the 4 surviving patients with a left VAD, minimal use of inotropic drugs was required during the period of full left VAD support. In the other 2, inotropic drugs were not required until the weaning process was begun. Bleeding led to at least one reoperation in 18 of the 30 patients (see Table 2), two reoperations in 2 patients, and four reoperations in 2 patients. Early in our experience, we left the sternum open and closed only the skin, a technique that facilitated reoperation in the intensive care unit. In the latter part of the study, we found that sternal closure decreased the amount of bleeding and allowed for endotracheal tube removal. Disseminated intravascular coagulation was documented in only 1 patient; coagulation disorders in the other patients were of diverse etiol-
ogy. Hemolysis (plasma hemoglobin level > 40 mg/dL) occurred in 2 patients, but cleared during the course of VAD perfusion, suggesting that hemolysis was unrelated to the VAD. Maximum plasma hemoglobin levels per patient ranged from 5 to 139 mg/dL (mean value, 15 mg/ dL). Thrombus was noted in three VAD sacs. In 1 patient with biventricular assist devices, we had intermittently clamped the right VAD cannulas to perform thermodilution cardiac output measurements. At the time of right VAD removal, there was a large thrombus (3 x 4 cm) in the right VAD sac and small thrombi on the right VAD valves. The other 2 patients had small thrombi (less than 3 mm in diameter) firmly adherent to the VAD sac. We subsequently discontinued the practice of briefly clamping the cannulas during pump-on/pump-off periods, and have had no visible thrombi present in the last 18 postcardiotomy patients who were perfused from eight to 528 hours. After 48 to 72 hours, a thin, fibrous layer composed largely of acellular elements forms on the sac and can be gently lifted away at the time of VAD removal. In spite of the finding of thrombus in some sacs, there were no documented instances of embolism from the sacs. Two patients had strokes during the perioperative period. In 1 of them, the postoperative course was complicated by renal failure, sepsis, and a large right-sided stroke. The etiology of the stroke was never determined because both the left and right VAD sacs were free from thrombus at removal. The second patient sustained a stroke of the left parietal lobe during the weaning period. Both VAD sacs were free from thrombus at the time the device was removed. However, as the patient sustained the stroke during the period of lowest flow, it is possible that the stroke was related to the device. The patient regained full function of his extremities within 6 months and was working full-time within a year of discharge. Postmortem examinations were performed on 15 of the 19 patients who died during the immediate postoperative period. Eleven patients had evidence of extensive myocardial infarction, which occurred during the perioperative period (Table 3). Five of these patients were known to have sustained acute myocardial infarctions preoperatively, which led to acute cardiogenic shock necessitating emergency cardiac operations. One patient had an extensive myocardial infarction during bypass grafting. Two
Table 3 . Survival Anion,q Patients With Myocardial Infarction Myocardial Infarction
Proven (autopsy or ECG and enzymes) Likely (positive ECG o r enzymes) Possible (inadequatedata) None
A
Survivors
Nonsurvivors
1
11
2
1
1
3
7
4
Total
11
19
Incidence of infarction (76)"
27
63
This is based on proven or likely diagnosis of myocardial infarction
ECG
=
electrocardiogram.
134
PENNINGTONET AL PIERCE-DONACHY VENTRICULAR ASSIST DEVICE
patients without acute myocardial infarctions had evidence at autopsy of either diffuse ischemic changes or focal interstitial myocarditis. Four patients did not have a postmortem examination. In 2 of them, right ventricular biopsy specimens demonstrated myocardial necrosis; 1 died in the operating room of a tear in the left atrium and the other of biventricular failure. Based on these pathological data, we estimated that at least 11 patients had myocardial injuries that in themselves were severe enough to preclude survival. The deaths of the other 8 patients were attributed primarily to bleeding, biventricular failure, and renal failure. Of the 11 survivors, 3 had some evidence of perioperative myocardial injury. One patient had elevated cardiac enzyme levels but no changes from the preoperative electrocardiogram, and 1 patient had electrocardiographic evidence of an infarction but normal postoperative enzyme levels. The latter patient lived for 6 months after discharge and then died of extensive myocardial infarctions in spite of patent coronary arteries. Only 1 discharged patient had positive electrocardiographic and enzymatic changes clearly indicating a perioperative acute myocardial infarction. Ten of the 11 survivors underwent echocardiographic or nuclear angiographic multigated acquisition evaluation approximately 1 month after operation. The left ventricular ejection fractions ranged from 32% to 72% (mean value, 53.5%), and 8 patients had normal ejection fractions.
