Extracorporeal Membrane Oxygenation for Infant Postcardiotomy Support: Significance of Shunt Management

Transcription

1 Extracorporeal Membrane Oxygenation for Infant Postcardiotomy Support: Significance of Shunt Management James J. Jaggers, MD, Joseph M. Forbess, MD, Ashish S. Shah, MD, Jon N. Meliones, MD, Paul M. Kirshbom, MD, Coleen E. Miller, MSN, and Ross M. Ungerleider, MD Pediatric Cardiovascular Program, Duke University Medical Center, Durham, North Carolina Background. After repair of complex congenital heart defects in infants and children, postcardiotomy cardiac failure requiring temporary circulatory support can occur. This is usually accomplished with the use of extracorporeal membrane oxygenation (). management of patients with single-ventricle physiology and aorto-pulmonary shunts can be particularly challenging. We retrospectively reviewed our experience with postcardiotomy support with particular attention to those children with single-ventricle palliation. Methods. Thirty-five consecutive children (age 1 to 820 days, median 19 days) out of 1,020 patients (3.4%) required mechanical support () after repair of congenital cardiac lesions from February 1994 to April Twenty-five patients underwent two ventricle repairs and 10 patients had single-ventricle palliation. Various parameters analyzed included strategies of shunt management, presence of presupport cardiac arrest, and timing of support initiation. Results. Overall hospital survival for these 35 patients was 61%. There were four additional late deaths. Hospital survival was the same for those patients in whom support was initiated for failure to wean from cardiopulmonary bypass in the operating room versus those patients in whom support was initiated after successful separation from cardiopulmonary bypass (6 of 10 vs 15 of 25 or 60% survival). In those patients with shuntdependent pulmonary circulation, survival was significantly improved in those patients in which the aortopulmonary shunt was left open (4 of 5 with open shunt vs 0 of 4 with occluded shunt (p 0.048). Conclusions. The ability to readily implement postcardiotomy support is vital to the management of children with complex congenital cardiac disease. Overall survival can be quite satisfactory if support is employed in a rational and expedient manner. In patients with singleventricle physiology and aorto-pulmonary shunts, leaving the shunt open during the period of support can result in markedly improved outcomes. (Ann Thorac Surg 2000;69: ) 2000 by The Society of Thoracic Surgeons After repair of complex cardiac defects in infants and children, postcardiotomy cardiac failure can occur. The most commonly employed technique of mechanical life support for postcardiotomy support in children is extracorporeal membrane oxygenation (). In many centers, can be deployed rapidly and has been shown to improve survival in an otherwise dismal situation. Successful hospital survival can be achieved in 40% to 50% in most series [1 9]. Management of patients with single-ventricle physiology and aorto-pulmonary shunts can be particularly challenging, and in some centers is a relative contraindication for postcardiotomy support with. Current practice suggests that the shunt should be either completely or partially occluded while on for fear that if the shunt is left open there will be significant overcirculation to the lungs. However, Presented at the Forty-sixth Annual Meeting of the Southern Thoracic Surgical Association, San Juan, Puerto Rico, Nov 4 6, Address reprint requests to Dr Jaggers, Division of Thoracic Surgery, Duke University Medical Center, Box 3474, Durham, NC 27710; jagge003@mc.duke.edu. complete