Extracorporeal membrane oxygenation for the circulatory support of children after repair of congenital heart disease

J THORAC CARDIOVASC SURG 1990;100:498-505

Extracorporeal membrane oxygenation for the circulatory support of children after repair of congenital heart disease We have treated 39 infants and children with congenital heart disease with extracorporeal membrane oxygenation during the past 5 years. Thirty-six were treated for low cardiac output or pulmonary vasoreactive crisis after repair of congenital heart defects. Twenty-two (61 %) survived. Most patients were cannulated from the neck via the right internal jugular vein and the right common carotid artery. Six patients were cannulated from the chest, including three who had separate drainage of the left side of the heart with a left atrial cannula. Two of these patients survived and were the only survivors of the nine patients cannulated in the operating room because they could not be weaned from cardiopulmonary bypass after open cardiac operations. We also reviewed 312 patients (the predictor study series) having open cardiac operations before the availability of extracorporeal membrane oxygenation; 27 of these patients died. Data were collected at 1 and 8 hours postoperatively to determine if any parameters might predict early mortality. With these parameters used as criteria, patients who went on extracorporeal membrane oxygenation were as sick as those who died before extracorporeal membrane oxygenation was available. The most common complication was bleeding related to heparinization. The mean transfusion requirement in survivors was 1.50 ± 1.13 m1jkgjhr, 5.63 ± 7.0 mljkgjhr in the nonsurvivors, and 7.46 ± 8.29 m1jkgjhr in those cannulated in the operating room because they could not be weaned from bypass. Four children had intracranial hemorrhage, and two of them died. There was one late death. Nine of the 22 survivors are entirely normal. All survivors who do not have Down's syndrome are considered to have normal central nervous system function. We conclude that extracorporeal membrane oxygenation can improve survival in patients with both pulmonary artery hypertension and low cardiac output after operations for congenital heart disease.

Michael D. Klein, MD, Kenneth W. Shaheen, MD , Grant C. Whittlesey, CCP, William W. Pinsky, MD, and Eduardo Arciniegas, MD, Detroit, Mich .

Extracorporeal membrane oxygenation (ECMO) has been used successfully to treat newborn infants with respiratory failure that is unresponsive to maximum ventilatory and pharmacologic support. We have also-used ECMO to treat children after repair of congenital heart From the Departments of Surgery (Pediatric General Surgery and CardiovascularSurgery) andPediatrics(Cardiology), WayneState University School of Medicine, and the Children's Hospital of Michigan, Detroit, Mich. Supported inpart by grants from the VarietyClub of Detroit. Received for publication June 29, 1988. Accepted for publication Oct. 30, 1989. Address for reprints: Michael D. Klein, MD, Department ofSurgery, Cllildren'sHospitalof Michigan,3901Beaubien Blvd., Detroit, MI 48201. 12/1 /18207

498

defects who have persistent low cardiac output or lifethreatening pulmonary vasoreactive crisis (reversible, severe pulmonary artery hypertension) unresponsive to other therapy. Patient population and methods "Predictor" patient group. To assist in the evaluation of candida tes for EC MO, we reviewed 312 children having operations with card iopulmonar y bypass for congenital heart disease between Janu ary 1984 and Jun e 1985. Th e medical records were reviewed and data collected for I hour, 8 hours, 16 hours, and 24 hours after cardiopulmonary bypass to determine if any parameters might predict early postoperative death . Twent y-seven patien ts (8.7%) with complex intracardiac defects died of low card iac outpu t during the early postoperative period. Among these were seven patient s with hypoplastic left heart syndrome, seven with complete at rioventricular septal defect, five with truncus arteriosus, three with complex transposition of the great arteries, two with total anoma lous

Volume 100 Number 4 October 1990

ECMO after repair of congenital heart disease

499

Table I. Comparison ofpatients placed on ECMO with "predictor" study patients dying after open repair of congenital heart defects Hour 8

Hour 1 ECMO Parameter

Age (rna) Weight (kg) Circulatory arrest time (min) Aorticcrossclamp time (min) Heart rate (beats/min) Systolic blood pressure (mm Hg) Left atrial pressure (mm Hg) Right atrial pressure (mm Hg) Urineoutput (ml/kg/rnin) Dopamine (J.Lg/kg/min) Fioz Paoz (mm Hg)

Ventilator rate pH

patients" 13.7 7.15 46.0 97.9 161 108 15 16 1.8 5.83 0.71 169 22.3 7.36

± 6.2 ± 2.41 ± 4.3 ± 12.2 ± 7 ± 17

NS NS

om NS

0.05 0.01

±2

NS

± 3 ± 1.04 ± 2.83 ± 0.14 ± 94 ± 4.2 ± 0.05

0.01 0.05 0.01 0.01

NS NS

0.01

Predictor study

ECMO

patiemsi

patients"