Comment This series demonstrates the increasing success of mechanical support of the failing ventricle after cardiac operations. Although our survival of 37% is not as high as desired, it has gradually improved. Seven of our first 11 patients died in the operating room, but of the subsequent 19 patients, 63% were weaned and 47% survived. We attribute our increasing success to improved patient selection, earlier application of devices, better control of postoperative bleeding, and more aggressive treatment of biventricular failure. Fifty-three percent of our patients required biventricular support: 9 received biventricular assist devices and 7 were supported with right VADs and IABs. The rate of weaning and the survival in the 16 patients who received biventricular support were substantially better than the results in patients with biventricular failure who received mechanical support of the left ventricle only. However, of the 6 patients with isolated left ventricular failure who received only a left VAD, 5 were weaned and survived (83%),thus indicating that some VAD candidates do not require biventricular devices. Isolated left ventricular failure does occur, and it can be treated effectively with a left VAD alone [lo]. We continue to prefer left atrial cannulation for patients with postcardiotomy heart failure to avoid injury to the left ventricle. Although Bernhard and co-workers [12] were successful with ventricular recovery after apex cannulation, we believe left atrial cannulation allows the ventricle more potential for recovery. Although apex cannulation allows for better ventricular decompression,
Ann Thorac Surg 1989;4713&5
it is apparent that ventricular recovery can occur without complete decompression, as evidenced by our results utilizing atrial cannulation. Two important factors are strong negative determinants of survival: perioperative myocardial infarction and renal failure [13, 141. Bleeding was a frequent problem in these patients, who had all undergone cardiac operations and periods of CPB. In spite of complete reversal of heparin anticoagulation with protamine, postoperative bleeding was frequent. Although disseminated intravascular coagulation was rarely documented, many patients have platelet disorders and other coagulation defects [15]. In the latter part of the study, we wired the sternum when possible, and the bleeding seemed to decrease. Meticulous surgical technique, the withholding of heparin, and liberal administration of platelets and fresh frozen plasma seem to have also helped decrease the incidence of bleeding. In 1987, we [13] reported the important effect of a perioperative myocardial infarction on survival of VAD patients, and our present data support that thesis. Of the 11 survivors in this series, only 1 had clear evidence of a perioperative infarction, whereas 11 of the 19 nonsurvivors had acute myocardial injuries severe enough to preclude survival. Previously, we [ 101 demonstrated that preoperative left ventricular function was not as important in determining survival as the effects of biventricular failure or perioperative infarction. Considering our present series, it continues to appear that acute perioperative myocardial infarction is an important negative determinant of survival. Renal failure requiring dialysis is also an important factor influencing survival. In this series of 30 patients, renal failure developed in 10 patients, 9 of whom had dialysis; there was 1 survivor. In an earlier evaluation of 27 VAD patients at our institution [14], 13 of 15 nonsurvivors had renal failure. It is clear that renal failure requiring dialysis in our VAD patients was highly predictive of death. Infection in patients with VADs is also an important event that can influence survival [16]. In this series, infections developed in 8 patients, pneumonia in 7 patients and empyema in l patient. Mediastinitis developed in 2 patients who had undergone multiple sternotomies for bleeding complications. Only 1 of the 8 patients with a major infection survived. All organisms isolated were typical nosocomial pathogens and not opportunistic agents. Surprisingly, the cannula exit sites were frequently uninvolved with infection and when infection did occur at the cannula sites, mediastinitis was not inevitable. Duration of support appears to be a factor in the development of infection; hence, weaning should be initiated after the first 24 hours of stabilization. Previous studies [ 111 suggested that hemodynamic measurements of atrial pressures and cardiac index with the VAD off were predictive of successful weaning. Therefore we perform pump-odpump-off measurements as soon as the hemodynamic data suggest that weaning might be successfully accomplished. Of the 11 patients discharged, 3 died. Only 1 of these deaths was cardiac related. The survivors are leading
Ann Thorac Surg
PENNINGTON ET AL PIERCE-DONACHY VENTRICULAR ASSIST DEVICE
1989;47130-5
active lives, and many have returned to work or school. Follow-up studies demonstrate normal or nearly normal ventricular function in more than half of the recipients. It is now apparent that many patients with postcardiotomy failure have a stunned myocardium that can undergo full recovery [S, 17). The PierceDonachy VAD has proved to be an effective device for postcardiotomy patients, and salvage rates are increasing as we improve patient selection and device applications. Supported by grant No. N01-HV12909 from the National Heart, Lung, and Blood Institute.