or partial occlusion of the shunt may result in severe pulmonary ischemia or in shunt thrombosis. Because of poor results with occluding the shunt in this situation, we adopted a policy in 1997 of leaving the shunt open during and increasing the flow rate to satisfy the requirements of both the systemic and pulmonary circulation. In this review, we report our experience with postcardiotomy support with particular attention to those children with singleventricle physiology and aorto-pulmonary shunts. Material and Methods Data were obtained in a retrospective manner from 35 patients who required postcardiotomy, mechanical support from February 1994 to April This group represents 3.4% of a total population of 1,029 cardiopulmonary bypass (CPB) patients operated on within this time period at our institution. The patients represented in this series are displayed in Table 1. Twenty-five of these patients had two-ventricle physiology and 10 patients 2000 by The Society of Thoracic Surgeons /00/$20.00 Published by Elsevier Science Inc PII S (00)

2 Ann Thorac Surg JAGGERS ET AL 2000;69: FOR POST CARDIOTOMY SUPPORT IN CHILDREN 1477 had single-ventricle physiology. Nine of the 10 patients with single-ventricle physiology had a modified B-T shunt and 1 patient had a bidirectional Glenn anastamosis. Of the 10 patients with single-ventricle physiology, 5 were hypoplastic left heart syndrome (HLHS) and had undergone Norwood stage I procedure. The decision to place a patient on was made by the attending surgeon, cardiologist, and intensivist, and was based on clinical findings of persistent hypotension, acidosis, hypoxemia, or pulmonary hypertension in the postoperative period. In the operating room, the decision to utilize was the decision of the surgeon and anesthesiologist and was based on the inability to wean or low cardiac output despite appropriate inotropic support and attempts to balance systemic and pulmonary circulation in the case of single-ventricle patients. The circuit that we use is based on the circuit described by Bartlett and associates [10] and consists of a venous line that leads to a 30-mL silicone bladder. This is a small compliant reservoir that allows servo-regulation of the roller pump. From the bladder, blood is drawn by the roller pump (Sorin CAPs System, Morindo, Italy) and pumped into a membrane oxygenator (Avecor Cardiovascular, Inc, Plymouth, MN). The blood then proceeds to the stainless steel heat exchanger and from there to the arterial inflow cannula. There is a bridge from the arterial to venous line for recirculation. In-line blood gas monitoring is also utilized. Anticoagulation is initiated with 100 U/kg body weight of heparin and maintained to an activated clotting time of 180 to 200 seconds. In those patients who were cannulated in the operating room, every attempt was made to separate from cardiopulmonary in order to reverse heparin effects and achieve hemostasis before reheparinizing and transferring to the circuit. In those patients who were successfully weaned and who later required support, cannulation was carried out via repeat sternotomy in all but 2 patients. In these 2 patients cannulation was carried out via the right common carotid artery and right internal jugular vein, as is our practice for noncardiac neonatal. Hematocrit was maintained at greater than 40% and platelet counts of greater than 100,000. Flow rates of 100 to 200 ml/kg/min were maintained depending upon the physiology and the mixed venous oxygen saturation. Venting of the left atrium in order to prevent left-sided distention was performed in only 1 patient with two-ventricle anatomy and, of course, is