Extracorporeal Membrane Oxygenation for Cardiac Failure After Congenital Heart Operation Stanley Ziomek, MD, James E. Harrell, Jr, MD, James W. Fasules, MD, Sherry C. Faulkner, CCP, Carl W. Chipman, RN, Michele Moss, MD, Elizabeth Frazier, MD, and Stephen H. Van Devanter, MD Departments of Cardiothoracic Surgery and Pediatric Cardiology, David M. Clark Cardiovascular Center, Arkansas Children’s Hospital, University of Arkansas for Medical Sciences, Little Rock, Arkansas

Despite continuing improvement in myocardial protection and surgical technique, the repair of complex congenital heart lesions can result in cardiopulmonary compromise refractory to conventional therapy. In a 29-month period, 24 patients (aged 14 hours to 6 years) were treated with extracorporeal membrane oxygenation (ECMO) 28 times for profound cardiopulmonary failure. Four patients required ECMO after each of two cardiopulmonary bypass procedures. Seventeen patients required ECMO to be initiated in the operating room: 12 (71%) were weaned successfully from ECMO, and 8 (47%)survived. Seven patients had ECMO initiated in the intensive care

E

xtracorporeal membrane oxygenation (ECMO) has become accepted therapy for children with acute respiratory failure refractory to maximal medical treatment [l-51. In spite of continuing improvements in myocardial protection and operative technique, 80% of the operative mortality of congenital heart operations is caused by myocardial dysfunction or pulmonary hypertension [&13]. The use of ECMO as lifesaving treatment for profound postoperative pediatric cardiac failure is increasing [4, 5, 1P231 and represents 4% of all ECMO cases [24]. We report our experience with 24 children treated with ECMO for postoperative cardiopulmonary failure after a congenital heart operation, the majority of whom had ECMO initiated in the operating room (OR).

Material and Methods Patient Population Data were obtained from 28 ECMO courses performed in 24 patients at Arkansas Children‘s Hospital who underwent cardiopulmonary bypass procedures between March 1989 and June 1991 (Table 1).Twelve were male and 12 were female. The patients’ ages ranged from 14 hours to 6 years (mean, 12.5 months), and weights ranged from 2.9 to 12 kg (mean, 5.7 kg). Extracorporeal membrane oxygenation was initiated if, in the opinion of the Presented at the Twenty-eighth Annual Meeting of The Society of Thoracic Surgeons, Orlando, FL, Feb S 5 , 1992. Address reprint requests to Dr Van Devanter, Department of Cardiothoracic Surgery, Arkansas Children’s Hospital, 800 Marshall, Little Rock, AR 72202.

0 1992 by The Society of Thoracic Surgeons

unit: 6 (86%)were weaned, and 5 (71%)survived. Serial echocardiograms demonstrated substantial recovery of cardiac function in 18 of 21 instances (86%)of ventricular failure from myocardial dysfunction. Overall, 18 of 24 patients (75%) were successfully weaned from ECMO including all 4 who underwent 2 ECMO treatments. We conclude that ECMO can successfully salvage children who have serious cardiopulmonary failure immediately after a congenital heart operation and that long-term survival is possible after two ECMO treatments. (Ann Thorac Surg 1992;54:861-8)

attending pediatric cardiologist and surgeon, death was imminent despite maximal conventional therapy (ie, unable to be separated from cardiopulmonary bypass or postbypass mean arterial blood pressure < 40 mm Hg, acidosis, and oliguria despite maximal inotropic and ventilatory support). Before committing to a course of ECMO in the OR, a search for a structural cause of cardiac failure was made with hernodynamic measurements and epicardial echocardiography. If no correctable structural defects were found, cardiopulmonary bypass was continued for 1 to 2 hours while the ECMO team was contacted and the machine was brought into the OR. If the patient‘s hemodynamics had not improved by this time, ECMO was initiated in the OR. Several patients have been salvaged by going back on bypass in the last 10 years. However in our recent experience, all patients unable to be weaned from bypass have ultimately required ECMO support. The predominant indications for ECMO were ventricular failure (23), pulmonary hypertension (6), and hypoxemia in (1). Two cases of ventricular failure were caused by refractory supraventricular tachycardia after Fontan procedures, 2 were due to volume restriction from small right ventricles after Rastelli procedures, and 1 was secondary to a left ventriculotomy made to close multiple ventricular septa1 defects. Two other cases of ventricular failure were caused by obliteration of the ventricular cavities from severe ventricular hypertrophy, one after repair of a truncus arteriosus (both ventricles) and another after reconstruction of pulmonary hypoplasia in a patient with Williams syndrome (right ventricle). A patient with tetralogy of Fallot, right pulmonary hypoplasia, and a congen0003-4975/92/$5.00

CDH with TOF

SV; TGV; VSD

Pulmonary atresia; h oplasticPA; s .ASDdp central shunt

Partial AVC; severe MR Complete AVC (type C); PDA Mixed TAPVR Truncus arteriosis (type n); branch PA stenosis

M

M

F

M

F

F

3.6

3.1

4.0

4.4

4.9

3.6 3.8

11

9.5

15 h

1 mo

6 mo

3 mo

9 mo

3d 3w

27 mo

14 mo

4

5

6

7

8"