13.7 6.4 69.1 106.6 149 86 17 24 1.12 14.13 0.9 168 25 7.26

± 26.7 ± 5.3 ± 31.9 ± 49.1 ± 21 ± 23 ±5 ± 7 ± 0.90 ± 8.38 ± 0.1 ± 134 ± 6 ± 0.11

NA NA NA NA 145 92 16 19 1.1 8.60 0.63 106 26.5 7.41

p<

± 27 ± 12

NS NS

± 2 ± 6 ± 0.67

0.01

± 3.18 ± 0.18 ± 22 ± 6.1 ± 0.09

0.01

NS NS NS

0.01

NS NS

Predictor study

patients'[

NA NA NA NA 144 83 21 21 1.51 19.4 0.77 75 26.1 7.33

± 41 ± 24

± 5 ± 9 ± 0.92 ± 10.1 ± 0.27 ± 33 ± 8.8 ± 0.18

NS, Not significant; NA, not applicable. • All patients. tNonsurviving patients.

pulmonary venous return, and one each with tricuspid atresia, complicated atrial septal defect (ostium primum, cleft mitral valve), and complicated tetralogy of Fallot (large atrial septal defect, left pulmonary artery stenosis). Of these, three patients could not come off bypass and died in the operating room. Twelveof the remaining 24 died by 8 hours postoperatively, and only five were alive 16 hours after operation. For this reason there are very few patients available with data at 16 and 24 hours postoperatively, and the analysis is, therefore, limited to data collected 1 and 8 hours postoperatively. Since many ofthe patients who lived had relatively simple anatomic problems that were not associated with mortality, we also analyzed a subgroup of survivors with the same diagnoses as the patients who died. Thus patients operated on for simple atrial septal defect, simple ventricular septal defect, and simple tetralogy of Fallot were not included in the statistical comparisons. We found statistically significant differences between patients who lived and patients who died for weight, circulatory arrest time, aortic crossclamp time, heart rate, systolic blood pressure, left atrial pressure, right atrial pressure, urine output, isoproterenol dose, dopamine dose, colloid administered, crystalloid administered, blood administered, inspired oxygen fraction (Fio-), arterial oxygen tension (Pao-), pH, and ventilator rate. Although not all statistically significant differences translated into a rule for predicting mortality, some differences were helpful in making a decision as to whether a patient should be placed on ECMO. Sixty-five percent of the nonsurvivors were ::s9 months old, whereas only 20% of the survivors were this young. Seventy percent of the nonsurvivors were on an Fioj of 2:0.8 at 1 hour, whereas only 9% of the survivors were on the same Fio-, Our best guidelines, however, combined two indicators of cardiac output, dopamine dose and urine output. Ninety-one percent of the nonsurvivors at 1 hour postoperatively had both a urine output of -s 1.2 ml/kg/hr and a dopamine infusion of 2:5 J,Lg/kg/min compared with only 3.1% of all the survivors. At 8 hours postoperatively, 82% of those who died required a

dopamine dose> 10 J,Lg/kg/min and had a urine output <2 ml/kg/hr, whereas only 2.8% of all the survivors fulfilled both these criteria. Table I compares the data for our ECM0 patients with the pre-ECMO patients in this "predictor" study. ECMO support postoperative patient group. We have treated 36 patients with low cardiac output or severe pulmonary artery vasoreactive crisis after open repair of congenital heart disease with ECMO between August 1984 and May 1989 (Table II). ECMO was initiated after electively performed intracardiac repair in 34 patients and after early reoperation for residual defects in two other patients. In addition, three patients were treated before repair of their congenital heart defect. One patient was placed on ECMO for a mistaken diagnosis of primary pulmonary hypertension of the newborn infant. Angiography while on ECMO demonstrated the correct diagnosis of obstructed total anomalous pulmonary venous return, and the patient underwent successful repair. Another patient with cor tria tria tum had profound cardiac failure during catheterization and was placed on ECMO. He underwent surgical repair and remained on ECMO for 42 hours postoperatively but died because of irreversible myocardial failure. The third patient had had repair of total anomalous pulmonary venous return at 1 week of age. Severe pulmonary venous obstruction proximal to the anastomosis associated with right heart failure was evident 8 weeks later, and he was placed on ECMO for 24 hours before unsuccessful reoperation.These three patients will not be discussed further, and the focus will remain on postoperative support with ECMO of 36 children after cardiac operations. The mean age of these 36 patients was 13.6 months (range 1 day to 7 years). Ten patients had Down's syndrome; five of these survived, one of whom also had imperforate anus and duodenal atresia. Other congenital anomalies were rare. One patient had a cleft palate and one had hypoparathyroidism. The critical nature of these patients' illness is documented in Table III. We classified patients as having biventricular failure, left ventricular failure, or right ventricular failure (which in

The Journal of Thoracic and Cardiovascular Surgery

5 0 0 Klein et al.

Table II. Patients treated with

EeMa after open repair of congenital heart defects ECMO

Pt. No.