not usually necessary in those with single-ventricle anatomy. Inotropic support was decreased slowly to maintain mean blood pressure at 40 to 50 mm Hg. High levels of positive end-expiratory pressure (PEEP) were not routinely employed, as this may artificially elevate central venous pressures and inhibit venous return to the pump. In those patients who had a shunt that was left open to perfuse the lungs, the ventilation was continued at a set rate of 10 to 14 breaths per minute and tidal volume of 10 to 15 ml/kg. Serial serum lactate levels, blood gases, and mixed venous oxygen saturation are typically followed to assess the efficacy of the. For the most part, patients who were on were heavily sedated with benzodiazepine and narcotic analgesia. Paralytic agents were not routinely employed. Weaning from was typically accomplished by slowly reducing the flow from the device while optimizing filling pressure in a fashion similar to weaning from cardiopulmonary bypass in the operating room. This generally can be accomplished in only a few minutes. Prolonged periods of reduced flow are avoided. If separation from the circuit is accomplished, cannulas are clamped and the pump is left to recirculate at a low flow across the bridge. The cannulas are removed after a period of 1 to 2 hours of hemodynamic stability. Patients charts were reviewed for significant events and complications including: hospital mortality, late mortality, late functional status, maximal serum lactate levels, pre- base deficit, time on, time to initiation of, indication for, the presence or absence of a systemic to pulmonary shunt, and the management of that shunt. Mean follow-up time was 18 months. Statistical Analysis Patients data were presented as a range of the mean or as a percentage of patients in a group. The statistical package (Statistica; Statsoft, Tulsa, OK) was used for all examinations. Differences between mean values were analyzed using Student s t test or Fisher s exact test where appropriate. All other dichotomous variables were analyzed using a 2 test. Differences were considered significant at p less than Results The mean age of patients requiring in this series was 89 days, with a median age of 19 days. Median weight was 4.7 kg; mean weight was 3.9 kg. The most common indication for was low cardiac output. Other indications included pulmonary hypertension, refractory hypoxemia, cardiopulmonary arrest, and arrhythmia. Overall hospital survival for this difficult group of patients was 21 of 35 (61%). Of these 21 patients, 4 died during the follow-up period, for an overall survival of 49%. Only 2 of 17 have any identifiable neurologic dysfunction at a mean follow-up of 18 months. Twelve patients were cannulated for in the operating room. Ten of those 12 patients were placed on because of failure to wean. Of these 10 patients, 6 of 10 (60%) survived. Eight of the 12 patients who were placed on in the operating room survived, which tended toward significant improvement in survival over those who were placed on in the intensive care unit. Of those patients who were placed on after weaning, 15 of 25 (60%) survived to be discharged from the hospital and 11 of 25 (44%) are alive long term. Thus, there was no difference in our experience in hospital outcome for patients placed on with respect to whether or not they could be weaned in the operating room; hospital survival was 60% for each group. There were 2 patients who