9

10 11

12"

13a

TAPVR repair Norwood procedure

179 147

Hypoxemia VF

Chest

Partial takedown Fontan; lacement of ,en& shunt Rastelli procedure; ligation BT

TGV; VSD; reduced RV volume; PV stenosis; dextrocardia; supinf ventricles

F

Neck

Chest

Revision Fontan with RA-RPA conduit

Tricuspid atresia; extreme reactive airway disease; d p Fontan d p Fontan revision

M

92 169

VF

VF

Chest

PH

41

VF VF

Neck Chest

87

55 56

PH

PH

67

VF

Neck Neck

65

VF 123

123

62

109 43

VF

VF

PH PH

ECMO Indication

ECMO Duration (h)

Neck

Neck

Chest

Chest

PA reconstruction

Patch RVOT obstruction Stansel procedure; aortic arch repair; atrial septectomy Takedown central shunt; Rastelli procedure; PA reconstruction; VSD, ASD repair AVC repair; MV replacement AVC repair; PDA ligation TAPVR repair Truncus repair

Fontan procedure

Chest

Neck Chest

Operation

Norwood procedure; cardiac transplant

Cannula Site

Stenosis of truncus repair

M

SV; TGV

F

12

5Y

3

A,

HLH

M

2.9

14 h

Mixed TAPVR HLH

Diagnosis

3.6 3.1

M M

Sex

7d

2d

Age

Weight *g)

1 2

Pa tien t No.

Table 1. Cases of Postoperative Extracorporeal Membrane Oxygenation

18

OR

3

OR

OR 8

OR

6

OR

OR

OR

OR

OR

OR OR

Interval to ECMO (h)

Yes

Yes

Yes

Yes

Yes Yes

Yes

Yes

No

Yes

No

No

Yes

Yes No

Weaned

Alive

Alive

Died

Alive

Alive

Alive

Died

Died

Died

Died

Died

Alive Died

Outcome

Postop cavitary obliteration from biventricular hypertrophy "suicide ventricles" Died of mediastinitis; re uired RV-PA conduit an% ASD, VSD closure postop

Died of myocardial failure

Died of m ocardial failure; renal falure Died of s e r i s ; RV ipfarct; renal fa ure, cardiac transplant after Norwood procedure Died of Candido pneumonia; postop refractory SVT; renal failure Died of Candida pneumonia; renal failure Died of bilateral pulmonary infractions

Comments

i4

2

b

El

p

z3

u

3Y

3Y

6Y 9d 18 mo

3 wk

3 mo

17 mo

17"

18

19= 20 21

22

23

24

~~

5.5

4.9

3.8

16 3.5 5

12

13

3

3.3 4.9

~~~

~~~~~

M

F

M

F F F

M

F

F

M M

~~~~

~~~~~~~~

Partial AVC; severe MR; valvular and supravalvular PS Recurrent MR s/p AVC repair s/p RVOT patch

Multi le VSDs; ASD; PDX

Pulmonary atresia with intact septum; PDA TGV; multi le VSDs; reduced KV volume Multi le VSDs, TR s/p%asteNi ' procedure Shone's complex; s/p MV re lacement, s/p sufaortic resection, s/p coarctation repair (in infancy) Mitral atresia HLH Williams syndrome; SIPaortic enlargement, s/p PA reconstruction Cor triahiatum; PAPVR

HLH VSD; RVOT obstruction

Mitral valvuloplasty

Cor triatriatum repair; PAPVR repair Re air multi le 6SDs; AS8 repair; PDA ligation AVC re air patch RVOP ;

Re air multiple 6SDs; TV valvuloplasty Redo MV replacement; redo resection subaortic stenosis; aortic valvuloplasty Fontan procedure Norwood procedure Redo reconstruction right and left PA branches

Patch RVOT; central shunt; PDA ligation Rastelli procedure

Norwood procedure VSD repair; infundibular resection

Chest

Chest

Neck

Neck

Neck Chest Neck

Chest

Chest

Neck

Chest

Chest Chest

VF

VF

VF

PH

VF VF VF

VF

VF

VF

VF

VF VF

39

116

198

160

84 48 170

198

65

130

69

43 17

OR

48

OR

OR

48 OR OR

OR

OR

48

2

OR 2

Yes

Yes

Yes

Yes

Yes Yes Yes

Yes

Yes

Yes

Yes

No No

Alive

Alive

Alive

Alive Alive Alive

Died

Alive

Died

Died Died

Postop refractory SVT; sepsis Sepsis Postop cavitary obliteration from suicide RV Rf;j! h ertrophy

Died of necrotizing Pseudomonas pneumonia

Died of intracranial bleed Died of myocardial failure; ECMO mitiated after 45 min of cardiac massage after resuscitation from postop cardiac tamponade Died of Staphylococcal sepsis, site undetermined