Age (mo)

Survivors I 13 2 0.5 22 3 4 4 5 26

Indica-

Diagnosis"

Operation

tioni

Hours postcardiotomy

Cannulation site! RIJ, RCC RIJ, RCC RA, LA,AO RA,LA,AO RIJ,RCC

Left heart venting

Hours on ECMO

Repair Arterial switch Repair ACBP I. Closure ASD, VSD 2. Left AVV repair Repair

RVF LVF BVF LVF LVF

23 4 0 0 2

144 91 148 144 117

RVF

5

RIJ,RCC

124

8

RIJ, RCC

112

9 23 24

88 96 77 144

Nonfatal complications

6

7

TOF D-TGA simple A VSD, partial ALCAPA 1. CCTGA+ VSD+ASD 2. Left AVV insufficiency TOF

7

9

AVSD, complete

Repair

PVRC

8 9 10 II

I d. 6 5 12

TAPVR ALCAPA A VSD, complete TOF

Repair ACBP Repair Repair

RVF LVF BVF LVF

27

RIJ, RCC RIJ, RCC RIJ, RCC RIJ,RCC

12

24

Repair

LVF

27

RIJ,RCC

116

13 14

5 25

Repair Repair

PVRC LVF

27 30

RIJ, RCC RIJ, RCC

94 69

Chest tube bleeding

15

4

VSD, muscular, multiple, PAB A VSD, complete DORV,PAB, subaortic stenosis D-TGA, VSD, PAB

LVF

35

RA,LA,AO

86

Chest tube bleeding

16

84

RVOTO,VSD

Complex' Mustard, VSD closure, PA angioplasty Repair

RVF

51

RIJ,RCC

61

17

3

D-TGA, VSD

BVF

56

112

18 19

6 16

AVSD, complete VSD

Mustard, VSD closure Repair Repair

Centrifugal pump failure, Staphylococcus aureus sepsis Chest tube bleeding

20 21

II 22

VSD DORV+ PAB

22

5

A VSD, complete§

Repair Rastelli, PA angioplasty Repair

Yes Yes

Yes

BVF PVRC

56 120

SVC and IVC from RA,AO RIJ, RCC RIJ, RCC

PVRC LVF

142 18

RIJ, RCC RIJ,RCC

124 127

PVRC

7

RIJ,RCC

162

87 93

Intracranial hemorrhage Oxygenator failure Arrhythmia

Chest tube bleeding, mediastinal tamponade Chest tube and mediastinal bleeding

Chest incisional bleeding Bleeding neck and left groin

Pseudomonas sepsis Oxygenator failure, intracranial hemorrhage Mediastinal and. neck bleeding

• ACBP. Aorta-coronary bypass;ALCAPA. anomalous left coronary artery from the pulmonary artery; ASD, atrial septal defect; AYSD. atrioventricular septal defect; AYY, atrioventricular valve;CCTGA, congenitallycorrected TGA; DORY, double-outlet right ventricle;HRHS, hypoplasticright heart syndrome;MY, mitral valve; PAB, pulmonary artery band; PS. pulmonary stenosis; RYOTO, right ventricular outflow tract obstruction; SY, single ventricle; TAPYR, total anomalous pulmonary venous return; TOF, tetralogy of Fallot; TGA, transposition of the great arteries; TY, tricuspid valve; YSD, ventricular septal defect; WT, Waterston shunt. tRYF, Right ventricular failure; LYF, left ventricular failure; BYF, biventricular failure; PYRC, pulmonary vasoreactive crisis. :j:RIJ, Right internal jugular vein; RCC, right common carotid artery; LA, left atrium; RA, right atrium; Ao, aorta; SYC, superior vena cava; IYC, inferior vena cava.

§This patient weaned from ECMO, was decannulated, extubated, and out of the intensive care unit when she had another PYRC and died.

II Intraoperative calcium overdose.

cUneventful repair was followed by hemolysis resulting from malfunctioning blood warmer in the operating room. Massive pulmonary arteriolar obstruction due to red cell debris and fibrin was noted at autopsy. some cases led to biventricular failure) after careful review of their clinical course and all available studies. Most had low cardiac output indicated by hypotension and oliguria despite adequate atrial filling pressures. All were receiving dopamine and

afterload-reducing agents such as nitroprusside or nitroglycerin, and many had the addition of other pharmacologic agents, including dobutamine, isoproterenol, or epinephrine. Enhancement of systemic venous return by an intermittent abdominal

Volume 100 Number 4 October 1990

ECMO after repair of congenital heartdisease 501

Table II. Cont'd. ECMO Pt. No.

Age (mo)

Hours Indica- postcar-

Diagnosis"

Nonsurvivors AVSD, complete 23 7 SV, PS, TAPVR 24 2

Cannulation site!