3 1478 JAGGERS ET AL Ann Thorac Surg FOR POST CARDIOTOMY SUPPORT IN CHILDREN 2000;69: Table 1. Patient Data Patient Diagnosis Age (days) Weight (kg) Procedure Phase of Indication for Time to (hours) Time on (days) Mortality Late Neurologioc Function 1 TOF Repair TOF OR Immediate Failure to wean 0 8 Survived Normal 2 HLHS Norwood ICU 8 h PO Low CO cardiac arrest ICU death 1 day 3 DORV/coarctation ASO; arch repair PICU 18 Low CO 18 6 Hospital death 60 days 4 HLHS 2 3 Norwood ICU 9 h PO Low CO 9 7 Late death 6 months 5 AVSD AVSD repair ICU 22 h PO Cardiac arrest/arrhythmia 22 5 ICU death 5 days OR after CPB Low CO 1 4 Survived Normal 6 DORV SubAS/PS/ IAA AS, IAA repair ASD/VSD closure Neurologically impaired 7 IAA/VSD LVOTO repair ICU 8 h PO Low CO 8 5 Survived Neurologically disabled 8 Heterotaxy/SV 60 5 BDG PO ICU 2 h Hypoxia/arrhythmia 2 15 ICU death 15 days 9 DORV; VSD 90 5 Repair DORV ICU POD 24 Low CO/cardiac arrest 24 4 Survived Normal 10 TOF with MAPCAS RVOT patch/asd repair OR 1.5 h after CPB Low CO ICU death 9 days 11 TAPVR (mixed) 30 3 Repair TAPVR 8 h PO ICU Pulmonary hypertension 8 7 Survived Normal 12 LVOTO Ross-Kono OR immediate Failure to wean 0 5 Survived Normal Normal 13 TOF RV-PA conduit; ASD; PA stents OR immediate Failure to wean 0 5 ICU death 5 days 14 TOF TOF repair ICU PO 2 days Low CO/seizures 48 2 ICU death 2 days 15 TOF 70 3 TOF repair OR after CPB Low CO/arrhythmia 1 5 Survived Normal 16 Tricuspid atresia A-P shunt OR immediate Failure to wean ICU death 2 days 17 Mitral MV repair OR immediate Failure to wean 0 5 Survived Normal insufficicency after AVSD repair 18 TAPVC/heterotaxy Repair TAPVC/ AP shunt ICU 12 h Low CO 12 8 ICU death 18 days 19 TO/AVSD Repair TOF ICU PO 30 h Bi-vent dysfunction r ICU death AVSD 61 days 20 Truncus arteriosus Truncus repair PICU 2 h Cardiac arrest 2 10 Late death 2 years Continued

4 Ann Thorac Surg JAGGERS ET AL 2000;69: FOR POST CARDIOTOMY SUPPORT IN CHILDREN 1479 Table 1. Continued Patient Diagnosis Age (days) Weight (kg) Procedure Phase of Indication for Time to (hours) Time on (days) Mortality Late Neurologioc Function 21 Mitral insufficiency MVR; VSD closure 22 AVSD/SV 6 8 Arch recon; DKS; PV stent; AP shunt 23 DORV/coarctation Rastelli; arch reconstruction 24 TGA ASO OR immediate Failure to wean 25 Pulmonary vein atresia Repair TAPVC/ atresia OR; off CPB Low CO 0 3 Survived Mild developmental delay PICU 4 h Low CO 4 5 Survived Normal ICU8hPO LowCO 8 3 Survived Normal ICU 24 h Pulmonary hypertension 26 ALCAPA 90 4 Repair ALCAPA OR/LV AD Failure to wean 27 HLHS Norwood 28 TAPVC TAPVC repair OR post-cpb Respiratory failure 0 4 Survived Normal 24 9 ICU death 9 days 0 2 ICU death 3 days ICU 15 h PO Low CO 15 5 ICU death 5 Days 0 1 Survived Normal 29 TOF TOF repair PICU 6 h Low CO 6 3 Late death 30 TGA 7 3 ASO OR immediate Failure to wean 0 3 Survived Normal 31 Coarct; VSD (multiple) 10 3 Coarct repair; PA band 32 HLHS 4 3 Norwood 33 Heterotaxy/SV A-P shunt OR to wean 34 HLHS 4 4 Norwood 35 IAA/VSD DSK/Arch repair/ RV/PA conduit ICU5hPO LowCO 5 8 Survived Normal OR immediate Failure to wean 0 11 ICU death 26 days Failure to wean 0 4 Survived Normal PICU 2 h Cardiac arrest 2 5 Survived Normal ICU 7 h PO Low CO 7 7 Late death Severe neurologic dysfunction ALCAPA anomalous left coronary artery from the pulmonary artery; AP shunt aortopumonary shunt; ASO arterial switch procedure; AVSD atrioventricular septal defect; Coarct coarctation of the aorta; CPB cardiopulmonary bypass; DSK Damus-Stansel-Kaye ; DORV double-outlet right ventricle; HLHS hypoplastic left heart syndrome; IAA interupted aortic arch; ICU intensive care unit Low CO low cardiac output; OR operating room; PA band pulmonary artery band; PV pulmonary veins; SV single ventricle; TAPVR total anomalous pulmonary venous connection; TGA transposition of the great arteries; TOF tetralogy of fallot; VSD ventricular septal defect.