AVC = atrioventricular canal; BT = Blalock-Taussig shunt; CDH = congenital diaphragmatic hernia; ECMO = extracorporeal membrane oxygenation; HLH = ASD = atrial septal defect; hypoplastic left heart syndrome; LV = left ventricular; MR = mitral regurgitation; MV = mitral valve; OR = operating room; PA = pulmonary artery; PAPVR = partial anomalous PH = pulmonary hypertension; PS = pulmonary stenosis; PV = pulmonary valve; RA = right ahium; RFA = right pulmonary venous return; FDA = patent ductus arteriosus; pulmonary artery; RV = right ventricular; RVOT = right ventricular outflow tract; Shone's complex = mitral stenosis + subaortic stenosis + coarctation; s/p = status post; sup-inf = supenor-inferior; SV = single ventricle; SVT = supraventricular tachycardia; TAPVR = total anomalous pulmonary venous return; TGV = transposition of great vessels; TOF = tetralogy of Fanot; TR = tricuspid regurgitation; TV = tricuspid valve; VF = ventricular failure; VSD = ventricular septal defect; Williams syndrome = supravalvular aortic stenosis + peripheral pulmonary artery stenosis.

Patient categorized as ECMO initiated in intensive care unit.

2d

16=

a

23 h 2 mo

14 15a

B

rn

W

m

M

864

ZIOMEK ET AL ECMO FOR CARDIAC FAILURE

ital diaphragmatic hernia was placed on postoperative ECMO support for hypoxemia.

Extracorporeal Membrane Oxygenation Technique The ECMO circuit and technique used in our series were similar to those described by Bartlett and associates [l]. Venoarterial ECMO was used in all patients. Fifteen patients were cannulated through the chest via the right atrium and ascending aorta. Nine patients were cannulated through the neck using the right internal jugular vein and right common carotid artery. Anticoagulation was begun with 100 U/kg heparin (beef lung) and adjusted to keep the activated clotting time between 210 and 230 seconds. Platelet transfusions were given if the platelet count was less than 120 X 1O’/L (12O,OOO/pL). The hematocrit was maintained greater than 0.40. In all patients, the sternum was left open and the wound covered with an iodinated plastic drape. Vancomycin and ceftazidime were used routinely for antibiotic prophylaxis. Every 48 hours, the wounds were cultured and then irrigated with an antibiotic solution. When necessary, clots were removed and bleeding points controlled with cauterization and a combination of thrombin spray and woven collagen. Patients were explored urgently to exclude tamponade for unexplained hypotension or increased central venous pressure. Once the patient was on ECMO, vasopressors were reduced to maintain a mean blood pressure of 45 to 60 mm Hg with a central venous pressure less than 6 mm Hg. Extracorporeal membrane oxygenation flow rates were begun at 120 to 150 mL * kg-’ * min-’. Ventricular overdistention was minimized by maintaining the central venous pressure less than 6 mm Hg. The ventilator was set at an inspired oxygen fraction less than 0.40, tidal volume of 15 mL/kg, positive end-expiratory pressure less than 7 mm Hg, and a rate of 10 breathdmin. Patients were rolled from side to side every 8 hours to minimize atelectasis and pooling of secretions. Intravascular volume was maintained with packed red blood cell transfusions and colloid solutions. While on ECMO, diuresis was stimulated with furosemide or bumetanide to eliminate fluid overload and improve pulmonary gas exchange. An ultrafiltration device (Amicon filter; W.R. Grace Co, Beverly, MA) was added to the ECMO circuit if urine output decreased to less than 2 mL . kg-’ . h-’ or if severe edema developed. The Sci-Med (Avecor, Minneapolis, MN) oxygenator was changed routinely every 7 to 9 days to lessen the risk of disseminated intravascular coagulopathy. The oxygen membrane was changed more frequently if necessary. The patients’ hemodynamics and mixed venous oxygen saturations were monitored, and myocardial function was evaluated with epicardial echocardiography and direct visual inspection through the plastic drape. Cranial ultrasonography was performed daily in neonates to detect intracranial bleeding. Criteria to begin weaning patients from ECMO included improved myocardial contractility on echocardiography, satisfactory pulmonary function, and reduced inotropic support. Over a period of 12 to 24 hours, ECMO flow rate was reduced to the minimum rated flow of the membrane oxygenator, which was 150 mL/min in the

Ann Thorac Surg 1992;54:861-8

smaller patients. If the patients tolerated reduction of the flow rate, manifested by stable hemodynamics and arterial blood gases, they were given at least 1 hour of a trial off by diverting flow through the bridge of the circuit. Cannulas were flushed every 5 to 10 minutes to maintain patency during this period. If the patients maintained adequate hemodynamic and pulmonary function during the trial off, decannulation was performed. The carotid artery was repaired whenever possible, and the internal jugular vein was routinely ligated in patients decannulated from the neck. Sternal closure usually was delayed 2 to 3 days to allow for further diuresis and reduction of tissue edema, thereby reducing mediastinal compression.

Statistical Analysis Data were presented as a range and mean or as a percentage of patients in a group. Differences between mean values were analyzed using Student’s t test. An F test was performed to ensure equality of variances before the Student’s t test was accepted. Differences of percentages between two groups were analyzed with Fisher’s exact test. Two-tailed analyses were employed, and differences were considered statistically significant at the 95% confidence level ( p < 0.05).