Left heart venting

Hours on ECMO

diotomy

Repair Fontan, repair TAPVR Repair

PVRC BVF

86 0

RH,RCC RH, RCC

135 6

Arrhythmia

BVF

0

RH, RCC

164

Chest tube bleeding

Repair

BVFlI

0

RH, RCC

215

Repair Fontan, proximal PA-Ao anastomosis Repair

BVF BVF

0 0

RH, RCC RH, RCC

143 21

Mediastinal bleeding, intracranial hemorrhage Chest tube bleeding Chest tube bleeding

BVF

0

RH, RCC

92

RVfC BVF

0 17

RA,LA,AO SVC and IVC fromRA,AO

LVF BVF

17 26

RH, RCC RH, RCC

87 47

25

7

26

7

27 28

5 32

AVSD, complete MVatresia, PAB

29

18

30 31

4 22

VSD, multiple +PAB TOF DORV, remote VSD, PAB

32 33

6 6

AVSD, complete VSDs, multiple

Repair Complex Mustard, VSD closure, LV-PA conduit Repair Repair

34

4

VSD+ASD

Repair

BVF

50

RH,RCC

148

35

30

Fontan, relief of subaortic stenosis

LVF

53

RH, RCC

15

36

30

DORV, straddling TV, subaortic stenosis HRHS, WT

1. Fontan, closure LVF WT 2. Reap for closure residual ASD

10

RH, RCC

145

AVSD, complete, PAB VSD

Nonfatal complications

tioni

Operation

compression device was also used in some patients.' No patient had primarily pulmonary parenchymal insufficiency. Completeness of the cardiac repair was confirmed in all patients by 2-D echocardiography and color flow Doppler echocardiography. This was done before ECMO in those patients cannulated in the intensive care unit and after ECMO in those cannulated in the operating room. The correctness of the noninvasive evaluation was confirmed at autopsy. The ECMO circuit was that described by Bartlett and colleagues.? Venoarterial bypass was used with neck cannulation in most cases. A catheter was passed to the right atrium from the right internal jugular vein for drainage, and an infusion cannula was passed via the right common carotid artery to the level of the aortic arch. Blood was drained by gravity through a servoregulating bladder (to turn off the pump if venous return was inadequate) to a roller pump. It was then pumped through a silicone membrane oxygenator and a heat exchanger before returning to the patient.

47 153

Chest tube bleeding, arrhythmia, seizures Chest tube bleeding Chest tube bleeding, intracranial hemorrhage

Chest tube and incisional bleeding, intracranial hemorrhage, seizures Mediastinal bleeding and tamponade

Six patients underwent open chest cannulation. In three of these, who could not come off bypass after intracardiac repair, cardiac distention was present after institution of ECMO by right atrial-aortic cannulation and was relieved by insertion of a left atrial venting cannula. The other three patients cannulated through the chest represented special anatomic situations. One had had a prior repair of a complex aortic coarctation with use of a left carotid flap angioplasty. Two had complex repairs involving a Mustard-type atrial baffle and underwent bicaval cannulation because it was believed that jugular venous cannulation alone could lead to superior vena caval obstruction and decreased venous return to the pump. A Kolobow, Sci-Med (Life Systems Inc., Minneapolis, Minn.) silicone rubber membrane lung of 0.8, 1.5, or 2.5 m 2 surface area, depending on the patient's size, was routinely used. Elecath (Electro-Catheter Corporation, Rahway, N. J.) catheters (with side holes and an end hole) were used for neck cannulation for venous cannulas from lOF to l6F. For larger chil-

The Journal. of Thoracic and Cardiovascular Surgery

5 0 2 Klein et al.

Table III. Pre-ECMO data* Group

MAP (mm Hg)

RAP (mmHg)

LAP (mm Hg)

(mlfkgfhr}

Dopamine (JJ,gjkgjmin)

pH

Pac02 (mmHg)

Pa02 (mmHg)

Survivors Nonsurvivors

61 ± 4 62 ± 4 55 ± 4

21 ± 6 20 ± 4 22 ± 4

18 ± 6 18 ± 3 19 ± 6

0.61 ± 0.74 1.7 ± 2.3 0.69 ± 0.5

12 ± 2 12 ± 2 13 ± 5

7.36 ± 0.11 7.33 ± 0.10 7.41 ± 0.10

37 ± 8 40 ± 6 30 ± 7

122 ± 63 133 ± 63 96 ± 55

All

Urine

MAP, Mean arterial pressure; RAP, right atrial pressure; LAP. left atrial pressure. • Average of 6 hours before cannulation. Patients cannulated in the operating room because they could not come off bypass do not have pre-ECMO data.