5 1480 JAGGERS ET AL Ann Thorac Surg FOR POST CARDIOTOMY SUPPORT IN CHILDREN 2000;69: Table 2. Analysis of Possible Risk Factors for Hospital Death Variable Survivors (n 21) Nonsurvivors (n 14) p Value Age (days) Weight (kg) Pre- base deficit (mmol/l) Maximal lactate level (mmol/l) Time to (hours) Time on (days) Table 4. Incidence of Complications in Survivors and Nonsurvivors Complications Hospital Survivors (21) Nonsurvivors (14) p Value Neurologic 10 (45%) 8 (57%) NS Coagulopathy 8 (36%) 7 (50%) NS Infection 8 (36%) 3 (21%) NS Renal failure 2 (9%) 7 (50%) 0.02 Pulmonary 6 (27%) 6 (42%) NS Thrombotic 4 (18%) 3 (21%) NS Table 3. Dichotomous Variables Compared With Overall Survival Variable Hospital Survival (%) p Value Shunt left open on 4/5 (80%) Initiation in ICU vs OR 13/23 (56%) 0.20 for cardiac arrest 3/6 (50%) 0.63 extracorporeal membrane oxygenation; care unit; OR operating room. ICU intensive were initially cannulated in the operating room for left ventricular assist device. One was a 21-day-old patient with transposition of the great arteries and postcardiotomy left ventricular dysfunction, and the other patient was a child with anomalous left coronary artery from the pulmonary artery (ALCAPA) in which postcardiotomy ventricular assist device support was planned before surgery. Both patients required conversion to for pulmonary dysfunction and right ventricular failure within the first 3 days of support. One of these has survived long term. The other died early on from severe coagulopathy and intracranial hemorrhage. Nine of the 35 patients had aorto-pulmonary shuntdependent pulmonary circulation (Table 3). Of those 9 patients, 4 survived to hospital discharge. Of these 9 patients, the first 4 in the series were placed on with the shunt occluded in order to prevent pulmonary over-circulation and ensure systemic perfusion. Of these 4 patients, none survived. Two had documented pulmonary infarcts at autopsy. In the next 5 patients with shunt-dependent pulmonary circulation, the shunt was left open to perfuse the lungs during support. flow rates were doubled to achieve perfusion of approximately 200 ml/kg/min. Patients were ventilated normally with respect to standard protocols for the management of infants after Norwood stage I procedure. In this group, 4 of 5 survived to discharge from the hospital. The only mortality in this group was an infant who underwent Norwood for HLHS and had preoperative renal insufficiency and renal hypoplasia with a creatinine of 2.0 mg/dl. His death was related to renal insufficiency, despite attempts at dialysis. There was also one late death in this group in a child who sustained neurologic impairment during the initial and subsequently died at the time of his bidirectional Glenn shunt with severe mesenteric and renal ischemia. There were 6 patients in this series who were cannulated for while undergoing cardiopulmonary resuscitation for postoperative cardiac arrest (Table 3). All of these patients were cannulated in the intensive care unit via a repeat sternotomy. Three out of 6 of these patients survived to discharge from the hospital, and 1 of the 3 survivors died at 16 months postoperatively. Both of the surviving patients were neurologically normal. The infant who died late was also neurologically normal but had severe cardiac dysfunction after repair of truncus arteriosus and died while awaiting cardiac transplantation. Table 4 displays the incidences of major complications in survivors versus nonsurvivors. There were no significant differences between the two groups except in renal dysfunction. Nonsurvivors were more likely to have suffered significant renal injury either during or in the postoperative period. Analysis of patient variables was carried out to try to determine if any variable seemed to correlate with improved survival (Table 2). Pre- base deficits and maximum lactate levels were chosen as possible indicators of the severity of poor tissue perfusion and compromised cardiac output. Neither of these variables, however, correlated with survival. Mean duration on Table 5. Review of Series in Literature Author/Year No. Patients Survival Walters/ /66 (58%) Black/ /25 (60%) Weinhaus/ /12 (42%) Del Nido/ (all cardiac arrest) 6/11 (55%) Kulic/ /64 (33%) Rogers/ /10 (70%) Langley/ /9 (22%) Ziomek/ /24 (58%) Raithel/ /65 (35%) Klein/ /39 (56%) Thuys/1998 (VAD only) 34 14/34 (41%) Jaggers/current series 35 21/35 (60%) Total /394 (47%)