Results Twenty-four patients underwent ECMO for postoperative cardiopulmonary failure. Of these, 4 patients underwent two separate treatments for a total of 28 ECMO courses. In 22 of the 28 courses (79%), weaning was successful. This represented 75% of the patients (18/24) receiving ECMO, of which 54% (13/24) were long-term survivors. Serial echocardiograms demonstrated substantial recovery in 18 of 21 (86%) instances of ventricular failure from myocardial dysfunction. Severe postoperative pulmonary hypertension unresponsive to conventional therapy was successfully reversed in all 6 patients (100%) with ECMO. The time from operation to institution of ECMO ranged from 0 to 48 hours (mean, 1.2 hours). Extracorporeal membrane oxygenation was initiated in the OR immediately after bypass in 17 patients (71%) of whom 12 were weaned and 8 were long-term survivors. Extracorporeal membrane oxygenation was started in the intensive care unit (ICU) in 7 patients, of whom 6 were weaned and 5 were long-term survivors. Four patients underwent two operations both of which required postoperative ECMO. All 4 were weaned successfully, and 3 were long-term survivors. The duration of ECMO ranged from 17 to 198 hours (mean, 96 hours). The only factor that statistically decreased long-term survival of postoperative ECMO support was the development of an infectious complication (2/8 survived with infection versus 11/16 survived without infection; p = 0.05). All patients in whom fatal infections developed were cannulated through the chest (6/15 chest cannulations developed infection versus 0/9 neck cannulations; p = 0.04). Weight less than 5 kg or single ventricular anatomy tended to increase mortality of ECMO support but did not reach statistical significance. The incidence of infection was not influenced by the duration of ECMO,

ZIOMEK ET AL ECMO FOR CARDIAC FAILURE

Ann Thorac Surg 1992;54:861-8

the age of the patient, or whether ECMO was started in the OR versus the ICU. Thirty-two percent of ECMO runs (6/19) started in the OR were complicated by infection compared with 22% (2/9) after ECMO initiated in the ICU (not significant; p = 0.3). There was a trend toward a lower salvage rate of patients who required ECMO support in the OR compared with patients whose condition deteriorated later in the ICU because patients who required ECMO support in the OR appeared to be at higher risk for operative death. Patients requiring ECMO in the OR weighed less than patients requiring ECMO later in the ICU (4.8 & 2.8 kg in OR versus 8.9 4.9 kg in ICU; p < 0.05) and underwent more complex operations (ECMO was initiated in the OR in all 4 Norwood procedures). At the time ECMO was discontinued, hypoxemia, a high mean airway pressure, or the requirement of an epinephrine drip were negative predictors of long-term survival ( p < 0.05).

*

Deaths Eleven patients (46%) died in this series. Seven deaths were early (on ECMO or shortly after decannulation) and four were late (>1week after decannulation). The causes of death were infection (6), failure of myocardial recovery (3), pulmonary infarction (l),and intracranial hemorrhage (1).Of the 6 patients failing to be weaned from ECMO, 3 died of failure of myocardial recovery, 2 of Cundidu pneumonitis, and 1 of intracranial bleeding. Patient 2 was unable to be separated from cardiopulmonary bypass because of pulmonary hypertension and severe hypoxemia after a Norwood procedure. The patient did not improve and was taken off ECMO support because of progressive myocardial failure. Patient 3 underwent cardiac transplantation 24 hours after an unsuccessful Norwood procedure. The patient sustained a right ventricular infarction of the allograft with attempted sternal closure, had development of renal failure, and eventually died of gram-negative sepsis. Patient 4, who had been receiving antibiotics for refractory otitis media for 1 month before operation, had development of overwhelming Cundidu sepsis on ECMO. Patient 5 had severe right pulmonary hypoplasia secondary to a congenital diaphragmatic hernia, and had development of Candida pneumonitis on ECMO. Patient 6 was weaned from ECMO after a DamusKaye-Stansel procedure with a central shunt for a single ventricle. However, bilateral pulmonary infarctions developed related to temporary occlusion of the central shunt to avoid excessive pulmonary flow while on ECMO. A thrombectomy of the shunt and resection of a pneumatocoele were performed but the patient ultimately died. Patient 7 required ECMO support in the OR for biventricular failure after a Rastelli repair and pulmonary artery reconstruction for pulmonary atresia with hypoplastic pulmonary arteries. Extracorporeal membrane oxygenation was stopped after 123 hours of support for failure of myocardial recovery. Patient 15 underwent open cardiac massage for 45 minutes before initiation of ECMO. The patient was decannulated because no myocardial activity was demonstrated on echocardiography after 17 hours of support. Patient 14 was a newborn less than 24 hours old at the time of operation in whom intracranial bleeding

865

developed after a Norwood procedure. The 4 patients who died late survived an average of 25 days before succumbing to infection. Deaths due to infection are discussed in the next section.