dren, 20F and 24F Argyle (Sherwood Medical, St. Louis, Mo.) chest tubes were employed. Arterial catheters ranging from 8F to 16F were fashioned from chest tubes, with the distal portion of the catheter, including the side holes, cut off. When chest cannulation was employed, William Harvey (C: R. Bard, Inc., Santa Ana, Calif.) and Pacifico (DLP, Grand Rapids, Mich.) venous cannulas and angled THI aortic perfusion cannulas (Sherwood Medical) were used. In seven older patients in whom high flows (> I L/min) were necessary, we used a centrifugal pump (Bio-Medicus, St. Paul, Minn.) with a BP-80 pump head since a bladder box is not required with this pump and bladders for 3/g-inch tubing were not commercially available. Goal of ECMO support. The goal of ECMO support in a patient after cardiac operation differs from that in a neonate with hypoxemia resulting from pulmonary hypertension. In the neonate, pump flows are adjusted to achieve satisfactory arterial oxygenation. In the patient with low cardiac output after cardiac operation, the goal is to maintain adequate tissue perfusion while providing complete or nearly complete cardiac bypass to prevent cardiac distention, minimize myocardial energy expenditure, and maximize potential functional cardiac recovery. Flows as high as 150 nil/kg/min are frequently needed to reduce both right atrial pressure and left atrial pressure, which must be measured continuously. High pump flows were maintained for at least 72 hours before attempting to wean from bypass. Since the patient's lungs are usually not significantly impaired and are also removing carbon dioxide, the high pump flow and an efficient membrane lung can cause the arterial carbon dioxide tension (Paco-) to fall below 20 mm Hg, and the oxygen tension (Po 2) to rise above 200 mm Hg. A gas blender with two flowmeters to the membrane lung for oxygen and carbon dioxide allowed control of the patient's Pao- and Paco- without changing the pump blood flow. Controlling hemorrhage and hypertension. Controlling hemorrhage' and hypertension" on ECMO is of great importance. Even in patients cannulated immediately after cardiopulmonary bypass, hemorrhage does not become significant until 72 hours. Although it is easy to transfuse a patient on ECMO, the massive transfusion leads to pulmonary impairment that is often not recoverable. Anticoagulation is controlled by maintaining the activated clotting time at 200 seconds with a continuous heparin infusion. When bleeding is a problem, the activated clotting time is maintained at an even lower level. Fibrin glue made on the surgical field from single donor cryoprecipitate and thrombin reconstituted with 10% calcium gluconate is used at all operative sites, especially the cannulation site. The prothrombin time is maintained as near normal as possible with vitamin K; IO nil/kg of fresh frozen plasma is given every 6 hours. The platelet count is maintained over l00,OOO/mm3 with platelet transfusions. Daily cranial ultrasounds are obtained in patients with an open fontanel to detect any intracranial hemorrhage.

Ultrafiltration. Ultrafiltration with the Minifilter (Amicon Corporation, Danvers, Mass.) is employed routinely to decrease intravascular volume, which has usually been expanded before ECMO to improve atrial filling and cardiac output and which continues to be expanded by transfusions of blood, fresh frozen plasma, and platelets. Ultrafiltration also decreases atrial pressures by decreasing intravascular volume, and is used to manage the many degrees of renal failure experienced by patients with low cardiac output. Weaning. Weaning the cardiac patient from ECMO is also different from the technique used in newborn infants with respiratory failure. In pulmonary failure the pump flow is decreased as lung recovery occurs. In cardiac failure the filling volumes and pressures have been kept low and must be restored (usually 15 to 20 mm Hg) over 6 to 8 hours while the pump flow is gradually turned down. We also add a moderate dose of inotropic infusion, such as 3 to 5 JLg/kg/min of dopamine, and carefully observe the pulse contour on the arterial tracing, as well as other signs of perfusion adequacy such as urine output. Objective assessment of cardiacfunction. Objective assessment of cardiac function during ECMO is difficult to obtain. The MUGA (multiple-gated acquisition) nuclear scan requires temporary discontinuation of ECMO for at least 3.minutes. Standard dilution techniques for the measurement of cardiac output are not applicable because appropriate sampling sites are not available with the patient on partial bypass. Our most useful tools have been the 2-D echocardiogram and the color flow Doppler echocardiogram. With these we can assess myocardial contractility and ventricular ejection in relationship to systolic blood pressure as pump flow is decreased.

Results

Myocardial function improved sufficiently in 22 of 36 patients (61%) to permit discontinuation of ECM0 support, decannulation, and eventual hospital discharge, except in one patient who died after discharge from the intensive care unit from presumed pulmonary vascular crisis. Twenty of the 27 patients cannulated 4 hours or more after operation survived (74%). Seven of the nine patients cannulated in the operating room becausethey couldnot comeoffbypass died.The twowhosurvived had open chest cannulation with separate decompression of the leftatrium. Twoof three additionalpatientswhowere returnedto the operatingroomforopenchestcannulation becauseof special anatomic considerations alsosurvived. Hemodynamic data during ECMo. Hemodynamic data during ECMO is presented in Table IV. Once

Volume 100 Number 4 October 1990

ECMO after repair ofcongenital heart disease

503

Table IV. On-ECMO data (average of all data while on ECMOj MAP (mm Hg)

Survivors Nonsurvivors

82 ± 7 69 ± 13

RAP

(mmHg) \2 ± 2

\2 ± 3

LAP (mmHg)

Urine (mljkg/hr)

ECMOj/ow (mljkg/min)

Blood administered (mljkg/hr)

11 ± 2 13 ± 4

4.3 ± 5.\

\59 ± \84 11\ ± 40

5.63 ± 7.0*

2.2 ± 2.2

1.50 ± 1.13

MAP, Mean arterial pressure; RAP, right atrial pressure; LAP, left atrial pressure. "Those patients requiring cannulation for ECMO in the operating room because they could not come off bypass had a blood requirement of 7.46 ± 8.29 mil kg/hr during ECMO.