6 Ann Thorac Surg JAGGERS ET AL 2000;69: FOR POST CARDIOTOMY SUPPORT IN CHILDREN 1481 for all patients was 5.6 days (median 5 days, range 0.7 to 16 days). The mean duration on for hospital survivors was 4.89 days, (median 4.4 days, range 1 to 10 days). Mean duration on for nonhospital survivors was significantly longer, 6.24 days (median 6 days, range 0.7 to 16 days; p 0.04). Comment Since the success of neonatal for respiratory distress syndrome in the 1980s, many groups have reported their experiences in the use of for postcardiotomy circulatory support (Table 5). Survival rates of nearly 50% can be expected. It is clear that myocardial dysfunction after complex congenital cardiac s are often reversible. can reverse the severe metabolic consequences of low cardiac output and tissue hypoxemia. Support of the infant or small child begins with pharmacologic inotropic therapy. If this fails to restore hemodynamic stability, data from this and previous reported series support the initiation of mechanical support with or other ventricular assist device. As with any form of circulatory support, results may be directly related to the threshold for which therapy is instituted. Many groups have reported a worse prognosis if patients were placed on in the operating room as a result of failure to wean. In this series, we found that of those patients cannulated in the operating room, 66% survived to hospital discharge, whereas 53% of those placed on in the intensive care unit survived. Furthermore, there was no difference in hospital survival between the group of patients who were placed on because they could not be weaned from CPB versus the remaining patients who were placed on after being first weaned, with 60% survival for each group. This experience validates the use of in properly selected instances when infants cannot be separated. In this study, the only variables that we were able to identify as predictive of survival were lack of significant renal impairment and leaving the shunt open during in those patients with shunt-dependent pulmonary circulation. The use of for postcardiotomy support in neonates and infants with aorto-pulmonary shunts is controversial, especially in patients with hypoplastic left heart syndrome (HLHS). While there is no consensus regarding the management of the shunt during the period, the popular convention has been to either occlude the shunt completely or partially in order to limit the pulmonary blood flow while on. There has been widespread concern that if the shunt is left open, there will be excessive pulmonary blood flow that may steal flow away from the systemic circulation and produce pulmonary injury. Some groups have even stated that the presence of an aorto-pulmonary shunt may be a contraindication to. Our experience in the early 1990s with occlusion of the shunt in infants placed on after Norwood stage I procedure was indeed dismal and supported this contention. We observed 2 patients with HLHS and a shunt-dependent pulmonary circulation after the Norwood who were unable to be weaned from because of severe pulmonary dysfunction, despite vigorous cardiac function. It became clear that the lungs had sustained a severe injury while on with the shunt occluded. These children ultimately died shortly after being taken off on postoperative days 5 and 7, respectively. An autopsy revealed severe ischemia to the lungs without emboli or infection. This experience prompted a study from our laboratory regarding the effects of no antegrade pulmonary blood flow during CPB (total CPB), on parameters of respiratory function, and pulmonary vasoreactivity [11]. In this study, pulmonary vascular resistance, A-a gradient, and pulmonary compliance were all adversely affected by no antegrade flow to the lungs during CPB. It may be that reopening an occluded shunt just before coming off sets the stage for an ischemia-reperfusion injury. In our laboratory study, we were able to measure significant injury to pulmonary endothelial function in the group maintained on CPB without antegrade pulmonary blood flow, suggesting that pulmonary blood flow (beyond what is available from the bronchial circulation) is necessary to preserve normal pulmonary vascular endothelial cell function. In June 1997, after a Norwood stage I procedure for HLHS, a neonate suffered shunt thrombosis and required emergency postcardiotomy support with. Because of the information generated by our previous clinical experience and from our laboratory investigation, we elected to alter our existing protocol for this patient. We left the shunt open and increased our flow rate while on to 200 ml/kg/min, presuming that in that manner we could support both the pulmonary and systemic circulation. We ventilated the infant in the standard manner we would use after a Norwood procedure in order to balance systemic and pulmonary blood flow. This infant made an uneventful recovery, has since undergone a Fontan procedure, and is functionally normal. In this article, we report on an additional 4 patients who were managed in a similar fashion. In all