Complications Fifteen complications developed in 11 patients (46%) in this series. Complications were infection (8), renal failure (4), right ventricular infarction (l), intracranial hemorrhage (l),and pulmonary infarction (1).Patients 4 and 18 had development of lethal pneumonitis (Candida and Pseudomonus) while on ECMO and had received long courses of antibiotics preoperatively for otitis media. Patient 16 had development of staphylococcal sepsis during ECMO support for right ventricular failure after operation for pulmonary atresia with an intact septum. Although the patient successfully recovered right ventricular function and was weaned from ECMO, the source of sepsis was not identified and the patient died 9 days later. Two other patients (19 and 20) had development of Cundidu sepsis from unidentified sources but eventually responded to amphotericin B and survived. Patient 11had development of mediastinitis (Staphylococcus uureus) after repair of a stenotic right ventricle-pulmonary arterial conduit and was never able to be weaned from the ventilator because of persistent sepsis. Four patients were treated for renal failure while on ECMO, patients 2 and 5 with ultrafiltration, patient 3 with peritoneal dialysis, and patient 4 with hemodialysis. Two of the 4 patients with renal failure were weaned from ECMO, but all eventually died.

Mechanical Dificulties Three mechanical problems occurred during the 28 ECMO courses in this series. Rupture of the pump-head tubing in the raceway occurred with the first patient in the series. Super Tygon tubing (Norton Performance Plastics, Akron, OH) was used in all patients afterward without incident. One shut-down of the ECMO circuit occurred because of a blown fuse in the pump computer. The faulty computer was replaced without coming off flow. A third patient sustained temporary dislodgement of an arterial cannula. None of these mechanical problems resulted in an adverse patient outcome.

Follow-up Follow-up of the 13 survivors ranged from 1 month to 18 months with a mean of 7 months. Two patients required a subsequent cardiac operation, one a redo mitral valve replacement 10 months postoperatively, and the other a repair of a stenotic right pulmonary vein 7 months postoperatively. These defects were unlikely to have been responsible for the patients requiring ECMO postoperatively because neither was detected with repeated intraoperative and postoperative echocardiography and hemodynamic measurements, and because these defects became evident more than 6 months after hospital discharge. Both are doing well at home. One patient is on antiseizure medication and another is receiving amiodarone for chronic atrial flutter. At present all survivors have activity levels normal for age, although 2 have mild developmental delay.

866

ZIOMEK ET AL ECMO FOR CARDIAC FAILURE

Ann Thorac Surg 1992;54:861-8

Table 2 . Extracorporeal Membrane Oxygenation for Cardiopulmonary Failure After Congenital Heart Operation (series with 10 or more patients reported) Series Kanter et al, 1987 [18] (St. Louis Univ Med

Total

Patients

Weaned OR from Long-Term ECMO ECMO

Survival

Survival

13

7 (54%)

5 (38%)

011

13

8 (62%)

5 (38%)

0/4

Rogers et al, 1989 [19] (Univ Pittsburg Med

10

8 (80%)

7 (70%)

0/1

Klein et al, 1990 [20] (Children‘s

39

22 (56%)

21 (54%)

2/9(22%)

ELSO Registry data (Oct 1991)

435

191 (44%)

N/A

N/A

Center) Weinhaus et al, 1988 [21] (St.

Louis Children’s Hasp)

Center)

Hosp of Michigan)

ELSO ECMO = extracorporeal membrane oxygenation; NIA = not available; real Life Support Organization; ating room.

=

ExtracorpoOR = oper-

Comment This study demonstrates ECMO can be used successfully in neonates and children for postoperative cardiopulmonary failure refractory to conventional medical therapy. Seventy-five percent of our patients (18/24) were weaned. The Extracorporeal Life Support Organization registry reported a 44% survival (weaned from ECMO) for postoperative cardiac patients [24]. Of the 18 patients weaned in this series, 13 (72%)were discharged home, representing a 54% overall survival rate. In comparison with previous reports [1€&21],this series includes a majority of patients begun on ECMO in the OR (Table 2). In our experience, long-term survival was best supporting myocardial dysfunction after repair of left to right shunts and severe pulmonary hypertension after repair of anomalous pulmonary venous return. Lowest patient survival with postoperative ECMO occurred after Norwood procedures. Extracorporeal membrane oxygenation has allowed us to perform corrective repairs in several patients we considered to be at extreme risk for death. Preoperative catheterization in patients 13 and 17 with transposition of the great vessels and ventricular septa1 defect revealed relatively hypoplastic right ventricles. Both underwent Rastelli repairs and have done well. Extracorporeal membrane oxygenation support allowed for gradual improvement in diastolic right ventricular function and progressive enlargement of both patients’ right ventricular chambers. Patient 9 had a complete atrioventricular canal and was transferred to our hospital at 9 months of age with severe pulmonary hypertension. After repair, the patient’s suprasystemic pulmonary artery pressure was managed successfully with ECMO. The intraaortic balloon pump [25, 261 and the left