ECMO SUpport was initiated, the condition of the patients usually improved dramatically. Mean arterial pressure rose, atrial filling pressures fell, and urine output increased. Most patients required approximately 150 nil/kg/min of pump flow,permitting right atrial pressure to decrease by a mean of 9 mm Hg and left atrial pressure by a mean of 6 mm Hg. The survivors were cannulated for ECMO a mean of 34.5 ± 37.0 hours after operation, and the nonsurvivors at 19.2 ± 27.7 hours after operation. The mean time on ECMO for all patients was 105.4 ± 45.7 hours, 109.9 ± 27.8 hours for survivors and 97.9 ± 66.8 hours for nonsurvivors. The nonsurvivors died 26.6 ± 59.2 hours after coming off ECMO. Thosewhosurvivedspentameanof27.6 ± 10.6 days in the hospital. Bleeding. Bleeding on ECMO was a common complication (17 patients). The average transfusion requirement during ECMO was 3.24 ± 5.14 ml/kg/hr; nonsurvivors and patients requiring cannulation because they could not come off bypass clearly had more bleeding (Table IV). Fivepatients required operative exploration of the neck or chest, or both, for bleeding while on ECMO. This was usually done in the intensive care unit, and removal of mediastinal clot clearly relieved tamponade in two patients. Intracranial hemorrhage. Intracranial hemorrhage occurred in four patients, two of whom died. One 6-month-old patient (patient 33) had seizures and left temporal lobe bleeding on the second day after cannulation for ECMO, and was decannulated because of the intracerebral hemorrhage. The second patient (patient 26), 7 months old, was noted to have a large left frontal hemorrhage on the routine cranial ultrasound on day 9 and also had seizures. He was also decannulated because ofthe intracerebral hemorrhage and died. One (patient 2) with intracerebral hemorrhage who survived was 2 weeks old and had an intraventricular hemorrhage on day 4. He was decannulated for this reason but lived. The second survivor with intracerebral hemorrhage (patient 19) was 16 months old and on the third day developed a right frontal hemorrhage that was larger on the fourth day and led to his decannulation. Mechanical or circuit complications. Mechanical or

circuit complications occurred in three patients. An early model impeller pump head failed in a patient who was being weaned from ECMO, and the patient was decannulated and lived. Two oxygenators failed and needed to be replaced during the ECMO run. Arrhythmias. Arrhythmias that began after ECMO occurred in three patients. Two of these were controlled with pacing, and one patient survived. One patient had a nodal rhythm that was temporarily controlled with core cooling while on ECMO, but effective myocardial function did not return, and the patient died. Two patients had bacteremia that was effectivelycontrolled with appropriate antibiotic therapy. Follow-up. Follow-up information is available on all but one survivor. Nine of the 19 patients (two patients were discharged lessthan 1 month before this writing) are entirely normal on evaluation in cardiology clinic when taking into consideration their associated problems such as Down's syndrome. The others have various degrees of cardiac disability requiring digoxin and diuretics consistent with their cardiac disease. No long-term problems have been attributable to ECMO.

Discussion Early postoperative survival after anatomic correction of congenital heart defects has improved dramatically. The risk of dying is now related in great part to the complexity of the underlying heart defect. Early death after adequate intracardiac repair is usually due to right, left, or biventricular failure, and, less frequently, pulmonary vasoreactive crisis, either sustained or paroxysmal. The latter may lead to unexpected and profound acute right ventricular failure, low cardiac output, hypoxemia, acidosis and death. Although hyperkalemic cold cardioplegia for intraoperative myocardial protection has markedly decreased the need for early inotropic support, there still remains a small group of children who have an anatomically satisfactory correction of the congenital heart defect and yet demonstrate low cardiac output or pulmonary vasoreactive crisis unresponsive to intensive ventilator and pharmacologic intervention. It is also recognized that low cardiac output is more common in children than in adults.' It is

5 0 4 Klein et al.