cases, the infants had no problem balancing their circulation while on as long as the circuit provided adequate flow. This experience supports the concept that a patient with single-ventricle, shunt-dependent circulation after a Norwood stage I procedure has an increased cardiac output requirement compared with an infant with a normal circulation. The need for a double-cardiac output (one for the lungs and one for the body) is occasionally more than the infant can produce after surgery, and they enter the spiral of hemodynamic collapse that is often described after this procedure. If provided with a normal cardiac output for their needs, these infants can balance their circulation without difficulty. In fact, we have found that intentional hypoxemia and hypercarbia are absolutely unnecessary and potentially harmful. We ventilate normally and have experienced no problems from pulmonary overcirculation. is not the only option for infants and children

7 1482 JAGGERS ET AL Ann Thorac Surg FOR POST CARDIOTOMY SUPPORT IN CHILDREN 2000;69: who require postcardiotomy support. A recent report suggests that mechanical support that does not utilize an oxygenator is also an option if the patient s own lungs can be used [12]. In this series, the most common lesion supported in this manner was HLHS (12 patients). Because this approach does not involve an oxygenator, the shunts of these patients with shunt-dependent pulmonary circulation must be left open to perfuse the lungs during the period of support. In this series, 7 of 12 patients could be weaned from support, but only 3 of 12 were discharged from the hospital. The elimination of the oxygenator may result in less inflammatory effects related to the device and requires less anticoagulation, which makes management of perioperative bleeding much easier. The lower hospital survival rate reported by Thuys and associates [12] might related to the timing of intervention because, in our experience, the mortality rate for infants placed on who have had a Norwood stage I procedure is near 100% if is instituted during a cardiac arrest. The simplicity of this system has led us to consider early intervention with ventricular assist device in those patients with low cardiac output after the Norwood procedure. Despite improvements in survival for patients requiring postcardiotomy support, the rate of complications remains high. In this series, there did not seem to be any difference in major complications between survivors and nonsurvivors except for significant renal dysfunction ( p 0.02) present in nonsurvivors. This finding is consistent with other reports [5]. The actual rate of neurologic complications is difficult to measure because it may be that some neurologic complications may not have been manifest in those patients who died while still on support. It is, however, gratifying that in the patients who survived long term, only two suffered obvious neurologic deficit. Bleeding is also a very common complication in all series, especially in those patients who require for failure to wean, because the heparin is unable to be reversed. In our practice, we make every attempt to wean long enough for heparin to be reversed and hemostasis obtained. We find that this lessens the severity of bleeding in the postoperative period. When this is not possible, we have found it reasonable to withhold heparin until bleeding is under control, with a back-up circuit (or oxygenator) available in case it is urgently required. Our more recent experience with ventricular assist device for infants with aorto-pulmonary shunts has enabled us to control bleeding in this group while still providing circulatory support. Although our hospital survival of 61% is comparable with other reported series, it is our impression from reviewing this experience that future results might be improved by altering postcardiotomy support strategies in the following ways. (1) Failure to wean is not a contraindication to the use of support and in appropriately selected cases. Support that is initiated in a timely manner before hemodynamic collapse has a better chance of immediate success and improved long-term outcome. (2) The biggest impediment to early initiation of is circuit set-up time. We think that it is prudent to keep a circuit available and primed with crystalloid solution for rapid deployment. (3) Infants with shunt-dependent circulation can have excellent outcome with if the shunt is left open and the infant is ventilated normally. In these instances, flow rates should be increased to satisfy both the systemic and pulmonary flow requirements. With normal ventilation, the infant will balance his own systemic-pulmonary circulations. Treatment with hypoxia and/or hypercarbia is not necessary and may