References 1. Spencer FC, Eiseman 8, Trinkle JK, Rossi NP. Assisted circulation for cardiac failure following intracardiac surgery with cardiopulmonary bypass. J Thorac Cardiovasc Surg 1965;49:56. 2. DeBakey ME. Left ventricular bypass pump for cardiac assistance: clinical experience. Am J Cardiol 1971;27:3. 3. Rose DM, Laschinger J, Grossi E, Krieger KH, Cunningham JN Jr, Spencer FC. Experimental and clinical results with a simplified left heart assist device for treatment of profound left ventricular dysfunction. World J Surg 1985;9:11-7. 4. Park SB, Liebler GA, Burkholder JA, et al. Mechanical support of the failing heart. Ann Thorac Surg 1986;42:627-31. 5. Gaines WE, Pierce WS, Donachy JH, et al. The Pennsylvania State University paracorporeal ventricular assist pump: optimal methods of use. World J Surg 1985;9:47-57. 6. Donachy JH, Landis DL, Rosenberg G, et al. Design and evaluation of a left ventricular assist device: the angle port pump. In: Unger F, ed. Assisted circulation. Berlin, Heidelberg, New York: Springer-Verlag, 1979:13E46. 7. Norman JC, Cooley DA, Igo SR, et al. Prognostic indices for
8.
9.
10.
11.
12.
13.
14.
15. 16. 17.
135
survival during postcardiotomy intra-aortic balloon pumping. J Thorac Cardiovasc Surg 1977;74:709. Pennington DG, Bernhard WF, Golding LR, Berger RL, Khuri SF, Watson JT. Long-term follow-up of postcardiotomy patients with profound cardiogenic shock treated with ventricular assist devices. Circulation 1985;72(Suppl 2):21626. Pennington DG, Samuels LD, Williams G, et al. Experience with the Pierce-Donachy ventricular assist device in postcardiotomy patients with cardiogenic shock. World J Surg 1985;9:37-46. Pennington DG, Merjavy JP, Swartz MT, et al. The importance of biventricular failure in patients with postoperative cardiogenic shock. Ann Thorac Surg 1985;39:16-26. Termuhlen DG, Swartz MT, Pennington DG, et al. Predictors for weaning patients from ventricular assist devices (VADs). ASAIO Trans 1987;33:683. Bernhard WF, Clay W, Gernes D, et al. Temporary and permanent left ventricular bypass: laboratory and clinical observations. World J Surg 1985;9:54. Pennington DG, McBride LR, Kanter KR, Swartz MT, Miller LW. The effect of acute perioperative myocardial infarction on survival of postcardiotomy patients supported with ventricular assist devices. Circulation (in press). Kanter KR, Swartz MT, Pennington DG, et al. Renal failure in patients with ventricular assist devices. ASAIO Trans 1987; 33:426-8. Joist JH, Pennington DG. Platelet reactions with artificial surfaces. ASAIO Trans 1987;33:341-4. McBride LR, Ruzevich SA, Pennington DG, et al. Infectious complications associated with ventricular assist device support. ASAIO Trans 1987;33:201. Ruzevich SA, Kanter KR, Pennington DG, Swartz MT, McBride LR, Termuhlen DG. Long-term follow-up of survivors of postcardiotomy circulatory support. ASAIO Trans 1987;33(Suppl 3):177.