to this group that we have directed mechanical support in the form of ECMO. Several forms of mechanical circulatory support have been applied to children, without significant or consistent success. Among these, intraaortic balloon pumping poses an increased risk oflocal arterial injury, is ineffective for right heart failure, and offers limited circulatory assistance at the rapid heart rates usually present in these infants. The University of Utah Group has had some success with specially designed balloons.? Intrapulmonary arterial balloon pumping, either alone or in combination with intraaortic balloon pumping, has been proposed but not yet tested clinically." Left-sided extracorporeal pump assist devices have demonstrated success in adults but ignore right ventricular failure.f Right-sided extracorporeal pump assist devices are theoretically attractive," but access to the pulmonary artery is difficult, and perfusion of the pulmonary artery can have a deleterious effect on the lung.!? Ventricular assist devices,Il,12 which are attached directly to the heart, offer the advantage of requiring little or no anticoagulation, but they require major operative procedures in the chest and could interfere with the anatomic repair. We have previously demonstrated that venovenous EeMO, in which the oxygenated blood is returned to the systemic venous system, can salvage newborn infants dying of respiratory failure.P This approach certainly unloads the right side of the heart, and it may have some applications for cardiac support in children since they so frequently begin with rightsided failure. Several other groups have reported their experience with ECMO for cardiac support in children. Baffes and colleagues, 14 in 1970, were the first to devote serious consideration to long-term extracorporeal circulation for the support of children with congenital heart disease. They did not, however, report specific survival figures, and they considered extracorporeal circulation as an adjunct to palliative operations more than as support for children having open procedures. In 1972, Hill and colleagues 15 first reported a survivor who was treated with ECMO in Diisseldorf after repair of tetralogy of Fallot. At 'the Ochsner Clinic l 6 one postoperative patient was weaned from ECMO, but died of renal failure, and one survived when ECMO supported him long enough to allow revision of a shunt. The group from St. Louis University'? reported the survival of six of 13 children supported with ECM0 after cardiac operations. Their patients were somewhat older than ours, but the results were equally encouraging. They cannulated patients through the chest, which avoided ligation of the internal jugular vein and common carotid artery. We have been concerned that cannulation through the chest requires a major operative procedure with

The Journal of Thoracic and Cardiovascular Surgery

attendant bleeding in the heparinized patient, necessitates reoperation to decannulate, risks exsanguination should the cannulas become dislodged, and has been associated with mediastinitis. Cannulation via the neck has been effective in our patients, although it does theoretically place the brain at risk with ligation of the right common carotid artery and internal jugular vein. Thus far we have not been able to ascribe neurologic damage to this technique. We can perform both cannulation and decannulation in the intensive care unit and can easily reexplore the cannulation site in the neck if bleeding should occur. Our patients are younger than the St. Louis University ECMO series, and this may explain why they tolerate neck cannulation relatively well. Some of these same cannulation considerations are noted by the recent University of Pittsburgh experience with cardiac ECMO, which includes 10 patients with seven survivors, 18 and the Washington University series, which includes 13 patients with five survivors.19 There has been some concern that mechanical circulatory support that drains only the right atrium cannot effectively decompress the left side of the heart. Eugene and colleagues?" demonstrated that in dogs with myocardial infarction, cardiac function improved much more if left ventricular drainage was added to ECMO support. In fact, the St. Louis University group used this in one patient who did not survive. The left atrial pressures in our patients demonstrated that the left side of the heart can be unloaded during ECMO, although not completely. Although our first choice remains neck cannulation, we use chest cannulation in specific circumstances. Our two surviving patients among those who could not be weaned from bypass in the operating room had chest cannulation with separate drainage of the left atrium. In any discussion of mechanical support for patients after cardiac operations, there is one unanswerable question. Would the patient have survived without such support? In Table III we have demonstrated how similar our survivors and nonsurvivors were before going on ECMO. We do not believe that the survivors would have lived without ECMO, because they were just as sick to start with as the nonsurvivors. We also compared our ECMO patients with patients dying in our predictor study (Table I). Our ECMO patients are not very different at all from these earlier patients who died before ECMO was available. There was no difference in age and weight, and, while the predictor study patients had longer circulatory arrest times, there was no difference in aortic crossclamp time. If there was a significant difference in the other variables tested at 1 hour postoperatively, there was no difference in those same variables tested at 8 hours postoperatively, and vice versa. The only consistent difference between the patients placed on ECMO and those who