have a negative effect on long-term outcome. (4) The use of ventricular assist device (we have coined the term NOMO-VAD for no membrane oxygenator-ventricular assist device) for infants after Norwood stage I procedure has significant advantages because the ventricular assist device supplies full cardiopulmonary support by utilizing the patient s own lungs for respiration and oxygenation. In these cases, postoperative bleeding will be more easily controlled because systemic heparinization is not necessary. Summary The timely application of for postcardiotomy support can result in improved survival. If used appropriately and expeditiously before end-organ injury or cardiac arrest has occurred, results can be excellent. Patients with shunt-dependent pulmonary circulation have been problematic using conventional protocols that include some form of shunt occlusion. Despite the limitations of this small review, the results of our experience, corroborated by data from our animal laboratory, support the strategy of leaving the shunt open to perfuse the lungs during the period and increasing flows to satisfy the needs of both the systemic and pulmonary circulation. This experience helps to redefine the requirements for treating patients after a Norwood stage I procedure with respect to ventilation and underscores the need these infants have for an adequate cardiac output. References 1. Ziomek S, Harrell JE, Fasules JW, et al. Extracorporeal membrane oxygenation for cardiac failure after congenital heart. Ann Thorac Surg 1992;54: Weinhaus L, Canter C, Noetzel M, McAlister W, Spray TL. Extracorporeal membrane oxygenation for circulatory support after repair of congenital heart defects. Ann Thorac Surg 1989;48: Walters HL, Hakimi M, Rice MD, Lyons JM, Whittlesey GC, Klein MD. Pediatric cardiac surgical : multivariate analysis of risk factors for hospital death. Ann Thorac Surg 1995;60: Rogers AJ, Trento A, Siewers RD, et al. Extracorporeal membrane oxygenation for post-cardiotomy cardiogenic shock in children. Ann Thorac Surg 1989;47: Raithel SC, Pennington G, Boegner E, Fiore A, Weber TR. Extracorporeal membrane oxygenation in children after cardiac surgery. Circulation 1992;86:II Kulik TJ, Moler FW, Palmisano JM, et al. Outcomeassociated factors in pediatric patients treated with extracorporeal membrane oxygenator after cardiac surgery. Circulation 1996;94:II Langley SM, Sheppard SV, Tsang VT, Monro JL, Lamb RK. When is extracorporeal life support worthwhile following

8 Ann Thorac Surg JAGGERS ET AL 2000;69: FOR POST CARDIOTOMY SUPPORT IN CHILDREN 1483 repair of congenital heart disease in children? Eur J Cardiothorac Surg 1998;13: Del Nido PJ, Dalton HJ, Thompson AE, Siewers RD. Extracorporeal membrane oxygenator rescue in children during cardiac arrest after cardiac surgery. Circulation 1992;86:II Black MD, Coles JG, Williams WG, et al. Determinants of success in pediatric cardiac patients undergoing extracorporeal membrane oxygenation. Ann Thorac Surg 1995;60: Bartlett RH, Andrews AF, Toomasian JM, Haiduc NJ, Gazzaninga AB. Extracorporeal membrane oxygenation for newborn respiratory failure: forty-five cases. Surgery 1982;92: Chai PJ, Williamson A, Lodge AJ, et al. Effects of ischemia on pulmonary dysfunction after cardiopulmonary bypass. Ann Thorac Surg 1999;67: Thuys CA, Mullaly RJ, Horton SB, et al. Centrifugal ventricular assist in children under 6 kg. Eur J Cardiothorac Surg 1998;13: DISCUSSION DR ROSS M. UNGERLEIDER (Durham, NC): This is a difficult group of patients and obviously the numbers would be small, but I think it is impressive to show that leaving the shunt open, despite conventional thought, really works quite well. Because you are using the patient s lungs, do you need an circuit? Would you be able to accomplish the same thing perhaps using a ventricular assist device with the shunt open and just using the lungs as your membrane oxygenator? DR JAGGERS: That is a good point, and in fact, Tom Karl s group in Melbourne, Australia, has published I believe last year a series of their patients in which they used VAD support primarily. In that series, they had 12 patients who were hypoplastic left heart syndrome, and their VAD support worked quite well in this group. They simply left the shunt open and ventilated the patient normally and did not have significant technical problems in that group. And since that time, we have also had the same thought and have had the opportunity to use VAD support in a hypoplastic left heart syndrome that had undergone stage I palliation and required postcardiotomy support. This child has survived and is doing well. So that is certainly an option, and also significantly improves the risks of postoperative bleeding and complications related to the assist device.