ventricular assist device [27, 281 have been used for myocardial dysfunction after congenital heart operations. However, both are limited to support of the left ventricle. Extracorporeal membrane oxygenation may be superior because it offers biventricular and pulmonary support and can be performed in neonates. A high incidence of clinically significant hemorrhage with postoperative ECMO has been frequently reported [18-211 and has led to a reluctance to initiate support in the OR [19]. Our experience has shown life-threatening hemorrhage can be avoided as there were no instances of uncontrolled mediastinal hemorrhage. Factors found to be helpful in controlling hemorrhage were (1) platelet transfusions before initiation of ECMO, (2) meticulous surgical hemostasis, (3) liberal use of topical hemostatic agents, (4) a low threshold for mediastinal exploration, and (5) tight control of the activated clotting times. Longterm survival of patients with ECMO initiated in the OR was similar to that of patients requiring ECMO in the ICU. Another concern about the use of postoperative ECMO has been the need for venting the failing left ventricle [21, 291. This was believed not to be necessary in any patients in this series. Ventricular overdistension was minimized by maintaining low central venous pressure, adequate venous outflow, and high ECMO flow rates. Infection lowered ECMO salvage rate by 22%, causing 4 of 18 patients successfully weaned to eventually die: 2 of generalized sepsis secondary to an unidentified source, 1 of pneumonitis, and 1 of mediastinitis. Patients requiring postoperative circulatory support appear to be at higher risk for infectious complications for a variety of reasons including (1) open wounds postoperatively, (2) presence of central lines and ECMO cannulas, (3) prolonged endotracheal ventilation, (4) fungal and Pseudomonas overgrowth from prolonged preoperative and postoperative antibiotics, and (5) potential immunosuppression from prolonged cardiopulmonary bypass. Despite this, it appears that mediastinitis was effectively prevented, as there was only 1 case in 28 ECMO courses. After reviewing our results from this series, we are taking measures to lower ECMO mortality from infection. Elective operations are now delayed at least 1 month after administration of antibiotics has been stopped. We will be making an attempt in the future to close the sternum within 24 to 36 hours while the patient is on ECMO, as long as there is minimal mediastinal hemorrhage. Because candidemia developed in 4 patients, changes in our intravenous and topical antibiotics are being considered. In addition, we are setting up a protocol to investigate the effects of prolonged cardiopulmonary bypass on the immune system, with an emphasis on T-cell function in view of the high incidence of fungal infections. Finally, although a clear explanation for a high rate of infectious complications in patients cannulated in the chest is not apparent, neck cannulation is preferred if technically possible. We postulate that chest cannulation may have increased systemic infectious complications because the raw surfaces of the mediastinum were more difficult to keep sterile with cannulas exiting from the open chest than from a separate site in the neck. Central shunts were occluded early in our experience to prevent excessive pulmonary flow. After the occurrence of bilateral pulmonary infarctions caused by occluding the

Ann Thorac Surg

1992;54861-8

central shunt, systemic to p u l m o n a r y flow through a shunt i s now preserved but restricted. Several patients early i n o u r experience were taken off ECMO when echocardiography showed no recovery of ventricular function after 12 to 24 hours. Presently, ECMO support is provided for patients w i t h o u t early ventricular recovery as l o n g as progressive acidosis or multiorgan failure does not develop. Our 23rd patient did not show return of any noticeable ventricular function until t h e fourth day of ECMO support, and i s a survivor w i t h normal ventricular function. In conclusion, ECMO has been a valuable resource i n the management of postoperative cardiopulmonary failure refractory to optimal medical management after repair or palliation of congenital heart defects. Although late infection remains a serious risk, o u r willingness to initiate support earlier has grown as mechanical risks of postoperative ECMO for the most p a r t have been avoided and benefits of early intervention have been observed. Extracorporeal membrane oxygenation can be used w i t h good results i n patients w i t h profound postoperative cardiop u l m o n a r y failure i n the OR and in those who require a second course of postoperative circulatory support.

References 1. Bartlett RH, Andrews AF, Toomasian JM, et al. Extracorporeal membrane oxygenation for newborn respiratory failure: forty-five cases. Surgery 1982;92425-33. 2. Krummel TM, Greenfield LJ, Kirkpatrick BV, et al. Clinical use of an extracorporeal membrane oxygenator in neonatal pulmonary failure. J Pediatr Surg 1982;17525-31. 3. Toomasian JM, Snedecor SM, Cornell RG, Cilley RE, Bartlett RH. National experience with extracorporeal membrane oxygenation for newborn respiratory failure: data from 715 cases. Trans Am SOCArtif Intern Organs 1988;34:140-7. 4. Trento A, Thompson A, Siewers R, et al. Extracorporeal membrane oxygenation in children. New trends. J Thorac Cardiovasc Surg 1988;96:542-7. 5. Redmond CR, Graves ED, Falterman KW, et al. Extracorporeal membrane oxygenation for respiratory and cardiac failure in infants and children. J Thorac Cardiovasc Surg 1987; 93~199-204. 6 . Kirklin JK, Castaneda AR, Kean JF, Fellows KE, Norwood WI. Surgical management of multiple ventricular defects. J Thorac Cardiovasc Surg 1980;80:485-93. 7. Mavroudis C, Weinstein G, Turley K, Ebert PA. Surgical management of complete atrioventricular canal. J Thorac Cardiovasc Surg 1982;83:670-9. 8. Arciniegas E, Farooki ZQ, Hakimi M, Perry BL, Green EW. Early and late results of total correction of tetralogy of Fallot. J Thorac Cardiovasc Surg 1980;80:770-8. 9. Castaneda AR, Trusler GA, Paul MH, Blackstone EH, Kirklin JW. The early results of treatment of simple transposition in the current era. J Thorac Cardiovasc Surg 1988;95:14-28. 10. Ebert PA, Turley K, Stranger P, Hoffman JEI, Heymann MA, Rudolph AM. Surgical treatment of truncus arteriosus in the first 6 months of life. Ann Surg 1984;200:451-6. 11. Mayer JE Jr, Helgason H, Jonas RA, et al. Extending the