Volume 100 Number 4 October 1990

died in the predictor series is that the predictor study patients received larger amounts of dopamine before they died. From this early experience with ECMO for cardiac support in children, we conclude that it can improve survival in patients with postoperative pulmonary vasoreactivecrisis and / or low cardiac output resulting from either right or left ventricular or bi-ventricular failure. Although we continue to apply ECMO in all situations where we believe the underlying problem is reversible, it appears more likely to be successful in patients who require mechanical support later than 6 hours after operation. Patients with predominantly left ventricular failure present earlier or cannot come off bypass in the operating room. They do not respond as well to ECMO and may benefit from direct left heart decompression during ECMO. Hemorrhage is the most significant complication of ECMO. The massive transfusion that is frequently required can have a deleterious effect on the lung and other organ systems. Control of hemorrhage is, thus, of the greatest importance. Many centers now have ECMO teams that routinely provide support for newborn infants with respiratory failure. With.very little reorganization, the same team can be employed to support patients after open heart operations. When the intracardiac repair is adequate and the myocardial failure is potentially reversible, postoperative ECMO support may salvage an important number of patients who otherwise would be highly likely to die oflow cardiac output or PVRC unresponsive to other methods of treatment. We wish to acknowledge the important contributions to this work made by Gary Amundson, MD, Marc L. Cullen, MD, Mehdi Hakimi, MD, Suzette LaVigne, Julie A. Long, MD, Steven B. Palder, MD, Linda L. Sell, MD, Daniel W. Ziegler, MD, and, above all, Arvin I. Philippart, MD. REFERENCES 1. Milliken JC, Laks H, George B. Use of a venous assist device after repair of complex lesions of the right heart. J Am Coil CardioI1986;8:922-9. 2. Bartlett RH, Gazzaniga AB, Toomasian J, Coran AG, Roloff D, Rucker R. Extracorporeal membrane oxygenation (ECMO) in neonatal respiratory failure. Ann Surg 1986;204:236-45. 3. Sell LL, Cullen ML, Whittlesey GC, et al. Hemorrhagic complications during extracorporea1 membrane oxygenation: prevention and treatment. J Pediatr Surg 1986; 21:1087-91. 4. Sell LL, Cullen ML, Lerner GR, Whittlesey GC, Shanley CJ, Klein MD. Hypertension during extracorporeal membrane oxygenation: cause, effect, and management. Surgery 1987;102:724-30.

ECMO after repairof congenital heart disease

50 5

5. Magovern JA, Pae WE Jr, Miller CA, Waldhausen JA. The mature and immature heart: response to normothermic ischemia. J Surg Res 1989;46:366-9. 6. Veasy LG, Blalock RC, Orth JL, Boucek MM. Intra-aortic balloon pumping in infants and children. Circulation 1983;68:1095-1100. 7. SpencePA, PenistonCM, MihicN, Jabr AK,SalernoTA. A rational approach to the selection of an assist device for the failing right ventricle. Ann Thorac Surg 1986;41: 606-8. 8. Magovern GJ, Park SB, Maher TD. Use of a centrifugal pump without anticoagulants for postoperative left ventricular assist. World J Surg 1985;9:25-36. 9: Fisher EIC, Willshaw P, Armentano RL, Delbo MIB, Pichel RH, Favaloro RG. Experimental acute right ventricular failure and right ventricular assist in the dog. J THORAC CARDIOVASC SURG 1985;90:580-5. 10. Toporoff B, Marini CP, Grubbs PE Jr, et al. Pulmonary complications of a roller pump right ventricular assist device. J Surg Res 1988;45:21-7. 11. Pierce WS, Parr GVS, Myers JL, Pae WE, Bull AP, Waldhausen JA. Ventricular-assist pumping in patients with cardiogenic shock after cardiac operations. N Engl J Med 1981;27:1606-10. 12. Taenaka Y, TakanoH,NakataniT,etal. Ventricular assist device (VAD) for children: in vitro and in vivo evaluation. Trans Am Soc Artif Intern Organs 1984;30:155-8. 13. Klein MD, Andrews AF, Wesley JR, et al. Venovenous perfusion in ECMO for newborn respiratory distress: a clinical comparison with venoarterial perfusion. Ann Surg 1985;201:520-6. 14. BaffesTG, Fridman JL, BicoffJP, Whitehill JL. Extracorporeal circulation for support of palliative cardiac surgery in infants. Ann Thorac Surg 1970;10:354-63. 15. Hill JD, de Leval MR, Fallat RJ, et al. Acute respiratory insufficiency: treatment with prolonged extracorporeal oxygenation. J THORAC CARDIOVASC SURG 1972;64:551-62. 16. Redmond CR, Graves ED, Falterman KW, Ochsner JL, Arensman RM. Extracorporeal membrane oxygenation for respiratory and cardiac failure in infants and children. J THORAC CARDIOVASC SURG 1987;93:199-204. 17. Kanter KR, Pennington DG, Weber TR, Zambie MA, Braun P, Martychenko V. Extracorporeal membrane oxygenation for postoperative cardiac support in children. J THORAC CARDIOVASC SURG 1987;93:27-35. 18. Rogers AJ, Trento A, Siewers RD, et al. Extracorporeal membrane oxygenation for postcardiotomy cardiogenic shock in children. Ann Thorac Surg 1989;47:903-6. 19. Weinhaus L, Canter C, Noetzel M, McAlister W, Spray TL. Extracorporeal membrane oxygenation for circulatory support after repair of congenital heart defects. Ann Thorae Surg 1989;48:206-12. 20. Eugene J, McColgan SJ, Moore-Jeffries EW, Ott RA, Haiduc NJ, Roohk HV. Cardiac assist by extracorporeal membrane oxygenation with in-line left ventricular venting. Trans Am Soc Artif Intern Organs 1984;30:98-101.