ZIOMEK ET AL ECMO FOR CARDIAC FAILURE

867

limits for modified Fontan procedures. J Thorac Cardiovasc Surg 1986;92:10214. 12. Meliones JN, Sider AR, Bove EL, Rosenthal A, Rosen DA. Longitudinal results after first-stage palliation for hypoplastic left heart syndrome. Circulation 1990;82:151-6. 13. Kirklin JW, Barratt-Boyes BG. Cardiac surgery. New York: Wiley Medical, 1986:514. 14. Soeter JR, Mamuja RT, Sprague AY, et al. Prolonged extracorporeal oxygenation for cardiorespiratory failure after tetralogy correction. J Thorac Cardiovasc Surg 1974;66:214-8. 15. Bartlett RH, Gazzaniga AB, Fong SW, Jefferies MR, Roohk HV, Haiduc N. Extracorporeal membrane oxygenator support for cardiopulmonary failure: experience in 28 cases. J Thorac Cardiovasc Surg 1977;73:375-86. 16. Galantowicz ME, Stolar CJH, King TC. Extracorporeal membrane oxygenation for perioperative support in pediatric heart transplantation. J Thorac Cardiovasc Surg 1991;102: 148-52. 17. Bartlett RH, Gazzaniga AB, Wetmore NE, et al. Extracorporeal membrane oxygenation (ECMO) in the treatment of cardiac and respiratory failure in children. Trans Am SOC Artif Intern Organs 1980;26578-8. 18. Kanter KR, Pennington G, 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. 19. Rogers AJ, Trento A, Sievers RD, et al. Extracorporeal membrane oxygenation for postcardiotomy cardiac shock in children. Ann Thorac Surg 1989;47:90M. 20. Klein MD, Shaheen KW, Whittlesey GC, et al. Extracorporeal membrane oxygenation for the circulatory support of children after repair of congenital heart disease. J Thorac Cardiovasc Surg 1990;100:498-505. 21. Weinhaus L, Canter C, Noetzel M, et al. Extracorporeal membrane oxygenation for Circulatory support after repair of congenital heart defects. Ann Thorac Surg 1989;48:206-12. 22. Pennington DG, Merjavy JP, Codd JE, et al. Extracorporeal membrane oxygenation for patients with cardiogenic shock. Circulation 1984;70:130. 23. Registry Report of the Extracorporeal Life Support Organization, October 1991. Ann Arbor, MI: ECMO Data Registry, 1991. 24. Meliones JN, Custer JR, Snedecor SM, Moler FW, ORourke PP, Delius RE. Extracorporeal life support for cardiac assist in pediatric patients: review of ELSO Registry data. Circulation 1991;84:168-72. 25. Pollock JC, Charlton MC, Williams WG, Edmons JF, Trusler GA. Intraaortic balloon pumping in children. Ann Thorac Surg 1980;29:522-8. 26. Veasy LG, Blalock RC, Orth JL, Boucek MM. Intraaortic balloon pumping in infants and children. Circulation 1983; 68:1095-100. 27. Karl TR, Sano S, Horton S, Mee RBB. Centrifugal pump left heart assist in pediatric cardiac operations. J Thorac Cardiovasc Surg 1991;102:624-30. 28. Drinkwater DC, Laks H. Clinical experience with centrifugal pump ventricular support at UCLA Medical Center. ASAIO Trans 1988;34:50.5-8. 29. 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. ASAIO Trans 1984;30:98-102.

DISCUSSION DR JOSEPH B. ZWISCHENBERGER (Galveston, TX): I rise to congratulate Dr Ziomek and the Little Rock group for their excellent results. I only want to mention that you referred to the ELSO Registry with a 44% initial weaning rate from ECMO. The same registry reports a 30% to 50% incidence of serious hemor-

rhagic complications in this particular subgroup of patients, and yet your abstract only mentions hemorrhagic complications in 1 patient. I do not recall your presentation focusing on those complications. I would like to hear how you manage your circuit, heparin dose and platelet use and whether or not you use Amicar

868

ZIOMEK ET AL ECMO FOR CARDIAC FAILURE

Ann Thorac Surg 1992;54:861-8

in the management of your patients to have such an outstandingly low bleeding complication rate.

artery cannula but we did not believe that it added anything to the overall support.

DR ZIOMEK Thank you for your kind comments. Almost every patient in our series had some amount of oozing from the mediastinum while on ECMO, but no patient had development of exsanguinating hemorrhage or had to be discontinued from ECMO as a result of mediastinal hemorrhage. Factors that we believe have allowed us to minimize bleeding complications with postcardiotomy ECMO start with transfusing 1 to 2 units of platelets while the patient is on bypass immediately before initiation of ECMO. As the perfusionists set up the ECMO circuit in the OR, meticulous care is undertaken to stop all mediastinal oozing, paying particular attention to ensure that the venous and arterial cannula sites are very secure. While the patients are on ECMO, we try to keep the platelet count greater than 120,00O/pLand also follow the fibrinogen level very closely. We give 10 mL of cryoprecipitate if the fibrinogen level falls to less than 150. Also, and this is very important, we leave the sternum open on all patients while they are on ECMO to allow us to identify and control bleeding sites with electrocautery or thrombin and Surgicel. This also substantially lessens the chance of cardiac tamponade developing while the patient is receiving postcardiotomy ECMO support.

DR MORONT: What measures exactly do you use as criteria during your weaning from ECMO?

DR PETER P. McKEOWN (Tampa, FL): My compliments again for an excellent series. I wonder if you could expand on what you mentioned about the mediastinal bleeding and as it relates perhaps to the site of cannulation. You said your preference at the moment would be for neck cannulation, and I wonder if you could talk about problems that you have had with neck cannulation, size of cannulas and flow differences, as opposed to those that you cannulated directly. DR ZIOMEK I am going to defer to Dr Harrell. DR HARRELL: Thank you, Dr Ziomek, and thank you for your comments, Dr McKeown. I thought I would elaborate just a little bit on some of the techniques that Dr Ziomek has not covered. For neck cannulation, we try to go with as small an arterial cannula as we can possibly put through the carotid artery in an effort to preserve the carotid for repair after decannulation. As a rule, we have not had serious problems with hemorrhage. If there is some oozing, it is usually from around the skin or the mediastinal tissues and is controlled with electrocautery or packed woven collagen. One of the reasons we like neck cannulation is it gets the cannulas out of the chest allowing better inspection of the heart and surgical field for hemostasis, if necessary. Achieving adequate pump flow through the neck cannulas has not been a problem. In addition, the activated clotting times are controlled fairly tightly, between about 210 and 230 seconds, and we believe that this is a reasonable range of control for heparin titration to the point that we have not had serious problems with hemorrhage. DR MICHAEL G . MORONT (Birmingham, AL): I have two questions for you. Number one, have you had any experience with the inability, during ECMO without a left atrial cannula, to decompress the left ventricle? DR HARRELL: That is also a very good question. Certainly we have had concerns about that, but in our series and experience we have not seen serious problems. We temporarily attempted left ventricular decompression on 1 patient with a pulmonary

DR HARRELL: It is a combination of factors. The absence of acidosis, evidence of good or improving ventricular function by echocardiography, satisfactory visual inspection of the heart through the open chest, and improving hemodynamics are the usual criteria. We normally like to do a "trial-off' period of at least an hour or two. If the patient tolerates the trial-off ECMO successfully, and if we are all satisfied, including the cardiologists, then we proceed with decannulation. DR THOMAS L. SPRAY (St. Louis, MO): I would like to again congratulate Dr Ziomek, Dr Harrell, and their associates on this series. This is a good result in terms of weaning and long-term survival, and it is also clear that if you look at the other reported series, the overall survival rates are getting to be about what you would expect with adult assist device technologies. This can be applied over a wide range. My question has to do with the issue of the hypoplastic left heart syndrome and the Norwood operation. We have found that ECMO really is not very useful in this setting because most of the patients are in trouble either from too much or too little pulmonary blood flow, and putting fully oxygenated arterial blood back into the aorta often causes the pulmonary resistance to drop so low that they flood the ventricle no matter what the ECMO flow. So I am very interested in this concept of limiting the amount of flow through the shunt while the patient is on ECMO, and if you do that, how then would you manage taking the patient off ECMO? Do you partially release the limitation in shunt flow or do you take it off completely? DR HARRELL: Dr Spray, thank you for your comments. First of all, our worst results were with the hypoplastic left heart patients and, I think, a lot of it has to do with the reactivity of the pulmonary vasculature. With the central shunt in place, all we do is take a metal Hemoclip and, watching the saturations and watching the hemodynamics, crimp down the clip to a point where we think we have reduced the pulmonary flow to a reasonable level and yet still maintain enough flow so that the shunt will not clot off, as occurred in 1 patient. Since we have started to restrict flow, we have only had 2 patients in whom we have tried to do this and, so far, it seems to have worked, as we were able to wean the patients off by completely removing the Hemoclip just before weaning. But we have had very limited experience with this technique so far. DR JOHN TERRANCE DAVIS (Columbus, OH): I have a quick question about what percentage of your practice this represents: is this greater than 1 per month? I was also interested in how many open heart procedures this is and what percentage of that this represents. DR HARRELL: In looking back over the series of about 2% years, it roughly represents about 350 pumps, which would be somewhere around 6% or 7%. Now, some people may say that this is perhaps a little on the high side, but I think that in our setting where we do a lot of ECMO support for other reasons, we have felt very comfortable using ECMO as a route for support, and we think it really has helped tremendously from the standpoint of supporting the marginal patient in coming off bypass.